Submitted by: Matt Burgess (matthew_at_linuxfromscratch_dot_org) Date: 2009-04-18 Initial Package Version: 7.2 Upstream Status: Rejected Origin: Based on Fedora's i18n patch at http://cvs.fedoraproject.org/viewvc/devel/coreutils/coreutils-i18n.patch. diff -Naur coreutils-7.2.orig/lib/linebuffer.h coreutils-7.2/lib/linebuffer.h --- coreutils-7.2.orig/lib/linebuffer.h 2009-03-11 09:57:26.000000000 +0000 +++ coreutils-7.2/lib/linebuffer.h 2009-04-18 10:41:11.000000000 +0100 @@ -21,6 +21,11 @@ # include +/* Get mbstate_t. */ +# if HAVE_WCHAR_H +# include +# endif + /* A `struct linebuffer' holds a line of text. */ struct linebuffer @@ -28,6 +33,9 @@ size_t size; /* Allocated. */ size_t length; /* Used. */ char *buffer; +# if HAVE_WCHAR_H + mbstate_t state; +# endif }; /* Initialize linebuffer LINEBUFFER for use. */ diff -Naur coreutils-7.2.orig/src/cut.c coreutils-7.2/src/cut.c --- coreutils-7.2.orig/src/cut.c 2009-03-29 18:43:41.000000000 +0100 +++ coreutils-7.2/src/cut.c 2009-04-18 10:41:11.000000000 +0100 @@ -28,6 +28,11 @@ #include #include #include + +/* Get mbstate_t, mbrtowc(). */ +#if HAVE_WCHAR_H +# include +#endif #include "system.h" #include "error.h" @@ -36,6 +41,18 @@ #include "quote.h" #include "xstrndup.h" +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC + installation; work around this configuration error. */ +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 +# undef MB_LEN_MAX +# define MB_LEN_MAX 16 +#endif + +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ +#if HAVE_MBRTOWC && defined mbstate_t +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) +#endif + /* The official name of this program (e.g., no `g' prefix). */ #define PROGRAM_NAME "cut" @@ -71,6 +88,52 @@ } \ while (0) +/* Refill the buffer BUF to get a multibyte character. */ +#define REFILL_BUFFER(BUF, BUFPOS, BUFLEN, STREAM) \ + do \ + { \ + if (BUFLEN < MB_LEN_MAX && !feof (STREAM) && !ferror (STREAM)) \ + { \ + memmove (BUF, BUFPOS, BUFLEN); \ + BUFLEN += fread (BUF + BUFLEN, sizeof(char), BUFSIZ, STREAM); \ + BUFPOS = BUF; \ + } \ + } \ + while (0) + +/* Get wide character on BUFPOS. BUFPOS is not included after that. + If byte sequence is not valid as a character, CONVFAIL is 1. Otherwise 0. */ +#define GET_NEXT_WC_FROM_BUFFER(WC, BUFPOS, BUFLEN, MBLENGTH, STATE, CONVFAIL) \ + do \ + { \ + mbstate_t state_bak; \ + \ + if (BUFLEN < 1) \ + { \ + WC = WEOF; \ + break; \ + } \ + \ + /* Get a wide character. */ \ + CONVFAIL = 0; \ + state_bak = STATE; \ + MBLENGTH = mbrtowc ((wchar_t *)&WC, BUFPOS, BUFLEN, &STATE); \ + \ + switch (MBLENGTH) \ + { \ + case (size_t)-1: \ + case (size_t)-2: \ + CONVFAIL++; \ + STATE = state_bak; \ + /* Fall througn. */ \ + \ + case 0: \ + MBLENGTH = 1; \ + break; \ + } \ + } \ + while (0) + struct range_pair { size_t lo; @@ -89,7 +152,7 @@ /* The number of bytes allocated for FIELD_1_BUFFER. */ static size_t field_1_bufsize; -/* The largest field or byte index used as an endpoint of a closed +/* The largest byte, character or field index used as an endpoint of a closed or degenerate range specification; this doesn't include the starting index of right-open-ended ranges. For example, with either range spec `2-5,9-', `2-3,5,9-' this variable would be set to 5. */ @@ -101,10 +164,11 @@ /* This is a bit vector. In byte mode, which bytes to output. + In character mode, which characters to output. In field mode, which DELIM-separated fields to output. - Both bytes and fields are numbered starting with 1, + Bytes, characters and fields are numbered starting with 1, so the zeroth bit of this array is unused. - A field or byte K has been selected if + A byte, character or field K has been selected if (K <= MAX_RANGE_ENDPOINT and is_printable_field(K)) || (EOL_RANGE_START > 0 && K >= EOL_RANGE_START). */ static unsigned char *printable_field; @@ -113,15 +177,25 @@ { undefined_mode, - /* Output characters that are in the given bytes. */ + /* Output bytes that are at the given positions. */ byte_mode, + /* Output characters that are at the given positions. */ + character_mode, + /* Output the given delimeter-separated fields. */ field_mode }; static enum operating_mode operating_mode; +/* If nonzero, when in byte mode, don't split multibyte characters. */ +static int byte_mode_character_aware; + +/* If nonzero, the function for single byte locale is work + if this program runs on multibyte locale. */ +static int force_singlebyte_mode; + /* If true do not output lines containing no delimeter characters. Otherwise, all such lines are printed. This option is valid only with field mode. */ @@ -133,6 +207,9 @@ /* The delimeter character for field mode. */ static unsigned char delim; +#if HAVE_WCHAR_H +static wchar_t wcdelim; +#endif /* True if the --output-delimiter=STRING option was specified. */ static bool output_delimiter_specified; @@ -206,7 +283,7 @@ -f, --fields=LIST select only these fields; also print any line\n\ that contains no delimiter character, unless\n\ the -s option is specified\n\ - -n (ignored)\n\ + -n with -b: don't split multibyte characters\n\ "), stdout); fputs (_("\ --complement complement the set of selected bytes, characters\n\ @@ -365,7 +442,7 @@ in_digits = false; /* Starting a range. */ if (dash_found) - FATAL_ERROR (_("invalid byte or field list")); + FATAL_ERROR (_("invalid byte, character or field list")); dash_found = true; fieldstr++; @@ -389,14 +466,16 @@ if (!rhs_specified) { /* `n-'. From `initial' to end of line. */ - eol_range_start = initial; + if (eol_range_start == 0 || + (eol_range_start != 0 && eol_range_start > initial)) + eol_range_start = initial; field_found = true; } else { /* `m-n' or `-n' (1-n). */ if (value < initial) - FATAL_ERROR (_("invalid decreasing range")); + FATAL_ERROR (_("invalid byte, character or field list")); /* Is there already a range going to end of line? */ if (eol_range_start != 0) @@ -476,6 +555,9 @@ if (operating_mode == byte_mode) error (0, 0, _("byte offset %s is too large"), quote (bad_num)); + else if (operating_mode == character_mode) + error (0, 0, + _("character offset %s is too large"), quote (bad_num)); else error (0, 0, _("field number %s is too large"), quote (bad_num)); @@ -486,7 +568,7 @@ fieldstr++; } else - FATAL_ERROR (_("invalid byte or field list")); + FATAL_ERROR (_("invalid byte, character or field list")); } max_range_endpoint = 0; @@ -579,6 +661,63 @@ } } +#if HAVE_MBRTOWC +/* This function is in use for the following case. + + 1. Read from the stream STREAM, printing to standard output any selected + characters. + + 2. Read from stream STREAM, printing to standard output any selected bytes, + without splitting multibyte characters. */ + +static void +cut_characters_or_cut_bytes_no_split (FILE *stream) +{ + int idx; /* number of bytes or characters in the line so far. */ + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ + char *bufpos; /* Next read position of BUF. */ + size_t buflen; /* The length of the byte sequence in buf. */ + wint_t wc; /* A gotten wide character. */ + size_t mblength; /* The byte size of a multibyte character which shows + as same character as WC. */ + mbstate_t state; /* State of the stream. */ + int convfail; /* 1, when conversion is failed. Otherwise 0. */ + + idx = 0; + buflen = 0; + bufpos = buf; + memset (&state, '\0', sizeof(mbstate_t)); + + while (1) + { + REFILL_BUFFER (buf, bufpos, buflen, stream); + + GET_NEXT_WC_FROM_BUFFER (wc, bufpos, buflen, mblength, state, convfail); + + if (wc == WEOF) + { + if (idx > 0) + putchar ('\n'); + break; + } + else if (wc == L'\n') + { + putchar ('\n'); + idx = 0; + } + else + { + idx += (operating_mode == byte_mode) ? mblength : 1; + if (print_kth (idx, NULL)) + fwrite (bufpos, mblength, sizeof(char), stdout); + } + + buflen -= mblength; + bufpos += mblength; + } +} +#endif + /* Read from stream STREAM, printing to standard output any selected fields. */ static void @@ -701,13 +840,192 @@ } } +#if HAVE_MBRTOWC +static void +cut_fields_mb (FILE *stream) +{ + int c; + unsigned int field_idx; + int found_any_selected_field; + int buffer_first_field; + int empty_input; + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ + char *bufpos; /* Next read position of BUF. */ + size_t buflen; /* The length of the byte sequence in buf. */ + wint_t wc = 0; /* A gotten wide character. */ + size_t mblength; /* The byte size of a multibyte character which shows + as same character as WC. */ + mbstate_t state; /* State of the stream. */ + int convfail; /* 1, when conversion is failed. Otherwise 0. */ + + found_any_selected_field = 0; + field_idx = 1; + bufpos = buf; + buflen = 0; + memset (&state, '\0', sizeof(mbstate_t)); + + c = getc (stream); + empty_input = (c == EOF); + if (c != EOF) + ungetc (c, stream); + else + wc = WEOF; + + /* To support the semantics of the -s flag, we may have to buffer + all of the first field to determine whether it is `delimited.' + But that is unnecessary if all non-delimited lines must be printed + and the first field has been selected, or if non-delimited lines + must be suppressed and the first field has *not* been selected. + That is because a non-delimited line has exactly one field. */ + buffer_first_field = (suppress_non_delimited ^ !print_kth (1, NULL)); + + while (1) + { + if (field_idx == 1 && buffer_first_field) + { + int len = 0; + + while (1) + { + REFILL_BUFFER (buf, bufpos, buflen, stream); + + GET_NEXT_WC_FROM_BUFFER + (wc, bufpos, buflen, mblength, state, convfail); + + if (wc == WEOF) + break; + + field_1_buffer = xrealloc (field_1_buffer, len + mblength); + memcpy (field_1_buffer + len, bufpos, mblength); + len += mblength; + buflen -= mblength; + bufpos += mblength; + + if (!convfail && (wc == L'\n' || wc == wcdelim)) + break; + } + + if (wc == WEOF) + break; + + /* If the first field extends to the end of line (it is not + delimited) and we are printing all non-delimited lines, + print this one. */ + if (convfail || (!convfail && wc != wcdelim)) + { + if (suppress_non_delimited) + { + /* Empty. */ + } + else + { + fwrite (field_1_buffer, sizeof (char), len, stdout); + /* Make sure the output line is newline terminated. */ + if (convfail || (!convfail && wc != L'\n')) + putchar ('\n'); + } + continue; + } + + if (print_kth (1, NULL)) + { + /* Print the field, but not the trailing delimiter. */ + fwrite (field_1_buffer, sizeof (char), len - 1, stdout); + found_any_selected_field = 1; + } + ++field_idx; + } + + if (wc != WEOF) + { + if (print_kth (field_idx, NULL)) + { + if (found_any_selected_field) + { + fwrite (output_delimiter_string, sizeof (char), + output_delimiter_length, stdout); + } + found_any_selected_field = 1; + } + + while (1) + { + REFILL_BUFFER (buf, bufpos, buflen, stream); + + GET_NEXT_WC_FROM_BUFFER + (wc, bufpos, buflen, mblength, state, convfail); + + if (wc == WEOF) + break; + else if (!convfail && (wc == wcdelim || wc == L'\n')) + { + buflen -= mblength; + bufpos += mblength; + break; + } + + if (print_kth (field_idx, NULL)) + fwrite (bufpos, mblength, sizeof(char), stdout); + + buflen -= mblength; + bufpos += mblength; + } + } + + if ((!convfail || wc == L'\n') && buflen < 1) + wc = WEOF; + + if (!convfail && wc == wcdelim) + ++field_idx; + else if (wc == WEOF || (!convfail && wc == L'\n')) + { + if (found_any_selected_field + || (!empty_input && !(suppress_non_delimited && field_idx == 1))) + putchar ('\n'); + if (wc == WEOF) + break; + field_idx = 1; + found_any_selected_field = 0; + } + } +} +#endif + static void cut_stream (FILE *stream) { - if (operating_mode == byte_mode) - cut_bytes (stream); +#if HAVE_MBRTOWC + if (MB_CUR_MAX > 1 && !force_singlebyte_mode) + { + switch (operating_mode) + { + case byte_mode: + if (byte_mode_character_aware) + cut_characters_or_cut_bytes_no_split (stream); + else + cut_bytes (stream); + break; + + case character_mode: + cut_characters_or_cut_bytes_no_split (stream); + break; + + case field_mode: + cut_fields_mb (stream); + break; + + default: + abort (); + } + } else - cut_fields (stream); +#endif + { + if (operating_mode == field_mode) + cut_fields (stream); + else + cut_bytes (stream); + } } /* Process file FILE to standard output. @@ -757,6 +1075,8 @@ bool ok; bool delim_specified = false; char *spec_list_string IF_LINT(= NULL); + char mbdelim[MB_LEN_MAX + 1]; + size_t delimlen = 0; initialize_main (&argc, &argv); set_program_name (argv[0]); @@ -779,7 +1099,6 @@ switch (optc) { case 'b': - case 'c': /* Build the byte list. */ if (operating_mode != undefined_mode) FATAL_ERROR (_("only one type of list may be specified")); @@ -787,6 +1106,14 @@ spec_list_string = optarg; break; + case 'c': + /* Build the character list. */ + if (operating_mode != undefined_mode) + FATAL_ERROR (_("only one type of list may be specified")); + operating_mode = character_mode; + spec_list_string = optarg; + break; + case 'f': /* Build the field list. */ if (operating_mode != undefined_mode) @@ -798,10 +1125,35 @@ case 'd': /* New delimiter. */ /* Interpret -d '' to mean `use the NUL byte as the delimiter.' */ - if (optarg[0] != '\0' && optarg[1] != '\0') - FATAL_ERROR (_("the delimiter must be a single character")); - delim = optarg[0]; - delim_specified = true; + { +#if HAVE_MBRTOWC + if(MB_CUR_MAX > 1) + { + mbstate_t state; + + memset (&state, '\0', sizeof(mbstate_t)); + delimlen = mbrtowc (&wcdelim, optarg, strnlen(optarg, MB_LEN_MAX), &state); + + if (delimlen == (size_t)-1 || delimlen == (size_t)-2) + ++force_singlebyte_mode; + else + { + delimlen = (delimlen < 1) ? 1 : delimlen; + if (wcdelim != L'\0' && *(optarg + delimlen) != '\0') + FATAL_ERROR (_("the delimiter must be a single character")); + memcpy (mbdelim, optarg, delimlen); + } + } + + if (MB_CUR_MAX <= 1 || force_singlebyte_mode) +#endif + { + if (optarg[0] != '\0' && optarg[1] != '\0') + FATAL_ERROR (_("the delimiter must be a single character")); + delim = (unsigned char) optarg[0]; + } + delim_specified = true; + } break; case OUTPUT_DELIMITER_OPTION: @@ -814,6 +1166,7 @@ break; case 'n': + byte_mode_character_aware = 1; break; case 's': @@ -836,7 +1189,7 @@ if (operating_mode == undefined_mode) FATAL_ERROR (_("you must specify a list of bytes, characters, or fields")); - if (delim != '\0' && operating_mode != field_mode) + if (delim_specified && operating_mode != field_mode) FATAL_ERROR (_("an input delimiter may be specified only\ when operating on fields")); @@ -863,15 +1216,34 @@ } if (!delim_specified) - delim = '\t'; + { + delim = '\t'; +#ifdef HAVE_MBRTOWC + wcdelim = L'\t'; + mbdelim[0] = '\t'; + mbdelim[1] = '\0'; + delimlen = 1; +#endif + } if (output_delimiter_string == NULL) { - static char dummy[2]; - dummy[0] = delim; - dummy[1] = '\0'; - output_delimiter_string = dummy; - output_delimiter_length = 1; +#ifdef HAVE_MBRTOWC + if (MB_CUR_MAX > 1 && !force_singlebyte_mode) + { + output_delimiter_string = xstrdup(mbdelim); + output_delimiter_length = delimlen; + } + + if (MB_CUR_MAX <= 1 || force_singlebyte_mode) +#endif + { + static char dummy[2]; + dummy[0] = delim; + dummy[1] = '\0'; + output_delimiter_string = dummy; + output_delimiter_length = 1; + } } if (optind == argc) diff -Naur coreutils-7.2.orig/src/expand.c coreutils-7.2/src/expand.c --- coreutils-7.2.orig/src/expand.c 2009-03-29 18:44:10.000000000 +0100 +++ coreutils-7.2/src/expand.c 2009-04-18 10:41:11.000000000 +0100 @@ -37,11 +37,28 @@ #include #include #include + +/* Get mbstate_t, mbrtowc(), wcwidth(). */ +#if HAVE_WCHAR_H +# include +#endif + #include "system.h" #include "error.h" #include "quote.h" #include "xstrndup.h" +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC + installation; work around this configuration error. */ +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 +# define MB_LEN_MAX 16 +#endif + +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ +#if HAVE_MBRTOWC && defined mbstate_t +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) +#endif + /* The official name of this program (e.g., no `g' prefix). */ #define PROGRAM_NAME "expand" @@ -175,6 +192,7 @@ stops = num_start + len - 1; } } + else { error (0, 0, _("tab size contains invalid character(s): %s"), @@ -357,6 +375,142 @@ } } +#if HAVE_MBRTOWC +static void +expand_multibyte (void) +{ + FILE *fp; /* Input strem. */ + mbstate_t i_state; /* Current shift state of the input stream. */ + mbstate_t i_state_bak; /* Back up the I_STATE. */ + mbstate_t o_state; /* Current shift state of the output stream. */ + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ + char *bufpos; /* Next read position of BUF. */ + size_t buflen = 0; /* The length of the byte sequence in buf. */ + wchar_t wc; /* A gotten wide character. */ + size_t mblength; /* The byte size of a multibyte character + which shows as same character as WC. */ + int tab_index = 0; /* Index in `tab_list' of next tabstop. */ + int column = 0; /* Column on screen of the next char. */ + int next_tab_column; /* Column the next tab stop is on. */ + int convert = 1; /* If nonzero, perform translations. */ + + fp = next_file ((FILE *) NULL); + if (fp == NULL) + return; + + memset (&o_state, '\0', sizeof(mbstate_t)); + memset (&i_state, '\0', sizeof(mbstate_t)); + + for (;;) + { + /* Refill the buffer BUF. */ + if (buflen < MB_LEN_MAX && !feof(fp) && !ferror(fp)) + { + memmove (buf, bufpos, buflen); + buflen += fread (buf + buflen, sizeof(char), BUFSIZ, fp); + bufpos = buf; + } + + /* No character is left in BUF. */ + if (buflen < 1) + { + fp = next_file (fp); + + if (fp == NULL) + break; /* No more files. */ + else + { + memset (&i_state, '\0', sizeof(mbstate_t)); + continue; + } + } + + /* Get a wide character. */ + i_state_bak = i_state; + mblength = mbrtowc (&wc, bufpos, buflen, &i_state); + + switch (mblength) + { + case (size_t)-1: /* illegal byte sequence. */ + case (size_t)-2: + mblength = 1; + i_state = i_state_bak; + if (convert) + { + ++column; + if (convert_entire_line == 0) + convert = 0; + } + putchar (*bufpos); + break; + + case 0: /* null. */ + mblength = 1; + if (convert && convert_entire_line == 0) + convert = 0; + putchar ('\0'); + break; + + default: + if (wc == L'\n') /* LF. */ + { + tab_index = 0; + column = 0; + convert = 1; + putchar ('\n'); + } + else if (wc == L'\t' && convert) /* Tab. */ + { + if (tab_size == 0) + { + /* Do not let tab_index == first_free_tab; + stop when it is 1 less. */ + while (tab_index < first_free_tab - 1 + && column >= tab_list[tab_index]) + tab_index++; + next_tab_column = tab_list[tab_index]; + if (tab_index < first_free_tab - 1) + tab_index++; + if (column >= next_tab_column) + next_tab_column = column + 1; + } + else + next_tab_column = column + tab_size - column % tab_size; + + while (column < next_tab_column) + { + putchar (' '); + ++column; + } + } + else /* Others. */ + { + if (convert) + { + if (wc == L'\b') + { + if (column > 0) + --column; + } + else + { + int width; /* The width of WC. */ + + width = wcwidth (wc); + column += (width > 0) ? width : 0; + if (convert_entire_line == 0) + convert = 0; + } + } + fwrite (bufpos, sizeof(char), mblength, stdout); + } + } + buflen -= mblength; + bufpos += mblength; + } +} +#endif + int main (int argc, char **argv) { @@ -421,7 +575,12 @@ file_list = (optind < argc ? &argv[optind] : stdin_argv); - expand (); +#if HAVE_MBRTOWC + if (MB_CUR_MAX > 1) + expand_multibyte (); + else +#endif + expand (); if (have_read_stdin && fclose (stdin) != 0) error (EXIT_FAILURE, errno, "-"); diff -Naur coreutils-7.2.orig/src/expand.c.orig coreutils-7.2/src/expand.c.orig --- coreutils-7.2.orig/src/expand.c.orig 1970-01-01 01:00:00.000000000 +0100 +++ coreutils-7.2/src/expand.c.orig 2009-03-29 18:44:10.000000000 +0100 @@ -0,0 +1,430 @@ +/* expand - convert tabs to spaces + Copyright (C) 89, 91, 1995-2006, 2008-2009 Free Software Foundation, Inc. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see . */ + +/* By default, convert all tabs to spaces. + Preserves backspace characters in the output; they decrement the + column count for tab calculations. + The default action is equivalent to -8. + + Options: + --tabs=tab1[,tab2[,...]] + -t tab1[,tab2[,...]] + -tab1[,tab2[,...]] If only one tab stop is given, set the tabs tab1 + columns apart instead of the default 8. Otherwise, + set the tabs at columns tab1, tab2, etc. (numbered from + 0); replace any tabs beyond the tab stops given with + single spaces. + --initial + -i Only convert initial tabs on each line to spaces. + + David MacKenzie */ + +#include + +#include +#include +#include +#include "system.h" +#include "error.h" +#include "quote.h" +#include "xstrndup.h" + +/* The official name of this program (e.g., no `g' prefix). */ +#define PROGRAM_NAME "expand" + +#define AUTHORS proper_name ("David MacKenzie") + +/* If true, convert blanks even after nonblank characters have been + read on the line. */ +static bool convert_entire_line; + +/* If nonzero, the size of all tab stops. If zero, use `tab_list' instead. */ +static uintmax_t tab_size; + +/* Array of the explicit column numbers of the tab stops; + after `tab_list' is exhausted, each additional tab is replaced + by a space. The first column is column 0. */ +static uintmax_t *tab_list; + +/* The number of allocated entries in `tab_list'. */ +static size_t n_tabs_allocated; + +/* The index of the first invalid element of `tab_list', + where the next element can be added. */ +static size_t first_free_tab; + +/* Null-terminated array of input filenames. */ +static char **file_list; + +/* Default for `file_list' if no files are given on the command line. */ +static char *stdin_argv[] = +{ + (char *) "-", NULL +}; + +/* True if we have ever read standard input. */ +static bool have_read_stdin; + +/* The desired exit status. */ +static int exit_status; + +static char const shortopts[] = "it:0::1::2::3::4::5::6::7::8::9::"; + +static struct option const longopts[] = +{ + {"tabs", required_argument, NULL, 't'}, + {"initial", no_argument, NULL, 'i'}, + {GETOPT_HELP_OPTION_DECL}, + {GETOPT_VERSION_OPTION_DECL}, + {NULL, 0, NULL, 0} +}; + +void +usage (int status) +{ + if (status != EXIT_SUCCESS) + fprintf (stderr, _("Try `%s --help' for more information.\n"), + program_name); + else + { + printf (_("\ +Usage: %s [OPTION]... [FILE]...\n\ +"), + program_name); + fputs (_("\ +Convert tabs in each FILE to spaces, writing to standard output.\n\ +With no FILE, or when FILE is -, read standard input.\n\ +\n\ +"), stdout); + fputs (_("\ +Mandatory arguments to long options are mandatory for short options too.\n\ +"), stdout); + fputs (_("\ + -i, --initial do not convert tabs after non blanks\n\ + -t, --tabs=NUMBER have tabs NUMBER characters apart, not 8\n\ +"), stdout); + fputs (_("\ + -t, --tabs=LIST use comma separated list of explicit tab positions\n\ +"), stdout); + fputs (HELP_OPTION_DESCRIPTION, stdout); + fputs (VERSION_OPTION_DESCRIPTION, stdout); + emit_bug_reporting_address (); + } + exit (status); +} + +/* Add tab stop TABVAL to the end of `tab_list'. */ + +static void +add_tab_stop (uintmax_t tabval) +{ + if (first_free_tab == n_tabs_allocated) + tab_list = X2NREALLOC (tab_list, &n_tabs_allocated); + tab_list[first_free_tab++] = tabval; +} + +/* Add the comma or blank separated list of tab stops STOPS + to the list of tab stops. */ + +static void +parse_tab_stops (char const *stops) +{ + bool have_tabval = false; + uintmax_t tabval IF_LINT (= 0); + char const *num_start IF_LINT (= NULL); + bool ok = true; + + for (; *stops; stops++) + { + if (*stops == ',' || isblank (to_uchar (*stops))) + { + if (have_tabval) + add_tab_stop (tabval); + have_tabval = false; + } + else if (ISDIGIT (*stops)) + { + if (!have_tabval) + { + tabval = 0; + have_tabval = true; + num_start = stops; + } + + /* Detect overflow. */ + if (!DECIMAL_DIGIT_ACCUMULATE (tabval, *stops - '0', uintmax_t)) + { + size_t len = strspn (num_start, "0123456789"); + char *bad_num = xstrndup (num_start, len); + error (0, 0, _("tab stop is too large %s"), quote (bad_num)); + free (bad_num); + ok = false; + stops = num_start + len - 1; + } + } + else + { + error (0, 0, _("tab size contains invalid character(s): %s"), + quote (stops)); + ok = false; + break; + } + } + + if (!ok) + exit (EXIT_FAILURE); + + if (have_tabval) + add_tab_stop (tabval); +} + +/* Check that the list of tab stops TABS, with ENTRIES entries, + contains only nonzero, ascending values. */ + +static void +validate_tab_stops (uintmax_t const *tabs, size_t entries) +{ + uintmax_t prev_tab = 0; + size_t i; + + for (i = 0; i < entries; i++) + { + if (tabs[i] == 0) + error (EXIT_FAILURE, 0, _("tab size cannot be 0")); + if (tabs[i] <= prev_tab) + error (EXIT_FAILURE, 0, _("tab sizes must be ascending")); + prev_tab = tabs[i]; + } +} + +/* Close the old stream pointer FP if it is non-NULL, + and return a new one opened to read the next input file. + Open a filename of `-' as the standard input. + Return NULL if there are no more input files. */ + +static FILE * +next_file (FILE *fp) +{ + static char *prev_file; + char *file; + + if (fp) + { + if (ferror (fp)) + { + error (0, errno, "%s", prev_file); + exit_status = EXIT_FAILURE; + } + if (STREQ (prev_file, "-")) + clearerr (fp); /* Also clear EOF. */ + else if (fclose (fp) != 0) + { + error (0, errno, "%s", prev_file); + exit_status = EXIT_FAILURE; + } + } + + while ((file = *file_list++) != NULL) + { + if (STREQ (file, "-")) + { + have_read_stdin = true; + prev_file = file; + return stdin; + } + fp = fopen (file, "r"); + if (fp) + { + prev_file = file; + return fp; + } + error (0, errno, "%s", file); + exit_status = EXIT_FAILURE; + } + return NULL; +} + +/* Change tabs to spaces, writing to stdout. + Read each file in `file_list', in order. */ + +static void +expand (void) +{ + /* Input stream. */ + FILE *fp = next_file (NULL); + + if (!fp) + return; + + for (;;) + { + /* Input character, or EOF. */ + int c; + + /* If true, perform translations. */ + bool convert = true; + + + /* The following variables have valid values only when CONVERT + is true: */ + + /* Column of next input character. */ + uintmax_t column = 0; + + /* Index in TAB_LIST of next tab stop to examine. */ + size_t tab_index = 0; + + + /* Convert a line of text. */ + + do + { + while ((c = getc (fp)) < 0 && (fp = next_file (fp))) + continue; + + if (convert) + { + if (c == '\t') + { + /* Column the next input tab stop is on. */ + uintmax_t next_tab_column; + + if (tab_size) + next_tab_column = column + (tab_size - column % tab_size); + else + for (;;) + if (tab_index == first_free_tab) + { + next_tab_column = column + 1; + break; + } + else + { + uintmax_t tab = tab_list[tab_index++]; + if (column < tab) + { + next_tab_column = tab; + break; + } + } + + if (next_tab_column < column) + error (EXIT_FAILURE, 0, _("input line is too long")); + + while (++column < next_tab_column) + if (putchar (' ') < 0) + error (EXIT_FAILURE, errno, _("write error")); + + c = ' '; + } + else if (c == '\b') + { + /* Go back one column, and force recalculation of the + next tab stop. */ + column -= !!column; + tab_index -= !!tab_index; + } + else + { + column++; + if (!column) + error (EXIT_FAILURE, 0, _("input line is too long")); + } + + convert &= convert_entire_line | !! isblank (c); + } + + if (c < 0) + return; + + if (putchar (c) < 0) + error (EXIT_FAILURE, errno, _("write error")); + } + while (c != '\n'); + } +} + +int +main (int argc, char **argv) +{ + int c; + + initialize_main (&argc, &argv); + set_program_name (argv[0]); + setlocale (LC_ALL, ""); + bindtextdomain (PACKAGE, LOCALEDIR); + textdomain (PACKAGE); + + atexit (close_stdout); + + have_read_stdin = false; + exit_status = EXIT_SUCCESS; + convert_entire_line = true; + tab_list = NULL; + first_free_tab = 0; + + while ((c = getopt_long (argc, argv, shortopts, longopts, NULL)) != -1) + { + switch (c) + { + case 'i': + convert_entire_line = false; + break; + + case 't': + parse_tab_stops (optarg); + break; + + case '0': case '1': case '2': case '3': case '4': + case '5': case '6': case '7': case '8': case '9': + if (optarg) + parse_tab_stops (optarg - 1); + else + { + char tab_stop[2]; + tab_stop[0] = c; + tab_stop[1] = '\0'; + parse_tab_stops (tab_stop); + } + break; + + case_GETOPT_HELP_CHAR; + + case_GETOPT_VERSION_CHAR (PROGRAM_NAME, AUTHORS); + + default: + usage (EXIT_FAILURE); + } + } + + validate_tab_stops (tab_list, first_free_tab); + + if (first_free_tab == 0) + tab_size = 8; + else if (first_free_tab == 1) + tab_size = tab_list[0]; + else + tab_size = 0; + + file_list = (optind < argc ? &argv[optind] : stdin_argv); + + expand (); + + if (have_read_stdin && fclose (stdin) != 0) + error (EXIT_FAILURE, errno, "-"); + + exit (exit_status); +} diff -Naur coreutils-7.2.orig/src/fold.c coreutils-7.2/src/fold.c --- coreutils-7.2.orig/src/fold.c 2009-03-29 18:43:41.000000000 +0100 +++ coreutils-7.2/src/fold.c 2009-04-18 10:41:11.000000000 +0100 @@ -22,11 +22,33 @@ #include #include +/* Get mbstate_t, mbrtowc(), wcwidth(). */ +#if HAVE_WCHAR_H +# include +#endif + +/* Get iswprint(), iswblank(), wcwidth(). */ +#if HAVE_WCTYPE_H +# include +#endif + #include "system.h" #include "error.h" #include "quote.h" #include "xstrtol.h" +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC + installation; work around this configuration error. */ +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 +# undef MB_LEN_MAX +# define MB_LEN_MAX 16 +#endif + +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ +#if HAVE_MBRTOWC && defined mbstate_t +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) +#endif + #define TAB_WIDTH 8 /* The official name of this program (e.g., no `g' prefix). */ @@ -34,20 +56,41 @@ #define AUTHORS proper_name ("David MacKenzie") +#define FATAL_ERROR(Message) \ + do \ + { \ + error (0, 0, (Message)); \ + usage (2); \ + } \ + while (0) + +enum operating_mode +{ + /* Fold texts by columns that are at the given positions. */ + column_mode, + + /* Fold texts by bytes that are at the given positions. */ + byte_mode, + + /* Fold texts by characters that are at the given positions. */ + character_mode, +}; + +/* The argument shows current mode. (Default: column_mode) */ +static enum operating_mode operating_mode; + /* If nonzero, try to break on whitespace. */ static bool break_spaces; -/* If nonzero, count bytes, not column positions. */ -static bool count_bytes; - /* If nonzero, at least one of the files we read was standard input. */ static bool have_read_stdin; -static char const shortopts[] = "bsw:0::1::2::3::4::5::6::7::8::9::"; +static char const shortopts[] = "bcsw:0::1::2::3::4::5::6::7::8::9::"; static struct option const longopts[] = { {"bytes", no_argument, NULL, 'b'}, + {"characters", no_argument, NULL, 'c'}, {"spaces", no_argument, NULL, 's'}, {"width", required_argument, NULL, 'w'}, {GETOPT_HELP_OPTION_DECL}, @@ -77,6 +120,7 @@ "), stdout); fputs (_("\ -b, --bytes count bytes rather than columns\n\ + -c, --characters count characters rather than columns\n\ -s, --spaces break at spaces\n\ -w, --width=WIDTH use WIDTH columns instead of 80\n\ "), stdout); @@ -94,7 +138,7 @@ static size_t adjust_column (size_t column, char c) { - if (!count_bytes) + if (operating_mode != byte_mode) { if (c == '\b') { @@ -117,30 +161,14 @@ to stdout, with maximum line length WIDTH. Return true if successful. */ -static bool -fold_file (char const *filename, size_t width) +static void +fold_text (FILE *istream, size_t width, int *saved_errno) { - FILE *istream; int c; size_t column = 0; /* Screen column where next char will go. */ size_t offset_out = 0; /* Index in `line_out' for next char. */ static char *line_out = NULL; static size_t allocated_out = 0; - int saved_errno; - - if (STREQ (filename, "-")) - { - istream = stdin; - have_read_stdin = true; - } - else - istream = fopen (filename, "r"); - - if (istream == NULL) - { - error (0, errno, "%s", filename); - return false; - } while ((c = getc (istream)) != EOF) { @@ -168,6 +196,15 @@ bool found_blank = false; size_t logical_end = offset_out; + /* If LINE_OUT has no wide character, + put a new wide character in LINE_OUT + if column is bigger than width. */ + if (offset_out == 0) + { + line_out[offset_out++] = c; + continue; + } + /* Look for the last blank. */ while (logical_end) { @@ -214,11 +251,225 @@ line_out[offset_out++] = c; } - saved_errno = errno; + *saved_errno = errno; + + if (offset_out) + fwrite (line_out, sizeof (char), (size_t) offset_out, stdout); + + free(line_out); +} + +#if HAVE_MBRTOWC +static void +fold_multibyte_text (FILE *istream, size_t width, int *saved_errno) +{ + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ + size_t buflen = 0; /* The length of the byte sequence in buf. */ + char *bufpos; /* Next read position of BUF. */ + wint_t wc; /* A gotten wide character. */ + size_t mblength; /* The byte size of a multibyte character which shows + as same character as WC. */ + mbstate_t state, state_bak; /* State of the stream. */ + int convfail; /* 1, when conversion is failed. Otherwise 0. */ + + char *line_out = NULL; + size_t offset_out = 0; /* Index in `line_out' for next char. */ + size_t allocated_out = 0; + + int increment; + size_t column = 0; + + size_t last_blank_pos; + size_t last_blank_column; + int is_blank_seen; + int last_blank_increment; + int is_bs_following_last_blank; + size_t bs_following_last_blank_num; + int is_cr_after_last_blank; + +#define CLEAR_FLAGS \ + do \ + { \ + last_blank_pos = 0; \ + last_blank_column = 0; \ + is_blank_seen = 0; \ + is_bs_following_last_blank = 0; \ + bs_following_last_blank_num = 0; \ + is_cr_after_last_blank = 0; \ + } \ + while (0) + +#define START_NEW_LINE \ + do \ + { \ + putchar ('\n'); \ + column = 0; \ + offset_out = 0; \ + CLEAR_FLAGS; \ + } \ + while (0) + + CLEAR_FLAGS; + memset (&state, '\0', sizeof(mbstate_t)); + + for (;; bufpos += mblength, buflen -= mblength) + { + if (buflen < MB_LEN_MAX && !feof (istream) && !ferror (istream)) + { + memmove (buf, bufpos, buflen); + buflen += fread (buf + buflen, sizeof(char), BUFSIZ, istream); + bufpos = buf; + } + + if (buflen < 1) + break; + + /* Get a wide character. */ + convfail = 0; + state_bak = state; + mblength = mbrtowc ((wchar_t *)&wc, bufpos, buflen, &state); + + switch (mblength) + { + case (size_t)-1: + case (size_t)-2: + convfail++; + state = state_bak; + /* Fall through. */ + + case 0: + mblength = 1; + break; + } + +rescan: + if (operating_mode == byte_mode) /* byte mode */ + increment = mblength; + else if (operating_mode == character_mode) /* character mode */ + increment = 1; + else /* column mode */ + { + if (convfail) + increment = 1; + else + { + switch (wc) + { + case L'\n': + fwrite (line_out, sizeof(char), offset_out, stdout); + START_NEW_LINE; + continue; + + case L'\b': + increment = (column > 0) ? -1 : 0; + break; + + case L'\r': + increment = -1 * column; + break; + + case L'\t': + increment = 8 - column % 8; + break; + + default: + increment = wcwidth (wc); + increment = (increment < 0) ? 0 : increment; + } + } + } + + if (column + increment > width && break_spaces && last_blank_pos) + { + fwrite (line_out, sizeof(char), last_blank_pos, stdout); + putchar ('\n'); + + offset_out = offset_out - last_blank_pos; + column = column - last_blank_column + ((is_cr_after_last_blank) + ? last_blank_increment : bs_following_last_blank_num); + memmove (line_out, line_out + last_blank_pos, offset_out); + CLEAR_FLAGS; + goto rescan; + } + + if (column + increment > width && column != 0) + { + fwrite (line_out, sizeof(char), offset_out, stdout); + START_NEW_LINE; + goto rescan; + } + + if (allocated_out < offset_out + mblength) + { + allocated_out += 1024; + line_out = xrealloc (line_out, allocated_out); + } + + memcpy (line_out + offset_out, bufpos, mblength); + offset_out += mblength; + column += increment; + + if (is_blank_seen && !convfail && wc == L'\r') + is_cr_after_last_blank = 1; + + if (is_bs_following_last_blank && !convfail && wc == L'\b') + ++bs_following_last_blank_num; + else + is_bs_following_last_blank = 0; + + if (break_spaces && !convfail && iswblank (wc)) + { + last_blank_pos = offset_out; + last_blank_column = column; + is_blank_seen = 1; + last_blank_increment = increment; + is_bs_following_last_blank = 1; + bs_following_last_blank_num = 0; + is_cr_after_last_blank = 0; + } + } + + *saved_errno = errno; if (offset_out) fwrite (line_out, sizeof (char), (size_t) offset_out, stdout); + free(line_out); +} +#endif + +/* Fold file FILENAME, or standard input if FILENAME is "-", + to stdout, with maximum line length WIDTH. + Return 0 if successful, 1 if an error occurs. */ + +static bool +fold_file (char *filename, size_t width) +{ + FILE *istream; + int saved_errno; + + if (STREQ (filename, "-")) + { + istream = stdin; + have_read_stdin = 1; + } + else + istream = fopen (filename, "r"); + + if (istream == NULL) + { + error (0, errno, "%s", filename); + return 1; + } + + /* Define how ISTREAM is being folded. */ +#if HAVE_MBRTOWC + if (MB_CUR_MAX > 1) + fold_multibyte_text (istream, width, &saved_errno); + else +#endif + fold_text (istream, width, &saved_errno); + if (ferror (istream)) { error (0, saved_errno, "%s", filename); @@ -251,7 +502,8 @@ atexit (close_stdout); - break_spaces = count_bytes = have_read_stdin = false; + operating_mode = column_mode; + break_spaces = have_read_stdin = false; while ((optc = getopt_long (argc, argv, shortopts, longopts, NULL)) != -1) { @@ -260,7 +512,15 @@ switch (optc) { case 'b': /* Count bytes rather than columns. */ - count_bytes = true; + if (operating_mode != column_mode) + FATAL_ERROR (_("only one way of folding may be specified")); + operating_mode = byte_mode; + break; + + case 'c': + if (operating_mode != column_mode) + FATAL_ERROR (_("only one way of folding may be specified")); + operating_mode = character_mode; break; case 's': /* Break at word boundaries. */ diff -Naur coreutils-7.2.orig/src/join.c coreutils-7.2/src/join.c --- coreutils-7.2.orig/src/join.c 2009-03-29 18:43:41.000000000 +0100 +++ coreutils-7.2/src/join.c 2009-04-18 10:41:11.000000000 +0100 @@ -22,16 +22,30 @@ #include #include +/* Get mbstate_t, mbrtowc(), mbrtowc(), wcwidth(). */ +#if HAVE_WCHAR_H +# include +#endif + +/* Get iswblank(), towupper. */ +#if HAVE_WCTYPE_H +# include +#endif + #include "system.h" #include "error.h" #include "linebuffer.h" -#include "memcasecmp.h" #include "quote.h" #include "stdio--.h" #include "xmemcoll.h" #include "xstrtol.h" #include "argmatch.h" +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ +#if HAVE_MBRTOWC && defined mbstate_t +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) +#endif + /* The official name of this program (e.g., no `g' prefix). */ #define PROGRAM_NAME "join" @@ -120,10 +134,12 @@ /* Last element in `outlist', where a new element can be added. */ static struct outlist *outlist_end = &outlist_head; -/* Tab character separating fields. If negative, fields are separated - by any nonempty string of blanks, otherwise by exactly one - tab character whose value (when cast to unsigned char) equals TAB. */ -static int tab = -1; +/* Tab character separating fields. If NULL, fields are separated + by any nonempty string of blanks. */ +static char *tab = NULL; + +/* The number of bytes used for tab. */ +static size_t tablen = 0; /* If nonzero, check that the input is correctly ordered. */ static enum @@ -237,10 +253,11 @@ if (ptr == lim) return; - if (0 <= tab) + if (tab != NULL) { + unsigned char t = tab[0]; char *sep; - for (; (sep = memchr (ptr, tab, lim - ptr)) != NULL; ptr = sep + 1) + for (; (sep = memchr (ptr, t, lim - ptr)) != NULL; ptr = sep + 1) extract_field (line, ptr, sep - ptr); } else @@ -267,6 +284,148 @@ extract_field (line, ptr, lim - ptr); } +#if HAVE_MBRTOWC +static void +xfields_multibyte (struct line *line) +{ + char *ptr = line->buf.buffer; + char const *lim = ptr + line->buf.length - 1; + wchar_t wc = 0; + size_t mblength = 1; + mbstate_t state, state_bak; + + memset (&state, 0, sizeof (mbstate_t)); + + if (ptr == lim) + return; + + if (tab != NULL) + { + unsigned char t = tab[0]; + char *sep = ptr; + for (; ptr < lim; ptr = sep + mblength) + { + sep = ptr; + while (sep < lim) + { + state_bak = state; + mblength = mbrtowc (&wc, sep, lim - sep + 1, &state); + + if (mblength == (size_t)-1 || mblength == (size_t)-2) + { + mblength = 1; + state = state_bak; + } + mblength = (mblength < 1) ? 1 : mblength; + + if (mblength == tablen && !memcmp (sep, tab, mblength)) + break; + else + { + sep += mblength; + continue; + } + } + + if (sep == lim) + break; + + extract_field (line, ptr, sep - ptr); + } + } + else + { + /* Skip leading blanks before the first field. */ + while(ptr < lim) + { + state_bak = state; + mblength = mbrtowc (&wc, ptr, lim - ptr + 1, &state); + + if (mblength == (size_t)-1 || mblength == (size_t)-2) + { + mblength = 1; + state = state_bak; + break; + } + mblength = (mblength < 1) ? 1 : mblength; + + if (!iswblank(wc)) + break; + ptr += mblength; + } + + do + { + char *sep; + state_bak = state; + mblength = mbrtowc (&wc, ptr, lim - ptr + 1, &state); + if (mblength == (size_t)-1 || mblength == (size_t)-2) + { + mblength = 1; + state = state_bak; + break; + } + mblength = (mblength < 1) ? 1 : mblength; + + sep = ptr + mblength; + while (sep != lim) + { + state_bak = state; + mblength = mbrtowc (&wc, sep, lim - sep + 1, &state); + if (mblength == (size_t)-1 || mblength == (size_t)-2) + { + mblength = 1; + state = state_bak; + break; + } + mblength = (mblength < 1) ? 1 : mblength; + + if (iswblank (wc)) + break; + + sep += mblength; + } + + extract_field (line, ptr, sep - ptr); + if (sep == lim) + return; + + state_bak = state; + mblength = mbrtowc (&wc, sep, lim - sep + 1, &state); + if (mblength == (size_t)-1 || mblength == (size_t)-2) + { + mblength = 1; + state = state_bak; + break; + } + mblength = (mblength < 1) ? 1 : mblength; + + ptr = sep + mblength; + while (ptr != lim) + { + state_bak = state; + mblength = mbrtowc (&wc, ptr, lim - ptr + 1, &state); + if (mblength == (size_t)-1 || mblength == (size_t)-2) + { + mblength = 1; + state = state_bak; + break; + } + mblength = (mblength < 1) ? 1 : mblength; + + if (!iswblank (wc)) + break; + + ptr += mblength; + } + } + while (ptr != lim); + } + + extract_field (line, ptr, lim - ptr); +} +#endif + static void freeline (struct line *line) { @@ -285,56 +444,115 @@ size_t jf_1, size_t jf_2) { /* Start of field to compare in each file. */ - char *beg1; - char *beg2; - - size_t len1; - size_t len2; /* Length of fields to compare. */ + char *beg[2]; + char *copy[2]; + size_t len[2]; /* Length of fields to compare. */ int diff; + int i, j; if (jf_1 < line1->nfields) { - beg1 = line1->fields[jf_1].beg; - len1 = line1->fields[jf_1].len; + beg[0] = line1->fields[jf_1].beg; + len[0] = line1->fields[jf_1].len; } else { - beg1 = NULL; - len1 = 0; + beg[0] = NULL; + len[0] = 0; } if (jf_2 < line2->nfields) { - beg2 = line2->fields[jf_2].beg; - len2 = line2->fields[jf_2].len; + beg[1] = line2->fields[jf_2].beg; + len[1] = line2->fields[jf_2].len; } else { - beg2 = NULL; - len2 = 0; + beg[1] = NULL; + len[1] = 0; } - if (len1 == 0) - return len2 == 0 ? 0 : -1; - if (len2 == 0) + if (len[0] == 0) + return len[1] == 0 ? 0 : -1; + if (len[1] == 0) return 1; if (ignore_case) { - /* FIXME: ignore_case does not work with NLS (in particular, - with multibyte chars). */ - diff = memcasecmp (beg1, beg2, MIN (len1, len2)); +#ifdef HAVE_MBRTOWC + if (MB_CUR_MAX > 1) + { + size_t mblength; + wchar_t wc, uwc; + mbstate_t state, state_bak; + + memset (&state, '\0', sizeof (mbstate_t)); + + for (i = 0; i < 2; i++) + { + copy[i] = alloca (len[i] + 1); + + for (j = 0; j < MIN (len[0], len[1]);) + { + state_bak = state; + mblength = mbrtowc (&wc, beg[i] + j, len[i] - j, &state); + + switch (mblength) + { + case (size_t) -1: + case (size_t) -2: + state = state_bak; + /* Fall through */ + case 0: + mblength = 1; + break; + + default: + uwc = towupper (wc); + + if (uwc != wc) + { + mbstate_t state_wc; + + memset (&state_wc, '\0', sizeof (mbstate_t)); + wcrtomb (copy[i] + j, uwc, &state_wc); + } + else + memcpy (copy[i] + j, beg[i] + j, mblength); + } + j += mblength; + } + copy[i][j] = '\0'; + } + } + else +#endif + { + for (i = 0; i < 2; i++) + { + copy[i] = alloca (len[i] + 1); + + for (j = 0; j < MIN (len[0], len[1]); j++) + copy[i][j] = toupper (beg[i][j]); + + copy[i][j] = '\0'; + } + } } else { - if (hard_LC_COLLATE) - return xmemcoll (beg1, len1, beg2, len2); - diff = memcmp (beg1, beg2, MIN (len1, len2)); + copy[0] = (unsigned char *) beg[0]; + copy[1] = (unsigned char *) beg[1]; } + if (hard_LC_COLLATE) + return xmemcoll ((char *) copy[0], len[0], (char *) copy[1], len[1]); + diff = memcmp (copy[0], copy[1], MIN (len[0], len[1])); + + if (diff) return diff; - return len1 < len2 ? -1 : len1 != len2; + return len[0] - len[1]; } /* Check that successive input lines PREV and CURRENT from input file @@ -415,6 +633,11 @@ return false; } +#if HAVE_MBRTOWC + if (MB_CUR_MAX > 1) + xfields_multibyte (line); + else +#endif xfields (line); if (prevline[which - 1]) @@ -516,11 +739,18 @@ /* Print the join of LINE1 and LINE2. */ +#define PUT_TAB_CHAR \ + do \ + { \ + (tab != NULL) ? \ + fwrite(tab, sizeof(char), tablen, stdout) : putchar (' '); \ + } \ + while (0) + static void prjoin (struct line const *line1, struct line const *line2) { const struct outlist *outlist; - char output_separator = tab < 0 ? ' ' : tab; outlist = outlist_head.next; if (outlist) @@ -536,12 +766,12 @@ if (o->file == 0) { if (line1 == &uni_blank) - { + { line = line2; field = join_field_2; } else - { + { line = line1; field = join_field_1; } @@ -555,7 +785,7 @@ o = o->next; if (o == NULL) break; - putchar (output_separator); + PUT_TAB_CHAR; } putchar ('\n'); } @@ -573,23 +803,23 @@ prfield (join_field_1, line1); for (i = 0; i < join_field_1 && i < line1->nfields; ++i) { - putchar (output_separator); + PUT_TAB_CHAR; prfield (i, line1); } for (i = join_field_1 + 1; i < line1->nfields; ++i) { - putchar (output_separator); + PUT_TAB_CHAR; prfield (i, line1); } for (i = 0; i < join_field_2 && i < line2->nfields; ++i) { - putchar (output_separator); + PUT_TAB_CHAR; prfield (i, line2); } for (i = join_field_2 + 1; i < line2->nfields; ++i) { - putchar (output_separator); + PUT_TAB_CHAR; prfield (i, line2); } putchar ('\n'); @@ -1020,20 +1250,41 @@ case 't': { - unsigned char newtab = optarg[0]; - if (! newtab) + char *newtab; + size_t newtablen; + if (! optarg[0]) error (EXIT_FAILURE, 0, _("empty tab")); - if (optarg[1]) + newtab = xstrdup (optarg); +#if HAVE_MBRTOWC + if (MB_CUR_MAX > 1) + { + mbstate_t state; + + memset (&state, 0, sizeof (mbstate_t)); + newtablen = mbrtowc (NULL, newtab, + strnlen (newtab, MB_LEN_MAX), + &state); + if (newtablen == (size_t) 0 + || newtablen == (size_t) -1 + || newtablen == (size_t) -2) + newtablen = 1; + } + else +#endif + newtablen = 1; + + if (newtablen == 1 && newtab[1]) + { + if (STREQ (newtab, "\\0")) + newtab[0] = '\0'; + } + if (tab != NULL && strcmp (tab, newtab)) { - if (STREQ (optarg, "\\0")) - newtab = '\0'; - else - error (EXIT_FAILURE, 0, _("multi-character tab %s"), - quote (optarg)); + free (newtab); + error (EXIT_FAILURE, 0, _("incompatible tabs")); } - if (0 <= tab && tab != newtab) - error (EXIT_FAILURE, 0, _("incompatible tabs")); tab = newtab; + tablen = newtablen; } break; diff -Naur coreutils-7.2.orig/src/pr.c coreutils-7.2/src/pr.c --- coreutils-7.2.orig/src/pr.c 2009-03-29 18:44:10.000000000 +0100 +++ coreutils-7.2/src/pr.c 2009-04-18 10:41:11.000000000 +0100 @@ -312,6 +312,32 @@ #include #include + +/* Get MB_LEN_MAX. */ +#include +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC + installation; work around this configuration error. */ +#if !defined MB_LEN_MAX || MB_LEN_MAX == 1 +# define MB_LEN_MAX 16 +#endif + +/* Get MB_CUR_MAX. */ +#include + +/* Solaris 2.5 has a bug: must be included before . */ +/* Get mbstate_t, mbrtowc(), wcwidth(). */ +#if HAVE_WCHAR_H +# include +#endif + +/* Get iswprint(). -- for wcwidth(). */ +#if HAVE_WCTYPE_H +# include +#endif +#if !defined iswprint && !HAVE_ISWPRINT +# define iswprint(wc) 1 +#endif + #include "system.h" #include "error.h" #include "mbswidth.h" @@ -321,6 +347,18 @@ #include "strftime.h" #include "xstrtol.h" +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ +#if HAVE_MBRTOWC && defined mbstate_t +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) +#endif + +#ifndef HAVE_DECL_WCWIDTH +"this configure-time declaration test was not run" +#endif +#if !HAVE_DECL_WCWIDTH +extern int wcwidth (); +#endif + /* The official name of this program (e.g., no `g' prefix). */ #define PROGRAM_NAME "pr" @@ -413,7 +451,20 @@ typedef struct COLUMN COLUMN; -static int char_to_clump (char c); +/* Funtion pointers to switch functions for single byte locale or for + multibyte locale. If multibyte functions do not exist in your sysytem, + these pointers always point the function for single byte locale. */ +static void (*print_char) (char c); +static int (*char_to_clump) (char c); + +/* Functions for single byte locale. */ +static void print_char_single (char c); +static int char_to_clump_single (char c); + +/* Functions for multibyte locale. */ +static void print_char_multi (char c); +static int char_to_clump_multi (char c); + static bool read_line (COLUMN *p); static bool print_page (void); static bool print_stored (COLUMN *p); @@ -423,6 +474,7 @@ static void pad_across_to (int position); static void add_line_number (COLUMN *p); static void getoptarg (char *arg, char switch_char, char *character, + int *character_length, int *character_width, int *number); void usage (int status); static void print_files (int number_of_files, char **av); @@ -437,7 +489,6 @@ static void pad_down (int lines); static void read_rest_of_line (COLUMN *p); static void skip_read (COLUMN *p, int column_number); -static void print_char (char c); static void cleanup (void); static void print_sep_string (void); static void separator_string (const char *optarg_S); @@ -449,7 +500,7 @@ we store the leftmost columns contiguously in buff. To print a line from buff, get the index of the first character from line_vector[i], and print up to line_vector[i + 1]. */ -static char *buff; +static unsigned char *buff; /* Index of the position in buff where the next character will be stored. */ @@ -553,7 +604,7 @@ static bool untabify_input = false; /* (-e) The input tab character. */ -static char input_tab_char = '\t'; +static char input_tab_char[MB_LEN_MAX] = "\t"; /* (-e) Tabstops are at chars_per_tab, 2*chars_per_tab, 3*chars_per_tab, ... where the leftmost column is 1. */ @@ -563,7 +614,10 @@ static bool tabify_output = false; /* (-i) The output tab character. */ -static char output_tab_char = '\t'; +static char output_tab_char[MB_LEN_MAX] = "\t"; + +/* (-i) The byte length of output tab character. */ +static int output_tab_char_length = 1; /* (-i) The width of the output tab. */ static int chars_per_output_tab = 8; @@ -637,7 +691,13 @@ static bool numbered_lines = false; /* (-n) Character which follows each line number. */ -static char number_separator = '\t'; +static char number_separator[MB_LEN_MAX] = "\t"; + +/* (-n) The byte length of the character which follows each line number. */ +static int number_separator_length = 1; + +/* (-n) The character width of the character which follows each line number. */ +static int number_separator_width = 0; /* (-n) line counting starts with 1st line of input file (not with 1st line of 1st page printed). */ @@ -690,6 +750,7 @@ -a|COLUMN|-m is a `space' and with the -J option a `tab'. */ static char *col_sep_string = (char *) ""; static int col_sep_length = 0; +static int col_sep_width = 0; static char *column_separator = (char *) " "; static char *line_separator = (char *) "\t"; @@ -846,6 +907,13 @@ col_sep_length = (int) strlen (optarg_S); col_sep_string = xmalloc (col_sep_length + 1); strcpy (col_sep_string, optarg_S); + +#if HAVE_MBRTOWC + if (MB_CUR_MAX > 1) + col_sep_width = mbswidth (col_sep_string, 0); + else +#endif + col_sep_width = col_sep_length; } int @@ -870,6 +938,21 @@ atexit (close_stdout); +/* Define which functions are used, the ones for single byte locale or the ones + for multibyte locale. */ +#if HAVE_MBRTOWC + if (MB_CUR_MAX > 1) + { + print_char = print_char_multi; + char_to_clump = char_to_clump_multi; + } + else +#endif + { + print_char = print_char_single; + char_to_clump = char_to_clump_single; + } + n_files = 0; file_names = (argc > 1 ? xmalloc ((argc - 1) * sizeof (char *)) @@ -946,8 +1029,12 @@ break; case 'e': if (optarg) - getoptarg (optarg, 'e', &input_tab_char, - &chars_per_input_tab); + { + int dummy_length, dummy_width; + + getoptarg (optarg, 'e', input_tab_char, &dummy_length, + &dummy_width, &chars_per_input_tab); + } /* Could check tab width > 0. */ untabify_input = true; break; @@ -960,8 +1047,12 @@ break; case 'i': if (optarg) - getoptarg (optarg, 'i', &output_tab_char, - &chars_per_output_tab); + { + int dummy_width; + + getoptarg (optarg, 'i', output_tab_char, &output_tab_char_length, + &dummy_width, &chars_per_output_tab); + } /* Could check tab width > 0. */ tabify_output = true; break; @@ -988,8 +1079,8 @@ case 'n': numbered_lines = true; if (optarg) - getoptarg (optarg, 'n', &number_separator, - &chars_per_number); + getoptarg (optarg, 'n', number_separator, &number_separator_length, + &number_separator_width, &chars_per_number); break; case 'N': skip_count = false; @@ -1028,7 +1119,7 @@ old_s = false; /* Reset an additional input of -s, -S dominates -s */ col_sep_string = bad_cast (""); - col_sep_length = 0; + col_sep_length = col_sep_width = 0; use_col_separator = true; if (optarg) separator_string (optarg); @@ -1185,10 +1276,45 @@ a number. */ static void -getoptarg (char *arg, char switch_char, char *character, int *number) +getoptarg (char *arg, char switch_char, char *character, int *character_length, + int *character_width, int *number) { if (!ISDIGIT (*arg)) - *character = *arg++; + { +#ifdef HAVE_MBRTOWC + if (MB_CUR_MAX > 1) /* for multibyte locale. */ + { + wchar_t wc; + size_t mblength; + int width; + mbstate_t state = {'\0'}; + + mblength = mbrtowc (&wc, arg, strnlen(arg, MB_LEN_MAX), &state); + + if (mblength == (size_t)-1 || mblength == (size_t)-2) + { + *character_length = 1; + *character_width = 1; + } + else + { + *character_length = (mblength < 1) ? 1 : mblength; + width = wcwidth (wc); + *character_width = (width < 0) ? 0 : width; + } + + strncpy (character, arg, *character_length); + arg += *character_length; + } + else /* for single byte locale. */ +#endif + { + *character = *arg++; + *character_length = 1; + *character_width = 1; + } + } + if (*arg) { long int tmp_long; @@ -1247,7 +1373,7 @@ else col_sep_string = column_separator; - col_sep_length = 1; + col_sep_length = col_sep_width = 1; use_col_separator = true; } /* It's rather pointless to define a TAB separator with column @@ -1278,11 +1404,11 @@ TAB_WIDTH (chars_per_input_tab, chars_per_number); */ /* Estimate chars_per_text without any margin and keep it constant. */ - if (number_separator == '\t') + if (number_separator[0] == '\t') number_width = chars_per_number + TAB_WIDTH (chars_per_default_tab, chars_per_number); else - number_width = chars_per_number + 1; + number_width = chars_per_number + number_separator_width; /* The number is part of the column width unless we are printing files in parallel. */ @@ -1297,7 +1423,7 @@ } chars_per_column = (chars_per_line - chars_used_by_number - - (columns - 1) * col_sep_length) / columns; + (columns - 1) * col_sep_width) / columns; if (chars_per_column < 1) error (EXIT_FAILURE, 0, _("page width too narrow")); @@ -1422,7 +1548,7 @@ /* Enlarge p->start_position of first column to use the same form of padding_not_printed with all columns. */ - h = h + col_sep_length; + h = h + col_sep_width; /* This loop takes care of all but the rightmost column. */ @@ -1456,7 +1582,7 @@ } else { - h = h_next + col_sep_length; + h = h_next + col_sep_width; h_next = h + chars_per_column; } } @@ -1746,9 +1872,9 @@ align_column (COLUMN *p) { padding_not_printed = p->start_position; - if (padding_not_printed - col_sep_length > 0) + if (padding_not_printed - col_sep_width > 0) { - pad_across_to (padding_not_printed - col_sep_length); + pad_across_to (padding_not_printed - col_sep_width); padding_not_printed = ANYWHERE; } @@ -2019,13 +2145,13 @@ /* May be too generous. */ buff = X2REALLOC (buff, &buff_allocated); } - buff[buff_current++] = c; + buff[buff_current++] = (unsigned char) c; } static void add_line_number (COLUMN *p) { - int i; + int i, j; char *s; int left_cut; @@ -2048,22 +2174,24 @@ /* Tabification is assumed for multiple columns, also for n-separators, but `default n-separator = TAB' hasn't been given priority over equal column_width also specified by POSIX. */ - if (number_separator == '\t') + if (number_separator[0] == '\t') { i = number_width - chars_per_number; while (i-- > 0) (p->char_func) (' '); } else - (p->char_func) (number_separator); + for (j = 0; j < number_separator_length; j++) + (p->char_func) (number_separator[j]); } else /* To comply with POSIX, we avoid any expansion of default TAB separator with a single column output. No column_width requirement has to be considered. */ { - (p->char_func) (number_separator); - if (number_separator == '\t') + for (j = 0; j < number_separator_length; j++) + (p->char_func) (number_separator[j]); + if (number_separator[0] == '\t') output_position = POS_AFTER_TAB (chars_per_output_tab, output_position); } @@ -2224,7 +2352,7 @@ while (goal - h_old > 1 && (h_new = POS_AFTER_TAB (chars_per_output_tab, h_old)) <= goal) { - putchar (output_tab_char); + fwrite (output_tab_char, sizeof(char), output_tab_char_length, stdout); h_old = h_new; } while (++h_old <= goal) @@ -2244,6 +2372,7 @@ { char *s; int l = col_sep_length; + int not_space_flag; s = col_sep_string; @@ -2257,6 +2386,7 @@ { for (; separators_not_printed > 0; --separators_not_printed) { + not_space_flag = 0; while (l-- > 0) { /* 3 types of sep_strings: spaces only, spaces and chars, @@ -2270,12 +2400,15 @@ } else { + not_space_flag = 1; if (spaces_not_printed > 0) print_white_space (); putchar (*s++); - ++output_position; } } + if (not_space_flag) + output_position += col_sep_width; + /* sep_string ends with some spaces */ if (spaces_not_printed > 0) print_white_space (); @@ -2303,7 +2436,7 @@ required number of tabs and spaces. */ static void -print_char (char c) +print_char_single (char c) { if (tabify_output) { @@ -2327,6 +2460,74 @@ putchar (c); } +#ifdef HAVE_MBRTOWC +static void +print_char_multi (char c) +{ + static size_t mbc_pos = 0; + static char mbc[MB_LEN_MAX] = {'\0'}; + static mbstate_t state = {'\0'}; + mbstate_t state_bak; + wchar_t wc; + size_t mblength; + int width; + + if (tabify_output) + { + state_bak = state; + mbc[mbc_pos++] = c; + mblength = mbrtowc (&wc, mbc, mbc_pos, &state); + + while (mbc_pos > 0) + { + switch (mblength) + { + case (size_t)-2: + state = state_bak; + return; + + case (size_t)-1: + state = state_bak; + ++output_position; + putchar (mbc[0]); + memmove (mbc, mbc + 1, MB_CUR_MAX - 1); + --mbc_pos; + break; + + case 0: + mblength = 1; + + default: + if (wc == L' ') + { + memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength); + --mbc_pos; + ++spaces_not_printed; + return; + } + else if (spaces_not_printed > 0) + print_white_space (); + + /* Nonprintables are assumed to have width 0, except L'\b'. */ + if ((width = wcwidth (wc)) < 1) + { + if (wc == L'\b') + --output_position; + } + else + output_position += width; + + fwrite (mbc, sizeof(char), mblength, stdout); + memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength); + mbc_pos -= mblength; + } + } + return; + } + putchar (c); +} +#endif + /* Skip to page PAGE before printing. PAGE may be larger than total number of pages. */ @@ -2506,9 +2707,9 @@ align_empty_cols = false; } - if (padding_not_printed - col_sep_length > 0) + if (padding_not_printed - col_sep_width > 0) { - pad_across_to (padding_not_printed - col_sep_length); + pad_across_to (padding_not_printed - col_sep_width); padding_not_printed = ANYWHERE; } @@ -2609,9 +2810,9 @@ } } - if (padding_not_printed - col_sep_length > 0) + if (padding_not_printed - col_sep_width > 0) { - pad_across_to (padding_not_printed - col_sep_length); + pad_across_to (padding_not_printed - col_sep_width); padding_not_printed = ANYWHERE; } @@ -2624,8 +2825,8 @@ if (spaces_not_printed == 0) { output_position = p->start_position + end_vector[line]; - if (p->start_position - col_sep_length == chars_per_margin) - output_position -= col_sep_length; + if (p->start_position - col_sep_width == chars_per_margin) + output_position -= col_sep_width; } return true; @@ -2644,7 +2845,7 @@ number of characters is 1.) */ static int -char_to_clump (char c) +char_to_clump_single (char c) { unsigned char uc = c; char *s = clump_buff; @@ -2654,10 +2855,10 @@ int chars; int chars_per_c = 8; - if (c == input_tab_char) + if (c == input_tab_char[0]) chars_per_c = chars_per_input_tab; - if (c == input_tab_char || c == '\t') + if (c == input_tab_char[0] || c == '\t') { width = TAB_WIDTH (chars_per_c, input_position); @@ -2738,6 +2939,154 @@ return chars; } +#ifdef HAVE_MBRTOWC +static int +char_to_clump_multi (char c) +{ + static size_t mbc_pos = 0; + static char mbc[MB_LEN_MAX] = {'\0'}; + static mbstate_t state = {'\0'}; + mbstate_t state_bak; + wchar_t wc; + size_t mblength; + int wc_width; + register char *s = clump_buff; + register int i, j; + char esc_buff[4]; + int width; + int chars; + int chars_per_c = 8; + + state_bak = state; + mbc[mbc_pos++] = c; + mblength = mbrtowc (&wc, mbc, mbc_pos, &state); + + width = 0; + chars = 0; + while (mbc_pos > 0) + { + switch (mblength) + { + case (size_t)-2: + state = state_bak; + return 0; + + case (size_t)-1: + state = state_bak; + mblength = 1; + + if (use_esc_sequence || use_cntrl_prefix) + { + width = +4; + chars = +4; + *s++ = '\\'; + sprintf (esc_buff, "%03o", mbc[0]); + for (i = 0; i <= 2; ++i) + *s++ = (int) esc_buff[i]; + } + else + { + width += 1; + chars += 1; + *s++ = mbc[0]; + } + break; + + case 0: + mblength = 1; + /* Fall through */ + + default: + if (memcmp (mbc, input_tab_char, mblength) == 0) + chars_per_c = chars_per_input_tab; + + if (memcmp (mbc, input_tab_char, mblength) == 0 || c == '\t') + { + int width_inc; + + width_inc = TAB_WIDTH (chars_per_c, input_position); + width += width_inc; + + if (untabify_input) + { + for (i = width_inc; i; --i) + *s++ = ' '; + chars += width_inc; + } + else + { + for (i = 0; i < mblength; i++) + *s++ = mbc[i]; + chars += mblength; + } + } + else if ((wc_width = wcwidth (wc)) < 1) + { + if (use_esc_sequence) + { + for (i = 0; i < mblength; i++) + { + width += 4; + chars += 4; + *s++ = '\\'; + sprintf (esc_buff, "%03o", c); + for (j = 0; j <= 2; ++j) + *s++ = (int) esc_buff[j]; + } + } + else if (use_cntrl_prefix) + { + if (wc < 0200) + { + width += 2; + chars += 2; + *s++ = '^'; + *s++ = wc ^ 0100; + } + else + { + for (i = 0; i < mblength; i++) + { + width += 4; + chars += 4; + *s++ = '\\'; + sprintf (esc_buff, "%03o", c); + for (j = 0; j <= 2; ++j) + *s++ = (int) esc_buff[j]; + } + } + } + else if (wc == L'\b') + { + width += -1; + chars += 1; + *s++ = c; + } + else + { + width += 0; + chars += mblength; + for (i = 0; i < mblength; i++) + *s++ = mbc[i]; + } + } + else + { + width += wc_width; + chars += mblength; + for (i = 0; i < mblength; i++) + *s++ = mbc[i]; + } + } + memmove (mbc, mbc + mblength, MB_CUR_MAX - mblength); + mbc_pos -= mblength; + } + + input_position += width; + return chars; +} +#endif + /* We've just printed some files and need to clean up things before looking for more options and printing the next batch of files. diff -Naur coreutils-7.2.orig/src/pr.c.orig coreutils-7.2/src/pr.c.orig --- coreutils-7.2.orig/src/pr.c.orig 1970-01-01 01:00:00.000000000 +0100 +++ coreutils-7.2/src/pr.c.orig 2009-03-29 18:44:10.000000000 +0100 @@ -0,0 +1,2876 @@ +/* pr -- convert text files for printing. + Copyright (C) 88, 91, 1995-2009 Free Software Foundation, Inc. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see . */ + +/* By Pete TerMaat, with considerable refinement by Roland Huebner. */ + +/* Things to watch: Sys V screws up on ... + pr -n -3 -s: /usr/dict/words + pr -m -o10 -n /usr/dict/words{,,,} + pr -6 -a -n -o5 /usr/dict/words + + Ideas: + + Keep a things_to_do list of functions to call when we know we have + something to print. Cleaner than current series of checks. + + Improve the printing of control prefixes. + + Expand the file name in the centered header line to a full file name. + + + Concept: + + If the input_tab_char differs from the default value TAB + (`-e[CHAR[...]]' is used), any input text tab is expanded to the + default width of 8 spaces (compare char_to_clump). - Same as SunOS + does. + + The treatment of the number_separator (compare add_line_number): + The default value TAB of the number_separator (`-n[SEP[...]]') doesn't + be thought to be an input character. An optional `-e'-input has no + effect. + - With single column output + only one POSIX requirement has to be met: + The default n-separator should be a TAB. The consequence is a + different width between the number and the text if the output position + of the separator changes, i.e. it depends upon the left margin used. + That's not nice but easy-to-use together with the defaults of other + utilities, e.g. sort or cut. - Same as SunOS does. + - With multicolumn output + two conflicting POSIX requirements exist: + First `default n-separator is TAB', second `output text columns shall + be of equal width'. Moreover POSIX specifies the number+separator a + part of the column, together with `-COLUMN' and `-a -COLUMN'. + (With -m output the number shall occupy each line only once. Exactly + the same situation as single column output exists.) + GNU pr gives priority to the 2nd requirement and observes POSIX + column definition. The n-separator TAB is expanded to the same number + of spaces in each column using the default value 8. Tabification is + only performed if it is compatible with the output position. + Consequence: The output text columns are of equal width. The layout + of a page does not change if the left margin varies. - Looks better + than the SunOS approach. + SunOS pr gives priority to the 1st requirement. n-separator TAB + width varies with each column. Only the width of text part of the + column is fixed. + Consequence: The output text columns don't have equal width. The + widths and the layout of the whole page varies with the left margin. + An overflow of the line length (without margin) over the input value + PAGE_WIDTH may occur. + + The interference of the POSIX-compliant small letter options -w and -s: + (`interference' means `setting a _separator_ with -s switches off the + column sturctur and the default - not generally - page_width, + acts on -w option') + options: text form / separator: equivalent new options: + -w l -s[x] + -------------------------------------------------------------------- + 1. -- -- columns / space -- + trunc. to page_width = 72 + 2. -- -s[:] full lines / TAB[:] -J --sep-string[=""|:] + no truncation + 3. -w l -- columns / space -W l + trunc. to page_width = l + 4. -w l -s[:] columns / no sep.[:] -W l --sep-string[=:] + trunc. to page_width = l + -------------------------------------------------------------------- + + + Options: + + Including version 1.22i: + Some SMALL LETTER options has been redefined with the object of a + better POSIX compliance. The output of some further cases has been + adapted to other UNIXes. A violation of downward compatibility has to + be accepted. + Some NEW CAPITAL LETTER options ( -J, -S, -W) has been introduced to + turn off unexpected interferences of small letter options (-s and -w + together with the three column options). + -N option and the second argument LAST_PAGE of +FIRST_PAGE offer more + flexibility; The detailed handling of form feeds set in the input + files requires -T option. + + Capital letter options dominate small letter ones. + + Some of the option-arguments cannot be specified as separate arguments + from the preceding option letter (already stated in POSIX specification). + + Form feeds in the input cause page breaks in the output. Multiple + form feeds produce empty pages. + + +FIRST_PAGE[:LAST_PAGE], --pages=FIRST_PAGE[:LAST_PAGE] + begin [stop] printing with page FIRST_[LAST_]PAGE + + -COLUMN, --columns=COLUMN + Produce output that is COLUMN columns wide and + print columns down, unless -a is used. Balance number of + lines in the columns on each page. + + -a, --across Print columns across rather than down, used + together with -COLUMN. The input + one + two + three + four + will be printed with `-a -3' as + one two three + four + + -b Balance columns on the last page. + -b is no longer an independent option. It's always used + together with -COLUMN (unless -a is used) to get a + consistent formulation with "FF set by hand" in input + files. Each formfeed found terminates the number of lines + to be read with the actual page. The situation for + printing columns down is equivalent to that on the last + page. So we need a balancing. + + Keeping -b as an underground option guarantees some + downward compatibility. Utilities using pr with -b + (a most frequently used form) still work as usual. + + -c, --show-control-chars + Print unprintable characters as control prefixes. + Control-g is printed as ^G (use hat notation) and + octal backslash notation. + + -d, --double-space Double space the output. + + -D FORMAT, --date-format=FORMAT Use FORMAT for the header date. + + -e[CHAR[WIDTH]], --expand-tabs[=CHAR[WIDTH]] + Expand tabs to spaces on input. Optional argument CHAR + is the input TAB character. (Default is TAB). Optional + argument WIDTH is the input TAB character's width. + (Default is 8.) + + -F, -f, --form-feed Use formfeeds instead of newlines to separate + pages. A three line HEADER is used, no TRAILER with -F, + without -F both HEADER and TRAILER are made of five lines. + + -h HEADER, --header=HEADER + Replace the filename in the header with the string HEADER. + A centered header is used. + + -i[CHAR[WIDTH]], --output-tabs[=CHAR[WIDTH]] + Replace spaces with tabs on output. Optional argument + CHAR is the output TAB character. (Default is TAB). + Optional argument WIDTH is the output TAB character's + width. (Default is 8) + + -J, --join-lines Merge lines of full length, turns off -W/-w + line truncation, no column alignment, --sep-string[=STRING] + sets separators, works with all column options + (-COLUMN | -a -COLUMN | -m). + -J has been introduced (together with -W and --sep-string) to + disentangle the old (POSIX compliant) options -w, -s + along with the 3 column options. + + -l PAGE_LENGTH, --length=PAGE_LENGTH + Set the page length to PAGE_LENGTH lines. Default is 66, + including 5 lines of HEADER and 5 lines of TRAILER + without -F, but only 3 lines of HEADER and no TRAILER + with -F (i.e the number of text lines defaults to 56 or + 63 respectively). + + -m, --merge Print files in parallel; pad_across_to align + columns; truncate lines and print separator strings; + Do it also with empty columns to get a continuous line + numbering and column marking by separators throughout + the whole merged file. + + Empty pages in some input files produce empty columns + [marked by separators] in the merged pages. Completely + empty merged pages show no column separators at all. + + The layout of a merged page is ruled by the largest form + feed distance of the single pages at that page. Shorter + columns will be filled up with empty lines. + + Together with -J option join lines of full length and + set separators when -S option is used. + + -n[SEP[DIGITS]], --number-lines[=SEP[DIGITS]] + Provide DIGITS digit line numbering (default for DIGITS + is 5). With multicolumn output the number occupies the + first DIGITS column positions of each text column or only + each line of -m output. + With single column output the number precedes each line + just as -m output. + Optional argument SEP is the character appended to the + line number to separate it from the text followed. + The default separator is a TAB. In a strict sense a TAB + is always printed with single column output only. The + TAB-width varies with the TAB-position, e.g. with the + left margin specified by -o option. + With multicolumn output priority is given to `equal width + of output columns' (a POSIX specification). The TAB-width + is fixed to the value of the 1st column and does not + change with different values of left margin. That means a + fixed number of spaces is always printed in the place of + a TAB. The tabification depends upon the output + position. + + Default counting of the line numbers starts with 1st + line of the input file (not the 1st line printed, + compare the --page option and -N option). + + -N NUMBER, --first-line-number=NUMBER + Start line counting with the number NUMBER at the 1st + line of first page printed (mostly not the 1st line of + the input file). + + -o MARGIN, --indent=MARGIN + Offset each line with a margin MARGIN spaces wide. + Total page width is the size of the margin plus the + PAGE_WIDTH set with -W/-w option. + + -r, --no-file-warnings + Omit warning when a file cannot be opened. + + -s[CHAR], --separator[=CHAR] + Separate columns by a single character CHAR, default for + CHAR is the TAB character without -w and 'no char' with -w. + Without `-s' default separator `space' is set. + -s[CHAR] turns off line truncation of all 3 column options + (-COLUMN|-a -COLUMN|-m) except -w is set. That is a POSIX + compliant formulation. The source code translates -s into + the new options -S and -J, also -W if required. + + -S STRING, --sep-string[=STRING] + Separate columns by any string STRING. The -S option + doesn't react upon the -W/-w option (unlike -s option + does). It defines a separator nothing else. + Without -S: Default separator TAB is used with -J and + `space' otherwise (same as -S" "). + With -S "": No separator is used. + Quotes should be used with blanks and some shell active + characters. + -S is problematic because in its obsolete form you + cannot use -S "STRING", but in its standard form you + must use -S "STRING" if STRING is empty. Use + --sep-string to avoid the ambiguity. + + -t, --omit-header Do not print headers or footers but retain form + feeds set in the input files. + + -T, --omit-pagination + Do not print headers or footers, eliminate any pagination + by form feeds set in the input files. + + -v, --show-nonprinting + Print unprintable characters as escape sequences. Use + octal backslash notation. Control-G becomes \007. + + -w PAGE_WIDTH, --width=PAGE_WIDTH + Set page width to PAGE_WIDTH characters for multiple + text-column output only (default for PAGE_WIDTH is 72). + -s[CHAR] turns off the default page width and any line + truncation. Lines of full length will be merged, + regardless of the column options set. A POSIX compliant + formulation. + + -W PAGE_WIDTH, --page-width=PAGE_WIDTH + Set the page width to PAGE_WIDTH characters. That's valid + with and without a column option. Text lines will be + truncated, unless -J is used. Together with one of the + column options (-COLUMN| -a -COLUMN| -m) column alignment + is always used. + Default is 72 characters. + Without -W PAGE_WIDTH + - but with one of the column options default truncation of + 72 characters is used (to keep downward compatibility + and to simplify most frequently met column tasks). + Column alignment and column separators are used. + - and without any of the column options NO line truncation + is used (to keep downward compatibility and to meet most + frequent tasks). That's equivalent to -W 72 -J . + + With/without -W PAGE_WIDTH the header line is always + truncated to avoid line overflow. + + (In pr versions newer than 1.14 -S option does no longer + affect -W option.) + +*/ + + +#include + +#include +#include +#include "system.h" +#include "error.h" +#include "mbswidth.h" +#include "quote.h" +#include "stat-time.h" +#include "stdio--.h" +#include "strftime.h" +#include "xstrtol.h" + +/* The official name of this program (e.g., no `g' prefix). */ +#define PROGRAM_NAME "pr" + +#define AUTHORS \ + proper_name ("Pete TerMaat"), \ + proper_name ("Roland Huebner") + +/* Used with start_position in the struct COLUMN described below. + If start_position == ANYWHERE, we aren't truncating columns and + can begin printing a column anywhere. Otherwise we must pad to + the horizontal position start_position. */ +#define ANYWHERE 0 + +/* Each column has one of these structures allocated for it. + If we're only dealing with one file, fp is the same for all + columns. + + The general strategy is to spend time setting up these column + structures (storing columns if necessary), after which printing + is a matter of flitting from column to column and calling + print_func. + + Parallel files, single files printing across in multiple + columns, and single files printing down in multiple columns all + fit the same printing loop. + + print_func Function used to print lines in this column. + If we're storing this column it will be + print_stored(), Otherwise it will be read_line(). + + char_func Function used to process characters in this column. + If we're storing this column it will be store_char(), + otherwise it will be print_char(). + + current_line Index of the current entry in line_vector, which + contains the index of the first character of the + current line in buff[]. + + lines_stored Number of lines in this column which are stored in + buff. + + lines_to_print If we're storing this column, lines_to_print is + the number of stored_lines which remain to be + printed. Otherwise it is the number of lines + we can print without exceeding lines_per_body. + + start_position The horizontal position we want to be in before we + print the first character in this column. + + numbered True means precede this column with a line number. */ + +/* FIXME: There are many unchecked integer overflows in this file, + that will cause this command to misbehave given large inputs or + options. Many of the "int" values below should be "size_t" or + something else like that. */ + +struct COLUMN; +struct COLUMN + { + FILE *fp; /* Input stream for this column. */ + char const *name; /* File name. */ + enum + { + OPEN, + FF_FOUND, /* used with -b option, set with \f, changed + to ON_HOLD after print_header */ + ON_HOLD, /* Hit a form feed. */ + CLOSED + } + status; /* Status of the file pointer. */ + + /* Func to print lines in this col. */ + bool (*print_func) (struct COLUMN *); + + /* Func to print/store chars in this col. */ + void (*char_func) (char); + + int current_line; /* Index of current place in line_vector. */ + int lines_stored; /* Number of lines stored in buff. */ + int lines_to_print; /* No. lines stored or space left on page. */ + int start_position; /* Horizontal position of first char. */ + bool numbered; + bool full_page_printed; /* True means printed without a FF found. */ + + /* p->full_page_printed controls a special case of "FF set by hand": + True means a full page has been printed without FF found. To avoid an + additional empty page we have to ignore a FF immediately following in + the next line. */ + }; + +typedef struct COLUMN COLUMN; + +static int char_to_clump (char c); +static bool read_line (COLUMN *p); +static bool print_page (void); +static bool print_stored (COLUMN *p); +static bool open_file (char *name, COLUMN *p); +static bool skip_to_page (uintmax_t page); +static void print_header (void); +static void pad_across_to (int position); +static void add_line_number (COLUMN *p); +static void getoptarg (char *arg, char switch_char, char *character, + int *number); +void usage (int status); +static void print_files (int number_of_files, char **av); +static void init_parameters (int number_of_files); +static void init_header (char const *filename, int desc); +static bool init_fps (int number_of_files, char **av); +static void init_funcs (void); +static void init_store_cols (void); +static void store_columns (void); +static void balance (int total_stored); +static void store_char (char c); +static void pad_down (int lines); +static void read_rest_of_line (COLUMN *p); +static void skip_read (COLUMN *p, int column_number); +static void print_char (char c); +static void cleanup (void); +static void print_sep_string (void); +static void separator_string (const char *optarg_S); + +/* All of the columns to print. */ +static COLUMN *column_vector; + +/* When printing a single file in multiple downward columns, + we store the leftmost columns contiguously in buff. + To print a line from buff, get the index of the first character + from line_vector[i], and print up to line_vector[i + 1]. */ +static char *buff; + +/* Index of the position in buff where the next character + will be stored. */ +static unsigned int buff_current; + +/* The number of characters in buff. + Used for allocation of buff and to detect overflow of buff. */ +static size_t buff_allocated; + +/* Array of indices into buff. + Each entry is an index of the first character of a line. + This is used when storing lines to facilitate shuffling when + we do column balancing on the last page. */ +static int *line_vector; + +/* Array of horizonal positions. + For each line in line_vector, end_vector[line] is the horizontal + position we are in after printing that line. We keep track of this + so that we know how much we need to pad to prepare for the next + column. */ +static int *end_vector; + +/* (-m) True means we're printing multiple files in parallel. */ +static bool parallel_files = false; + +/* (-m) True means a line starts with some empty columns (some files + already CLOSED or ON_HOLD) which we have to align. */ +static bool align_empty_cols; + +/* (-m) True means we have not yet found any printable column in a line. + align_empty_cols = true has to be maintained. */ +static bool empty_line; + +/* (-m) False means printable column output precedes a form feed found. + Column alignment is done only once. No additional action with that form + feed. + True means we found only a form feed in a column. Maybe we have to do + some column alignment with that form feed. */ +static bool FF_only; + +/* (-[0-9]+) True means we're given an option explicitly specifying + number of columns. Used to detect when this option is used with -m + and when translating old options to new/long options. */ +static bool explicit_columns = false; + +/* (-t|-T) False means we aren't printing headers and footers. */ +static bool extremities = true; + +/* (-t) True means we retain all FF set by hand in input files. + False is set with -T option. */ +static bool keep_FF = false; +static bool print_a_FF = false; + +/* True means we need to print a header as soon as we know we've got input + to print after it. */ +static bool print_a_header; + +/* (-f) True means use formfeeds instead of newlines to separate pages. */ +static bool use_form_feed = false; + +/* True means we have read the standard input. */ +static bool have_read_stdin = false; + +/* True means the -a flag has been given. */ +static bool print_across_flag = false; + +/* True means we're printing one file in multiple (>1) downward columns. */ +static bool storing_columns = true; + +/* (-b) True means balance columns on the last page as Sys V does. */ +/* That's no longer an independent option. With storing_columns = true + balance_columns = true is used too (s. function init_parameters). + We get a consistent formulation with "FF set by hand" in input files. */ +static bool balance_columns = false; + +/* (-l) Number of lines on a page, including header and footer lines. */ +static int lines_per_page = 66; + +/* Number of lines in the header and footer can be reset to 0 using + the -t flag. */ +enum { lines_per_header = 5 }; +static int lines_per_body; +enum { lines_per_footer = 5 }; + +/* (-w|-W) Width in characters of the page. Does not include the width of + the margin. */ +static int chars_per_line = 72; + +/* (-w|W) True means we truncate lines longer than chars_per_column. */ +static bool truncate_lines = false; + +/* (-J) True means we join lines without any line truncation. -J + dominates -w option. */ +static bool join_lines = false; + +/* Number of characters in a column. Based on col_sep_length and + page width. */ +static int chars_per_column; + +/* (-e) True means convert tabs to spaces on input. */ +static bool untabify_input = false; + +/* (-e) The input tab character. */ +static char input_tab_char = '\t'; + +/* (-e) Tabstops are at chars_per_tab, 2*chars_per_tab, 3*chars_per_tab, ... + where the leftmost column is 1. */ +static int chars_per_input_tab = 8; + +/* (-i) True means convert spaces to tabs on output. */ +static bool tabify_output = false; + +/* (-i) The output tab character. */ +static char output_tab_char = '\t'; + +/* (-i) The width of the output tab. */ +static int chars_per_output_tab = 8; + +/* Keeps track of pending white space. When we hit a nonspace + character after some whitespace, we print whitespace, tabbing + if necessary to get to output_position + spaces_not_printed. */ +static int spaces_not_printed; + +/* (-o) Number of spaces in the left margin (tabs used when possible). */ +static int chars_per_margin = 0; + +/* Position where the next character will fall. + Leftmost position is 0 + chars_per_margin. + Rightmost position is chars_per_margin + chars_per_line - 1. + This is important for converting spaces to tabs on output. */ +static int output_position; + +/* Horizontal position relative to the current file. + (output_position depends on where we are on the page; + input_position depends on where we are in the file.) + Important for converting tabs to spaces on input. */ +static int input_position; + +/* True if there were any failed opens so we can exit with nonzero + status. */ +static bool failed_opens = false; + +/* The number of spaces taken up if we print a tab character with width + c_ from position h_. */ +#define TAB_WIDTH(c_, h_) ((c_) - ((h_) % (c_))) + +/* The horizontal position we'll be at after printing a tab character + of width c_ from the position h_. */ +#define POS_AFTER_TAB(c_, h_) ((h_) + TAB_WIDTH (c_, h_)) + +/* (-NNN) Number of columns of text to print. */ +static int columns = 1; + +/* (+NNN:MMM) Page numbers on which to begin and stop printing. + first_page_number = 0 will be used to check input only. */ +static uintmax_t first_page_number = 0; +static uintmax_t last_page_number = UINTMAX_MAX; + +/* Number of files open (not closed, not on hold). */ +static int files_ready_to_read = 0; + +/* Current page number. Displayed in header. */ +static uintmax_t page_number; + +/* Current line number. Displayed when -n flag is specified. + + When printing files in parallel (-m flag), line numbering is as follows: + 1 foo goo moo + 2 hoo too zoo + + When printing files across (-a flag), ... + 1 foo 2 moo 3 goo + 4 hoo 5 too 6 zoo + + Otherwise, line numbering is as follows: + 1 foo 3 goo 5 too + 2 moo 4 hoo 6 zoo */ +static int line_number; + +/* With line_number overflow, we use power_10 to cut off the higher-order + digits of the line_number */ +static int power_10; + +/* (-n) True means lines should be preceded by numbers. */ +static bool numbered_lines = false; + +/* (-n) Character which follows each line number. */ +static char number_separator = '\t'; + +/* (-n) line counting starts with 1st line of input file (not with 1st + line of 1st page printed). */ +static int line_count = 1; + +/* (-n) True means counting of skipped lines starts with 1st line of + input file. False means -N option is used in addition, counting of + skipped lines not required. */ +static bool skip_count = true; + +/* (-N) Counting starts with start_line_number = NUMBER at 1st line of + first page printed, usually not 1st page of input file. */ +static int start_line_num = 1; + +/* (-n) Width in characters of a line number. */ +static int chars_per_number = 5; + +/* Used when widening the first column to accommodate numbers -- only + needed when printing files in parallel. Includes width of both the + number and the number_separator. */ +static int number_width; + +/* Buffer sprintf uses to format a line number. */ +static char *number_buff; + +/* (-v) True means unprintable characters are printed as escape sequences. + control-g becomes \007. */ +static bool use_esc_sequence = false; + +/* (-c) True means unprintable characters are printed as control prefixes. + control-g becomes ^G. */ +static bool use_cntrl_prefix = false; + +/* (-d) True means output is double spaced. */ +static bool double_space = false; + +/* Number of files opened initially in init_files. Should be 1 + unless we're printing multiple files in parallel. */ +static int total_files = 0; + +/* (-r) True means don't complain if we can't open a file. */ +static bool ignore_failed_opens = false; + +/* (-S) True means we separate columns with a specified string. + -S option does not affect line truncation nor column alignment. */ +static bool use_col_separator = false; + +/* String used to separate columns if the -S option has been specified. + Default without -S but together with one of the column options + -a|COLUMN|-m is a `space' and with the -J option a `tab'. */ +static char *col_sep_string = (char *) ""; +static int col_sep_length = 0; +static char *column_separator = (char *) " "; +static char *line_separator = (char *) "\t"; + +/* Number of separator characters waiting to be printed as soon as we + know that we have any input remaining to be printed. */ +static int separators_not_printed; + +/* Position we need to pad to, as soon as we know that we have input + remaining to be printed. */ +static int padding_not_printed; + +/* True means we should pad the end of the page. Remains false until we + know we have a page to print. */ +static bool pad_vertically; + +/* (-h) String of characters used in place of the filename in the header. */ +static char *custom_header; + +/* (-D) Date format for the header. */ +static char const *date_format; + +/* Date and file name for the header. */ +static char *date_text; +static char const *file_text; + +/* Output columns available, not counting the date and file name. */ +static int header_width_available; + +static char *clump_buff; + +/* True means we read the line no. lines_per_body in skip_read + called by skip_to_page. That variable controls the coincidence of a + "FF set by hand" and "full_page_printed", see above the definition of + structure COLUMN. */ +static bool last_line = false; + +/* For long options that have no equivalent short option, use a + non-character as a pseudo short option, starting with CHAR_MAX + 1. */ +enum +{ + COLUMNS_OPTION = CHAR_MAX + 1, + PAGES_OPTION +}; + +static char const short_options[] = + "-0123456789D:FJN:S::TW:abcde::fh:i::l:mn::o:rs::tvw:"; + +static struct option const long_options[] = +{ + {"pages", required_argument, NULL, PAGES_OPTION}, + {"columns", required_argument, NULL, COLUMNS_OPTION}, + {"across", no_argument, NULL, 'a'}, + {"show-control-chars", no_argument, NULL, 'c'}, + {"double-space", no_argument, NULL, 'd'}, + {"date-format", required_argument, NULL, 'D'}, + {"expand-tabs", optional_argument, NULL, 'e'}, + {"form-feed", no_argument, NULL, 'f'}, + {"header", required_argument, NULL, 'h'}, + {"output-tabs", optional_argument, NULL, 'i'}, + {"join-lines", no_argument, NULL, 'J'}, + {"length", required_argument, NULL, 'l'}, + {"merge", no_argument, NULL, 'm'}, + {"number-lines", optional_argument, NULL, 'n'}, + {"first-line-number", required_argument, NULL, 'N'}, + {"indent", required_argument, NULL, 'o'}, + {"no-file-warnings", no_argument, NULL, 'r'}, + {"separator", optional_argument, NULL, 's'}, + {"sep-string", optional_argument, NULL, 'S'}, + {"omit-header", no_argument, NULL, 't'}, + {"omit-pagination", no_argument, NULL, 'T'}, + {"show-nonprinting", no_argument, NULL, 'v'}, + {"width", required_argument, NULL, 'w'}, + {"page-width", required_argument, NULL, 'W'}, + {GETOPT_HELP_OPTION_DECL}, + {GETOPT_VERSION_OPTION_DECL}, + {NULL, 0, NULL, 0} +}; + +/* Return the number of columns that have either an open file or + stored lines. */ + +static int +cols_ready_to_print (void) +{ + COLUMN *q; + int i; + int n; + + n = 0; + for (q = column_vector, i = 0; i < columns; ++q, ++i) + if (q->status == OPEN || + q->status == FF_FOUND || /* With -b: To print a header only */ + (storing_columns && q->lines_stored > 0 && q->lines_to_print > 0)) + ++n; + return n; +} + +/* Estimate first_ / last_page_number + using option +FIRST_PAGE:LAST_PAGE */ + +static bool +first_last_page (int oi, char c, char const *pages) +{ + char *p; + uintmax_t first; + uintmax_t last = UINTMAX_MAX; + strtol_error err = xstrtoumax (pages, &p, 10, &first, ""); + if (err != LONGINT_OK && err != LONGINT_INVALID_SUFFIX_CHAR) + xstrtol_fatal (err, oi, c, long_options, pages); + + if (p == pages || !first) + return false; + + if (*p == ':') + { + char const *p1 = p + 1; + err = xstrtoumax (p1, &p, 10, &last, ""); + if (err != LONGINT_OK) + xstrtol_fatal (err, oi, c, long_options, pages); + if (p1 == p || last < first) + return false; + } + + if (*p) + return false; + + first_page_number = first; + last_page_number = last; + return true; +} + +/* Parse column count string S, and if it's valid (1 or larger and + within range of the type of `columns') set the global variables + columns and explicit_columns and return true. + Otherwise, exit with a diagnostic. */ +static void +parse_column_count (char const *s) +{ + long int tmp_long; + if (xstrtol (s, NULL, 10, &tmp_long, "") != LONGINT_OK + || !(1 <= tmp_long && tmp_long <= INT_MAX)) + error (EXIT_FAILURE, 0, + _("invalid number of columns: %s"), quote (s)); + + columns = tmp_long; + explicit_columns = true; +} + +/* Estimate length of col_sep_string with option -S. */ + +static void +separator_string (const char *optarg_S) +{ + col_sep_length = (int) strlen (optarg_S); + col_sep_string = xmalloc (col_sep_length + 1); + strcpy (col_sep_string, optarg_S); +} + +int +main (int argc, char **argv) +{ + int n_files; + bool old_options = false; + bool old_w = false; + bool old_s = false; + char **file_names; + + /* Accumulate the digits of old-style options like -99. */ + char *column_count_string = NULL; + size_t n_digits = 0; + size_t n_alloc = 0; + + initialize_main (&argc, &argv); + set_program_name (argv[0]); + setlocale (LC_ALL, ""); + bindtextdomain (PACKAGE, LOCALEDIR); + textdomain (PACKAGE); + + atexit (close_stdout); + + n_files = 0; + file_names = (argc > 1 + ? xmalloc ((argc - 1) * sizeof (char *)) + : NULL); + + for (;;) + { + int oi = -1; + int c = getopt_long (argc, argv, short_options, long_options, &oi); + if (c == -1) + break; + + if (ISDIGIT (c)) + { + /* Accumulate column-count digits specified via old-style options. */ + if (n_digits + 1 >= n_alloc) + column_count_string + = X2REALLOC (column_count_string, &n_alloc); + column_count_string[n_digits++] = c; + column_count_string[n_digits] = '\0'; + continue; + } + + n_digits = 0; + + switch (c) + { + case 1: /* Non-option argument. */ + /* long option --page dominates old `+FIRST_PAGE ...'. */ + if (! (first_page_number == 0 + && *optarg == '+' && first_last_page (-2, '+', optarg + 1))) + file_names[n_files++] = optarg; + break; + + case PAGES_OPTION: /* --pages=FIRST_PAGE[:LAST_PAGE] */ + { /* dominates old opt +... */ + if (! optarg) + error (EXIT_FAILURE, 0, + _("`--pages=FIRST_PAGE[:LAST_PAGE]' missing argument")); + else if (! first_last_page (oi, 0, optarg)) + error (EXIT_FAILURE, 0, _("invalid page range %s"), + quote (optarg)); + break; + } + + case COLUMNS_OPTION: /* --columns=COLUMN */ + { + parse_column_count (optarg); + + /* If there was a prior column count specified via the + short-named option syntax, e.g., -9, ensure that this + long-name-specified value overrides it. */ + free (column_count_string); + column_count_string = NULL; + n_alloc = 0; + break; + } + + case 'a': + print_across_flag = true; + storing_columns = false; + break; + case 'b': + balance_columns = true; + break; + case 'c': + use_cntrl_prefix = true; + break; + case 'd': + double_space = true; + break; + case 'D': + date_format = optarg; + break; + case 'e': + if (optarg) + getoptarg (optarg, 'e', &input_tab_char, + &chars_per_input_tab); + /* Could check tab width > 0. */ + untabify_input = true; + break; + case 'f': + case 'F': + use_form_feed = true; + break; + case 'h': + custom_header = optarg; + break; + case 'i': + if (optarg) + getoptarg (optarg, 'i', &output_tab_char, + &chars_per_output_tab); + /* Could check tab width > 0. */ + tabify_output = true; + break; + case 'J': + join_lines = true; + break; + case 'l': + { + long int tmp_long; + if (xstrtol (optarg, NULL, 10, &tmp_long, "") != LONGINT_OK + || tmp_long <= 0 || tmp_long > INT_MAX) + { + error (EXIT_FAILURE, 0, + _("`-l PAGE_LENGTH' invalid number of lines: %s"), + quote (optarg)); + } + lines_per_page = tmp_long; + break; + } + case 'm': + parallel_files = true; + storing_columns = false; + break; + case 'n': + numbered_lines = true; + if (optarg) + getoptarg (optarg, 'n', &number_separator, + &chars_per_number); + break; + case 'N': + skip_count = false; + { + long int tmp_long; + if (xstrtol (optarg, NULL, 10, &tmp_long, "") != LONGINT_OK + || tmp_long > INT_MAX) + { + error (EXIT_FAILURE, 0, + _("`-N NUMBER' invalid starting line number: %s"), + quote (optarg)); + } + start_line_num = tmp_long; + break; + } + case 'o': + { + long int tmp_long; + if (xstrtol (optarg, NULL, 10, &tmp_long, "") != LONGINT_OK + || tmp_long < 0 || tmp_long > INT_MAX) + error (EXIT_FAILURE, 0, + _("`-o MARGIN' invalid line offset: %s"), quote (optarg)); + chars_per_margin = tmp_long; + break; + } + case 'r': + ignore_failed_opens = true; + break; + case 's': + old_options = true; + old_s = true; + if (!use_col_separator && optarg) + separator_string (optarg); + break; + case 'S': + old_s = false; + /* Reset an additional input of -s, -S dominates -s */ + col_sep_string = bad_cast (""); + col_sep_length = 0; + use_col_separator = true; + if (optarg) + separator_string (optarg); + break; + case 't': + extremities = false; + keep_FF = true; + break; + case 'T': + extremities = false; + keep_FF = false; + break; + case 'v': + use_esc_sequence = true; + break; + case 'w': + old_options = true; + old_w = true; + { + long int tmp_long; + if (xstrtol (optarg, NULL, 10, &tmp_long, "") != LONGINT_OK + || tmp_long <= 0 || tmp_long > INT_MAX) + error (EXIT_FAILURE, 0, + _("`-w PAGE_WIDTH' invalid number of characters: %s"), + quote (optarg)); + if (!truncate_lines) + chars_per_line = tmp_long; + break; + } + case 'W': + old_w = false; /* dominates -w */ + truncate_lines = true; + { + long int tmp_long; + if (xstrtol (optarg, NULL, 10, &tmp_long, "") != LONGINT_OK + || tmp_long <= 0 || tmp_long > INT_MAX) + error (EXIT_FAILURE, 0, + _("`-W PAGE_WIDTH' invalid number of characters: %s"), + quote (optarg)); + chars_per_line = tmp_long; + break; + } + case_GETOPT_HELP_CHAR; + case_GETOPT_VERSION_CHAR (PROGRAM_NAME, AUTHORS); + default: + usage (EXIT_FAILURE); + break; + } + } + + if (column_count_string) + { + parse_column_count (column_count_string); + free (column_count_string); + } + + if (! date_format) + date_format = (getenv ("POSIXLY_CORRECT") && !hard_locale (LC_TIME) + ? "%b %e %H:%M %Y" + : "%Y-%m-%d %H:%M"); + + /* Now we can set a reasonable initial value: */ + if (first_page_number == 0) + first_page_number = 1; + + if (parallel_files & explicit_columns) + error (EXIT_FAILURE, 0, + _("cannot specify number of columns when printing in parallel")); + + if (parallel_files & print_across_flag) + error (EXIT_FAILURE, 0, + _("cannot specify both printing across and printing in parallel")); + +/* Translate some old short options to new/long options. + To meet downward compatibility with other UNIX pr utilities + and some POSIX specifications. */ + + if (old_options) + { + if (old_w) + { + if (parallel_files | explicit_columns) + { + /* activate -W */ + truncate_lines = true; + if (old_s) + /* adapt HP-UX and SunOS: -s = no separator; + activate -S */ + use_col_separator = true; + } + else + /* old -w sets width with columns only + activate -J */ + join_lines = true; + } + else if (!use_col_separator) + { + /* No -S option read */ + if (old_s & (parallel_files | explicit_columns)) + { + if (!truncate_lines) + { + /* old -s (without -w and -W) annuls column alignment, + uses fields, activate -J */ + join_lines = true; + if (col_sep_length > 0) + /* activate -S */ + use_col_separator = true; + } + else + /* with -W */ + /* adapt HP-UX and SunOS: -s = no separator; + activate -S */ + use_col_separator = true; + } + } + } + + for (; optind < argc; optind++) + { + file_names[n_files++] = argv[optind]; + } + + if (n_files == 0) + { + /* No file arguments specified; read from standard input. */ + print_files (0, NULL); + } + else + { + if (parallel_files) + print_files (n_files, file_names); + else + { + int i; + for (i = 0; i < n_files; i++) + print_files (1, &file_names[i]); + } + } + + cleanup (); + + if (have_read_stdin && fclose (stdin) == EOF) + error (EXIT_FAILURE, errno, _("standard input")); + if (failed_opens) + exit (EXIT_FAILURE); + exit (EXIT_SUCCESS); +} + +/* Parse options of the form -scNNN. + + Example: -nck, where 'n' is the option, c is the optional number + separator, and k is the optional width of the field used when printing + a number. */ + +static void +getoptarg (char *arg, char switch_char, char *character, int *number) +{ + if (!ISDIGIT (*arg)) + *character = *arg++; + if (*arg) + { + long int tmp_long; + if (xstrtol (arg, NULL, 10, &tmp_long, "") != LONGINT_OK + || tmp_long <= 0 || tmp_long > INT_MAX) + { + error (0, 0, + _("`-%c' extra characters or invalid number in the argument: %s"), + switch_char, quote (arg)); + usage (EXIT_FAILURE); + } + *number = tmp_long; + } +} + +/* Set parameters related to formatting. */ + +static void +init_parameters (int number_of_files) +{ + int chars_used_by_number = 0; + + lines_per_body = lines_per_page - lines_per_header - lines_per_footer; + if (lines_per_body <= 0) + { + extremities = false; + keep_FF = true; + } + if (extremities == false) + lines_per_body = lines_per_page; + + if (double_space) + lines_per_body = lines_per_body / 2; + + /* If input is stdin, cannot print parallel files. BSD dumps core + on this. */ + if (number_of_files == 0) + parallel_files = false; + + if (parallel_files) + columns = number_of_files; + + /* One file, multi columns down: -b option is set to get a consistent + formulation with "FF set by hand" in input files. */ + if (storing_columns) + balance_columns = true; + + /* Tabification is assumed for multiple columns. */ + if (columns > 1) + { + if (!use_col_separator) + { + /* Use default separator */ + if (join_lines) + col_sep_string = line_separator; + else + col_sep_string = column_separator; + + col_sep_length = 1; + use_col_separator = true; + } + /* It's rather pointless to define a TAB separator with column + alignment */ + else if (!join_lines && *col_sep_string == '\t') + col_sep_string = column_separator; + + truncate_lines = true; + tabify_output = true; + } + else + storing_columns = false; + + /* -J dominates -w in any case */ + if (join_lines) + truncate_lines = false; + + if (numbered_lines) + { + int tmp_i; + int chars_per_default_tab = 8; + + line_count = start_line_num; + + /* To allow input tab-expansion (-e sensitive) use: + if (number_separator == input_tab_char) + number_width = chars_per_number + + TAB_WIDTH (chars_per_input_tab, chars_per_number); */ + + /* Estimate chars_per_text without any margin and keep it constant. */ + if (number_separator == '\t') + number_width = chars_per_number + + TAB_WIDTH (chars_per_default_tab, chars_per_number); + else + number_width = chars_per_number + 1; + + /* The number is part of the column width unless we are + printing files in parallel. */ + if (parallel_files) + chars_used_by_number = number_width; + + /* We use power_10 to cut off the higher-order digits of the + line_number in function add_line_number */ + tmp_i = chars_per_number; + for (power_10 = 1; tmp_i > 0; --tmp_i) + power_10 = 10 * power_10; + } + + chars_per_column = (chars_per_line - chars_used_by_number - + (columns - 1) * col_sep_length) / columns; + + if (chars_per_column < 1) + error (EXIT_FAILURE, 0, _("page width too narrow")); + + if (numbered_lines) + { + free (number_buff); + number_buff = xmalloc (2 * chars_per_number); + } + + /* Pick the maximum between the tab width and the width of an + escape sequence. + The width of an escape sequence (4) isn't the lower limit any longer. + We've to use 8 as the lower limit, if we use chars_per_default_tab = 8 + to expand a tab which is not an input_tab-char. */ + free (clump_buff); + clump_buff = xmalloc (MAX (8, chars_per_input_tab)); +} + +/* Open the necessary files, + maintaining a COLUMN structure for each column. + + With multiple files, each column p has a different p->fp. + With single files, each column p has the same p->fp. + Return false if (number_of_files > 0) and no files can be opened, + true otherwise. + + With each column/file p, p->full_page_printed is initialized, + see also open_file. */ + +static bool +init_fps (int number_of_files, char **av) +{ + int i, files_left; + COLUMN *p; + FILE *firstfp; + char const *firstname; + + total_files = 0; + + free (column_vector); + column_vector = xnmalloc (columns, sizeof (COLUMN)); + + if (parallel_files) + { + files_left = number_of_files; + for (p = column_vector; files_left--; ++p, ++av) + { + if (! open_file (*av, p)) + { + --p; + --columns; + } + } + if (columns == 0) + return false; + init_header ("", -1); + } + else + { + p = column_vector; + if (number_of_files > 0) + { + if (! open_file (*av, p)) + return false; + init_header (*av, fileno (p->fp)); + p->lines_stored = 0; + } + else + { + p->name = _("standard input"); + p->fp = stdin; + have_read_stdin = true; + p->status = OPEN; + p->full_page_printed = false; + ++total_files; + init_header ("", -1); + p->lines_stored = 0; + } + + firstname = p->name; + firstfp = p->fp; + for (i = columns - 1, ++p; i; --i, ++p) + { + p->name = firstname; + p->fp = firstfp; + p->status = OPEN; + p->full_page_printed = false; + p->lines_stored = 0; + } + } + files_ready_to_read = total_files; + return true; +} + +/* Determine print_func and char_func, the functions + used by each column for printing and/or storing. + + Determine the horizontal position desired when we begin + printing a column (p->start_position). */ + +static void +init_funcs (void) +{ + int i, h, h_next; + COLUMN *p; + + h = chars_per_margin; + + if (!truncate_lines) + h_next = ANYWHERE; + else + { + /* When numbering lines of parallel files, we enlarge the + first column to accomodate the number. Looks better than + the Sys V approach. */ + if (parallel_files & numbered_lines) + h_next = h + chars_per_column + number_width; + else + h_next = h + chars_per_column; + } + + /* Enlarge p->start_position of first column to use the same form of + padding_not_printed with all columns. */ + h = h + col_sep_length; + + /* This loop takes care of all but the rightmost column. */ + + for (p = column_vector, i = 1; i < columns; ++p, ++i) + { + if (storing_columns) /* One file, multi columns down. */ + { + p->char_func = store_char; + p->print_func = print_stored; + } + else + /* One file, multi columns across; or parallel files. */ + { + p->char_func = print_char; + p->print_func = read_line; + } + + /* Number only the first column when printing files in + parallel. */ + p->numbered = numbered_lines && (!parallel_files || i == 1); + p->start_position = h; + + /* If we don't truncate lines, all start_positions are + ANYWHERE, except the first column's start_position when + using a margin. */ + + if (!truncate_lines) + { + h = ANYWHERE; + h_next = ANYWHERE; + } + else + { + h = h_next + col_sep_length; + h_next = h + chars_per_column; + } + } + + /* The rightmost column. + + Doesn't need to be stored unless we intend to balance + columns on the last page. */ + if (storing_columns & balance_columns) + { + p->char_func = store_char; + p->print_func = print_stored; + } + else + { + p->char_func = print_char; + p->print_func = read_line; + } + + p->numbered = numbered_lines && (!parallel_files || i == 1); + p->start_position = h; +} + +/* Open a file. Return true if successful. + + With each file p, p->full_page_printed is initialized, + see also init_fps. */ + +static bool +open_file (char *name, COLUMN *p) +{ + if (STREQ (name, "-")) + { + p->name = _("standard input"); + p->fp = stdin; + have_read_stdin = true; + } + else + { + p->name = name; + p->fp = fopen (name, "r"); + } + if (p->fp == NULL) + { + failed_opens = true; + if (!ignore_failed_opens) + error (0, errno, "%s", name); + return false; + } + p->status = OPEN; + p->full_page_printed = false; + ++total_files; + return true; +} + +/* Close the file in P. + + If we aren't dealing with multiple files in parallel, we change + the status of all columns in the column list to reflect the close. */ + +static void +close_file (COLUMN *p) +{ + COLUMN *q; + int i; + + if (p->status == CLOSED) + return; + if (ferror (p->fp)) + error (EXIT_FAILURE, errno, "%s", p->name); + if (fileno (p->fp) != STDIN_FILENO && fclose (p->fp) != 0) + error (EXIT_FAILURE, errno, "%s", p->name); + + if (!parallel_files) + { + for (q = column_vector, i = columns; i; ++q, --i) + { + q->status = CLOSED; + if (q->lines_stored == 0) + { + q->lines_to_print = 0; + } + } + } + else + { + p->status = CLOSED; + p->lines_to_print = 0; + } + + --files_ready_to_read; +} + +/* Put a file on hold until we start a new page, + since we've hit a form feed. + + If we aren't dealing with parallel files, we must change the + status of all columns in the column list. */ + +static void +hold_file (COLUMN *p) +{ + COLUMN *q; + int i; + + if (!parallel_files) + for (q = column_vector, i = columns; i; ++q, --i) + { + if (storing_columns) + q->status = FF_FOUND; + else + q->status = ON_HOLD; + } + else + p->status = ON_HOLD; + + p->lines_to_print = 0; + --files_ready_to_read; +} + +/* Undo hold_file -- go through the column list and change any + ON_HOLD columns to OPEN. Used at the end of each page. */ + +static void +reset_status (void) +{ + int i = columns; + COLUMN *p; + + for (p = column_vector; i; --i, ++p) + if (p->status == ON_HOLD) + { + p->status = OPEN; + files_ready_to_read++; + } + + if (storing_columns) + { + if (column_vector->status == CLOSED) + /* We use the info to output an error message in skip_to_page. */ + files_ready_to_read = 0; + else + files_ready_to_read = 1; + } +} + +/* Print a single file, or multiple files in parallel. + + Set up the list of columns, opening the necessary files. + Allocate space for storing columns, if necessary. + Skip to first_page_number, if user has asked to skip leading pages. + Determine which functions are appropriate to store/print lines + in each column. + Print the file(s). */ + +static void +print_files (int number_of_files, char **av) +{ + init_parameters (number_of_files); + if (! init_fps (number_of_files, av)) + return; + if (storing_columns) + init_store_cols (); + + if (first_page_number > 1) + { + if (!skip_to_page (first_page_number)) + return; + else + page_number = first_page_number; + } + else + page_number = 1; + + init_funcs (); + + line_number = line_count; + while (print_page ()) + ; +} + +/* Initialize header information. + If DESC is non-negative, it is a file descriptor open to + FILENAME for reading. */ + +static void +init_header (char const *filename, int desc) +{ + char *buf = NULL; + struct stat st; + struct timespec t; + int ns; + struct tm *tm; + + /* If parallel files or standard input, use current date. */ + if (STREQ (filename, "-")) + desc = -1; + if (0 <= desc && fstat (desc, &st) == 0) + t = get_stat_mtime (&st); + else + { + static struct timespec timespec; + if (! timespec.tv_sec) + gettime (×pec); + t = timespec; + } + + ns = t.tv_nsec; + tm = localtime (&t.tv_sec); + if (tm == NULL) + { + buf = xmalloc (INT_BUFSIZE_BOUND (long int) + + MAX (10, INT_BUFSIZE_BOUND (int))); + sprintf (buf, "%ld.%09d", (long int) t.tv_sec, ns); + } + else + { + size_t bufsize = nstrftime (NULL, SIZE_MAX, date_format, tm, 0, ns) + 1; + buf = xmalloc (bufsize); + nstrftime (buf, bufsize, date_format, tm, 0, ns); + } + + free (date_text); + date_text = buf; + file_text = custom_header ? custom_header : desc < 0 ? "" : filename; + header_width_available = (chars_per_line + - mbswidth (date_text, 0) + - mbswidth (file_text, 0)); +} + +/* Set things up for printing a page + + Scan through the columns ... + Determine which are ready to print + (i.e., which have lines stored or open files) + Set p->lines_to_print appropriately + (to p->lines_stored if we're storing, or lines_per_body + if we're reading straight from the file) + Keep track of this total so we know when to stop printing */ + +static void +init_page (void) +{ + int j; + COLUMN *p; + + if (storing_columns) + { + store_columns (); + for (j = columns - 1, p = column_vector; j; --j, ++p) + { + p->lines_to_print = p->lines_stored; + } + + /* Last column. */ + if (balance_columns) + { + p->lines_to_print = p->lines_stored; + } + /* Since we're not balancing columns, we don't need to store + the rightmost column. Read it straight from the file. */ + else + { + if (p->status == OPEN) + { + p->lines_to_print = lines_per_body; + } + else + p->lines_to_print = 0; + } + } + else + for (j = columns, p = column_vector; j; --j, ++p) + if (p->status == OPEN) + { + p->lines_to_print = lines_per_body; + } + else + p->lines_to_print = 0; +} + +/* Align empty columns and print separators. + Empty columns will be formed by files with status ON_HOLD or CLOSED + when printing multiple files in parallel. */ + +static void +align_column (COLUMN *p) +{ + padding_not_printed = p->start_position; + if (padding_not_printed - col_sep_length > 0) + { + pad_across_to (padding_not_printed - col_sep_length); + padding_not_printed = ANYWHERE; + } + + if (use_col_separator) + print_sep_string (); + + if (p->numbered) + add_line_number (p); +} + +/* Print one page. + + As long as there are lines left on the page and columns ready to print, + Scan across the column list + if the column has stored lines or the file is open + pad to the appropriate spot + print the column + pad the remainder of the page with \n or \f as requested + reset the status of all files -- any files which where on hold because + of formfeeds are now put back into the lineup. */ + +static bool +print_page (void) +{ + int j; + int lines_left_on_page; + COLUMN *p; + + /* Used as an accumulator (with | operator) of successive values of + pad_vertically. The trick is to set pad_vertically + to false before each run through the inner loop, then after that + loop, it tells us whether a line was actually printed (whether a + newline needs to be output -- or two for double spacing). But those + values have to be accumulated (in pv) so we can invoke pad_down + properly after the outer loop completes. */ + bool pv; + + init_page (); + + if (cols_ready_to_print () == 0) + return false; + + if (extremities) + print_a_header = true; + + /* Don't pad unless we know a page was printed. */ + pad_vertically = false; + pv = false; + + lines_left_on_page = lines_per_body; + if (double_space) + lines_left_on_page *= 2; + + while (lines_left_on_page > 0 && cols_ready_to_print () > 0) + { + output_position = 0; + spaces_not_printed = 0; + separators_not_printed = 0; + pad_vertically = false; + align_empty_cols = false; + empty_line = true; + + for (j = 1, p = column_vector; j <= columns; ++j, ++p) + { + input_position = 0; + if (p->lines_to_print > 0 || p->status == FF_FOUND) + { + FF_only = false; + padding_not_printed = p->start_position; + if (!(p->print_func) (p)) + read_rest_of_line (p); + pv |= pad_vertically; + + --p->lines_to_print; + if (p->lines_to_print <= 0) + { + if (cols_ready_to_print () <= 0) + break; + } + + /* File p changed its status to ON_HOLD or CLOSED */ + if (parallel_files && p->status != OPEN) + { + if (empty_line) + align_empty_cols = true; + else if (p->status == CLOSED || + (p->status == ON_HOLD && FF_only)) + align_column (p); + } + } + else if (parallel_files) + { + /* File status ON_HOLD or CLOSED */ + if (empty_line) + align_empty_cols = true; + else + align_column (p); + } + + /* We need it also with an empty column */ + if (use_col_separator) + ++separators_not_printed; + } + + if (pad_vertically) + { + putchar ('\n'); + --lines_left_on_page; + } + + if (cols_ready_to_print () <= 0 && !extremities) + break; + + if (double_space & pv) + { + putchar ('\n'); + --lines_left_on_page; + } + } + + if (lines_left_on_page == 0) + for (j = 1, p = column_vector; j <= columns; ++j, ++p) + if (p->status == OPEN) + p->full_page_printed = true; + + pad_vertically = pv; + + if (pad_vertically & extremities) + pad_down (lines_left_on_page + lines_per_footer); + else if (keep_FF & print_a_FF) + { + putchar ('\f'); + print_a_FF = false; + } + + if (last_page_number < page_number) + return false; /* Stop printing with LAST_PAGE */ + + reset_status (); /* Change ON_HOLD to OPEN. */ + + return true; /* More pages to go. */ +} + +/* Allocate space for storing columns. + + This is necessary when printing multiple columns from a single file. + Lines are stored consecutively in buff, separated by '\0'. + + The following doesn't apply any longer - any tuning possible? + (We can't use a fixed offset since with the '-s' flag lines aren't + truncated.) + + We maintain a list (line_vector) of pointers to the beginnings + of lines in buff. We allocate one more than the number of lines + because the last entry tells us the index of the last character, + which we need to know in order to print the last line in buff. */ + +static void +init_store_cols (void) +{ + int total_lines = lines_per_body * columns; + int chars_if_truncate = total_lines * (chars_per_column + 1); + + free (line_vector); + /* FIXME: here's where it was allocated. */ + line_vector = xmalloc ((total_lines + 1) * sizeof (int *)); + + free (end_vector); + end_vector = xmalloc (total_lines * sizeof (int *)); + + free (buff); + buff_allocated = (use_col_separator + ? 2 * chars_if_truncate + : chars_if_truncate); /* Tune this. */ + buff = xmalloc (buff_allocated); +} + +/* Store all but the rightmost column. + (Used when printing a single file in multiple downward columns) + + For each column + set p->current_line to be the index in line_vector of the + first line in the column + For each line in the column + store the line in buff + add to line_vector the index of the line's first char + buff_start is the index in buff of the first character in the + current line. */ + +static void +store_columns (void) +{ + int i, j; + unsigned int line = 0; + unsigned int buff_start; + int last_col; /* The rightmost column which will be saved in buff */ + COLUMN *p; + + buff_current = 0; + buff_start = 0; + + if (balance_columns) + last_col = columns; + else + last_col = columns - 1; + + for (i = 1, p = column_vector; i <= last_col; ++i, ++p) + p->lines_stored = 0; + + for (i = 1, p = column_vector; i <= last_col && files_ready_to_read; + ++i, ++p) + { + p->current_line = line; + for (j = lines_per_body; j && files_ready_to_read; --j) + + if (p->status == OPEN) /* Redundant. Clean up. */ + { + input_position = 0; + + if (!read_line (p)) + read_rest_of_line (p); + + if (p->status == OPEN + || buff_start != buff_current) + { + ++p->lines_stored; + line_vector[line] = buff_start; + end_vector[line++] = input_position; + buff_start = buff_current; + } + } + } + + /* Keep track of the location of the last char in buff. */ + line_vector[line] = buff_start; + + if (balance_columns) + balance (line); +} + +static void +balance (int total_stored) +{ + COLUMN *p; + int i, lines; + int first_line = 0; + + for (i = 1, p = column_vector; i <= columns; ++i, ++p) + { + lines = total_stored / columns; + if (i <= total_stored % columns) + ++lines; + + p->lines_stored = lines; + p->current_line = first_line; + + first_line += lines; + } +} + +/* Store a character in the buffer. */ + +static void +store_char (char c) +{ + if (buff_current >= buff_allocated) + { + /* May be too generous. */ + buff = X2REALLOC (buff, &buff_allocated); + } + buff[buff_current++] = c; +} + +static void +add_line_number (COLUMN *p) +{ + int i; + char *s; + int left_cut; + + /* Cutting off the higher-order digits is more informative than + lower-order cut off*/ + if (line_number < power_10) + sprintf (number_buff, "%*d", chars_per_number, line_number); + else + { + left_cut = line_number % power_10; + sprintf (number_buff, "%0*d", chars_per_number, left_cut); + } + line_number++; + s = number_buff; + for (i = chars_per_number; i > 0; i--) + (p->char_func) (*s++); + + if (columns > 1) + { + /* Tabification is assumed for multiple columns, also for n-separators, + but `default n-separator = TAB' hasn't been given priority over + equal column_width also specified by POSIX. */ + if (number_separator == '\t') + { + i = number_width - chars_per_number; + while (i-- > 0) + (p->char_func) (' '); + } + else + (p->char_func) (number_separator); + } + else + /* To comply with POSIX, we avoid any expansion of default TAB + separator with a single column output. No column_width requirement + has to be considered. */ + { + (p->char_func) (number_separator); + if (number_separator == '\t') + output_position = POS_AFTER_TAB (chars_per_output_tab, + output_position); + } + + if (truncate_lines & !parallel_files) + input_position += number_width; +} + +/* Print (or store) padding until the current horizontal position + is position. */ + +static void +pad_across_to (int position) +{ + int h = output_position; + + if (tabify_output) + spaces_not_printed = position - output_position; + else + { + while (++h <= position) + putchar (' '); + output_position = position; + } +} + +/* Pad to the bottom of the page. + + If the user has requested a formfeed, use one. + Otherwise, use newlines. */ + +static void +pad_down (int lines) +{ + int i; + + if (use_form_feed) + putchar ('\f'); + else + for (i = lines; i; --i) + putchar ('\n'); +} + +/* Read the rest of the line. + + Read from the current column's file until an end of line is + hit. Used when we've truncated a line and we no longer need + to print or store its characters. */ + +static void +read_rest_of_line (COLUMN *p) +{ + int c; + FILE *f = p->fp; + + while ((c = getc (f)) != '\n') + { + if (c == '\f') + { + if ((c = getc (f)) != '\n') + ungetc (c, f); + if (keep_FF) + print_a_FF = true; + hold_file (p); + break; + } + else if (c == EOF) + { + close_file (p); + break; + } + } +} + +/* Read a line with skip_to_page. + + Read from the current column's file until an end of line is + hit. Used when we read full lines to skip pages. + With skip_to_page we have to check for FF-coincidence which is done + in function read_line otherwise. + Count lines of skipped pages to find the line number of 1st page + printed relative to 1st line of input file (start_line_num). */ + +static void +skip_read (COLUMN *p, int column_number) +{ + int c; + FILE *f = p->fp; + int i; + bool single_ff = false; + COLUMN *q; + + /* Read 1st character in a line or any character succeeding a FF */ + if ((c = getc (f)) == '\f' && p->full_page_printed) + /* A FF-coincidence with a previous full_page_printed. + To avoid an additional empty page, eliminate the FF */ + if ((c = getc (f)) == '\n') + c = getc (f); + + p->full_page_printed = false; + + /* 1st character a FF means a single FF without any printable + characters. Don't count it as a line with -n option. */ + if (c == '\f') + single_ff = true; + + /* Preparing for a FF-coincidence: Maybe we finish that page + without a FF found */ + if (last_line) + p->full_page_printed = true; + + while (c != '\n') + { + if (c == '\f') + { + /* No FF-coincidence possible, + no catching up of a FF-coincidence with next page */ + if (last_line) + { + if (!parallel_files) + for (q = column_vector, i = columns; i; ++q, --i) + q->full_page_printed = false; + else + p->full_page_printed = false; + } + + if ((c = getc (f)) != '\n') + ungetc (c, f); + hold_file (p); + break; + } + else if (c == EOF) + { + close_file (p); + break; + } + c = getc (f); + } + + if (skip_count) + if ((!parallel_files || column_number == 1) && !single_ff) + ++line_count; +} + +/* If we're tabifying output, + + When print_char encounters white space it keeps track + of our desired horizontal position and delays printing + until this function is called. */ + +static void +print_white_space (void) +{ + int h_new; + int h_old = output_position; + int goal = h_old + spaces_not_printed; + + while (goal - h_old > 1 + && (h_new = POS_AFTER_TAB (chars_per_output_tab, h_old)) <= goal) + { + putchar (output_tab_char); + h_old = h_new; + } + while (++h_old <= goal) + putchar (' '); + + output_position = goal; + spaces_not_printed = 0; +} + +/* Print column separators. + + We keep a count until we know that we'll be printing a line, + then print_sep_string() is called. */ + +static void +print_sep_string (void) +{ + char *s; + int l = col_sep_length; + + s = col_sep_string; + + if (separators_not_printed <= 0) + { + /* We'll be starting a line with chars_per_margin, anything else? */ + if (spaces_not_printed > 0) + print_white_space (); + } + else + { + for (; separators_not_printed > 0; --separators_not_printed) + { + while (l-- > 0) + { + /* 3 types of sep_strings: spaces only, spaces and chars, + chars only */ + if (*s == ' ') + { + /* We're tabifying output; consecutive spaces in + sep_string may have to be converted to tabs */ + s++; + ++spaces_not_printed; + } + else + { + if (spaces_not_printed > 0) + print_white_space (); + putchar (*s++); + ++output_position; + } + } + /* sep_string ends with some spaces */ + if (spaces_not_printed > 0) + print_white_space (); + } + } +} + +/* Print (or store, depending on p->char_func) a clump of N + characters. */ + +static void +print_clump (COLUMN *p, int n, char *clump) +{ + while (n--) + (p->char_func) (*clump++); +} + +/* Print a character. + + Update the following comment: process-char hasn't been used any + longer. + If we're tabifying, all tabs have been converted to spaces by + process_char(). Keep a count of consecutive spaces, and when + a nonspace is encountered, call print_white_space() to print the + required number of tabs and spaces. */ + +static void +print_char (char c) +{ + if (tabify_output) + { + if (c == ' ') + { + ++spaces_not_printed; + return; + } + else if (spaces_not_printed > 0) + print_white_space (); + + /* Nonprintables are assumed to have width 0, except '\b'. */ + if (! isprint (to_uchar (c))) + { + if (c == '\b') + --output_position; + } + else + ++output_position; + } + putchar (c); +} + +/* Skip to page PAGE before printing. + PAGE may be larger than total number of pages. */ + +static bool +skip_to_page (uintmax_t page) +{ + uintmax_t n; + int i; + int j; + COLUMN *p; + + for (n = 1; n < page; ++n) + { + for (i = 1; i < lines_per_body; ++i) + { + for (j = 1, p = column_vector; j <= columns; ++j, ++p) + if (p->status == OPEN) + skip_read (p, j); + } + last_line = true; + for (j = 1, p = column_vector; j <= columns; ++j, ++p) + if (p->status == OPEN) + skip_read (p, j); + + if (storing_columns) /* change FF_FOUND to ON_HOLD */ + for (j = 1, p = column_vector; j <= columns; ++j, ++p) + if (p->status != CLOSED) + p->status = ON_HOLD; + + reset_status (); + last_line = false; + + if (files_ready_to_read < 1) + { + /* It's very helpful, normally the total number of pages is + not known in advance. */ + error (0, 0, + _("starting page number %"PRIuMAX + " exceeds page count %"PRIuMAX), + page, n); + break; + } + } + return files_ready_to_read > 0; +} + +/* Print a header. + + Formfeeds are assumed to use up two lines at the beginning of + the page. */ + +static void +print_header (void) +{ + char page_text[256 + INT_STRLEN_BOUND (page_number)]; + int available_width; + int lhs_spaces; + int rhs_spaces; + + output_position = 0; + pad_across_to (chars_per_margin); + print_white_space (); + + if (page_number == 0) + error (EXIT_FAILURE, 0, _("page number overflow")); + + /* The translator must ensure that formatting the translation of + "Page %"PRIuMAX does not generate more than (sizeof page_text - 1) + bytes. */ + sprintf (page_text, _("Page %"PRIuMAX), page_number++); + available_width = header_width_available - mbswidth (page_text, 0); + available_width = MAX (0, available_width); + lhs_spaces = available_width >> 1; + rhs_spaces = available_width - lhs_spaces; + + printf ("\n\n%*.*s%s%*.*s%s%*.*s%s\n\n\n", + chars_per_margin, chars_per_margin, " ", + date_text, lhs_spaces, lhs_spaces, " ", + file_text, rhs_spaces, rhs_spaces, " ", page_text); + + print_a_header = false; + output_position = 0; +} + +/* Print (or store, if p->char_func is store_char()) a line. + + Read a character to determine whether we have a line or not. + (We may hit EOF, \n, or \f) + + Once we know we have a line, + set pad_vertically = true, meaning it's safe + to pad down at the end of the page, since we do have a page. + print a header if needed. + pad across to padding_not_printed if needed. + print any separators which need to be printed. + print a line number if it needs to be printed. + + Print the clump which corresponds to the first character. + + Enter a loop and keep printing until an end of line condition + exists, or until we exceed chars_per_column. + + Return false if we exceed chars_per_column before reading + an end of line character, true otherwise. */ + +static bool +read_line (COLUMN *p) +{ + int c; + int chars IF_LINT (= 0); + int last_input_position; + int j, k; + COLUMN *q; + + /* read 1st character in each line or any character succeeding a FF: */ + c = getc (p->fp); + + last_input_position = input_position; + + if (c == '\f' && p->full_page_printed) + if ((c = getc (p->fp)) == '\n') + c = getc (p->fp); + p->full_page_printed = false; + + switch (c) + { + case '\f': + if ((c = getc (p->fp)) != '\n') + ungetc (c, p->fp); + FF_only = true; + if (print_a_header & !storing_columns) + { + pad_vertically = true; + print_header (); + } + else if (keep_FF) + print_a_FF = true; + hold_file (p); + return true; + case EOF: + close_file (p); + return true; + case '\n': + break; + default: + chars = char_to_clump (c); + } + + if (truncate_lines && input_position > chars_per_column) + { + input_position = last_input_position; + return false; + } + + if (p->char_func != store_char) + { + pad_vertically = true; + + if (print_a_header & !storing_columns) + print_header (); + + if (parallel_files & align_empty_cols) + { + /* We have to align empty columns at the beginning of a line. */ + k = separators_not_printed; + separators_not_printed = 0; + for (j = 1, q = column_vector; j <= k; ++j, ++q) + { + align_column (q); + separators_not_printed += 1; + } + padding_not_printed = p->start_position; + if (truncate_lines) + spaces_not_printed = chars_per_column; + else + spaces_not_printed = 0; + align_empty_cols = false; + } + + if (padding_not_printed - col_sep_length > 0) + { + pad_across_to (padding_not_printed - col_sep_length); + padding_not_printed = ANYWHERE; + } + + if (use_col_separator) + print_sep_string (); + } + + if (p->numbered) + add_line_number (p); + + empty_line = false; + if (c == '\n') + return true; + + print_clump (p, chars, clump_buff); + + for (;;) + { + c = getc (p->fp); + + switch (c) + { + case '\n': + return true; + case '\f': + if ((c = getc (p->fp)) != '\n') + ungetc (c, p->fp); + if (keep_FF) + print_a_FF = true; + hold_file (p); + return true; + case EOF: + close_file (p); + return true; + } + + last_input_position = input_position; + chars = char_to_clump (c); + if (truncate_lines && input_position > chars_per_column) + { + input_position = last_input_position; + return false; + } + + print_clump (p, chars, clump_buff); + } +} + +/* Print a line from buff. + + If this function has been called, we know we have "something to + print". But it remains to be seen whether we have a real text page + or an empty page (a single form feed) with/without a header only. + Therefore first we set pad_vertically to true and print a header + if necessary. + If FF_FOUND and we are using -t|-T option we omit any newline by + setting pad_vertically to false (see print_page). + Otherwise we pad across if necessary, print separators if necessary + and text of COLUMN *p. + + Return true, meaning there is no need to call read_rest_of_line. */ + +static bool +print_stored (COLUMN *p) +{ + COLUMN *q; + int i; + + int line = p->current_line++; + char *first = &buff[line_vector[line]]; + /* FIXME + UMR: Uninitialized memory read: + * This is occurring while in: + print_stored [pr.c:2239] + * Reading 4 bytes from 0x5148c in the heap. + * Address 0x5148c is 4 bytes into a malloc'd block at 0x51488 of 676 bytes + * This block was allocated from: + malloc [rtlib.o] + xmalloc [xmalloc.c:94] + init_store_cols [pr.c:1648] + */ + char *last = &buff[line_vector[line + 1]]; + + pad_vertically = true; + + if (print_a_header) + print_header (); + + if (p->status == FF_FOUND) + { + for (i = 1, q = column_vector; i <= columns; ++i, ++q) + q->status = ON_HOLD; + if (column_vector->lines_to_print <= 0) + { + if (!extremities) + pad_vertically = false; + return true; /* print a header only */ + } + } + + if (padding_not_printed - col_sep_length > 0) + { + pad_across_to (padding_not_printed - col_sep_length); + padding_not_printed = ANYWHERE; + } + + if (use_col_separator) + print_sep_string (); + + while (first != last) + print_char (*first++); + + if (spaces_not_printed == 0) + { + output_position = p->start_position + end_vector[line]; + if (p->start_position - col_sep_length == chars_per_margin) + output_position -= col_sep_length; + } + + return true; +} + +/* Convert a character to the proper format and return the number of + characters in the resulting clump. Increment input_position by + the width of the clump. + + Tabs are converted to clumps of spaces. + Nonprintable characters may be converted to clumps of escape + sequences or control prefixes. + + Note: the width of a clump is not necessarily equal to the number of + characters in clump_buff. (e.g, the width of '\b' is -1, while the + number of characters is 1.) */ + +static int +char_to_clump (char c) +{ + unsigned char uc = c; + char *s = clump_buff; + int i; + char esc_buff[4]; + int width; + int chars; + int chars_per_c = 8; + + if (c == input_tab_char) + chars_per_c = chars_per_input_tab; + + if (c == input_tab_char || c == '\t') + { + width = TAB_WIDTH (chars_per_c, input_position); + + if (untabify_input) + { + for (i = width; i; --i) + *s++ = ' '; + chars = width; + } + else + { + *s = c; + chars = 1; + } + + } + else if (! isprint (uc)) + { + if (use_esc_sequence) + { + width = 4; + chars = 4; + *s++ = '\\'; + sprintf (esc_buff, "%03o", uc); + for (i = 0; i <= 2; ++i) + *s++ = esc_buff[i]; + } + else if (use_cntrl_prefix) + { + if (uc < 0200) + { + width = 2; + chars = 2; + *s++ = '^'; + *s++ = c ^ 0100; + } + else + { + width = 4; + chars = 4; + *s++ = '\\'; + sprintf (esc_buff, "%03o", uc); + for (i = 0; i <= 2; ++i) + *s++ = esc_buff[i]; + } + } + else if (c == '\b') + { + width = -1; + chars = 1; + *s = c; + } + else + { + width = 0; + chars = 1; + *s = c; + } + } + else + { + width = 1; + chars = 1; + *s = c; + } + + /* Too many backspaces must put us in position 0 -- never negative. */ + if (width < 0 && input_position == 0) + { + chars = 0; + input_position = 0; + } + else if (width < 0 && input_position <= -width) + input_position = 0; + else + input_position += width; + + return chars; +} + +/* We've just printed some files and need to clean up things before + looking for more options and printing the next batch of files. + + Free everything we've xmalloc'ed, except `header'. */ + +static void +cleanup (void) +{ + free (number_buff); + free (clump_buff); + free (column_vector); + free (line_vector); + free (end_vector); + free (buff); +} + +/* Complain, print a usage message, and die. */ + +void +usage (int status) +{ + if (status != EXIT_SUCCESS) + fprintf (stderr, _("Try `%s --help' for more information.\n"), + program_name); + else + { + printf (_("\ +Usage: %s [OPTION]... [FILE]...\n\ +"), + program_name); + + fputs (_("\ +Paginate or columnate FILE(s) for printing.\n\ +\n\ +"), stdout); + fputs (_("\ +Mandatory arguments to long options are mandatory for short options too.\n\ +"), stdout); + fputs (_("\ + +FIRST_PAGE[:LAST_PAGE], --pages=FIRST_PAGE[:LAST_PAGE]\n\ + begin [stop] printing with page FIRST_[LAST_]PAGE\n\ + -COLUMN, --columns=COLUMN\n\ + output COLUMN columns and print columns down,\n\ + unless -a is used. Balance number of lines in the\n\ + columns on each page.\n\ +"), stdout); + fputs (_("\ + -a, --across print columns across rather than down, used together\n\ + with -COLUMN\n\ + -c, --show-control-chars\n\ + use hat notation (^G) and octal backslash notation\n\ + -d, --double-space\n\ + double space the output\n\ +"), stdout); + fputs (_("\ + -D, --date-format=FORMAT\n\ + use FORMAT for the header date\n\ + -e[CHAR[WIDTH]], --expand-tabs[=CHAR[WIDTH]]\n\ + expand input CHARs (TABs) to tab WIDTH (8)\n\ + -F, -f, --form-feed\n\ + use form feeds instead of newlines to separate pages\n\ + (by a 3-line page header with -F or a 5-line header\n\ + and trailer without -F)\n\ +"), stdout); + fputs (_("\ + -h, --header=HEADER\n\ + use a centered HEADER instead of filename in page header,\n\ + -h \"\" prints a blank line, don't use -h\"\"\n\ + -i[CHAR[WIDTH]], --output-tabs[=CHAR[WIDTH]]\n\ + replace spaces with CHARs (TABs) to tab WIDTH (8)\n\ + -J, --join-lines merge full lines, turns off -W line truncation, no column\n\ + alignment, --sep-string[=STRING] sets separators\n\ +"), stdout); + fputs (_("\ + -l, --length=PAGE_LENGTH\n\ + set the page length to PAGE_LENGTH (66) lines\n\ + (default number of lines of text 56, and with -F 63)\n\ + -m, --merge print all files in parallel, one in each column,\n\ + truncate lines, but join lines of full length with -J\n\ +"), stdout); + fputs (_("\ + -n[SEP[DIGITS]], --number-lines[=SEP[DIGITS]]\n\ + number lines, use DIGITS (5) digits, then SEP (TAB),\n\ + default counting starts with 1st line of input file\n\ + -N, --first-line-number=NUMBER\n\ + start counting with NUMBER at 1st line of first\n\ + page printed (see +FIRST_PAGE)\n\ +"), stdout); + fputs (_("\ + -o, --indent=MARGIN\n\ + offset each line with MARGIN (zero) spaces, do not\n\ + affect -w or -W, MARGIN will be added to PAGE_WIDTH\n\ + -r, --no-file-warnings\n\ + omit warning when a file cannot be opened\n\ +"), stdout); + fputs (_("\ + -s[CHAR],--separator[=CHAR]\n\ + separate columns by a single character, default for CHAR\n\ + is the character without -w and \'no char\' with -w\n\ + -s[CHAR] turns off line truncation of all 3 column\n\ + options (-COLUMN|-a -COLUMN|-m) except -w is set\n\ +"), stdout); + fputs (_("\ + -SSTRING, --sep-string[=STRING]\n\ + separate columns by STRING,\n\ + without -S: Default separator with -J and \n\ + otherwise (same as -S\" \"), no effect on column options\n\ + -t, --omit-header omit page headers and trailers\n\ +"), stdout); + fputs (_("\ + -T, --omit-pagination\n\ + omit page headers and trailers, eliminate any pagination\n\ + by form feeds set in input files\n\ + -v, --show-nonprinting\n\ + use octal backslash notation\n\ + -w, --width=PAGE_WIDTH\n\ + set page width to PAGE_WIDTH (72) characters for\n\ + multiple text-column output only, -s[char] turns off (72)\n\ +"), stdout); + fputs (_("\ + -W, --page-width=PAGE_WIDTH\n\ + set page width to PAGE_WIDTH (72) characters always,\n\ + truncate lines, except -J option is set, no interference\n\ + with -S or -s\n\ +"), stdout); + fputs (HELP_OPTION_DESCRIPTION, stdout); + fputs (VERSION_OPTION_DESCRIPTION, stdout); + fputs (_("\ +\n\ +-t is implied if PAGE_LENGTH <= 10. With no FILE, or when\n\ +FILE is -, read standard input.\n\ +"), stdout); + emit_bug_reporting_address (); + } + exit (status); +} diff -Naur coreutils-7.2.orig/src/sort.c coreutils-7.2/src/sort.c --- coreutils-7.2.orig/src/sort.c 2009-03-29 18:44:10.000000000 +0100 +++ coreutils-7.2/src/sort.c 2009-04-18 10:41:11.000000000 +0100 @@ -22,10 +22,19 @@ #include +#include #include #include #include #include +#if HAVE_WCHAR_H +# include +#endif +/* Get isw* functions. */ +#if HAVE_WCTYPE_H +# include +#endif + #include "system.h" #include "argmatch.h" #include "error.h" @@ -121,14 +130,38 @@ /* Thousands separator; if -1, then there isn't one. */ static int thousands_sep; +static int force_general_numcompare = 0; + /* Nonzero if the corresponding locales are hard. */ static bool hard_LC_COLLATE; -#if HAVE_NL_LANGINFO +#if HAVE_LANGINFO_CODESET static bool hard_LC_TIME; #endif #define NONZERO(x) ((x) != 0) +/* get a multibyte character's byte length. */ +#define GET_BYTELEN_OF_CHAR(LIM, PTR, MBLENGTH, STATE) \ + do \ + { \ + wchar_t wc; \ + mbstate_t state_bak; \ + \ + state_bak = STATE; \ + mblength = mbrtowc (&wc, PTR, LIM - PTR, &STATE); \ + \ + switch (MBLENGTH) \ + { \ + case (size_t)-1: \ + case (size_t)-2: \ + STATE = state_bak; \ + /* Fall through. */ \ + case 0: \ + MBLENGTH = 1; \ + } \ + } \ + while (0) + /* The kind of blanks for '-b' to skip in various options. */ enum blanktype { bl_start, bl_end, bl_both }; @@ -264,13 +297,11 @@ they were read if all keys compare equal. */ static bool stable; -/* If TAB has this value, blanks separate fields. */ -enum { TAB_DEFAULT = CHAR_MAX + 1 }; - -/* Tab character separating fields. If TAB_DEFAULT, then fields are +/* Tab character separating fields. If tab_length is 0, then fields are separated by the empty string between a non-blank character and a blank character. */ -static int tab = TAB_DEFAULT; +static char tab[MB_LEN_MAX + 1]; +static size_t tab_length = 0; /* Flag to remove consecutive duplicate lines from the output. Only the last of a sequence of equal lines will be output. */ @@ -702,6 +733,44 @@ update_proc (pid); } +/* Function pointers. */ +static void +(*inittables) (void); +static char * +(*begfield) (const struct line*, const struct keyfield *); +static char * +(*limfield) (const struct line*, const struct keyfield *); +static int +(*getmonth) (char const *, size_t); +static int +(*keycompare) (const struct line *, const struct line *); +static int +(*numcompare) (const char *, const char *); + +/* Test for white space multibyte character. + Set LENGTH the byte length of investigated multibyte character. */ +#if HAVE_MBRTOWC +static int +ismbblank (const char *str, size_t len, size_t *length) +{ + size_t mblength; + wchar_t wc; + mbstate_t state; + + memset (&state, '\0', sizeof(mbstate_t)); + mblength = mbrtowc (&wc, str, len, &state); + + if (mblength == (size_t)-1 || mblength == (size_t)-2) + { + *length = 1; + return 0; + } + + *length = (mblength < 1) ? 1 : mblength; + return iswblank (wc); +} +#endif + /* Clean up any remaining temporary files. */ static void @@ -1083,7 +1152,7 @@ free (node); } -#if HAVE_NL_LANGINFO +#if HAVE_LANGINFO_CODESET static int struct_month_cmp (const void *m1, const void *m2) @@ -1098,7 +1167,7 @@ /* Initialize the character class tables. */ static void -inittables (void) +inittables_uni (void) { size_t i; @@ -1110,7 +1179,7 @@ fold_toupper[i] = toupper (i); } -#if HAVE_NL_LANGINFO +#if HAVE_LANGINFO_CODESET /* If we're not in the "C" locale, read different names for months. */ if (hard_LC_TIME) { @@ -1192,6 +1261,64 @@ xstrtol_fatal (e, oi, c, long_options, s); } +#if HAVE_MBRTOWC +static void +inittables_mb (void) +{ + int i, j, k, l; + char *name, *s; + size_t s_len, mblength; + char mbc[MB_LEN_MAX]; + wchar_t wc, pwc; + mbstate_t state_mb, state_wc; + + for (i = 0; i < MONTHS_PER_YEAR; i++) + { + s = (char *) nl_langinfo (ABMON_1 + i); + s_len = strlen (s); + monthtab[i].name = name = (char *) xmalloc (s_len + 1); + monthtab[i].val = i + 1; + + memset (&state_mb, '\0', sizeof (mbstate_t)); + memset (&state_wc, '\0', sizeof (mbstate_t)); + + for (j = 0; j < s_len;) + { + if (!ismbblank (s + j, s_len - j, &mblength)) + break; + j += mblength; + } + + for (k = 0; j < s_len;) + { + mblength = mbrtowc (&wc, (s + j), (s_len - j), &state_mb); + assert (mblength != (size_t)-1 && mblength != (size_t)-2); + if (mblength == 0) + break; + + pwc = towupper (wc); + if (pwc == wc) + { + memcpy (mbc, s + j, mblength); + j += mblength; + } + else + { + j += mblength; + mblength = wcrtomb (mbc, pwc, &state_wc); + assert (mblength != (size_t)0 && mblength != (size_t)-1); + } + + for (l = 0; l < mblength; l++) + name[k++] = mbc[l]; + } + name[k] = '\0'; + } + qsort ((void *) monthtab, MONTHS_PER_YEAR, + sizeof (struct month), struct_month_cmp); +} +#endif + /* Specify the amount of main memory to use when sorting. */ static void specify_sort_size (int oi, char c, char const *s) @@ -1402,7 +1529,7 @@ by KEY in LINE. */ static char * -begfield (const struct line *line, const struct keyfield *key) +begfield_uni (const struct line *line, const struct keyfield *key) { char *ptr = line->text, *lim = ptr + line->length - 1; size_t sword = key->sword; @@ -1411,10 +1538,10 @@ /* The leading field separator itself is included in a field when -t is absent. */ - if (tab != TAB_DEFAULT) + if (tab_length) while (ptr < lim && sword--) { - while (ptr < lim && *ptr != tab) + while (ptr < lim && *ptr != tab[0]) ++ptr; if (ptr < lim) ++ptr; @@ -1440,11 +1567,70 @@ return ptr; } +#if HAVE_MBRTOWC +static char * +begfield_mb (const struct line *line, const struct keyfield *key) +{ + int i; + char *ptr = line->text, *lim = ptr + line->length - 1; + size_t sword = key->sword; + size_t schar = key->schar; + size_t mblength; + mbstate_t state; + + memset (&state, '\0', sizeof(mbstate_t)); + + if (tab_length) + while (ptr < lim && sword--) + { + while (ptr < lim && memcmp (ptr, tab, tab_length) != 0) + { + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); + ptr += mblength; + } + if (ptr < lim) + { + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); + ptr += mblength; + } + } + else + while (ptr < lim && sword--) + { + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) + ptr += mblength; + if (ptr < lim) + { + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); + ptr += mblength; + } + while (ptr < lim && !ismbblank (ptr, lim - ptr, &mblength)) + ptr += mblength; + } + + if (key->skipsblanks) + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) + ptr += mblength; + + for (i = 0; i < schar; i++) + { + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); + + if (ptr + mblength > lim) + break; + else + ptr += mblength; + } + + return ptr; +} +#endif + /* Return the limit of (a pointer to the first character after) the field in LINE specified by KEY. */ static char * -limfield (const struct line *line, const struct keyfield *key) +limfield_uni (const struct line *line, const struct keyfield *key) { char *ptr = line->text, *lim = ptr + line->length - 1; size_t eword = key->eword, echar = key->echar; @@ -1459,10 +1645,10 @@ `beginning' is the first character following the delimiting TAB. Otherwise, leave PTR pointing at the first `blank' character after the preceding field. */ - if (tab != TAB_DEFAULT) + if (tab_length) while (ptr < lim && eword--) { - while (ptr < lim && *ptr != tab) + while (ptr < lim && *ptr != tab[0]) ++ptr; if (ptr < lim && (eword | echar)) ++ptr; @@ -1508,10 +1694,10 @@ */ /* Make LIM point to the end of (one byte past) the current field. */ - if (tab != TAB_DEFAULT) + if (tab_length) { char *newlim; - newlim = memchr (ptr, tab, lim - ptr); + newlim = memchr (ptr, tab[0], lim - ptr); if (newlim) lim = newlim; } @@ -1542,6 +1728,107 @@ return ptr; } +#if HAVE_MBRTOWC +static char * +limfield_mb (const struct line *line, const struct keyfield *key) +{ + char *ptr = line->text, *lim = ptr + line->length - 1; + size_t eword = key->eword, echar = key->echar; + int i; + size_t mblength; + mbstate_t state; + + memset (&state, '\0', sizeof(mbstate_t)); + + if (tab_length) + while (ptr < lim && eword--) + { + while (ptr < lim && memcmp (ptr, tab, tab_length) != 0) + { + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); + ptr += mblength; + } + if (ptr < lim && (eword | echar)) + { + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); + ptr += mblength; + } + } + else + while (ptr < lim && eword--) + { + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) + ptr += mblength; + if (ptr < lim) + { + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); + ptr += mblength; + } + while (ptr < lim && !ismbblank (ptr, lim - ptr, &mblength)) + ptr += mblength; + } + + +# ifdef POSIX_UNSPECIFIED + /* Make LIM point to the end of (one byte past) the current field. */ + if (tab_length) + { + char *newlim, *p; + + newlim = NULL; + for (p = ptr; p < lim;) + { + if (memcmp (p, tab, tab_length) == 0) + { + newlim = p; + break; + } + + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); + p += mblength; + } + } + else + { + char *newlim; + newlim = ptr; + + while (newlim < lim && ismbblank (newlim, lim - newlim, &mblength)) + newlim += mblength; + if (ptr < lim) + { + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); + ptr += mblength; + } + while (newlim < lim && !ismbblank (newlim, lim - newlim, &mblength)) + newlim += mblength; + lim = newlim; + } +# endif + + /* If we're skipping leading blanks, don't start counting characters + * until after skipping past any leading blanks. */ + if (key->skipsblanks) + while (ptr < lim && ismbblank (ptr, lim - ptr, &mblength)) + ptr += mblength; + + memset (&state, '\0', sizeof(mbstate_t)); + + /* Advance PTR by ECHAR (if possible), but no further than LIM. */ + for (i = 0; i < echar; i++) + { + GET_BYTELEN_OF_CHAR (lim, ptr, mblength, state); + + if (ptr + mblength > lim) + break; + else + ptr += mblength; + } + + return ptr; +} +#endif + /* Fill BUF reading from FP, moving buf->left bytes from the end of buf->buf to the beginning first. If EOF is reached and the file wasn't terminated by a newline, supply one. Set up BUF's line @@ -1624,8 +1911,24 @@ else { if (key->skipsblanks) - while (blanks[to_uchar (*line_start)]) - line_start++; + { +#if HAVE_MBRTOWC + if (MB_CUR_MAX > 1) + { + size_t mblength; + mbstate_t state; + memset (&state, '\0', sizeof(mbstate_t)); + while (line_start < line->keylim && + ismbblank (line_start, + line->keylim - line_start, + &mblength)) + line_start += mblength; + } + else +#endif + while (blanks[to_uchar (*line_start)]) + line_start++; + } line->keybeg = line_start; } } @@ -1663,7 +1966,7 @@ hideously fast. */ static int -numcompare (const char *a, const char *b) +numcompare_uni (const char *a, const char *b) { while (blanks[to_uchar (*a)]) a++; @@ -1673,6 +1976,25 @@ return strnumcmp (a, b, decimal_point, thousands_sep); } +#if HAVE_MBRTOWC +static int +numcompare_mb (const char *a, const char *b) +{ + size_t mblength, len; + len = strlen (a); /* okay for UTF-8 */ + while (*a && ismbblank (a, len > MB_CUR_MAX ? MB_CUR_MAX : len, &mblength)) + { + a += mblength; + len -= mblength; + } + len = strlen (b); /* okay for UTF-8 */ + while (*b && ismbblank (b, len > MB_CUR_MAX ? MB_CUR_MAX : len, &mblength)) + b += mblength; + + return strnumcmp (a, b, decimal_point, thousands_sep); +} +#endif /* HAV_EMBRTOWC */ + static int general_numcompare (const char *sa, const char *sb) { @@ -1706,7 +2028,7 @@ Return 0 if the name in S is not recognized. */ static int -getmonth (char const *month, size_t len) +getmonth_uni (char const *month, size_t len) { size_t lo = 0; size_t hi = MONTHS_PER_YEAR; @@ -1887,11 +2209,79 @@ return diff; } +#if HAVE_MBRTOWC +static int +getmonth_mb (const char *s, size_t len) +{ + char *month; + register size_t i; + register int lo = 0, hi = MONTHS_PER_YEAR, result; + char *tmp; + size_t wclength, mblength; + const char **pp; + const wchar_t **wpp; + wchar_t *month_wcs; + mbstate_t state; + + while (len > 0 && ismbblank (s, len, &mblength)) + { + s += mblength; + len -= mblength; + } + + if (len == 0) + return 0; + + month = (char *) alloca (len + 1); + + tmp = (char *) alloca (len + 1); + memcpy (tmp, s, len); + tmp[len] = '\0'; + pp = (const char **)&tmp; + month_wcs = (wchar_t *) alloca ((len + 1) * sizeof (wchar_t)); + memset (&state, '\0', sizeof(mbstate_t)); + + wclength = mbsrtowcs (month_wcs, pp, len + 1, &state); + assert (wclength != (size_t)-1 && *pp == NULL); + + for (i = 0; i < wclength; i++) + { + month_wcs[i] = towupper(month_wcs[i]); + if (iswblank (month_wcs[i])) + { + month_wcs[i] = L'\0'; + break; + } + } + + wpp = (const wchar_t **)&month_wcs; + + mblength = wcsrtombs (month, wpp, len + 1, &state); + assert (mblength != (-1) && *wpp == NULL); + + do + { + int ix = (lo + hi) / 2; + + if (strncmp (month, monthtab[ix].name, strlen (monthtab[ix].name)) < 0) + hi = ix; + else + lo = ix; + } + while (hi - lo > 1); + + result = (!strncmp (month, monthtab[lo].name, strlen (monthtab[lo].name)) + ? monthtab[lo].val : 0); + + return result; +} +#endif + /* Compare two lines A and B trying every key in sequence until there are no more keys or a difference is found. */ static int -keycompare (const struct line *a, const struct line *b) +keycompare_uni (const struct line *a, const struct line *b) { struct keyfield const *key = keylist; @@ -2066,6 +2456,179 @@ return key->reverse ? -diff : diff; } +#if HAVE_MBRTOWC +static int +keycompare_mb (const struct line *a, const struct line *b) +{ + struct keyfield *key = keylist; + + /* For the first iteration only, the key positions have been + precomputed for us. */ + char *texta = a->keybeg; + char *textb = b->keybeg; + char *lima = a->keylim; + char *limb = b->keylim; + + size_t mblength_a, mblength_b; + wchar_t wc_a, wc_b; + mbstate_t state_a, state_b; + + int diff; + + memset (&state_a, '\0', sizeof(mbstate_t)); + memset (&state_b, '\0', sizeof(mbstate_t)); + + for (;;) + { + unsigned char *translate = (unsigned char *) key->translate; + bool const *ignore = key->ignore; + + /* Find the lengths. */ + size_t lena = lima <= texta ? 0 : lima - texta; + size_t lenb = limb <= textb ? 0 : limb - textb; + + /* Actually compare the fields. */ + if (key->random) + diff = compare_random (texta, lena, textb, lenb); + else if (key->numeric | key->general_numeric) + { + char savea = *lima, saveb = *limb; + + *lima = *limb = '\0'; + if (force_general_numcompare) + diff = general_numcompare (texta, textb); + else + diff = ((key->numeric ? numcompare : general_numcompare) + (texta, textb)); + *lima = savea, *limb = saveb; + } + else if (key->month) + diff = getmonth (texta, lena) - getmonth (textb, lenb); + else + { + if (ignore || translate) + { + char *copy_a = (char *) alloca (lena + 1 + lenb + 1); + char *copy_b = copy_a + lena + 1; + size_t new_len_a, new_len_b; + size_t i, j; + + /* Ignore and/or translate chars before comparing. */ +# define IGNORE_CHARS(NEW_LEN, LEN, TEXT, COPY, WC, MBLENGTH, STATE) \ + do \ + { \ + wchar_t uwc; \ + char mbc[MB_LEN_MAX]; \ + mbstate_t state_wc; \ + \ + for (NEW_LEN = i = 0; i < LEN;) \ + { \ + mbstate_t state_bak; \ + \ + state_bak = STATE; \ + MBLENGTH = mbrtowc (&WC, TEXT + i, LEN - i, &STATE); \ + \ + if (MBLENGTH == (size_t)-2 || MBLENGTH == (size_t)-1 \ + || MBLENGTH == 0) \ + { \ + if (MBLENGTH == (size_t)-2 || MBLENGTH == (size_t)-1) \ + STATE = state_bak; \ + if (!ignore) \ + COPY[NEW_LEN++] = TEXT[i++]; \ + continue; \ + } \ + \ + if (ignore) \ + { \ + if ((ignore == nonprinting && !iswprint (WC)) \ + || (ignore == nondictionary \ + && !iswalnum (WC) && !iswblank (WC))) \ + { \ + i += MBLENGTH; \ + continue; \ + } \ + } \ + \ + if (translate) \ + { \ + \ + uwc = towupper(WC); \ + if (WC == uwc) \ + { \ + memcpy (mbc, TEXT + i, MBLENGTH); \ + i += MBLENGTH; \ + } \ + else \ + { \ + i += MBLENGTH; \ + WC = uwc; \ + memset (&state_wc, '\0', sizeof (mbstate_t)); \ + \ + MBLENGTH = wcrtomb (mbc, WC, &state_wc); \ + assert (MBLENGTH != (size_t)-1 && MBLENGTH != 0); \ + } \ + \ + for (j = 0; j < MBLENGTH; j++) \ + COPY[NEW_LEN++] = mbc[j]; \ + } \ + else \ + for (j = 0; j < MBLENGTH; j++) \ + COPY[NEW_LEN++] = TEXT[i++]; \ + } \ + COPY[NEW_LEN] = '\0'; \ + } \ + while (0) + IGNORE_CHARS (new_len_a, lena, texta, copy_a, + wc_a, mblength_a, state_a); + IGNORE_CHARS (new_len_b, lenb, textb, copy_b, + wc_b, mblength_b, state_b); + diff = xmemcoll (copy_a, new_len_a, copy_b, new_len_b); + } + else if (lena == 0) + diff = - NONZERO (lenb); + else if (lenb == 0) + goto greater; + else + diff = xmemcoll (texta, lena, textb, lenb); + } + + if (diff) + goto not_equal; + + key = key->next; + if (! key) + break; + + /* Find the beginning and limit of the next field. */ + if (key->eword != -1) + lima = limfield (a, key), limb = limfield (b, key); + else + lima = a->text + a->length - 1, limb = b->text + b->length - 1; + + if (key->sword != -1) + texta = begfield (a, key), textb = begfield (b, key); + else + { + texta = a->text, textb = b->text; + if (key->skipsblanks) + { + while (texta < lima && ismbblank (texta, lima - texta, &mblength_a)) + texta += mblength_a; + while (textb < limb && ismbblank (textb, limb - textb, &mblength_b)) + textb += mblength_b; + } + } + } + + return 0; + +greater: + diff = 1; +not_equal: + return key->reverse ? -diff : diff; +} +#endif + /* Compare two lines A and B, returning negative, zero, or positive depending on whether A compares less than, equal to, or greater than B. */ @@ -3056,7 +3619,7 @@ initialize_exit_failure (SORT_FAILURE); hard_LC_COLLATE = hard_locale (LC_COLLATE); -#if HAVE_NL_LANGINFO +#if HAVE_LANGINFO_CODESET hard_LC_TIME = hard_locale (LC_TIME); #endif @@ -3077,6 +3640,27 @@ thousands_sep = -1; } +#if HAVE_MBRTOWC + if (MB_CUR_MAX > 1) + { + inittables = inittables_mb; + begfield = begfield_mb; + limfield = limfield_mb; + getmonth = getmonth_mb; + keycompare = keycompare_mb; + numcompare = numcompare_mb; + } + else +#endif + { + inittables = inittables_uni; + begfield = begfield_uni; + limfield = limfield_uni; + getmonth = getmonth_uni; + keycompare = keycompare_uni; + numcompare = numcompare_uni; + } + have_read_stdin = false; inittables (); @@ -3334,13 +3918,35 @@ case 't': { - char newtab = optarg[0]; - if (! newtab) + char newtab[MB_LEN_MAX + 1]; + size_t newtab_length = 1; + strncpy (newtab, optarg, MB_LEN_MAX); + if (! newtab[0]) error (SORT_FAILURE, 0, _("empty tab")); - if (optarg[1]) +#if HAVE_MBRTOWC + if (MB_CUR_MAX > 1) + { + wchar_t wc; + mbstate_t state; + size_t i; + + memset (&state, '\0', sizeof (mbstate_t)); + newtab_length = mbrtowc (&wc, newtab, strnlen (newtab, + MB_LEN_MAX), + &state); + switch (newtab_length) + { + case (size_t) -1: + case (size_t) -2: + case 0: + newtab_length = 1; + } + } +#endif + if (newtab_length == 1 && optarg[1]) { if (STREQ (optarg, "\\0")) - newtab = '\0'; + newtab[0] = '\0'; else { /* Provoke with `sort -txx'. Complain about @@ -3351,9 +3957,12 @@ quote (optarg)); } } - if (tab != TAB_DEFAULT && tab != newtab) + if (tab_length + && (tab_length != newtab_length + || memcmp (tab, newtab, tab_length) != 0)) error (SORT_FAILURE, 0, _("incompatible tabs")); - tab = newtab; + memcpy (tab, newtab, newtab_length); + tab_length = newtab_length; } break; diff -Naur coreutils-7.2.orig/src/sort.c.orig coreutils-7.2/src/sort.c.orig --- coreutils-7.2.orig/src/sort.c.orig 1970-01-01 01:00:00.000000000 +0100 +++ coreutils-7.2/src/sort.c.orig 2009-03-29 18:44:10.000000000 +0100 @@ -0,0 +1,3569 @@ +/* sort - sort lines of text (with all kinds of options). + Copyright (C) 1988, 1991-2008 Free Software Foundation, Inc. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see . + + Written December 1988 by Mike Haertel. + The author may be reached (Email) at the address mike@gnu.ai.mit.edu, + or (US mail) as Mike Haertel c/o Free Software Foundation. + + Ørn E. Hansen added NLS support in 1997. */ + +#include + +#include +#include +#include +#include +#include "system.h" +#include "argmatch.h" +#include "error.h" +#include "filevercmp.h" +#include "hash.h" +#include "md5.h" +#include "physmem.h" +#include "posixver.h" +#include "quote.h" +#include "quotearg.h" +#include "randread.h" +#include "readtokens0.h" +#include "stdio--.h" +#include "stdlib--.h" +#include "strnumcmp.h" +#include "xmemcoll.h" +#include "xmemxfrm.h" +#include "xstrtol.h" + +#if HAVE_SYS_RESOURCE_H +# include +#endif +#ifndef RLIMIT_DATA +struct rlimit { size_t rlim_cur; }; +# define getrlimit(Resource, Rlp) (-1) +#endif + +/* The official name of this program (e.g., no `g' prefix). */ +#define PROGRAM_NAME "sort" + +#define AUTHORS \ + proper_name ("Mike Haertel"), \ + proper_name ("Paul Eggert") + +#if HAVE_LANGINFO_CODESET +# include +#endif + +/* Use SA_NOCLDSTOP as a proxy for whether the sigaction machinery is + present. */ +#ifndef SA_NOCLDSTOP +# define SA_NOCLDSTOP 0 +/* No sigprocmask. Always 'return' zero. */ +# define sigprocmask(How, Set, Oset) (0) +# define sigset_t int +# if ! HAVE_SIGINTERRUPT +# define siginterrupt(sig, flag) /* empty */ +# endif +#endif + +#if !defined OPEN_MAX && defined NR_OPEN +# define OPEN_MAX NR_OPEN +#endif +#if !defined OPEN_MAX +# define OPEN_MAX 20 +#endif + +#define UCHAR_LIM (UCHAR_MAX + 1) + +#ifndef DEFAULT_TMPDIR +# define DEFAULT_TMPDIR "/tmp" +#endif + +/* Exit statuses. */ +enum + { + /* POSIX says to exit with status 1 if invoked with -c and the + input is not properly sorted. */ + SORT_OUT_OF_ORDER = 1, + + /* POSIX says any other irregular exit must exit with a status + code greater than 1. */ + SORT_FAILURE = 2 + }; + +enum + { + /* The number of times we should try to fork a compression process + (we retry if the fork call fails). We don't _need_ to compress + temp files, this is just to reduce disk access, so this number + can be small. */ + MAX_FORK_TRIES_COMPRESS = 2, + + /* The number of times we should try to fork a decompression process. + If we can't fork a decompression process, we can't sort, so this + number should be big. */ + MAX_FORK_TRIES_DECOMPRESS = 8 + }; + +/* The representation of the decimal point in the current locale. */ +static int decimal_point; + +/* Thousands separator; if -1, then there isn't one. */ +static int thousands_sep; + +/* Nonzero if the corresponding locales are hard. */ +static bool hard_LC_COLLATE; +#if HAVE_NL_LANGINFO +static bool hard_LC_TIME; +#endif + +#define NONZERO(x) ((x) != 0) + +/* The kind of blanks for '-b' to skip in various options. */ +enum blanktype { bl_start, bl_end, bl_both }; + +/* The character marking end of line. Default to \n. */ +static char eolchar = '\n'; + +/* Lines are held in core as counted strings. */ +struct line +{ + char *text; /* Text of the line. */ + size_t length; /* Length including final newline. */ + char *keybeg; /* Start of first key. */ + char *keylim; /* Limit of first key. */ +}; + +/* Input buffers. */ +struct buffer +{ + char *buf; /* Dynamically allocated buffer, + partitioned into 3 regions: + - input data; + - unused area; + - an array of lines, in reverse order. */ + size_t used; /* Number of bytes used for input data. */ + size_t nlines; /* Number of lines in the line array. */ + size_t alloc; /* Number of bytes allocated. */ + size_t left; /* Number of bytes left from previous reads. */ + size_t line_bytes; /* Number of bytes to reserve for each line. */ + bool eof; /* An EOF has been read. */ +}; + +struct keyfield +{ + size_t sword; /* Zero-origin 'word' to start at. */ + size_t schar; /* Additional characters to skip. */ + size_t eword; /* Zero-origin first word after field. */ + size_t echar; /* Additional characters in field. */ + bool const *ignore; /* Boolean array of characters to ignore. */ + char const *translate; /* Translation applied to characters. */ + bool skipsblanks; /* Skip leading blanks when finding start. */ + bool skipeblanks; /* Skip leading blanks when finding end. */ + bool numeric; /* Flag for numeric comparison. Handle + strings of digits with optional decimal + point, but no exponential notation. */ + bool random; /* Sort by random hash of key. */ + bool general_numeric; /* Flag for general, numeric comparison. + Handle numbers in exponential notation. */ + bool month; /* Flag for comparison by month name. */ + bool reverse; /* Reverse the sense of comparison. */ + bool version; /* sort by version number */ + struct keyfield *next; /* Next keyfield to try. */ +}; + +struct month +{ + char const *name; + int val; +}; + +/* FIXME: None of these tables work with multibyte character sets. + Also, there are many other bugs when handling multibyte characters. + One way to fix this is to rewrite `sort' to use wide characters + internally, but doing this with good performance is a bit + tricky. */ + +/* Table of blanks. */ +static bool blanks[UCHAR_LIM]; + +/* Table of non-printing characters. */ +static bool nonprinting[UCHAR_LIM]; + +/* Table of non-dictionary characters (not letters, digits, or blanks). */ +static bool nondictionary[UCHAR_LIM]; + +/* Translation table folding lower case to upper. */ +static char fold_toupper[UCHAR_LIM]; + +#define MONTHS_PER_YEAR 12 + +/* Table mapping month names to integers. + Alphabetic order allows binary search. */ +static struct month monthtab[] = +{ + {"APR", 4}, + {"AUG", 8}, + {"DEC", 12}, + {"FEB", 2}, + {"JAN", 1}, + {"JUL", 7}, + {"JUN", 6}, + {"MAR", 3}, + {"MAY", 5}, + {"NOV", 11}, + {"OCT", 10}, + {"SEP", 9} +}; + +/* During the merge phase, the number of files to merge at once. */ +#define NMERGE_DEFAULT 16 + +/* Minimum size for a merge or check buffer. */ +#define MIN_MERGE_BUFFER_SIZE (2 + sizeof (struct line)) + +/* Minimum sort size; the code might not work with smaller sizes. */ +#define MIN_SORT_SIZE (nmerge * MIN_MERGE_BUFFER_SIZE) + +/* The number of bytes needed for a merge or check buffer, which can + function relatively efficiently even if it holds only one line. If + a longer line is seen, this value is increased. */ +static size_t merge_buffer_size = MAX (MIN_MERGE_BUFFER_SIZE, 256 * 1024); + +/* The approximate maximum number of bytes of main memory to use, as + specified by the user. Zero if the user has not specified a size. */ +static size_t sort_size; + +/* The guessed size for non-regular files. */ +#define INPUT_FILE_SIZE_GUESS (1024 * 1024) + +/* Array of directory names in which any temporary files are to be created. */ +static char const **temp_dirs; + +/* Number of temporary directory names used. */ +static size_t temp_dir_count; + +/* Number of allocated slots in temp_dirs. */ +static size_t temp_dir_alloc; + +/* Flag to reverse the order of all comparisons. */ +static bool reverse; + +/* Flag for stable sort. This turns off the last ditch bytewise + comparison of lines, and instead leaves lines in the same order + they were read if all keys compare equal. */ +static bool stable; + +/* If TAB has this value, blanks separate fields. */ +enum { TAB_DEFAULT = CHAR_MAX + 1 }; + +/* Tab character separating fields. If TAB_DEFAULT, then fields are + separated by the empty string between a non-blank character and a blank + character. */ +static int tab = TAB_DEFAULT; + +/* Flag to remove consecutive duplicate lines from the output. + Only the last of a sequence of equal lines will be output. */ +static bool unique; + +/* Nonzero if any of the input files are the standard input. */ +static bool have_read_stdin; + +/* List of key field comparisons to be tried. */ +static struct keyfield *keylist; + +/* Program used to (de)compress temp files. Must accept -d. */ +static char const *compress_program; + +/* Maximum number of files to merge in one go. If more than this + number are present, temp files will be used. */ +static unsigned int nmerge = NMERGE_DEFAULT; + +static void sortlines_temp (struct line *, size_t, struct line *); + +/* Report MESSAGE for FILE, then clean up and exit. + If FILE is null, it represents standard output. */ + +static void die (char const *, char const *) ATTRIBUTE_NORETURN; +static void +die (char const *message, char const *file) +{ + error (0, errno, "%s: %s", message, file ? file : _("standard output")); + exit (SORT_FAILURE); +} + +void +usage (int status) +{ + if (status != EXIT_SUCCESS) + fprintf (stderr, _("Try `%s --help' for more information.\n"), + program_name); + else + { + printf (_("\ +Usage: %s [OPTION]... [FILE]...\n\ + or: %s [OPTION]... --files0-from=F\n\ +"), + program_name, program_name); + fputs (_("\ +Write sorted concatenation of all FILE(s) to standard output.\n\ +\n\ +"), stdout); + fputs (_("\ +Mandatory arguments to long options are mandatory for short options too.\n\ +"), stdout); + fputs (_("\ +Ordering options:\n\ +\n\ +"), stdout); + fputs (_("\ + -b, --ignore-leading-blanks ignore leading blanks\n\ + -d, --dictionary-order consider only blanks and alphanumeric characters\n\ + -f, --ignore-case fold lower case to upper case characters\n\ +"), stdout); + fputs (_("\ + -g, --general-numeric-sort compare according to general numerical value\n\ + -i, --ignore-nonprinting consider only printable characters\n\ + -M, --month-sort compare (unknown) < `JAN' < ... < `DEC'\n\ +"), stdout); + fputs (_("\ + -n, --numeric-sort compare according to string numerical value\n\ + -R, --random-sort sort by random hash of keys\n\ + --random-source=FILE get random bytes from FILE (default /dev/urandom)\n\ + -r, --reverse reverse the result of comparisons\n\ +"), stdout); + fputs (_("\ + --sort=WORD sort according to WORD:\n\ + general-numeric -g, month -M, numeric -n,\n\ + random -R, version -V\n\ + -V, --version-sort sort by numeric version\n\ +\n\ +"), stdout); + fputs (_("\ +Other options:\n\ +\n\ +"), stdout); + fputs (_("\ + --batch-size=NMERGE merge at most NMERGE inputs at once;\n\ + for more use temp files\n\ +"), stdout); + fputs (_("\ + -c, --check, --check=diagnose-first check for sorted input; do not sort\n\ + -C, --check=quiet, --check=silent like -c, but do not report first bad line\n\ + --compress-program=PROG compress temporaries with PROG;\n\ + decompress them with PROG -d\n\ + --files0-from=F read input from the files specified by\n\ + NUL-terminated names in file F;\n\ + If F is - then read names from standard input\n\ +"), stdout); + fputs (_("\ + -k, --key=POS1[,POS2] start a key at POS1 (origin 1), end it at POS2\n\ + (default end of line)\n\ + -m, --merge merge already sorted files; do not sort\n\ +"), stdout); + fputs (_("\ + -o, --output=FILE write result to FILE instead of standard output\n\ + -s, --stable stabilize sort by disabling last-resort comparison\n\ + -S, --buffer-size=SIZE use SIZE for main memory buffer\n\ +"), stdout); + printf (_("\ + -t, --field-separator=SEP use SEP instead of non-blank to blank transition\n\ + -T, --temporary-directory=DIR use DIR for temporaries, not $TMPDIR or %s;\n\ + multiple options specify multiple directories\n\ + -u, --unique with -c, check for strict ordering;\n\ + without -c, output only the first of an equal run\n\ +"), DEFAULT_TMPDIR); + fputs (_("\ + -z, --zero-terminated end lines with 0 byte, not newline\n\ +"), stdout); + fputs (HELP_OPTION_DESCRIPTION, stdout); + fputs (VERSION_OPTION_DESCRIPTION, stdout); + fputs (_("\ +\n\ +POS is F[.C][OPTS], where F is the field number and C the character position\n\ +in the field; both are origin 1. If neither -t nor -b is in effect, characters\n\ +in a field are counted from the beginning of the preceding whitespace. OPTS is\n\ +one or more single-letter ordering options, which override global ordering\n\ +options for that key. If no key is given, use the entire line as the key.\n\ +\n\ +SIZE may be followed by the following multiplicative suffixes:\n\ +"), stdout); + fputs (_("\ +% 1% of memory, b 1, K 1024 (default), and so on for M, G, T, P, E, Z, Y.\n\ +\n\ +With no FILE, or when FILE is -, read standard input.\n\ +\n\ +*** WARNING ***\n\ +The locale specified by the environment affects sort order.\n\ +Set LC_ALL=C to get the traditional sort order that uses\n\ +native byte values.\n\ +"), stdout ); + emit_bug_reporting_address (); + } + + exit (status); +} + +/* For long options that have no equivalent short option, use a + non-character as a pseudo short option, starting with CHAR_MAX + 1. */ +enum +{ + CHECK_OPTION = CHAR_MAX + 1, + COMPRESS_PROGRAM_OPTION, + FILES0_FROM_OPTION, + NMERGE_OPTION, + RANDOM_SOURCE_OPTION, + SORT_OPTION +}; + +static char const short_options[] = "-bcCdfgik:mMno:rRsS:t:T:uVy:z"; + +static struct option const long_options[] = +{ + {"ignore-leading-blanks", no_argument, NULL, 'b'}, + {"check", optional_argument, NULL, CHECK_OPTION}, + {"compress-program", required_argument, NULL, COMPRESS_PROGRAM_OPTION}, + {"dictionary-order", no_argument, NULL, 'd'}, + {"ignore-case", no_argument, NULL, 'f'}, + {"files0-from", required_argument, NULL, FILES0_FROM_OPTION}, + {"general-numeric-sort", no_argument, NULL, 'g'}, + {"ignore-nonprinting", no_argument, NULL, 'i'}, + {"key", required_argument, NULL, 'k'}, + {"merge", no_argument, NULL, 'm'}, + {"month-sort", no_argument, NULL, 'M'}, + {"numeric-sort", no_argument, NULL, 'n'}, + {"version-sort", no_argument, NULL, 'V'}, + {"random-sort", no_argument, NULL, 'R'}, + {"random-source", required_argument, NULL, RANDOM_SOURCE_OPTION}, + {"sort", required_argument, NULL, SORT_OPTION}, + {"output", required_argument, NULL, 'o'}, + {"reverse", no_argument, NULL, 'r'}, + {"stable", no_argument, NULL, 's'}, + {"batch-size", required_argument, NULL, NMERGE_OPTION}, + {"buffer-size", required_argument, NULL, 'S'}, + {"field-separator", required_argument, NULL, 't'}, + {"temporary-directory", required_argument, NULL, 'T'}, + {"unique", no_argument, NULL, 'u'}, + {"zero-terminated", no_argument, NULL, 'z'}, + {GETOPT_HELP_OPTION_DECL}, + {GETOPT_VERSION_OPTION_DECL}, + {NULL, 0, NULL, 0}, +}; + +#define CHECK_TABLE \ + _ct_("quiet", 'C') \ + _ct_("silent", 'C') \ + _ct_("diagnose-first", 'c') + +static char const *const check_args[] = +{ +#define _ct_(_s, _c) _s, + CHECK_TABLE NULL +#undef _ct_ +}; +static char const check_types[] = +{ +#define _ct_(_s, _c) _c, + CHECK_TABLE +#undef _ct_ +}; + +#define SORT_TABLE \ + _st_("general-numeric", 'g') \ + _st_("month", 'M') \ + _st_("numeric", 'n') \ + _st_("random", 'R') \ + _st_("version", 'V') + +static char const *const sort_args[] = +{ +#define _st_(_s, _c) _s, + SORT_TABLE NULL +#undef _st_ +}; +static char const sort_types[] = +{ +#define _st_(_s, _c) _c, + SORT_TABLE +#undef _st_ +}; + +/* The set of signals that are caught. */ +static sigset_t caught_signals; + +/* Critical section status. */ +struct cs_status +{ + bool valid; + sigset_t sigs; +}; + +/* Enter a critical section. */ +static struct cs_status +cs_enter (void) +{ + struct cs_status status; + status.valid = (sigprocmask (SIG_BLOCK, &caught_signals, &status.sigs) == 0); + return status; +} + +/* Leave a critical section. */ +static void +cs_leave (struct cs_status status) +{ + if (status.valid) + { + /* Ignore failure when restoring the signal mask. */ + sigprocmask (SIG_SETMASK, &status.sigs, NULL); + } +} + +/* The list of temporary files. */ +struct tempnode +{ + struct tempnode *volatile next; + pid_t pid; /* If compressed, the pid of compressor, else zero */ + char name[1]; /* Actual size is 1 + file name length. */ +}; +static struct tempnode *volatile temphead; +static struct tempnode *volatile *temptail = &temphead; + +struct sortfile +{ + char const *name; + pid_t pid; /* If compressed, the pid of compressor, else zero */ +}; + +/* A table where we store compression process states. We clean up all + processes in a timely manner so as not to exhaust system resources, + so we store the info on whether the process is still running, or has + been reaped here. */ +static Hash_table *proctab; + +enum { INIT_PROCTAB_SIZE = 47 }; + +enum procstate { ALIVE, ZOMBIE }; + +/* A proctab entry. The COUNT field is there in case we fork a new + compression process that has the same PID as an old zombie process + that is still in the table (because the process to decompress the + temp file it was associated with hasn't started yet). */ +struct procnode +{ + pid_t pid; + enum procstate state; + size_t count; +}; + +static size_t +proctab_hasher (const void *entry, size_t tabsize) +{ + const struct procnode *node = entry; + return node->pid % tabsize; +} + +static bool +proctab_comparator (const void *e1, const void *e2) +{ + const struct procnode *n1 = e1, *n2 = e2; + return n1->pid == n2->pid; +} + +/* The total number of forked processes (compressors and decompressors) + that have not been reaped yet. */ +static size_t nprocs; + +/* The number of child processes we'll allow before we try to reap some. */ +enum { MAX_PROCS_BEFORE_REAP = 2 }; + +/* If 0 < PID, wait for the child process with that PID to exit. + If PID is -1, clean up a random child process which has finished and + return the process ID of that child. If PID is -1 and no processes + have quit yet, return 0 without waiting. */ + +static pid_t +reap (pid_t pid) +{ + int status; + pid_t cpid = waitpid (pid, &status, pid < 0 ? WNOHANG : 0); + + if (cpid < 0) + error (SORT_FAILURE, errno, _("waiting for %s [-d]"), + compress_program); + else if (0 < cpid) + { + if (! WIFEXITED (status) || WEXITSTATUS (status)) + error (SORT_FAILURE, 0, _("%s [-d] terminated abnormally"), + compress_program); + --nprocs; + } + + return cpid; +} + +/* Add the PID of a running compression process to proctab, or update + the entry COUNT and STATE fields if it's already there. This also + creates the table for us the first time it's called. */ + +static void +register_proc (pid_t pid) +{ + struct procnode test, *node; + + if (! proctab) + { + proctab = hash_initialize (INIT_PROCTAB_SIZE, NULL, + proctab_hasher, + proctab_comparator, + free); + if (! proctab) + xalloc_die (); + } + + test.pid = pid; + node = hash_lookup (proctab, &test); + if (node) + { + node->state = ALIVE; + ++node->count; + } + else + { + node = xmalloc (sizeof *node); + node->pid = pid; + node->state = ALIVE; + node->count = 1; + hash_insert (proctab, node); + } +} + +/* This is called when we reap a random process. We don't know + whether we have reaped a compression process or a decompression + process until we look in the table. If there's an ALIVE entry for + it, then we have reaped a compression process, so change the state + to ZOMBIE. Otherwise, it's a decompression processes, so ignore it. */ + +static void +update_proc (pid_t pid) +{ + struct procnode test, *node; + + test.pid = pid; + node = hash_lookup (proctab, &test); + if (node) + node->state = ZOMBIE; +} + +/* This is for when we need to wait for a compression process to exit. + If it has a ZOMBIE entry in the table then it's already dead and has + been reaped. Note that if there's an ALIVE entry for it, it still may + already have died and been reaped if a second process was created with + the same PID. This is probably exceedingly rare, but to be on the safe + side we will have to wait for any compression process with this PID. */ + +static void +wait_proc (pid_t pid) +{ + struct procnode test, *node; + + test.pid = pid; + node = hash_lookup (proctab, &test); + if (node->state == ALIVE) + reap (pid); + + node->state = ZOMBIE; + if (! --node->count) + { + hash_delete (proctab, node); + free (node); + } +} + +/* Keep reaping finished children as long as there are more to reap. + This doesn't block waiting for any of them, it only reaps those + that are already dead. */ + +static void +reap_some (void) +{ + pid_t pid; + + while (0 < nprocs && (pid = reap (-1))) + update_proc (pid); +} + +/* Clean up any remaining temporary files. */ + +static void +cleanup (void) +{ + struct tempnode const *node; + + for (node = temphead; node; node = node->next) + unlink (node->name); + temphead = NULL; +} + +/* Cleanup actions to take when exiting. */ + +static void +exit_cleanup (void) +{ + if (temphead) + { + /* Clean up any remaining temporary files in a critical section so + that a signal handler does not try to clean them too. */ + struct cs_status cs = cs_enter (); + cleanup (); + cs_leave (cs); + } + + close_stdout (); +} + +/* Create a new temporary file, returning its newly allocated tempnode. + Store into *PFD the file descriptor open for writing. + If the creation fails, return NULL and store -1 into *PFD if the + failure is due to file descriptor exhaustion and + SURVIVE_FD_EXHAUSTION; otherwise, die. */ + +static struct tempnode * +create_temp_file (int *pfd, bool survive_fd_exhaustion) +{ + static char const slashbase[] = "/sortXXXXXX"; + static size_t temp_dir_index; + int fd; + int saved_errno; + char const *temp_dir = temp_dirs[temp_dir_index]; + size_t len = strlen (temp_dir); + struct tempnode *node = + xmalloc (offsetof (struct tempnode, name) + len + sizeof slashbase); + char *file = node->name; + struct cs_status cs; + + memcpy (file, temp_dir, len); + memcpy (file + len, slashbase, sizeof slashbase); + node->next = NULL; + node->pid = 0; + if (++temp_dir_index == temp_dir_count) + temp_dir_index = 0; + + /* Create the temporary file in a critical section, to avoid races. */ + cs = cs_enter (); + fd = mkstemp (file); + if (0 <= fd) + { + *temptail = node; + temptail = &node->next; + } + saved_errno = errno; + cs_leave (cs); + errno = saved_errno; + + if (fd < 0) + { + if (! (survive_fd_exhaustion && errno == EMFILE)) + error (SORT_FAILURE, errno, _("cannot create temporary file in %s"), + quote (temp_dir)); + free (node); + node = NULL; + } + + *pfd = fd; + return node; +} + +/* Return a stream for FILE, opened with mode HOW. A null FILE means + standard output; HOW should be "w". When opening for input, "-" + means standard input. To avoid confusion, do not return file + descriptors STDIN_FILENO, STDOUT_FILENO, or STDERR_FILENO when + opening an ordinary FILE. Return NULL if unsuccessful. */ + +static FILE * +stream_open (const char *file, const char *how) +{ + if (!file) + return stdout; + if (STREQ (file, "-") && *how == 'r') + { + have_read_stdin = true; + return stdin; + } + return fopen (file, how); +} + +/* Same as stream_open, except always return a non-null value; die on + failure. */ + +static FILE * +xfopen (const char *file, const char *how) + { + FILE *fp = stream_open (file, how); + if (!fp) + die (_("open failed"), file); + return fp; +} + +/* Close FP, whose name is FILE, and report any errors. */ + +static void +xfclose (FILE *fp, char const *file) +{ + switch (fileno (fp)) + { + case STDIN_FILENO: + /* Allow reading stdin from tty more than once. */ + if (feof (fp)) + clearerr (fp); + break; + + case STDOUT_FILENO: + /* Don't close stdout just yet. close_stdout does that. */ + if (fflush (fp) != 0) + die (_("fflush failed"), file); + break; + + default: + if (fclose (fp) != 0) + die (_("close failed"), file); + break; + } +} + +static void +dup2_or_die (int oldfd, int newfd) +{ + if (dup2 (oldfd, newfd) < 0) + error (SORT_FAILURE, errno, _("dup2 failed")); +} + +/* Fork a child process for piping to and do common cleanup. The + TRIES parameter tells us how many times to try to fork before + giving up. Return the PID of the child, or -1 (setting errno) + on failure. */ + +static pid_t +pipe_fork (int pipefds[2], size_t tries) +{ +#if HAVE_WORKING_FORK + struct tempnode *saved_temphead; + int saved_errno; + unsigned int wait_retry = 1; + pid_t pid IF_LINT (= -1); + struct cs_status cs; + + if (pipe (pipefds) < 0) + return -1; + + while (tries--) + { + /* This is so the child process won't delete our temp files + if it receives a signal before exec-ing. */ + cs = cs_enter (); + saved_temphead = temphead; + temphead = NULL; + + pid = fork (); + saved_errno = errno; + if (pid) + temphead = saved_temphead; + + cs_leave (cs); + errno = saved_errno; + + if (0 <= pid || errno != EAGAIN) + break; + else + { + sleep (wait_retry); + wait_retry *= 2; + reap_some (); + } + } + + if (pid < 0) + { + saved_errno = errno; + close (pipefds[0]); + close (pipefds[1]); + errno = saved_errno; + } + else if (pid == 0) + { + close (STDIN_FILENO); + close (STDOUT_FILENO); + } + else + ++nprocs; + + return pid; + +#else /* ! HAVE_WORKING_FORK */ + return -1; +#endif +} + +/* Create a temporary file and start a compression program to filter output + to that file. Set *PFP to the file handle and if PPID is non-NULL, + set *PPID to the PID of the newly-created process. If the creation + fails, return NULL if the failure is due to file descriptor + exhaustion and SURVIVE_FD_EXHAUSTION; otherwise, die. */ + +static char * +maybe_create_temp (FILE **pfp, pid_t *ppid, bool survive_fd_exhaustion) +{ + int tempfd; + struct tempnode *node = create_temp_file (&tempfd, survive_fd_exhaustion); + char *name; + + if (! node) + return NULL; + + name = node->name; + + if (compress_program) + { + int pipefds[2]; + + node->pid = pipe_fork (pipefds, MAX_FORK_TRIES_COMPRESS); + if (0 < node->pid) + { + close (tempfd); + close (pipefds[0]); + tempfd = pipefds[1]; + + register_proc (node->pid); + } + else if (node->pid == 0) + { + close (pipefds[1]); + dup2_or_die (tempfd, STDOUT_FILENO); + close (tempfd); + dup2_or_die (pipefds[0], STDIN_FILENO); + close (pipefds[0]); + + if (execlp (compress_program, compress_program, (char *) NULL) < 0) + error (SORT_FAILURE, errno, _("couldn't execute %s"), + compress_program); + } + else + node->pid = 0; + } + + *pfp = fdopen (tempfd, "w"); + if (! *pfp) + die (_("couldn't create temporary file"), name); + + if (ppid) + *ppid = node->pid; + + return name; +} + +/* Create a temporary file and start a compression program to filter output + to that file. Set *PFP to the file handle and if *PPID is non-NULL, + set it to the PID of the newly-created process. Die on failure. */ + +static char * +create_temp (FILE **pfp, pid_t *ppid) +{ + return maybe_create_temp (pfp, ppid, false); +} + +/* Open a compressed temp file and start a decompression process through + which to filter the input. PID must be the valid processes ID of the + process used to compress the file. Return NULL (setting errno to + EMFILE) if we ran out of file descriptors, and die on any other + kind of failure. */ + +static FILE * +open_temp (const char *name, pid_t pid) +{ + int tempfd, pipefds[2]; + FILE *fp = NULL; + + wait_proc (pid); + + tempfd = open (name, O_RDONLY); + if (tempfd < 0) + return NULL; + + switch (pipe_fork (pipefds, MAX_FORK_TRIES_DECOMPRESS)) + { + case -1: + if (errno != EMFILE) + error (SORT_FAILURE, errno, _("couldn't create process for %s -d"), + compress_program); + close (tempfd); + errno = EMFILE; + break; + + case 0: + close (pipefds[0]); + dup2_or_die (tempfd, STDIN_FILENO); + close (tempfd); + dup2_or_die (pipefds[1], STDOUT_FILENO); + close (pipefds[1]); + + execlp (compress_program, compress_program, "-d", (char *) NULL); + error (SORT_FAILURE, errno, _("couldn't execute %s -d"), + compress_program); + + default: + close (tempfd); + close (pipefds[1]); + + fp = fdopen (pipefds[0], "r"); + if (! fp) + { + int saved_errno = errno; + close (pipefds[0]); + errno = saved_errno; + } + break; + } + + return fp; +} + +static void +write_bytes (const char *buf, size_t n_bytes, FILE *fp, const char *output_file) +{ + if (fwrite (buf, 1, n_bytes, fp) != n_bytes) + die (_("write failed"), output_file); +} + +/* Append DIR to the array of temporary directory names. */ +static void +add_temp_dir (char const *dir) +{ + if (temp_dir_count == temp_dir_alloc) + temp_dirs = X2NREALLOC (temp_dirs, &temp_dir_alloc); + + temp_dirs[temp_dir_count++] = dir; +} + +/* Remove NAME from the list of temporary files. */ + +static void +zaptemp (const char *name) +{ + struct tempnode *volatile *pnode; + struct tempnode *node; + struct tempnode *next; + int unlink_status; + int unlink_errno = 0; + struct cs_status cs; + + for (pnode = &temphead; (node = *pnode)->name != name; pnode = &node->next) + continue; + + /* Unlink the temporary file in a critical section to avoid races. */ + next = node->next; + cs = cs_enter (); + unlink_status = unlink (name); + unlink_errno = errno; + *pnode = next; + cs_leave (cs); + + if (unlink_status != 0) + error (0, unlink_errno, _("warning: cannot remove: %s"), name); + if (! next) + temptail = pnode; + free (node); +} + +#if HAVE_NL_LANGINFO + +static int +struct_month_cmp (const void *m1, const void *m2) +{ + struct month const *month1 = m1; + struct month const *month2 = m2; + return strcmp (month1->name, month2->name); +} + +#endif + +/* Initialize the character class tables. */ + +static void +inittables (void) +{ + size_t i; + + for (i = 0; i < UCHAR_LIM; ++i) + { + blanks[i] = !! isblank (i); + nonprinting[i] = ! isprint (i); + nondictionary[i] = ! isalnum (i) && ! isblank (i); + fold_toupper[i] = toupper (i); + } + +#if HAVE_NL_LANGINFO + /* If we're not in the "C" locale, read different names for months. */ + if (hard_LC_TIME) + { + for (i = 0; i < MONTHS_PER_YEAR; i++) + { + char const *s; + size_t s_len; + size_t j; + char *name; + + s = (char *) nl_langinfo (ABMON_1 + i); + s_len = strlen (s); + monthtab[i].name = name = xmalloc (s_len + 1); + monthtab[i].val = i + 1; + + for (j = 0; j < s_len; j++) + name[j] = fold_toupper[to_uchar (s[j])]; + name[j] = '\0'; + } + qsort ((void *) monthtab, MONTHS_PER_YEAR, + sizeof *monthtab, struct_month_cmp); + } +#endif +} + +/* Specify how many inputs may be merged at once. + This may be set on the command-line with the + --batch-size option. */ +static void +specify_nmerge (int oi, char c, char const *s) +{ + uintmax_t n; + struct rlimit rlimit; + enum strtol_error e = xstrtoumax (s, NULL, 10, &n, NULL); + + /* Try to find out how many file descriptors we'll be able + to open. We need at least nmerge + 3 (STDIN_FILENO, + STDOUT_FILENO and STDERR_FILENO). */ + unsigned int max_nmerge = ((getrlimit (RLIMIT_NOFILE, &rlimit) == 0 + ? rlimit.rlim_cur + : OPEN_MAX) + - 3); + + if (e == LONGINT_OK) + { + nmerge = n; + if (nmerge != n) + e = LONGINT_OVERFLOW; + else + { + if (nmerge < 2) + { + error (0, 0, _("invalid --%s argument %s"), + long_options[oi].name, quote(s)); + error (SORT_FAILURE, 0, + _("minimum --%s argument is %s"), + long_options[oi].name, quote("2")); + } + else if (max_nmerge < nmerge) + { + e = LONGINT_OVERFLOW; + } + else + return; + } + } + + if (e == LONGINT_OVERFLOW) + { + char max_nmerge_buf[INT_BUFSIZE_BOUND (unsigned int)]; + error (0, 0, _("--%s argument %s too large"), + long_options[oi].name, quote(s)); + error (SORT_FAILURE, 0, + _("maximum --%s argument with current rlimit is %s"), + long_options[oi].name, + uinttostr (max_nmerge, max_nmerge_buf)); + } + else + xstrtol_fatal (e, oi, c, long_options, s); +} + +/* Specify the amount of main memory to use when sorting. */ +static void +specify_sort_size (int oi, char c, char const *s) +{ + uintmax_t n; + char *suffix; + enum strtol_error e = xstrtoumax (s, &suffix, 10, &n, "EgGkKmMPtTYZ"); + + /* The default unit is KiB. */ + if (e == LONGINT_OK && ISDIGIT (suffix[-1])) + { + if (n <= UINTMAX_MAX / 1024) + n *= 1024; + else + e = LONGINT_OVERFLOW; + } + + /* A 'b' suffix means bytes; a '%' suffix means percent of memory. */ + if (e == LONGINT_INVALID_SUFFIX_CHAR && ISDIGIT (suffix[-1]) && ! suffix[1]) + switch (suffix[0]) + { + case 'b': + e = LONGINT_OK; + break; + + case '%': + { + double mem = physmem_total () * n / 100; + + /* Use "<", not "<=", to avoid problems with rounding. */ + if (mem < UINTMAX_MAX) + { + n = mem; + e = LONGINT_OK; + } + else + e = LONGINT_OVERFLOW; + } + break; + } + + if (e == LONGINT_OK) + { + /* If multiple sort sizes are specified, take the maximum, so + that option order does not matter. */ + if (n < sort_size) + return; + + sort_size = n; + if (sort_size == n) + { + sort_size = MAX (sort_size, MIN_SORT_SIZE); + return; + } + + e = LONGINT_OVERFLOW; + } + + xstrtol_fatal (e, oi, c, long_options, s); +} + +/* Return the default sort size. */ +static size_t +default_sort_size (void) +{ + /* Let MEM be available memory or 1/8 of total memory, whichever + is greater. */ + double avail = physmem_available (); + double total = physmem_total (); + double mem = MAX (avail, total / 8); + struct rlimit rlimit; + + /* Let SIZE be MEM, but no more than the maximum object size or + system resource limits. Avoid the MIN macro here, as it is not + quite right when only one argument is floating point. Don't + bother to check for values like RLIM_INFINITY since in practice + they are not much less than SIZE_MAX. */ + size_t size = SIZE_MAX; + if (mem < size) + size = mem; + if (getrlimit (RLIMIT_DATA, &rlimit) == 0 && rlimit.rlim_cur < size) + size = rlimit.rlim_cur; +#ifdef RLIMIT_AS + if (getrlimit (RLIMIT_AS, &rlimit) == 0 && rlimit.rlim_cur < size) + size = rlimit.rlim_cur; +#endif + + /* Leave a large safety margin for the above limits, as failure can + occur when they are exceeded. */ + size /= 2; + +#ifdef RLIMIT_RSS + /* Leave a 1/16 margin for RSS to leave room for code, stack, etc. + Exceeding RSS is not fatal, but can be quite slow. */ + if (getrlimit (RLIMIT_RSS, &rlimit) == 0 && rlimit.rlim_cur / 16 * 15 < size) + size = rlimit.rlim_cur / 16 * 15; +#endif + + /* Use no less than the minimum. */ + return MAX (size, MIN_SORT_SIZE); +} + +/* Return the sort buffer size to use with the input files identified + by FPS and FILES, which are alternate names of the same files. + NFILES gives the number of input files; NFPS may be less. Assume + that each input line requires LINE_BYTES extra bytes' worth of line + information. Do not exceed the size bound specified by the user + (or a default size bound, if the user does not specify one). */ + +static size_t +sort_buffer_size (FILE *const *fps, size_t nfps, + char *const *files, size_t nfiles, + size_t line_bytes) +{ + /* A bound on the input size. If zero, the bound hasn't been + determined yet. */ + static size_t size_bound; + + /* In the worst case, each input byte is a newline. */ + size_t worst_case_per_input_byte = line_bytes + 1; + + /* Keep enough room for one extra input line and an extra byte. + This extra room might be needed when preparing to read EOF. */ + size_t size = worst_case_per_input_byte + 1; + + size_t i; + + for (i = 0; i < nfiles; i++) + { + struct stat st; + off_t file_size; + size_t worst_case; + + if ((i < nfps ? fstat (fileno (fps[i]), &st) + : STREQ (files[i], "-") ? fstat (STDIN_FILENO, &st) + : stat (files[i], &st)) + != 0) + die (_("stat failed"), files[i]); + + if (S_ISREG (st.st_mode)) + file_size = st.st_size; + else + { + /* The file has unknown size. If the user specified a sort + buffer size, use that; otherwise, guess the size. */ + if (sort_size) + return sort_size; + file_size = INPUT_FILE_SIZE_GUESS; + } + + if (! size_bound) + { + size_bound = sort_size; + if (! size_bound) + size_bound = default_sort_size (); + } + + /* Add the amount of memory needed to represent the worst case + where the input consists entirely of newlines followed by a + single non-newline. Check for overflow. */ + worst_case = file_size * worst_case_per_input_byte + 1; + if (file_size != worst_case / worst_case_per_input_byte + || size_bound - size <= worst_case) + return size_bound; + size += worst_case; + } + + return size; +} + +/* Initialize BUF. Reserve LINE_BYTES bytes for each line; LINE_BYTES + must be at least sizeof (struct line). Allocate ALLOC bytes + initially. */ + +static void +initbuf (struct buffer *buf, size_t line_bytes, size_t alloc) +{ + /* Ensure that the line array is properly aligned. If the desired + size cannot be allocated, repeatedly halve it until allocation + succeeds. The smaller allocation may hurt overall performance, + but that's better than failing. */ + for (;;) + { + alloc += sizeof (struct line) - alloc % sizeof (struct line); + buf->buf = malloc (alloc); + if (buf->buf) + break; + alloc /= 2; + if (alloc <= line_bytes + 1) + xalloc_die (); + } + + buf->line_bytes = line_bytes; + buf->alloc = alloc; + buf->used = buf->left = buf->nlines = 0; + buf->eof = false; +} + +/* Return one past the limit of the line array. */ + +static inline struct line * +buffer_linelim (struct buffer const *buf) +{ + return (struct line *) (buf->buf + buf->alloc); +} + +/* Return a pointer to the first character of the field specified + by KEY in LINE. */ + +static char * +begfield (const struct line *line, const struct keyfield *key) +{ + char *ptr = line->text, *lim = ptr + line->length - 1; + size_t sword = key->sword; + size_t schar = key->schar; + + /* The leading field separator itself is included in a field when -t + is absent. */ + + if (tab != TAB_DEFAULT) + while (ptr < lim && sword--) + { + while (ptr < lim && *ptr != tab) + ++ptr; + if (ptr < lim) + ++ptr; + } + else + while (ptr < lim && sword--) + { + while (ptr < lim && blanks[to_uchar (*ptr)]) + ++ptr; + while (ptr < lim && !blanks[to_uchar (*ptr)]) + ++ptr; + } + + /* If we're ignoring leading blanks when computing the Start + of the field, skip past them here. */ + if (key->skipsblanks) + while (ptr < lim && blanks[to_uchar (*ptr)]) + ++ptr; + + /* Advance PTR by SCHAR (if possible), but no further than LIM. */ + ptr = MIN (lim, ptr + schar); + + return ptr; +} + +/* Return the limit of (a pointer to the first character after) the field + in LINE specified by KEY. */ + +static char * +limfield (const struct line *line, const struct keyfield *key) +{ + char *ptr = line->text, *lim = ptr + line->length - 1; + size_t eword = key->eword, echar = key->echar; + + if (echar == 0) + eword++; /* Skip all of end field. */ + + /* Move PTR past EWORD fields or to one past the last byte on LINE, + whichever comes first. If there are more than EWORD fields, leave + PTR pointing at the beginning of the field having zero-based index, + EWORD. If a delimiter character was specified (via -t), then that + `beginning' is the first character following the delimiting TAB. + Otherwise, leave PTR pointing at the first `blank' character after + the preceding field. */ + if (tab != TAB_DEFAULT) + while (ptr < lim && eword--) + { + while (ptr < lim && *ptr != tab) + ++ptr; + if (ptr < lim && (eword | echar)) + ++ptr; + } + else + while (ptr < lim && eword--) + { + while (ptr < lim && blanks[to_uchar (*ptr)]) + ++ptr; + while (ptr < lim && !blanks[to_uchar (*ptr)]) + ++ptr; + } + +#ifdef POSIX_UNSPECIFIED + /* The following block of code makes GNU sort incompatible with + standard Unix sort, so it's ifdef'd out for now. + The POSIX spec isn't clear on how to interpret this. + FIXME: request clarification. + + From: kwzh@gnu.ai.mit.edu (Karl Heuer) + Date: Thu, 30 May 96 12:20:41 -0400 + [Translated to POSIX 1003.1-2001 terminology by Paul Eggert.] + + [...]I believe I've found another bug in `sort'. + + $ cat /tmp/sort.in + a b c 2 d + pq rs 1 t + $ textutils-1.15/src/sort -k1.7,1.7 skipeblanks) + while (ptr < lim && blanks[to_uchar (*ptr)]) + ++ptr; + + /* Advance PTR by ECHAR (if possible), but no further than LIM. */ + ptr = MIN (lim, ptr + echar); + } + + return ptr; +} + +/* Fill BUF reading from FP, moving buf->left bytes from the end + of buf->buf to the beginning first. If EOF is reached and the + file wasn't terminated by a newline, supply one. Set up BUF's line + table too. FILE is the name of the file corresponding to FP. + Return true if some input was read. */ + +static bool +fillbuf (struct buffer *buf, FILE *fp, char const *file) +{ + struct keyfield const *key = keylist; + char eol = eolchar; + size_t line_bytes = buf->line_bytes; + size_t mergesize = merge_buffer_size - MIN_MERGE_BUFFER_SIZE; + + if (buf->eof) + return false; + + if (buf->used != buf->left) + { + memmove (buf->buf, buf->buf + buf->used - buf->left, buf->left); + buf->used = buf->left; + buf->nlines = 0; + } + + for (;;) + { + char *ptr = buf->buf + buf->used; + struct line *linelim = buffer_linelim (buf); + struct line *line = linelim - buf->nlines; + size_t avail = (char *) linelim - buf->nlines * line_bytes - ptr; + char *line_start = buf->nlines ? line->text + line->length : buf->buf; + + while (line_bytes + 1 < avail) + { + /* Read as many bytes as possible, but do not read so many + bytes that there might not be enough room for the + corresponding line array. The worst case is when the + rest of the input file consists entirely of newlines, + except that the last byte is not a newline. */ + size_t readsize = (avail - 1) / (line_bytes + 1); + size_t bytes_read = fread (ptr, 1, readsize, fp); + char *ptrlim = ptr + bytes_read; + char *p; + avail -= bytes_read; + + if (bytes_read != readsize) + { + if (ferror (fp)) + die (_("read failed"), file); + if (feof (fp)) + { + buf->eof = true; + if (buf->buf == ptrlim) + return false; + if (ptrlim[-1] != eol) + *ptrlim++ = eol; + } + } + + /* Find and record each line in the just-read input. */ + while ((p = memchr (ptr, eol, ptrlim - ptr))) + { + ptr = p + 1; + line--; + line->text = line_start; + line->length = ptr - line_start; + mergesize = MAX (mergesize, line->length); + avail -= line_bytes; + + if (key) + { + /* Precompute the position of the first key for + efficiency. */ + line->keylim = (key->eword == SIZE_MAX + ? p + : limfield (line, key)); + + if (key->sword != SIZE_MAX) + line->keybeg = begfield (line, key); + else + { + if (key->skipsblanks) + while (blanks[to_uchar (*line_start)]) + line_start++; + line->keybeg = line_start; + } + } + + line_start = ptr; + } + + ptr = ptrlim; + if (buf->eof) + break; + } + + buf->used = ptr - buf->buf; + buf->nlines = buffer_linelim (buf) - line; + if (buf->nlines != 0) + { + buf->left = ptr - line_start; + merge_buffer_size = mergesize + MIN_MERGE_BUFFER_SIZE; + return true; + } + + { + /* The current input line is too long to fit in the buffer. + Double the buffer size and try again, keeping it properly + aligned. */ + size_t line_alloc = buf->alloc / sizeof (struct line); + buf->buf = x2nrealloc (buf->buf, &line_alloc, sizeof (struct line)); + buf->alloc = line_alloc * sizeof (struct line); + } + } +} + +/* Compare strings A and B as numbers without explicitly converting them to + machine numbers. Comparatively slow for short strings, but asymptotically + hideously fast. */ + +static int +numcompare (const char *a, const char *b) +{ + while (blanks[to_uchar (*a)]) + a++; + while (blanks[to_uchar (*b)]) + b++; + + return strnumcmp (a, b, decimal_point, thousands_sep); +} + +static int +general_numcompare (const char *sa, const char *sb) +{ + /* FIXME: add option to warn about failed conversions. */ + /* FIXME: maybe add option to try expensive FP conversion + only if A and B can't be compared more cheaply/accurately. */ + + char *ea; + char *eb; + double a = strtod (sa, &ea); + double b = strtod (sb, &eb); + + /* Put conversion errors at the start of the collating sequence. */ + if (sa == ea) + return sb == eb ? 0 : -1; + if (sb == eb) + return 1; + + /* Sort numbers in the usual way, where -0 == +0. Put NaNs after + conversion errors but before numbers; sort them by internal + bit-pattern, for lack of a more portable alternative. */ + return (a < b ? -1 + : a > b ? 1 + : a == b ? 0 + : b == b ? -1 + : a == a ? 1 + : memcmp ((char *) &a, (char *) &b, sizeof a)); +} + +/* Return an integer in 1..12 of the month name MONTH with length LEN. + Return 0 if the name in S is not recognized. */ + +static int +getmonth (char const *month, size_t len) +{ + size_t lo = 0; + size_t hi = MONTHS_PER_YEAR; + char const *monthlim = month + len; + + for (;;) + { + if (month == monthlim) + return 0; + if (!blanks[to_uchar (*month)]) + break; + ++month; + } + + do + { + size_t ix = (lo + hi) / 2; + char const *m = month; + char const *n = monthtab[ix].name; + + for (;; m++, n++) + { + if (!*n) + return monthtab[ix].val; + if (m == monthlim || fold_toupper[to_uchar (*m)] < to_uchar (*n)) + { + hi = ix; + break; + } + else if (fold_toupper[to_uchar (*m)] > to_uchar (*n)) + { + lo = ix + 1; + break; + } + } + } + while (lo < hi); + + return 0; +} + +/* A source of random data. */ +static struct randread_source *randread_source; + +/* Return the Ith randomly-generated state. The caller must invoke + random_state (H) for all H less than I before invoking random_state + (I). */ + +static struct md5_ctx +random_state (size_t i) +{ + /* An array of states resulting from the random data, and counts of + its used and allocated members. */ + static struct md5_ctx *state; + static size_t used; + static size_t allocated; + + struct md5_ctx *s = &state[i]; + + if (used <= i) + { + unsigned char buf[MD5_DIGEST_SIZE]; + + used++; + + if (allocated <= i) + { + state = X2NREALLOC (state, &allocated); + s = &state[i]; + } + + randread (randread_source, buf, sizeof buf); + md5_init_ctx (s); + md5_process_bytes (buf, sizeof buf, s); + } + + return *s; +} + +/* Compare the hashes of TEXTA with length LENGTHA to those of TEXTB + with length LENGTHB. Return negative if less, zero if equal, + positive if greater. */ + +static int +cmp_hashes (char const *texta, size_t lena, + char const *textb, size_t lenb) +{ + /* Try random hashes until a pair of hashes disagree. But if the + first pair of random hashes agree, check whether the keys are + identical and if so report no difference. */ + int diff; + size_t i; + for (i = 0; ; i++) + { + uint32_t dig[2][MD5_DIGEST_SIZE / sizeof (uint32_t)]; + struct md5_ctx s[2]; + s[0] = s[1] = random_state (i); + md5_process_bytes (texta, lena, &s[0]); md5_finish_ctx (&s[0], dig[0]); + md5_process_bytes (textb, lenb, &s[1]); md5_finish_ctx (&s[1], dig[1]); + diff = memcmp (dig[0], dig[1], sizeof dig[0]); + if (diff != 0) + break; + if (i == 0 && lena == lenb && memcmp (texta, textb, lena) == 0) + break; + } + + return diff; +} + +/* Compare the keys TEXTA (of length LENA) and TEXTB (of length LENB) + using one or more random hash functions. */ + +static int +compare_random (char *restrict texta, size_t lena, + char *restrict textb, size_t lenb) +{ + int diff; + + if (! hard_LC_COLLATE) + diff = cmp_hashes (texta, lena, textb, lenb); + else + { + /* Transform the text into the basis of comparison, so that byte + strings that would otherwise considered to be equal are + considered equal here even if their bytes differ. */ + + char *buf = NULL; + char stackbuf[4000]; + size_t tlena = xmemxfrm (stackbuf, sizeof stackbuf, texta, lena); + bool a_fits = tlena <= sizeof stackbuf; + size_t tlenb = xmemxfrm ((a_fits ? stackbuf + tlena : NULL), + (a_fits ? sizeof stackbuf - tlena : 0), + textb, lenb); + + if (a_fits && tlena + tlenb <= sizeof stackbuf) + buf = stackbuf; + else + { + /* Adding 1 to the buffer size lets xmemxfrm run a bit + faster by avoiding the need for an extra buffer copy. */ + buf = xmalloc (tlena + tlenb + 1); + xmemxfrm (buf, tlena + 1, texta, lena); + xmemxfrm (buf + tlena, tlenb + 1, textb, lenb); + } + + diff = cmp_hashes (buf, tlena, buf + tlena, tlenb); + + if (buf != stackbuf) + free (buf); + } + + return diff; +} + +/* Compare the keys TEXTA (of length LENA) and TEXTB (of length LENB) + using filevercmp. See lib/filevercmp.h for function description. */ + +static int +compare_version (char *restrict texta, size_t lena, + char *restrict textb, size_t lenb) +{ + int diff; + + /* It is necessary to save the character after the end of the field. + "filevercmp" works with NUL terminated strings. Our blocks of + text are not necessarily terminated with a NUL byte. */ + char sv_a = texta[lena]; + char sv_b = textb[lenb]; + + texta[lena] = '\0'; + textb[lenb] = '\0'; + + diff = filevercmp (texta, textb); + + texta[lena] = sv_a; + textb[lenb] = sv_b; + + return diff; +} + +/* Compare two lines A and B trying every key in sequence until there + are no more keys or a difference is found. */ + +static int +keycompare (const struct line *a, const struct line *b) +{ + struct keyfield const *key = keylist; + + /* For the first iteration only, the key positions have been + precomputed for us. */ + char *texta = a->keybeg; + char *textb = b->keybeg; + char *lima = a->keylim; + char *limb = b->keylim; + + int diff; + + for (;;) + { + char const *translate = key->translate; + bool const *ignore = key->ignore; + + /* Find the lengths. */ + size_t lena = lima <= texta ? 0 : lima - texta; + size_t lenb = limb <= textb ? 0 : limb - textb; + + /* Actually compare the fields. */ + + if (key->random) + diff = compare_random (texta, lena, textb, lenb); + else if (key->numeric | key->general_numeric) + { + char savea = *lima, saveb = *limb; + + *lima = *limb = '\0'; + diff = ((key->numeric ? numcompare : general_numcompare) + (texta, textb)); + *lima = savea, *limb = saveb; + } + else if (key->version) + diff = compare_version (texta, lena, textb, lenb); + else if (key->month) + diff = getmonth (texta, lena) - getmonth (textb, lenb); + /* Sorting like this may become slow, so in a simple locale the user + can select a faster sort that is similar to ascii sort. */ + else if (hard_LC_COLLATE) + { + if (ignore || translate) + { + char buf[4000]; + size_t size = lena + 1 + lenb + 1; + char *copy_a = (size <= sizeof buf ? buf : xmalloc (size)); + char *copy_b = copy_a + lena + 1; + size_t new_len_a, new_len_b, i; + + /* Ignore and/or translate chars before comparing. */ + for (new_len_a = new_len_b = i = 0; i < MAX (lena, lenb); i++) + { + if (i < lena) + { + copy_a[new_len_a] = (translate + ? translate[to_uchar (texta[i])] + : texta[i]); + if (!ignore || !ignore[to_uchar (texta[i])]) + ++new_len_a; + } + if (i < lenb) + { + copy_b[new_len_b] = (translate + ? translate[to_uchar (textb[i])] + : textb [i]); + if (!ignore || !ignore[to_uchar (textb[i])]) + ++new_len_b; + } + } + + diff = xmemcoll (copy_a, new_len_a, copy_b, new_len_b); + + if (sizeof buf < size) + free (copy_a); + } + else if (lena == 0) + diff = - NONZERO (lenb); + else if (lenb == 0) + goto greater; + else + diff = xmemcoll (texta, lena, textb, lenb); + } + else if (ignore) + { +#define CMP_WITH_IGNORE(A, B) \ + do \ + { \ + for (;;) \ + { \ + while (texta < lima && ignore[to_uchar (*texta)]) \ + ++texta; \ + while (textb < limb && ignore[to_uchar (*textb)]) \ + ++textb; \ + if (! (texta < lima && textb < limb)) \ + break; \ + diff = to_uchar (A) - to_uchar (B); \ + if (diff) \ + goto not_equal; \ + ++texta; \ + ++textb; \ + } \ + \ + diff = (texta < lima) - (textb < limb); \ + } \ + while (0) + + if (translate) + CMP_WITH_IGNORE (translate[to_uchar (*texta)], + translate[to_uchar (*textb)]); + else + CMP_WITH_IGNORE (*texta, *textb); + } + else if (lena == 0) + diff = - NONZERO (lenb); + else if (lenb == 0) + goto greater; + else + { + if (translate) + { + while (texta < lima && textb < limb) + { + diff = (to_uchar (translate[to_uchar (*texta++)]) + - to_uchar (translate[to_uchar (*textb++)])); + if (diff) + goto not_equal; + } + } + else + { + diff = memcmp (texta, textb, MIN (lena, lenb)); + if (diff) + goto not_equal; + } + diff = lena < lenb ? -1 : lena != lenb; + } + + if (diff) + goto not_equal; + + key = key->next; + if (! key) + break; + + /* Find the beginning and limit of the next field. */ + if (key->eword != SIZE_MAX) + lima = limfield (a, key), limb = limfield (b, key); + else + lima = a->text + a->length - 1, limb = b->text + b->length - 1; + + if (key->sword != SIZE_MAX) + texta = begfield (a, key), textb = begfield (b, key); + else + { + texta = a->text, textb = b->text; + if (key->skipsblanks) + { + while (texta < lima && blanks[to_uchar (*texta)]) + ++texta; + while (textb < limb && blanks[to_uchar (*textb)]) + ++textb; + } + } + } + + return 0; + + greater: + diff = 1; + not_equal: + return key->reverse ? -diff : diff; +} + +/* Compare two lines A and B, returning negative, zero, or positive + depending on whether A compares less than, equal to, or greater than B. */ + +static int +compare (const struct line *a, const struct line *b) +{ + int diff; + size_t alen, blen; + + /* First try to compare on the specified keys (if any). + The only two cases with no key at all are unadorned sort, + and unadorned sort -r. */ + if (keylist) + { + diff = keycompare (a, b); + if (diff | unique | stable) + return diff; + } + + /* If the keys all compare equal (or no keys were specified) + fall through to the default comparison. */ + alen = a->length - 1, blen = b->length - 1; + + if (alen == 0) + diff = - NONZERO (blen); + else if (blen == 0) + diff = 1; + else if (hard_LC_COLLATE) + diff = xmemcoll (a->text, alen, b->text, blen); + else if (! (diff = memcmp (a->text, b->text, MIN (alen, blen)))) + diff = alen < blen ? -1 : alen != blen; + + return reverse ? -diff : diff; +} + +/* Check that the lines read from FILE_NAME come in order. Return + true if they are in order. If CHECKONLY == 'c', also print a + diagnostic (FILE_NAME, line number, contents of line) to stderr if + they are not in order. */ + +static bool +check (char const *file_name, char checkonly) +{ + FILE *fp = xfopen (file_name, "r"); + struct buffer buf; /* Input buffer. */ + struct line temp; /* Copy of previous line. */ + size_t alloc = 0; + uintmax_t line_number = 0; + struct keyfield const *key = keylist; + bool nonunique = ! unique; + bool ordered = true; + + initbuf (&buf, sizeof (struct line), + MAX (merge_buffer_size, sort_size)); + temp.text = NULL; + + while (fillbuf (&buf, fp, file_name)) + { + struct line const *line = buffer_linelim (&buf); + struct line const *linebase = line - buf.nlines; + + /* Make sure the line saved from the old buffer contents is + less than or equal to the first line of the new buffer. */ + if (alloc && nonunique <= compare (&temp, line - 1)) + { + found_disorder: + { + if (checkonly == 'c') + { + struct line const *disorder_line = line - 1; + uintmax_t disorder_line_number = + buffer_linelim (&buf) - disorder_line + line_number; + char hr_buf[INT_BUFSIZE_BOUND (uintmax_t)]; + fprintf (stderr, _("%s: %s:%s: disorder: "), + program_name, file_name, + umaxtostr (disorder_line_number, hr_buf)); + write_bytes (disorder_line->text, disorder_line->length, + stderr, _("standard error")); + } + + ordered = false; + break; + } + } + + /* Compare each line in the buffer with its successor. */ + while (linebase < --line) + if (nonunique <= compare (line, line - 1)) + goto found_disorder; + + line_number += buf.nlines; + + /* Save the last line of the buffer. */ + if (alloc < line->length) + { + do + { + alloc *= 2; + if (! alloc) + { + alloc = line->length; + break; + } + } + while (alloc < line->length); + + temp.text = xrealloc (temp.text, alloc); + } + memcpy (temp.text, line->text, line->length); + temp.length = line->length; + if (key) + { + temp.keybeg = temp.text + (line->keybeg - line->text); + temp.keylim = temp.text + (line->keylim - line->text); + } + } + + xfclose (fp, file_name); + free (buf.buf); + free (temp.text); + return ordered; +} + +/* Open FILES (there are NFILES of them) and store the resulting array + of stream pointers into (*PFPS). Allocate the array. Return the + number of successfully opened files, setting errno if this value is + less than NFILES. */ + +static size_t +open_input_files (struct sortfile *files, size_t nfiles, FILE ***pfps) +{ + FILE **fps = *pfps = xnmalloc (nfiles, sizeof *fps); + int i; + + /* Open as many input files as we can. */ + for (i = 0; i < nfiles; i++) + { + fps[i] = (files[i].pid + ? open_temp (files[i].name, files[i].pid) + : stream_open (files[i].name, "r")); + if (!fps[i]) + break; + } + + return i; +} + +/* Merge lines from FILES onto OFP. NTEMPS is the number of temporary + files (all of which are at the start of the FILES array), and + NFILES is the number of files; 0 <= NTEMPS <= NFILES <= NMERGE. + FPS is the vector of open stream corresponding to the files. + Close input and output streams before returning. + OUTPUT_FILE gives the name of the output file. If it is NULL, + the output file is standard output. */ + +static void +mergefps (struct sortfile *files, size_t ntemps, size_t nfiles, + FILE *ofp, char const *output_file, FILE **fps) +{ + struct buffer *buffer = xnmalloc (nfiles, sizeof *buffer); + /* Input buffers for each file. */ + struct line saved; /* Saved line storage for unique check. */ + struct line const *savedline = NULL; + /* &saved if there is a saved line. */ + size_t savealloc = 0; /* Size allocated for the saved line. */ + struct line const **cur = xnmalloc (nfiles, sizeof *cur); + /* Current line in each line table. */ + struct line const **base = xnmalloc (nfiles, sizeof *base); + /* Base of each line table. */ + size_t *ord = xnmalloc (nfiles, sizeof *ord); + /* Table representing a permutation of fps, + such that cur[ord[0]] is the smallest line + and will be next output. */ + size_t i; + size_t j; + size_t t; + struct keyfield const *key = keylist; + saved.text = NULL; + + /* Read initial lines from each input file. */ + for (i = 0; i < nfiles; ) + { + initbuf (&buffer[i], sizeof (struct line), + MAX (merge_buffer_size, sort_size / nfiles)); + if (fillbuf (&buffer[i], fps[i], files[i].name)) + { + struct line const *linelim = buffer_linelim (&buffer[i]); + cur[i] = linelim - 1; + base[i] = linelim - buffer[i].nlines; + i++; + } + else + { + /* fps[i] is empty; eliminate it from future consideration. */ + xfclose (fps[i], files[i].name); + if (i < ntemps) + { + ntemps--; + zaptemp (files[i].name); + } + free (buffer[i].buf); + --nfiles; + for (j = i; j < nfiles; ++j) + { + files[j] = files[j + 1]; + fps[j] = fps[j + 1]; + } + } + } + + /* Set up the ord table according to comparisons among input lines. + Since this only reorders two items if one is strictly greater than + the other, it is stable. */ + for (i = 0; i < nfiles; ++i) + ord[i] = i; + for (i = 1; i < nfiles; ++i) + if (0 < compare (cur[ord[i - 1]], cur[ord[i]])) + t = ord[i - 1], ord[i - 1] = ord[i], ord[i] = t, i = 0; + + /* Repeatedly output the smallest line until no input remains. */ + while (nfiles) + { + struct line const *smallest = cur[ord[0]]; + + /* If uniquified output is turned on, output only the first of + an identical series of lines. */ + if (unique) + { + if (savedline && compare (savedline, smallest)) + { + savedline = NULL; + write_bytes (saved.text, saved.length, ofp, output_file); + } + if (!savedline) + { + savedline = &saved; + if (savealloc < smallest->length) + { + do + if (! savealloc) + { + savealloc = smallest->length; + break; + } + while ((savealloc *= 2) < smallest->length); + + saved.text = xrealloc (saved.text, savealloc); + } + saved.length = smallest->length; + memcpy (saved.text, smallest->text, saved.length); + if (key) + { + saved.keybeg = + saved.text + (smallest->keybeg - smallest->text); + saved.keylim = + saved.text + (smallest->keylim - smallest->text); + } + } + } + else + write_bytes (smallest->text, smallest->length, ofp, output_file); + + /* Check if we need to read more lines into core. */ + if (base[ord[0]] < smallest) + cur[ord[0]] = smallest - 1; + else + { + if (fillbuf (&buffer[ord[0]], fps[ord[0]], files[ord[0]].name)) + { + struct line const *linelim = buffer_linelim (&buffer[ord[0]]); + cur[ord[0]] = linelim - 1; + base[ord[0]] = linelim - buffer[ord[0]].nlines; + } + else + { + /* We reached EOF on fps[ord[0]]. */ + for (i = 1; i < nfiles; ++i) + if (ord[i] > ord[0]) + --ord[i]; + --nfiles; + xfclose (fps[ord[0]], files[ord[0]].name); + if (ord[0] < ntemps) + { + ntemps--; + zaptemp (files[ord[0]].name); + } + free (buffer[ord[0]].buf); + for (i = ord[0]; i < nfiles; ++i) + { + fps[i] = fps[i + 1]; + files[i] = files[i + 1]; + buffer[i] = buffer[i + 1]; + cur[i] = cur[i + 1]; + base[i] = base[i + 1]; + } + for (i = 0; i < nfiles; ++i) + ord[i] = ord[i + 1]; + continue; + } + } + + /* The new line just read in may be larger than other lines + already in main memory; push it back in the queue until we + encounter a line larger than it. Optimize for the common + case where the new line is smallest. */ + { + size_t lo = 1; + size_t hi = nfiles; + size_t probe = lo; + size_t ord0 = ord[0]; + size_t count_of_smaller_lines; + + while (lo < hi) + { + int cmp = compare (cur[ord0], cur[ord[probe]]); + if (cmp < 0 || (cmp == 0 && ord0 < ord[probe])) + hi = probe; + else + lo = probe + 1; + probe = (lo + hi) / 2; + } + + count_of_smaller_lines = lo - 1; + for (j = 0; j < count_of_smaller_lines; j++) + ord[j] = ord[j + 1]; + ord[count_of_smaller_lines] = ord0; + } + + /* Free up some resources every once in a while. */ + if (MAX_PROCS_BEFORE_REAP < nprocs) + reap_some (); + } + + if (unique && savedline) + { + write_bytes (saved.text, saved.length, ofp, output_file); + free (saved.text); + } + + xfclose (ofp, output_file); + free(fps); + free(buffer); + free(ord); + free(base); + free(cur); +} + +/* Merge lines from FILES onto OFP. NTEMPS is the number of temporary + files (all of which are at the start of the FILES array), and + NFILES is the number of files; 0 <= NTEMPS <= NFILES <= NMERGE. + Close input and output files before returning. + OUTPUT_FILE gives the name of the output file. + + Return the number of files successfully merged. This number can be + less than NFILES if we ran low on file descriptors, but in this + case it is never less than 2. */ + +static size_t +mergefiles (struct sortfile *files, size_t ntemps, size_t nfiles, + FILE *ofp, char const *output_file) +{ + FILE **fps; + size_t nopened = open_input_files (files, nfiles, &fps); + if (nopened < nfiles && nopened < 2) + die (_("open failed"), files[nopened].name); + mergefps (files, ntemps, nopened, ofp, output_file, fps); + return nopened; +} + +/* Merge into T the two sorted arrays of lines LO (with NLO members) + and HI (with NHI members). T, LO, and HI point just past their + respective arrays, and the arrays are in reverse order. NLO and + NHI must be positive, and HI - NHI must equal T - (NLO + NHI). */ + +static inline void +mergelines (struct line *t, + struct line const *lo, size_t nlo, + struct line const *hi, size_t nhi) +{ + for (;;) + if (compare (lo - 1, hi - 1) <= 0) + { + *--t = *--lo; + if (! --nlo) + { + /* HI - NHI equalled T - (NLO + NHI) when this function + began. Therefore HI must equal T now, and there is no + need to copy from HI to T. */ + return; + } + } + else + { + *--t = *--hi; + if (! --nhi) + { + do + *--t = *--lo; + while (--nlo); + + return; + } + } +} + +/* Sort the array LINES with NLINES members, using TEMP for temporary space. + NLINES must be at least 2. + The input and output arrays are in reverse order, and LINES and + TEMP point just past the end of their respective arrays. + + Use a recursive divide-and-conquer algorithm, in the style + suggested by Knuth volume 3 (2nd edition), exercise 5.2.4-23. Use + the optimization suggested by exercise 5.2.4-10; this requires room + for only 1.5*N lines, rather than the usual 2*N lines. Knuth + writes that this memory optimization was originally published by + D. A. Bell, Comp J. 1 (1958), 75. */ + +static void +sortlines (struct line *lines, size_t nlines, struct line *temp) +{ + if (nlines == 2) + { + if (0 < compare (&lines[-1], &lines[-2])) + { + struct line tmp = lines[-1]; + lines[-1] = lines[-2]; + lines[-2] = tmp; + } + } + else + { + size_t nlo = nlines / 2; + size_t nhi = nlines - nlo; + struct line *lo = lines; + struct line *hi = lines - nlo; + struct line *sorted_lo = temp; + + sortlines (hi, nhi, temp); + if (1 < nlo) + sortlines_temp (lo, nlo, sorted_lo); + else + sorted_lo[-1] = lo[-1]; + + mergelines (lines, sorted_lo, nlo, hi, nhi); + } +} + +/* Like sortlines (LINES, NLINES, TEMP), except output into TEMP + rather than sorting in place. */ + +static void +sortlines_temp (struct line *lines, size_t nlines, struct line *temp) +{ + if (nlines == 2) + { + /* Declare `swap' as int, not bool, to work around a bug + + in the IBM xlc 6.0.0.0 compiler in 64-bit mode. */ + int swap = (0 < compare (&lines[-1], &lines[-2])); + temp[-1] = lines[-1 - swap]; + temp[-2] = lines[-2 + swap]; + } + else + { + size_t nlo = nlines / 2; + size_t nhi = nlines - nlo; + struct line *lo = lines; + struct line *hi = lines - nlo; + struct line *sorted_hi = temp - nlo; + + sortlines_temp (hi, nhi, sorted_hi); + if (1 < nlo) + sortlines (lo, nlo, temp); + + mergelines (temp, lo, nlo, sorted_hi, nhi); + } +} + +/* Scan through FILES[NTEMPS .. NFILES-1] looking for a file that is + the same as OUTFILE. If found, merge the found instances (and perhaps + some other files) into a temporary file so that it can in turn be + merged into OUTFILE without destroying OUTFILE before it is completely + read. Return the new value of NFILES, which differs from the old if + some merging occurred. + + This test ensures that an otherwise-erroneous use like + "sort -m -o FILE ... FILE ..." copies FILE before writing to it. + It's not clear that POSIX requires this nicety. + Detect common error cases, but don't try to catch obscure cases like + "cat ... FILE ... | sort -m -o FILE" + where traditional "sort" doesn't copy the input and where + people should know that they're getting into trouble anyway. + Catching these obscure cases would slow down performance in + common cases. */ + +static size_t +avoid_trashing_input (struct sortfile *files, size_t ntemps, + size_t nfiles, char const *outfile) +{ + size_t i; + bool got_outstat = false; + struct stat outstat; + + for (i = ntemps; i < nfiles; i++) + { + bool is_stdin = STREQ (files[i].name, "-"); + bool same; + struct stat instat; + + if (outfile && STREQ (outfile, files[i].name) && !is_stdin) + same = true; + else + { + if (! got_outstat) + { + if ((outfile + ? stat (outfile, &outstat) + : fstat (STDOUT_FILENO, &outstat)) + != 0) + break; + got_outstat = true; + } + + same = (((is_stdin + ? fstat (STDIN_FILENO, &instat) + : stat (files[i].name, &instat)) + == 0) + && SAME_INODE (instat, outstat)); + } + + if (same) + { + FILE *tftp; + pid_t pid; + char *temp = create_temp (&tftp, &pid); + size_t num_merged = 0; + while (i + num_merged < nfiles) + { + num_merged += mergefiles (&files[i], 0, nfiles - i, tftp, temp); + files[i].name = temp; + files[i].pid = pid; + + memmove(&files[i], &files[i + num_merged], + num_merged * sizeof *files); + ntemps += 1; + nfiles -= num_merged - 1;; + i += num_merged; + } + } + } + + return nfiles; +} + +/* Merge the input FILES. NTEMPS is the number of files at the + start of FILES that are temporary; it is zero at the top level. + NFILES is the total number of files. Put the output in + OUTPUT_FILE; a null OUTPUT_FILE stands for standard output. */ + +static void +merge (struct sortfile *files, size_t ntemps, size_t nfiles, + char const *output_file) +{ + while (nmerge < nfiles) + { + /* Number of input files processed so far. */ + size_t in; + + /* Number of output files generated so far. */ + size_t out; + + /* nfiles % NMERGE; this counts input files that are left over + after all full-sized merges have been done. */ + size_t remainder; + + /* Number of easily-available slots at the next loop iteration. */ + size_t cheap_slots; + + /* Do as many NMERGE-size merges as possible. In the case that + nmerge is bogus, increment by the maximum number of file + descriptors allowed. */ + for (out = in = 0; nmerge <= nfiles - in; out++) + { + FILE *tfp; + pid_t pid; + char *temp = create_temp (&tfp, &pid); + size_t num_merged = mergefiles (&files[in], MIN (ntemps, nmerge), + nmerge, tfp, temp); + ntemps -= MIN (ntemps, num_merged); + files[out].name = temp; + files[out].pid = pid; + in += num_merged; + } + + remainder = nfiles - in; + cheap_slots = nmerge - out % nmerge; + + if (cheap_slots < remainder) + { + /* So many files remain that they can't all be put into the last + NMERGE-sized output window. Do one more merge. Merge as few + files as possible, to avoid needless I/O. */ + size_t nshortmerge = remainder - cheap_slots + 1; + FILE *tfp; + pid_t pid; + char *temp = create_temp (&tfp, &pid); + size_t num_merged = mergefiles (&files[in], MIN (ntemps, nshortmerge), + nshortmerge, tfp, temp); + ntemps -= MIN (ntemps, num_merged); + files[out].name = temp; + files[out++].pid = pid; + in += num_merged; + } + + /* Put the remaining input files into the last NMERGE-sized output + window, so they will be merged in the next pass. */ + memmove(&files[out], &files[in], (nfiles - in) * sizeof *files); + ntemps += out; + nfiles -= in - out; + } + + nfiles = avoid_trashing_input (files, ntemps, nfiles, output_file); + + /* We aren't guaranteed that this final mergefiles will work, therefore we + try to merge into the output, and then merge as much as we can into a + temp file if we can't. Repeat. */ + + for (;;) + { + /* Merge directly into the output file if possible. */ + FILE **fps; + size_t nopened = open_input_files (files, nfiles, &fps); + + if (nopened == nfiles) + { + FILE *ofp = stream_open (output_file, "w"); + if (ofp) + { + mergefps (files, ntemps, nfiles, ofp, output_file, fps); + break; + } + if (errno != EMFILE || nopened <= 2) + die (_("open failed"), output_file); + } + else if (nopened <= 2) + die (_("open failed"), files[nopened].name); + + /* We ran out of file descriptors. Close one of the input + files, to gain a file descriptor. Then create a temporary + file with our spare file descriptor. Retry if that failed + (e.g., some other process could open a file between the time + we closed and tried to create). */ + FILE *tfp; + pid_t pid; + char *temp; + do + { + nopened--; + xfclose (fps[nopened], files[nopened].name); + temp = maybe_create_temp (&tfp, &pid, ! (nopened <= 2)); + } + while (!temp); + + /* Merge into the newly allocated temporary. */ + mergefps (&files[0], MIN (ntemps, nopened), nopened, tfp, temp, fps); + ntemps -= MIN (ntemps, nopened); + files[0].name = temp; + files[0].pid = pid; + + memmove (&files[1], &files[nopened], (nfiles - nopened) * sizeof *files); + ntemps++; + nfiles -= nopened - 1; + } +} + +/* Sort NFILES FILES onto OUTPUT_FILE. */ + +static void +sort (char * const *files, size_t nfiles, char const *output_file) +{ + struct buffer buf; + size_t ntemps = 0; + bool output_file_created = false; + + buf.alloc = 0; + + while (nfiles) + { + char const *temp_output; + char const *file = *files; + FILE *fp = xfopen (file, "r"); + FILE *tfp; + size_t bytes_per_line = (2 * sizeof (struct line) + - sizeof (struct line) / 2); + + if (! buf.alloc) + initbuf (&buf, bytes_per_line, + sort_buffer_size (&fp, 1, files, nfiles, bytes_per_line)); + buf.eof = false; + files++; + nfiles--; + + while (fillbuf (&buf, fp, file)) + { + struct line *line; + struct line *linebase; + + if (buf.eof && nfiles + && (bytes_per_line + 1 + < (buf.alloc - buf.used - bytes_per_line * buf.nlines))) + { + /* End of file, but there is more input and buffer room. + Concatenate the next input file; this is faster in + the usual case. */ + buf.left = buf.used; + break; + } + + line = buffer_linelim (&buf); + linebase = line - buf.nlines; + if (1 < buf.nlines) + sortlines (line, buf.nlines, linebase); + if (buf.eof && !nfiles && !ntemps && !buf.left) + { + xfclose (fp, file); + tfp = xfopen (output_file, "w"); + temp_output = output_file; + output_file_created = true; + } + else + { + ++ntemps; + temp_output = create_temp (&tfp, NULL); + } + + do + { + line--; + write_bytes (line->text, line->length, tfp, temp_output); + if (unique) + while (linebase < line && compare (line, line - 1) == 0) + line--; + } + while (linebase < line); + + xfclose (tfp, temp_output); + + /* Free up some resources every once in a while. */ + if (MAX_PROCS_BEFORE_REAP < nprocs) + reap_some (); + + if (output_file_created) + goto finish; + } + xfclose (fp, file); + } + + finish: + free (buf.buf); + + if (! output_file_created) + { + size_t i; + struct tempnode *node = temphead; + struct sortfile *tempfiles = xnmalloc (ntemps, sizeof *tempfiles); + for (i = 0; node; i++) + { + tempfiles[i].name = node->name; + tempfiles[i].pid = node->pid; + node = node->next; + } + merge (tempfiles, ntemps, ntemps, output_file); + free (tempfiles); + } +} + +/* Insert a malloc'd copy of key KEY_ARG at the end of the key list. */ + +static void +insertkey (struct keyfield *key_arg) +{ + struct keyfield **p; + struct keyfield *key = xmemdup (key_arg, sizeof *key); + + for (p = &keylist; *p; p = &(*p)->next) + continue; + *p = key; + key->next = NULL; +} + +/* Report a bad field specification SPEC, with extra info MSGID. */ + +static void badfieldspec (char const *, char const *) + ATTRIBUTE_NORETURN; +static void +badfieldspec (char const *spec, char const *msgid) +{ + error (SORT_FAILURE, 0, _("%s: invalid field specification %s"), + _(msgid), quote (spec)); + abort (); +} + +/* Report incompatible options. */ + +static void incompatible_options (char const *) ATTRIBUTE_NORETURN; +static void +incompatible_options (char const *opts) +{ + error (SORT_FAILURE, 0, _("options `-%s' are incompatible"), opts); + abort (); +} + +/* Check compatibility of ordering options. */ + +static void +check_ordering_compatibility (void) +{ + struct keyfield const *key; + + for (key = keylist; key; key = key->next) + if ((1 < (key->random + key->numeric + key->general_numeric + key->month + + key->version + !!key->ignore)) + || (key->random && key->translate)) + { + /* The following is too big, but guaranteed to be "big enough". */ + char opts[sizeof short_options]; + char *p = opts; + if (key->ignore == nondictionary) + *p++ = 'd'; + if (key->translate) + *p++ = 'f'; + if (key->general_numeric) + *p++ = 'g'; + if (key->ignore == nonprinting) + *p++ = 'i'; + if (key->month) + *p++ = 'M'; + if (key->numeric) + *p++ = 'n'; + if (key->version) + *p++ = 'V'; + if (key->random) + *p++ = 'R'; + *p = '\0'; + incompatible_options (opts); + } +} + +/* Parse the leading integer in STRING and store the resulting value + (which must fit into size_t) into *VAL. Return the address of the + suffix after the integer. If the value is too large, silently + substitute SIZE_MAX. If MSGID is NULL, return NULL after + failure; otherwise, report MSGID and exit on failure. */ + +static char const * +parse_field_count (char const *string, size_t *val, char const *msgid) +{ + char *suffix; + uintmax_t n; + + switch (xstrtoumax (string, &suffix, 10, &n, "")) + { + case LONGINT_OK: + case LONGINT_INVALID_SUFFIX_CHAR: + *val = n; + if (*val == n) + break; + /* Fall through. */ + case LONGINT_OVERFLOW: + case LONGINT_OVERFLOW | LONGINT_INVALID_SUFFIX_CHAR: + *val = SIZE_MAX; + break; + + case LONGINT_INVALID: + if (msgid) + error (SORT_FAILURE, 0, _("%s: invalid count at start of %s"), + _(msgid), quote (string)); + return NULL; + } + + return suffix; +} + +/* Handle interrupts and hangups. */ + +static void +sighandler (int sig) +{ + if (! SA_NOCLDSTOP) + signal (sig, SIG_IGN); + + cleanup (); + + signal (sig, SIG_DFL); + raise (sig); +} + +/* Set the ordering options for KEY specified in S. + Return the address of the first character in S that + is not a valid ordering option. + BLANKTYPE is the kind of blanks that 'b' should skip. */ + +static char * +set_ordering (const char *s, struct keyfield *key, enum blanktype blanktype) +{ + while (*s) + { + switch (*s) + { + case 'b': + if (blanktype == bl_start || blanktype == bl_both) + key->skipsblanks = true; + if (blanktype == bl_end || blanktype == bl_both) + key->skipeblanks = true; + break; + case 'd': + key->ignore = nondictionary; + break; + case 'f': + key->translate = fold_toupper; + break; + case 'g': + key->general_numeric = true; + break; + case 'i': + /* Option order should not matter, so don't let -i override + -d. -d implies -i, but -i does not imply -d. */ + if (! key->ignore) + key->ignore = nonprinting; + break; + case 'M': + key->month = true; + break; + case 'n': + key->numeric = true; + break; + case 'R': + key->random = true; + break; + case 'r': + key->reverse = true; + break; + case 'V': + key->version = true; + break; + default: + return (char *) s; + } + ++s; + } + return (char *) s; +} + +static struct keyfield * +key_init (struct keyfield *key) +{ + memset (key, 0, sizeof *key); + key->eword = SIZE_MAX; + return key; +} + +int +main (int argc, char **argv) +{ + struct keyfield *key; + struct keyfield key_buf; + struct keyfield gkey; + char const *s; + int c = 0; + char checkonly = 0; + bool mergeonly = false; + char *random_source = NULL; + bool need_random = false; + size_t nfiles = 0; + bool posixly_correct = (getenv ("POSIXLY_CORRECT") != NULL); + bool obsolete_usage = (posix2_version () < 200112); + char **files; + char *files_from = NULL; + struct Tokens tok; + char const *outfile = NULL; + + initialize_main (&argc, &argv); + set_program_name (argv[0]); + setlocale (LC_ALL, ""); + bindtextdomain (PACKAGE, LOCALEDIR); + textdomain (PACKAGE); + + initialize_exit_failure (SORT_FAILURE); + + hard_LC_COLLATE = hard_locale (LC_COLLATE); +#if HAVE_NL_LANGINFO + hard_LC_TIME = hard_locale (LC_TIME); +#endif + + /* Get locale's representation of the decimal point. */ + { + struct lconv const *locale = localeconv (); + + /* If the locale doesn't define a decimal point, or if the decimal + point is multibyte, use the C locale's decimal point. FIXME: + add support for multibyte decimal points. */ + decimal_point = to_uchar (locale->decimal_point[0]); + if (! decimal_point || locale->decimal_point[1]) + decimal_point = '.'; + + /* FIXME: add support for multibyte thousands separators. */ + thousands_sep = to_uchar (*locale->thousands_sep); + if (! thousands_sep || locale->thousands_sep[1]) + thousands_sep = -1; + } + + have_read_stdin = false; + inittables (); + + { + size_t i; + static int const sig[] = + { + /* The usual suspects. */ + SIGALRM, SIGHUP, SIGINT, SIGPIPE, SIGQUIT, SIGTERM, +#ifdef SIGPOLL + SIGPOLL, +#endif +#ifdef SIGPROF + SIGPROF, +#endif +#ifdef SIGVTALRM + SIGVTALRM, +#endif +#ifdef SIGXCPU + SIGXCPU, +#endif +#ifdef SIGXFSZ + SIGXFSZ, +#endif + }; + enum { nsigs = sizeof sig / sizeof sig[0] }; + +#if SA_NOCLDSTOP + struct sigaction act; + + sigemptyset (&caught_signals); + for (i = 0; i < nsigs; i++) + { + sigaction (sig[i], NULL, &act); + if (act.sa_handler != SIG_IGN) + sigaddset (&caught_signals, sig[i]); + } + + act.sa_handler = sighandler; + act.sa_mask = caught_signals; + act.sa_flags = 0; + + for (i = 0; i < nsigs; i++) + if (sigismember (&caught_signals, sig[i])) + sigaction (sig[i], &act, NULL); +#else + for (i = 0; i < nsigs; i++) + if (signal (sig[i], SIG_IGN) != SIG_IGN) + { + signal (sig[i], sighandler); + siginterrupt (sig[i], 1); + } +#endif + } + + /* The signal mask is known, so it is safe to invoke exit_cleanup. */ + atexit (exit_cleanup); + + gkey.sword = gkey.eword = SIZE_MAX; + gkey.ignore = NULL; + gkey.translate = NULL; + gkey.numeric = gkey.general_numeric = gkey.random = gkey.version = false; + gkey.month = gkey.reverse = false; + gkey.skipsblanks = gkey.skipeblanks = false; + + files = xnmalloc (argc, sizeof *files); + + for (;;) + { + /* Parse an operand as a file after "--" was seen; or if + pedantic and a file was seen, unless the POSIX version + predates 1003.1-2001 and -c was not seen and the operand is + "-o FILE" or "-oFILE". */ + int oi = -1; + + if (c == -1 + || (posixly_correct && nfiles != 0 + && ! (obsolete_usage + && ! checkonly + && optind != argc + && argv[optind][0] == '-' && argv[optind][1] == 'o' + && (argv[optind][2] || optind + 1 != argc))) + || ((c = getopt_long (argc, argv, short_options, + long_options, &oi)) + == -1)) + { + if (argc <= optind) + break; + files[nfiles++] = argv[optind++]; + } + else switch (c) + { + case 1: + key = NULL; + if (optarg[0] == '+') + { + bool minus_pos_usage = (optind != argc && argv[optind][0] == '-' + && ISDIGIT (argv[optind][1])); + obsolete_usage |= minus_pos_usage & ~posixly_correct; + if (obsolete_usage) + { + /* Treat +POS1 [-POS2] as a key if possible; but silently + treat an operand as a file if it is not a valid +POS1. */ + key = key_init (&key_buf); + s = parse_field_count (optarg + 1, &key->sword, NULL); + if (s && *s == '.') + s = parse_field_count (s + 1, &key->schar, NULL); + if (! (key->sword | key->schar)) + key->sword = SIZE_MAX; + if (! s || *set_ordering (s, key, bl_start)) + key = NULL; + else + { + if (minus_pos_usage) + { + char const *optarg1 = argv[optind++]; + s = parse_field_count (optarg1 + 1, &key->eword, + N_("invalid number after `-'")); + if (*s == '.') + s = parse_field_count (s + 1, &key->echar, + N_("invalid number after `.'")); + if (*set_ordering (s, key, bl_end)) + badfieldspec (optarg1, + N_("stray character in field spec")); + } + insertkey (key); + } + } + } + if (! key) + files[nfiles++] = optarg; + break; + + case SORT_OPTION: + c = XARGMATCH ("--sort", optarg, sort_args, sort_types); + /* Fall through. */ + case 'b': + case 'd': + case 'f': + case 'g': + case 'i': + case 'M': + case 'n': + case 'r': + case 'R': + case 'V': + { + char str[2]; + str[0] = c; + str[1] = '\0'; + set_ordering (str, &gkey, bl_both); + } + break; + + case CHECK_OPTION: + c = (optarg + ? XARGMATCH ("--check", optarg, check_args, check_types) + : 'c'); + /* Fall through. */ + case 'c': + case 'C': + if (checkonly && checkonly != c) + incompatible_options ("cC"); + checkonly = c; + break; + + case COMPRESS_PROGRAM_OPTION: + if (compress_program && !STREQ (compress_program, optarg)) + error (SORT_FAILURE, 0, _("multiple compress programs specified")); + compress_program = optarg; + break; + + case FILES0_FROM_OPTION: + files_from = optarg; + break; + + case 'k': + key = key_init (&key_buf); + + /* Get POS1. */ + s = parse_field_count (optarg, &key->sword, + N_("invalid number at field start")); + if (! key->sword--) + { + /* Provoke with `sort -k0' */ + badfieldspec (optarg, N_("field number is zero")); + } + if (*s == '.') + { + s = parse_field_count (s + 1, &key->schar, + N_("invalid number after `.'")); + if (! key->schar--) + { + /* Provoke with `sort -k1.0' */ + badfieldspec (optarg, N_("character offset is zero")); + } + } + if (! (key->sword | key->schar)) + key->sword = SIZE_MAX; + s = set_ordering (s, key, bl_start); + if (*s != ',') + { + key->eword = SIZE_MAX; + key->echar = 0; + } + else + { + /* Get POS2. */ + s = parse_field_count (s + 1, &key->eword, + N_("invalid number after `,'")); + if (! key->eword--) + { + /* Provoke with `sort -k1,0' */ + badfieldspec (optarg, N_("field number is zero")); + } + if (*s == '.') + { + s = parse_field_count (s + 1, &key->echar, + N_("invalid number after `.'")); + } + s = set_ordering (s, key, bl_end); + } + if (*s) + badfieldspec (optarg, N_("stray character in field spec")); + insertkey (key); + break; + + case 'm': + mergeonly = true; + break; + + case NMERGE_OPTION: + specify_nmerge (oi, c, optarg); + break; + + case 'o': + if (outfile && !STREQ (outfile, optarg)) + error (SORT_FAILURE, 0, _("multiple output files specified")); + outfile = optarg; + break; + + case RANDOM_SOURCE_OPTION: + if (random_source && !STREQ (random_source, optarg)) + error (SORT_FAILURE, 0, _("multiple random sources specified")); + random_source = optarg; + break; + + case 's': + stable = true; + break; + + case 'S': + specify_sort_size (oi, c, optarg); + break; + + case 't': + { + char newtab = optarg[0]; + if (! newtab) + error (SORT_FAILURE, 0, _("empty tab")); + if (optarg[1]) + { + if (STREQ (optarg, "\\0")) + newtab = '\0'; + else + { + /* Provoke with `sort -txx'. Complain about + "multi-character tab" instead of "multibyte tab", so + that the diagnostic's wording does not need to be + changed once multibyte characters are supported. */ + error (SORT_FAILURE, 0, _("multi-character tab %s"), + quote (optarg)); + } + } + if (tab != TAB_DEFAULT && tab != newtab) + error (SORT_FAILURE, 0, _("incompatible tabs")); + tab = newtab; + } + break; + + case 'T': + add_temp_dir (optarg); + break; + + case 'u': + unique = true; + break; + + case 'y': + /* Accept and ignore e.g. -y0 for compatibility with Solaris 2.x + through Solaris 7. It is also accepted by many non-Solaris + "sort" implementations, e.g., AIX 5.2, HP-UX 11i v2, IRIX 6.5. + -y is marked as obsolete starting with Solaris 8 (1999), but is + still accepted as of Solaris 10 prerelease (2004). + + Solaris 2.5.1 "sort -y 100" reads the input file "100", but + emulate Solaris 8 and 9 "sort -y 100" which ignores the "100", + and which in general ignores the argument after "-y" if it + consists entirely of digits (it can even be empty). */ + if (optarg == argv[optind - 1]) + { + char const *p; + for (p = optarg; ISDIGIT (*p); p++) + continue; + optind -= (*p != '\0'); + } + break; + + case 'z': + eolchar = 0; + break; + + case_GETOPT_HELP_CHAR; + + case_GETOPT_VERSION_CHAR (PROGRAM_NAME, AUTHORS); + + default: + usage (SORT_FAILURE); + } + } + + if (files_from) + { + FILE *stream; + + /* When using --files0-from=F, you may not specify any files + on the command-line. */ + if (nfiles) + { + error (0, 0, _("extra operand %s"), quote (files[0])); + fprintf (stderr, "%s\n", + _("file operands cannot be combined with --files0-from")); + usage (SORT_FAILURE); + } + + if (STREQ (files_from, "-")) + stream = stdin; + else + { + stream = fopen (files_from, "r"); + if (stream == NULL) + error (SORT_FAILURE, errno, _("cannot open %s for reading"), + quote (files_from)); + } + + readtokens0_init (&tok); + + if (! readtokens0 (stream, &tok) || fclose (stream) != 0) + error (SORT_FAILURE, 0, _("cannot read file names from %s"), + quote (files_from)); + + if (tok.n_tok) + { + size_t i; + free (files); + files = tok.tok; + nfiles = tok.n_tok; + for (i = 0; i < nfiles; i++) + { + if (STREQ (files[i], "-")) + error (SORT_FAILURE, 0, _("when reading file names from stdin, " + "no file name of %s allowed"), + quote (files[i])); + else if (files[i][0] == '\0') + { + /* Using the standard `filename:line-number:' prefix here is + not totally appropriate, since NUL is the separator, not NL, + but it might be better than nothing. */ + unsigned long int file_number = i + 1; + error (SORT_FAILURE, 0, + _("%s:%lu: invalid zero-length file name"), + quotearg_colon (files_from), file_number); + } + } + } + else + error (SORT_FAILURE, 0, _("no input from %s"), + quote (files_from)); + } + + /* Inheritance of global options to individual keys. */ + for (key = keylist; key; key = key->next) + { + if (! (key->ignore + || key->translate + || (key->skipsblanks + | key->reverse + | key->skipeblanks + | key->month + | key->numeric + | key->version + | key->general_numeric + | key->random))) + { + key->ignore = gkey.ignore; + key->translate = gkey.translate; + key->skipsblanks = gkey.skipsblanks; + key->skipeblanks = gkey.skipeblanks; + key->month = gkey.month; + key->numeric = gkey.numeric; + key->general_numeric = gkey.general_numeric; + key->random = gkey.random; + key->reverse = gkey.reverse; + key->version = gkey.version; + } + + need_random |= key->random; + } + + if (!keylist && (gkey.ignore + || gkey.translate + || (gkey.skipsblanks + | gkey.skipeblanks + | gkey.month + | gkey.numeric + | gkey.general_numeric + | gkey.random + | gkey.version))) + { + insertkey (&gkey); + need_random |= gkey.random; + } + + check_ordering_compatibility (); + + reverse = gkey.reverse; + + if (need_random) + { + randread_source = randread_new (random_source, MD5_DIGEST_SIZE); + if (! randread_source) + die (_("open failed"), random_source); + } + + if (temp_dir_count == 0) + { + char const *tmp_dir = getenv ("TMPDIR"); + add_temp_dir (tmp_dir ? tmp_dir : DEFAULT_TMPDIR); + } + + if (nfiles == 0) + { + static char *minus = (char *) "-"; + nfiles = 1; + free (files); + files = − + } + + /* Need to re-check that we meet the minimum requirement for memory + usage with the final value for NMERGE. */ + if (0 < sort_size) + sort_size = MAX (sort_size, MIN_SORT_SIZE); + + if (checkonly) + { + if (nfiles > 1) + error (SORT_FAILURE, 0, _("extra operand %s not allowed with -%c"), + quote (files[1]), checkonly); + + if (outfile) + { + static char opts[] = {0, 'o', 0}; + opts[0] = checkonly; + incompatible_options (opts); + } + + /* POSIX requires that sort return 1 IFF invoked with -c or -C and the + input is not properly sorted. */ + exit (check (files[0], checkonly) ? EXIT_SUCCESS : SORT_OUT_OF_ORDER); + } + + if (mergeonly) + { + struct sortfile *sortfiles = xcalloc (nfiles, sizeof *sortfiles); + size_t i; + + for (i = 0; i < nfiles; ++i) + sortfiles[i].name = files[i]; + + merge (sortfiles, 0, nfiles, outfile); + IF_LINT (free (sortfiles)); + } + else + sort (files, nfiles, outfile); + + if (have_read_stdin && fclose (stdin) == EOF) + die (_("close failed"), "-"); + + exit (EXIT_SUCCESS); +} diff -Naur coreutils-7.2.orig/src/unexpand.c coreutils-7.2/src/unexpand.c --- coreutils-7.2.orig/src/unexpand.c 2009-03-29 18:44:10.000000000 +0100 +++ coreutils-7.2/src/unexpand.c 2009-04-18 10:41:11.000000000 +0100 @@ -38,11 +38,28 @@ #include #include #include + +/* Get mbstate_t, mbrtowc(), wcwidth(). */ +#if HAVE_WCHAR_H +# include +#endif + #include "system.h" #include "error.h" #include "quote.h" #include "xstrndup.h" +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC + installation; work around this configuration error. */ +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 +# define MB_LEN_MAX 16 +#endif + +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ +#if HAVE_MBRTOWC && defined mbstate_t +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) +#endif + /* The official name of this program (e.g., no `g' prefix). */ #define PROGRAM_NAME "unexpand" @@ -102,6 +119,208 @@ {NULL, 0, NULL, 0} }; +static FILE *next_file (FILE *fp); + +#if HAVE_MBRTOWC +static void +unexpand_multibyte (void) +{ + FILE *fp; /* Input stream. */ + mbstate_t i_state; /* Current shift state of the input stream. */ + mbstate_t i_state_bak; /* Back up the I_STATE. */ + mbstate_t o_state; /* Current shift state of the output stream. */ + char buf[MB_LEN_MAX + BUFSIZ]; /* For spooling a read byte sequence. */ + char *bufpos; /* Next read position of BUF. */ + size_t buflen = 0; /* The length of the byte sequence in buf. */ + wint_t wc; /* A gotten wide character. */ + size_t mblength; /* The byte size of a multibyte character + which shows as same character as WC. */ + + /* Index in `tab_list' of next tabstop: */ + int tab_index = 0; /* For calculating width of pending tabs. */ + int print_tab_index = 0; /* For printing as many tabs as possible. */ + unsigned int column = 0; /* Column on screen of next char. */ + int next_tab_column; /* Column the next tab stop is on. */ + int convert = 1; /* If nonzero, perform translations. */ + unsigned int pending = 0; /* Pending columns of blanks. */ + + fp = next_file ((FILE *) NULL); + if (fp == NULL) + return; + + memset (&o_state, '\0', sizeof(mbstate_t)); + memset (&i_state, '\0', sizeof(mbstate_t)); + + for (;;) + { + if (buflen < MB_LEN_MAX && !feof(fp) && !ferror(fp)) + { + memmove (buf, bufpos, buflen); + buflen += fread (buf + buflen, sizeof(char), BUFSIZ, fp); + bufpos = buf; + } + + /* Get a wide character. */ + if (buflen < 1) + { + mblength = 1; + wc = WEOF; + } + else + { + i_state_bak = i_state; + mblength = mbrtowc ((wchar_t *)&wc, bufpos, buflen, &i_state); + } + + if (mblength == (size_t)-1 || mblength == (size_t)-2) + { + i_state = i_state_bak; + wc = L'\0'; + } + + if (wc == L' ' && convert && column < INT_MAX) + { + ++pending; + ++column; + } + else if (wc == L'\t' && convert) + { + if (tab_size == 0) + { + /* Do not let tab_index == first_free_tab; + stop when it is 1 less. */ + while (tab_index < first_free_tab - 1 + && column >= tab_list[tab_index]) + tab_index++; + next_tab_column = tab_list[tab_index]; + if (tab_index < first_free_tab - 1) + tab_index++; + if (column >= next_tab_column) + { + convert = 0; /* Ran out of tab stops. */ + goto flush_pend_mb; + } + } + else + { + next_tab_column = column + tab_size - column % tab_size; + } + pending += next_tab_column - column; + column = next_tab_column; + } + else + { +flush_pend_mb: + /* Flush pending spaces. Print as many tabs as possible, + then print the rest as spaces. */ + if (pending == 1) + { + putchar (' '); + pending = 0; + } + column -= pending; + while (pending > 0) + { + if (tab_size == 0) + { + /* Do not let print_tab_index == first_free_tab; + stop when it is 1 less. */ + while (print_tab_index < first_free_tab - 1 + && column >= tab_list[print_tab_index]) + print_tab_index++; + next_tab_column = tab_list[print_tab_index]; + if (print_tab_index < first_free_tab - 1) + print_tab_index++; + } + else + { + next_tab_column = + column + tab_size - column % tab_size; + } + if (next_tab_column - column <= pending) + { + putchar ('\t'); + pending -= next_tab_column - column; + column = next_tab_column; + } + else + { + --print_tab_index; + column += pending; + while (pending != 0) + { + putchar (' '); + pending--; + } + } + } + + if (wc == WEOF) + { + fp = next_file (fp); + if (fp == NULL) + break; /* No more files. */ + else + { + memset (&i_state, '\0', sizeof(mbstate_t)); + continue; + } + } + + if (mblength == (size_t)-1 || mblength == (size_t)-2) + { + if (convert) + { + ++column; + if (convert_entire_line == 0) + convert = 0; + } + mblength = 1; + putchar (buf[0]); + } + else if (mblength == 0) + { + if (convert && convert_entire_line == 0) + convert = 0; + mblength = 1; + putchar ('\0'); + } + else + { + if (convert) + { + if (wc == L'\b') + { + if (column > 0) + --column; + } + else + { + int width; /* The width of WC. */ + + width = wcwidth (wc); + column += (width > 0) ? width : 0; + if (convert_entire_line == 0) + convert = 0; + } + } + + if (wc == L'\n') + { + tab_index = print_tab_index = 0; + column = pending = 0; + convert = 1; + } + fwrite (bufpos, sizeof(char), mblength, stdout); + } + } + buflen -= mblength; + bufpos += mblength; + } +} +#endif + + void usage (int status) { @@ -523,7 +742,12 @@ file_list = (optind < argc ? &argv[optind] : stdin_argv); - unexpand (); +#if HAVE_MBRTOWC + if (MB_CUR_MAX > 1) + unexpand_multibyte (); + else +#endif + unexpand (); if (have_read_stdin && fclose (stdin) != 0) error (EXIT_FAILURE, errno, "-"); diff -Naur coreutils-7.2.orig/src/unexpand.c.orig coreutils-7.2/src/unexpand.c.orig --- coreutils-7.2.orig/src/unexpand.c.orig 1970-01-01 01:00:00.000000000 +0100 +++ coreutils-7.2/src/unexpand.c.orig 2009-03-29 18:44:10.000000000 +0100 @@ -0,0 +1,532 @@ +/* unexpand - convert blanks to tabs + Copyright (C) 89, 91, 1995-2006, 2008-2009 Free Software Foundation, Inc. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see . */ + +/* By default, convert only maximal strings of initial blanks and tabs + into tabs. + Preserves backspace characters in the output; they decrement the + column count for tab calculations. + The default action is equivalent to -8. + + Options: + --tabs=tab1[,tab2[,...]] + -t tab1[,tab2[,...]] + -tab1[,tab2[,...]] If only one tab stop is given, set the tabs tab1 + columns apart instead of the default 8. Otherwise, + set the tabs at columns tab1, tab2, etc. (numbered from + 0); preserve any blanks beyond the tab stops given. + --all + -a Use tabs wherever they would replace 2 or more blanks, + not just at the beginnings of lines. + + David MacKenzie */ + +#include + +#include +#include +#include +#include "system.h" +#include "error.h" +#include "quote.h" +#include "xstrndup.h" + +/* The official name of this program (e.g., no `g' prefix). */ +#define PROGRAM_NAME "unexpand" + +#define AUTHORS proper_name ("David MacKenzie") + +/* If true, convert blanks even after nonblank characters have been + read on the line. */ +static bool convert_entire_line; + +/* If nonzero, the size of all tab stops. If zero, use `tab_list' instead. */ +static size_t tab_size; + +/* The maximum distance between tab stops. */ +static size_t max_column_width; + +/* Array of the explicit column numbers of the tab stops; + after `tab_list' is exhausted, the rest of the line is printed + unchanged. The first column is column 0. */ +static uintmax_t *tab_list; + +/* The number of allocated entries in `tab_list'. */ +static size_t n_tabs_allocated; + +/* The index of the first invalid element of `tab_list', + where the next element can be added. */ +static size_t first_free_tab; + +/* Null-terminated array of input filenames. */ +static char **file_list; + +/* Default for `file_list' if no files are given on the command line. */ +static char *stdin_argv[] = +{ + (char *) "-", NULL +}; + +/* True if we have ever read standard input. */ +static bool have_read_stdin; + +/* The desired exit status. */ +static int exit_status; + +/* For long options that have no equivalent short option, use a + non-character as a pseudo short option, starting with CHAR_MAX + 1. */ +enum +{ + CONVERT_FIRST_ONLY_OPTION = CHAR_MAX + 1 +}; + +static struct option const longopts[] = +{ + {"tabs", required_argument, NULL, 't'}, + {"all", no_argument, NULL, 'a'}, + {"first-only", no_argument, NULL, CONVERT_FIRST_ONLY_OPTION}, + {GETOPT_HELP_OPTION_DECL}, + {GETOPT_VERSION_OPTION_DECL}, + {NULL, 0, NULL, 0} +}; + +void +usage (int status) +{ + if (status != EXIT_SUCCESS) + fprintf (stderr, _("Try `%s --help' for more information.\n"), + program_name); + else + { + printf (_("\ +Usage: %s [OPTION]... [FILE]...\n\ +"), + program_name); + fputs (_("\ +Convert blanks in each FILE to tabs, writing to standard output.\n\ +With no FILE, or when FILE is -, read standard input.\n\ +\n\ +"), stdout); + fputs (_("\ +Mandatory arguments to long options are mandatory for short options too.\n\ +"), stdout); + fputs (_("\ + -a, --all convert all blanks, instead of just initial blanks\n\ + --first-only convert only leading sequences of blanks (overrides -a)\n\ + -t, --tabs=N have tabs N characters apart instead of 8 (enables -a)\n\ + -t, --tabs=LIST use comma separated LIST of tab positions (enables -a)\n\ +"), stdout); + fputs (HELP_OPTION_DESCRIPTION, stdout); + fputs (VERSION_OPTION_DESCRIPTION, stdout); + emit_bug_reporting_address (); + } + exit (status); +} + +/* Add tab stop TABVAL to the end of `tab_list'. */ + +static void +add_tab_stop (uintmax_t tabval) +{ + uintmax_t prev_column = first_free_tab ? tab_list[first_free_tab - 1] : 0; + uintmax_t column_width = prev_column <= tabval ? tabval - prev_column : 0; + + if (first_free_tab == n_tabs_allocated) + tab_list = X2NREALLOC (tab_list, &n_tabs_allocated); + tab_list[first_free_tab++] = tabval; + + if (max_column_width < column_width) + { + if (SIZE_MAX < column_width) + error (EXIT_FAILURE, 0, _("tabs are too far apart")); + max_column_width = column_width; + } +} + +/* Add the comma or blank separated list of tab stops STOPS + to the list of tab stops. */ + +static void +parse_tab_stops (char const *stops) +{ + bool have_tabval = false; + uintmax_t tabval IF_LINT (= 0); + char const *num_start IF_LINT (= NULL); + bool ok = true; + + for (; *stops; stops++) + { + if (*stops == ',' || isblank (to_uchar (*stops))) + { + if (have_tabval) + add_tab_stop (tabval); + have_tabval = false; + } + else if (ISDIGIT (*stops)) + { + if (!have_tabval) + { + tabval = 0; + have_tabval = true; + num_start = stops; + } + + /* Detect overflow. */ + if (!DECIMAL_DIGIT_ACCUMULATE (tabval, *stops - '0', uintmax_t)) + { + size_t len = strspn (num_start, "0123456789"); + char *bad_num = xstrndup (num_start, len); + error (0, 0, _("tab stop is too large %s"), quote (bad_num)); + free (bad_num); + ok = false; + stops = num_start + len - 1; + } + } + else + { + error (0, 0, _("tab size contains invalid character(s): %s"), + quote (stops)); + ok = false; + break; + } + } + + if (!ok) + exit (EXIT_FAILURE); + + if (have_tabval) + add_tab_stop (tabval); +} + +/* Check that the list of tab stops TABS, with ENTRIES entries, + contains only nonzero, ascending values. */ + +static void +validate_tab_stops (uintmax_t const *tabs, size_t entries) +{ + uintmax_t prev_tab = 0; + size_t i; + + for (i = 0; i < entries; i++) + { + if (tabs[i] == 0) + error (EXIT_FAILURE, 0, _("tab size cannot be 0")); + if (tabs[i] <= prev_tab) + error (EXIT_FAILURE, 0, _("tab sizes must be ascending")); + prev_tab = tabs[i]; + } +} + +/* Close the old stream pointer FP if it is non-NULL, + and return a new one opened to read the next input file. + Open a filename of `-' as the standard input. + Return NULL if there are no more input files. */ + +static FILE * +next_file (FILE *fp) +{ + static char *prev_file; + char *file; + + if (fp) + { + if (ferror (fp)) + { + error (0, errno, "%s", prev_file); + exit_status = EXIT_FAILURE; + } + if (STREQ (prev_file, "-")) + clearerr (fp); /* Also clear EOF. */ + else if (fclose (fp) != 0) + { + error (0, errno, "%s", prev_file); + exit_status = EXIT_FAILURE; + } + } + + while ((file = *file_list++) != NULL) + { + if (STREQ (file, "-")) + { + have_read_stdin = true; + prev_file = file; + return stdin; + } + fp = fopen (file, "r"); + if (fp) + { + prev_file = file; + return fp; + } + error (0, errno, "%s", file); + exit_status = EXIT_FAILURE; + } + return NULL; +} + +/* Change blanks to tabs, writing to stdout. + Read each file in `file_list', in order. */ + +static void +unexpand (void) +{ + /* Input stream. */ + FILE *fp = next_file (NULL); + + /* The array of pending blanks. In non-POSIX locales, blanks can + include characters other than spaces, so the blanks must be + stored, not merely counted. */ + char *pending_blank; + + if (!fp) + return; + + /* The worst case is a non-blank character, then one blank, then a + tab stop, then MAX_COLUMN_WIDTH - 1 blanks, then a non-blank; so + allocate MAX_COLUMN_WIDTH bytes to store the blanks. */ + pending_blank = xmalloc (max_column_width); + + for (;;) + { + /* Input character, or EOF. */ + int c; + + /* If true, perform translations. */ + bool convert = true; + + + /* The following variables have valid values only when CONVERT + is true: */ + + /* Column of next input character. */ + uintmax_t column = 0; + + /* Column the next input tab stop is on. */ + uintmax_t next_tab_column = 0; + + /* Index in TAB_LIST of next tab stop to examine. */ + size_t tab_index = 0; + + /* If true, the first pending blank came just before a tab stop. */ + bool one_blank_before_tab_stop = false; + + /* If true, the previous input character was a blank. This is + initially true, since initial strings of blanks are treated + as if the line was preceded by a blank. */ + bool prev_blank = true; + + /* Number of pending columns of blanks. */ + size_t pending = 0; + + + /* Convert a line of text. */ + + do + { + while ((c = getc (fp)) < 0 && (fp = next_file (fp))) + continue; + + if (convert) + { + bool blank = !! isblank (c); + + if (blank) + { + if (next_tab_column <= column) + { + if (tab_size) + next_tab_column = + column + (tab_size - column % tab_size); + else + for (;;) + if (tab_index == first_free_tab) + { + convert = false; + break; + } + else + { + uintmax_t tab = tab_list[tab_index++]; + if (column < tab) + { + next_tab_column = tab; + break; + } + } + } + + if (convert) + { + if (next_tab_column < column) + error (EXIT_FAILURE, 0, _("input line is too long")); + + if (c == '\t') + { + column = next_tab_column; + + /* Discard pending blanks, unless it was a single + blank just before the previous tab stop. */ + if (! (pending == 1 && one_blank_before_tab_stop)) + { + pending = 0; + one_blank_before_tab_stop = false; + } + } + else + { + column++; + + if (! (prev_blank && column == next_tab_column)) + { + /* It is not yet known whether the pending blanks + will be replaced by tabs. */ + if (column == next_tab_column) + one_blank_before_tab_stop = true; + pending_blank[pending++] = c; + prev_blank = true; + continue; + } + + /* Replace the pending blanks by a tab or two. */ + pending_blank[0] = c = '\t'; + pending = one_blank_before_tab_stop; + } + } + } + else if (c == '\b') + { + /* Go back one column, and force recalculation of the + next tab stop. */ + column -= !!column; + next_tab_column = column; + tab_index -= !!tab_index; + } + else + { + column++; + if (!column) + error (EXIT_FAILURE, 0, _("input line is too long")); + } + + if (pending) + { + if (fwrite (pending_blank, 1, pending, stdout) != pending) + error (EXIT_FAILURE, errno, _("write error")); + pending = 0; + one_blank_before_tab_stop = false; + } + + prev_blank = blank; + convert &= convert_entire_line | blank; + } + + if (c < 0) + { + free (pending_blank); + return; + } + + if (putchar (c) < 0) + error (EXIT_FAILURE, errno, _("write error")); + } + while (c != '\n'); + } +} + +int +main (int argc, char **argv) +{ + bool have_tabval = false; + uintmax_t tabval IF_LINT (= 0); + int c; + + /* If true, cancel the effect of any -a (explicit or implicit in -t), + so that only leading blanks will be considered. */ + bool convert_first_only = false; + + initialize_main (&argc, &argv); + set_program_name (argv[0]); + setlocale (LC_ALL, ""); + bindtextdomain (PACKAGE, LOCALEDIR); + textdomain (PACKAGE); + + atexit (close_stdout); + + have_read_stdin = false; + exit_status = EXIT_SUCCESS; + convert_entire_line = false; + tab_list = NULL; + first_free_tab = 0; + + while ((c = getopt_long (argc, argv, ",0123456789at:", longopts, NULL)) + != -1) + { + switch (c) + { + case '?': + usage (EXIT_FAILURE); + case 'a': + convert_entire_line = true; + break; + case 't': + convert_entire_line = true; + parse_tab_stops (optarg); + break; + case CONVERT_FIRST_ONLY_OPTION: + convert_first_only = true; + break; + case ',': + if (have_tabval) + add_tab_stop (tabval); + have_tabval = false; + break; + case_GETOPT_HELP_CHAR; + case_GETOPT_VERSION_CHAR (PROGRAM_NAME, AUTHORS); + default: + if (!have_tabval) + { + tabval = 0; + have_tabval = true; + } + if (!DECIMAL_DIGIT_ACCUMULATE (tabval, c - '0', uintmax_t)) + error (EXIT_FAILURE, 0, _("tab stop value is too large")); + break; + } + } + + if (convert_first_only) + convert_entire_line = false; + + if (have_tabval) + add_tab_stop (tabval); + + validate_tab_stops (tab_list, first_free_tab); + + if (first_free_tab == 0) + tab_size = max_column_width = 8; + else if (first_free_tab == 1) + tab_size = tab_list[0]; + else + tab_size = 0; + + file_list = (optind < argc ? &argv[optind] : stdin_argv); + + unexpand (); + + if (have_read_stdin && fclose (stdin) != 0) + error (EXIT_FAILURE, errno, "-"); + + exit (exit_status); +} diff -Naur coreutils-7.2.orig/src/uniq.c coreutils-7.2/src/uniq.c --- coreutils-7.2.orig/src/uniq.c 2009-03-29 18:43:41.000000000 +0100 +++ coreutils-7.2/src/uniq.c 2009-04-18 10:41:11.000000000 +0100 @@ -22,6 +22,16 @@ #include #include +/* Get mbstate_t, mbrtowc(). */ +#if HAVE_WCHAR_H +# include +#endif + +/* Get isw* functions. */ +#if HAVE_WCTYPE_H +# include +#endif + #include "system.h" #include "argmatch.h" #include "linebuffer.h" @@ -30,7 +40,19 @@ #include "quote.h" #include "xmemcoll.h" #include "xstrtol.h" -#include "memcasecmp.h" +#include "xmemcoll.h" + +/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC + installation; work around this configuration error. */ +#if !defined MB_LEN_MAX || MB_LEN_MAX < 2 +# define MB_LEN_MAX 16 +#endif + +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ +#if HAVE_MBRTOWC && defined mbstate_t +# define mbrtowc(pwc, s, n, ps) (mbrtowc) (pwc, s, n, 0) +#endif + /* The official name of this program (e.g., no `g' prefix). */ #define PROGRAM_NAME "uniq" @@ -106,6 +128,10 @@ /* Select whether/how to delimit groups of duplicate lines. */ static enum delimit_method delimit_groups; +/* Function pointers. */ +static char * +(*find_field) (struct linebuffer *line); + static struct option const longopts[] = { {"count", no_argument, NULL, 'c'}, @@ -202,7 +228,7 @@ return a pointer to the beginning of the line's field to be compared. */ static char * -find_field (struct linebuffer const *line) +find_field_uni (struct linebuffer *line) { size_t count; char const *lp = line->buffer; @@ -223,6 +249,83 @@ return line->buffer + i; } +#if HAVE_MBRTOWC + +# define MBCHAR_TO_WCHAR(WC, MBLENGTH, LP, POS, SIZE, STATEP, CONVFAIL) \ + do \ + { \ + mbstate_t state_bak; \ + \ + CONVFAIL = 0; \ + state_bak = *STATEP; \ + \ + MBLENGTH = mbrtowc (&WC, LP + POS, SIZE - POS, STATEP); \ + \ + switch (MBLENGTH) \ + { \ + case (size_t)-2: \ + case (size_t)-1: \ + *STATEP = state_bak; \ + CONVFAIL++; \ + /* Fall through */ \ + case 0: \ + MBLENGTH = 1; \ + } \ + } \ + while (0) + +static char * +find_field_multi (struct linebuffer *line) +{ + size_t count; + char *lp = line->buffer; + size_t size = line->length - 1; + size_t pos; + size_t mblength; + wchar_t wc; + mbstate_t *statep; + int convfail; + + pos = 0; + statep = &(line->state); + + /* skip fields. */ + for (count = 0; count < skip_fields && pos < size; count++) + { + while (pos < size) + { + MBCHAR_TO_WCHAR (wc, mblength, lp, pos, size, statep, convfail); + + if (convfail || !iswblank (wc)) + { + pos += mblength; + break; + } + pos += mblength; + } + + while (pos < size) + { + MBCHAR_TO_WCHAR (wc, mblength, lp, pos, size, statep, convfail); + + if (!convfail && iswblank (wc)) + break; + + pos += mblength; + } + } + + /* skip fields. */ + for (count = 0; count < skip_chars && pos < size; count++) + { + MBCHAR_TO_WCHAR (wc, mblength, lp, pos, size, statep, convfail); + pos += mblength; + } + + return lp + pos; +} +#endif + /* Return false if two strings OLD and NEW match, true if not. OLD and NEW point not to the beginnings of the lines but rather to the beginnings of the fields to compare. @@ -231,6 +334,8 @@ static bool different (char *old, char *new, size_t oldlen, size_t newlen) { + char *copy_old, *copy_new; + if (check_chars < oldlen) oldlen = check_chars; if (check_chars < newlen) @@ -238,15 +343,93 @@ if (ignore_case) { - /* FIXME: This should invoke strcoll somehow. */ - return oldlen != newlen || memcasecmp (old, new, oldlen); + size_t i; + + copy_old = alloca (oldlen + 1); + copy_new = alloca (oldlen + 1); + + for (i = 0; i < oldlen; i++) + { + copy_old[i] = toupper (old[i]); + copy_new[i] = toupper (new[i]); + } } - else if (hard_LC_COLLATE) - return xmemcoll (old, oldlen, new, newlen) != 0; else - return oldlen != newlen || memcmp (old, new, oldlen); + { + copy_old = (char *)old; + copy_new = (char *)new; + } + + return xmemcoll (copy_old, oldlen, copy_new, newlen); } +#if HAVE_MBRTOWC +static int +different_multi (const char *old, const char *new, size_t oldlen, size_t newlen, mbstate_t oldstate, mbstate_t newstate) +{ + size_t i, j, chars; + const char *str[2]; + char *copy[2]; + size_t len[2]; + mbstate_t state[2]; + size_t mblength; + wchar_t wc, uwc; + mbstate_t state_bak; + + str[0] = old; + str[1] = new; + len[0] = oldlen; + len[1] = newlen; + state[0] = oldstate; + state[1] = newstate; + + for (i = 0; i < 2; i++) + { + copy[i] = alloca (len[i] + 1); + + for (j = 0, chars = 0; j < len[i] && chars < check_chars; chars++) + { + state_bak = state[i]; + mblength = mbrtowc (&wc, str[i] + j, len[i] - j, &(state[i])); + + switch (mblength) + { + case (size_t)-1: + case (size_t)-2: + state[i] = state_bak; + /* Fall through */ + case 0: + mblength = 1; + break; + + default: + if (ignore_case) + { + uwc = towupper (wc); + + if (uwc != wc) + { + mbstate_t state_wc; + + memset (&state_wc, '\0', sizeof(mbstate_t)); + wcrtomb (copy[i] + j, uwc, &state_wc); + } + else + memcpy (copy[i] + j, str[i] + j, mblength); + } + else + memcpy (copy[i] + j, str[i] + j, mblength); + } + j += mblength; + } + copy[i][j] = '\0'; + len[i] = j; + } + + return xmemcoll (copy[0], len[0], copy[1], len[1]); +} +#endif + /* Output the line in linebuffer LINE to standard output provided that the switches say it should be output. MATCH is true if the line matches the previous line. @@ -299,15 +482,43 @@ { char *prevfield IF_LINT (= NULL); size_t prevlen IF_LINT (= 0); +#if HAVE_MBRTOWC + mbstate_t prevstate; + + memset (&prevstate, '\0', sizeof (mbstate_t)); +#endif while (!feof (stdin)) { char *thisfield; size_t thislen; +#if HAVE_MBRTOWC + mbstate_t thisstate; +#endif + if (readlinebuffer_delim (thisline, stdin, delimiter) == 0) break; thisfield = find_field (thisline); thislen = thisline->length - 1 - (thisfield - thisline->buffer); +#if HAVE_MBRTOWC + if (MB_CUR_MAX > 1) + { + thisstate = thisline->state; + + if (prevline->length == 0 || different_multi + (thisfield, prevfield, thislen, prevlen, thisstate, prevstate)) + { + fwrite (thisline->buffer, sizeof (char), + thisline->length, stdout); + + SWAP_LINES (prevline, thisline); + prevfield = thisfield; + prevlen = thislen; + prevstate = thisstate; + } + } + else +#endif if (prevline->length == 0 || different (thisfield, prevfield, thislen, prevlen)) { @@ -326,17 +537,26 @@ size_t prevlen; uintmax_t match_count = 0; bool first_delimiter = true; +#if HAVE_MBRTOWC + mbstate_t prevstate; +#endif if (readlinebuffer_delim (prevline, stdin, delimiter) == 0) goto closefiles; prevfield = find_field (prevline); prevlen = prevline->length - 1 - (prevfield - prevline->buffer); +#if HAVE_MBRTOWC + prevstate = prevline->state; +#endif while (!feof (stdin)) { bool match; char *thisfield; size_t thislen; +#if HAVE_MBRTOWC + mbstate_t thisstate; +#endif if (readlinebuffer_delim (thisline, stdin, delimiter) == 0) { if (ferror (stdin)) @@ -345,6 +565,15 @@ } thisfield = find_field (thisline); thislen = thisline->length - 1 - (thisfield - thisline->buffer); +#if HAVE_MBRTOWC + if (MB_CUR_MAX > 1) + { + thisstate = thisline->state; + match = !different_multi (thisfield, prevfield, + thislen, prevlen, thisstate, prevstate); + } + else +#endif match = !different (thisfield, prevfield, thislen, prevlen); match_count += match; @@ -377,6 +606,9 @@ SWAP_LINES (prevline, thisline); prevfield = thisfield; prevlen = thislen; +#if HAVE_MBRTOWC + prevstate = thisstate; +#endif if (!match) match_count = 0; } @@ -422,6 +654,19 @@ atexit (close_stdout); +#if HAVE_MBRTOWC + if (MB_CUR_MAX > 1) + { + find_field = find_field_multi; + } + else +#endif + { + find_field = find_field_uni; + } + + + skip_chars = 0; skip_fields = 0; check_chars = SIZE_MAX; diff -Naur coreutils-7.2.orig/tests/Makefile.am coreutils-7.2/tests/Makefile.am --- coreutils-7.2.orig/tests/Makefile.am 2009-03-31 10:10:39.000000000 +0100 +++ coreutils-7.2/tests/Makefile.am 2009-04-18 10:41:11.000000000 +0100 @@ -203,6 +203,7 @@ misc/sort-compress \ misc/sort-continue \ misc/sort-files0-from \ + misc/sort-mb-tests \ misc/sort-merge \ misc/sort-merge-fdlimit \ misc/sort-rand \ @@ -426,6 +427,10 @@ $(root_tests) pr_data = \ + misc/mb1.X \ + misc/mb1.I \ + misc/mb2.X \ + misc/mb2.I \ pr/0F \ pr/0FF \ pr/0FFnt \ diff -Naur coreutils-7.2.orig/tests/Makefile.am.orig coreutils-7.2/tests/Makefile.am.orig --- coreutils-7.2.orig/tests/Makefile.am.orig 1970-01-01 01:00:00.000000000 +0100 +++ coreutils-7.2/tests/Makefile.am.orig 2009-03-31 10:10:39.000000000 +0100 @@ -0,0 +1,591 @@ +## Process this file with automake to produce Makefile.in -*-Makefile-*-. + +# Sort in traditional ASCII order, regardless of the current locale; +# otherwise we may get into trouble with distinct strings that the +# current locale considers to be equal. +ASSORT = LC_ALL=C sort + +EXTRA_DIST = \ + Coreutils.pm \ + CuTmpdir.pm \ + README \ + check.mk \ + envvar-check \ + lang-default \ + other-fs-tmpdir \ + require-perl \ + sample-test \ + test-lib.sh \ + $(pr_data) + +root_tests = \ + chown/basic \ + cp/cp-a-selinux \ + cp/preserve-gid \ + cp/special-bits \ + dd/skip-seek-past-dev \ + install/install-C-root \ + ls/capability \ + ls/nameless-uid \ + misc/chcon \ + misc/selinux \ + misc/truncate-owned-by-other \ + mkdir/writable-under-readonly \ + mv/sticky-to-xpart \ + rm/fail-2eperm \ + rm/no-give-up \ + rm/one-file-system \ + tail-2/append-only \ + touch/now-owned-by-other + +.PHONY: check-root +check-root: + $(MAKE) check TESTS='$(root_tests)' + +check-recursive: root-hint + +# Advertise `check-root' target. +.PHONY: root-hint +root-hint: + @echo '***********************************************************' + @echo "NOTICE: Some tests may be run only as root." + @echo " See the 'Running tests as root' section in README." + @echo '***********************************************************' + +EXTRA_DIST += $(TESTS) + +# Do not choose a name that is a shell keyword like 'if', or a +# commonly-used utility like 'cat' or 'test', as the name of a test. +# Otherwise, VPATH builds will fail on hosts like Solaris, since they +# will expand 'if test ...' to 'if .../test ...', and the '.../test' +# will execute the test script rather than the standard utility. + +# Notes on the ordering of these tests: +# Place early in the list tests of the tools that +# are most commonly used in test scripts themselves. +# E.g., nearly every test script uses rm and chmod. +# help-version comes early because it's a basic sanity test. +# Put seq early, since lots of other tests use it. +# Put tests that sleep early, but not all together, so in parallel builds +# they share time with tests that burn CPU, not with others that sleep. +# Put head-elide-tail early, because it's long-running. + +TESTS = \ + misc/help-version \ + misc/invalid-opt \ + rm/ext3-perf \ + rm/cycle \ + cp/link-heap \ + chmod/no-x \ + chgrp/basic \ + rm/dangling-symlink \ + misc/ls-time \ + rm/deep-1 \ + rm/deep-2 \ + rm/dir-no-w \ + rm/dir-nonrecur \ + rm/dot-rel \ + rm/isatty \ + rm/empty-inacc \ + rm/empty-name \ + rm/f-1 \ + rm/fail-eacces \ + rm/fail-eperm \ + tail-2/assert \ + rm/hash \ + rm/i-1 \ + rm/i-never \ + rm/i-no-r \ + tail-2/infloop-1 \ + rm/ignorable \ + rm/inaccessible \ + rm/interactive-always \ + rm/interactive-once \ + rm/ir-1 \ + rm/r-1 \ + rm/r-2 \ + rm/r-3 \ + rm/r-4 \ + rm/readdir-bug \ + rm/rm1 \ + touch/empty-file \ + rm/rm2 \ + rm/rm3 \ + rm/rm4 \ + rm/rm5 \ + rm/sunos-1 \ + rm/unread2 \ + rm/unread3 \ + rm/unreadable \ + rm/v-slash \ + chgrp/default-no-deref \ + chgrp/deref \ + chgrp/no-x \ + chgrp/posix-H \ + chgrp/recurse \ + misc/ptx \ + misc/test \ + misc/seq \ + misc/seq-long-double \ + misc/head \ + misc/head-elide-tail \ + tail-2/tail-n0f \ + misc/ls-misc \ + misc/date \ + misc/ptx-overrun \ + misc/xstrtol \ + misc/od \ + misc/mktemp \ + misc/arch \ + misc/pr \ + misc/join \ + pr/pr-tests \ + misc/df-P \ + misc/pwd-option \ + misc/pwd-unreadable-parent \ + misc/chcon-fail \ + misc/cut \ + misc/wc \ + misc/wc-files0-from \ + misc/wc-files0 \ + misc/cat-proc \ + misc/cat-buf \ + misc/base64 \ + misc/basename \ + misc/close-stdout \ + misc/comm \ + misc/csplit \ + misc/date-sec \ + misc/dircolors \ + misc/df \ + misc/dirname \ + misc/expand \ + misc/expr \ + misc/factor \ + misc/false-status \ + misc/fmt \ + misc/fmt-long-line \ + misc/fold \ + misc/groups-dash \ + misc/groups-version \ + misc/head-c \ + misc/head-pos \ + misc/id-context \ + misc/md5sum \ + misc/md5sum-newline \ + misc/mknod \ + misc/nice \ + misc/nl \ + misc/nohup \ + misc/od-N \ + misc/od-multiple-t \ + misc/od-x8 \ + misc/paste \ + misc/pathchk1 \ + misc/printf \ + misc/printf-cov \ + misc/printf-hex \ + misc/printf-surprise \ + misc/pwd-long \ + misc/readlink-fp-loop \ + misc/runcon-no-reorder \ + misc/sha1sum \ + misc/sha1sum-vec \ + misc/sha224sum \ + misc/sha256sum \ + misc/sha384sum \ + misc/sha512sum \ + misc/shred-exact \ + misc/shred-passes \ + misc/shred-remove \ + misc/shuf \ + misc/sort \ + misc/sort-compress \ + misc/sort-continue \ + misc/sort-files0-from \ + misc/sort-merge \ + misc/sort-merge-fdlimit \ + misc/sort-rand \ + misc/sort-version \ + misc/split-a \ + misc/split-fail \ + misc/split-l \ + misc/stat-fmt \ + misc/stat-printf \ + misc/stty \ + misc/stty-invalid \ + misc/stty-row-col \ + misc/sum \ + misc/sum-sysv \ + misc/tac \ + misc/tac-continue \ + misc/tail \ + misc/tee \ + misc/tee-dash \ + misc/test-diag \ + misc/timeout \ + misc/timeout-parameters \ + misc/tr \ + misc/truncate-dangling-symlink \ + misc/truncate-dir-fail \ + misc/truncate-fail-diag \ + misc/truncate-fifo \ + misc/truncate-no-create-missing \ + misc/truncate-overflow \ + misc/truncate-parameters \ + misc/truncate-relative \ + misc/tsort \ + misc/tty-eof \ + misc/unexpand \ + misc/uniq \ + misc/xattr \ + chmod/c-option \ + chmod/equal-x \ + chmod/equals \ + chmod/inaccessible \ + chmod/octal \ + chmod/setgid \ + chmod/silent \ + chmod/thru-dangling \ + chmod/umask-x \ + chmod/usage \ + chown/deref \ + chown/preserve-root \ + chown/separator \ + cp/abuse \ + cp/acl \ + cp/backup-1 \ + cp/backup-dir \ + cp/backup-is-src \ + cp/cp-HL \ + cp/cp-deref \ + cp/cp-i \ + cp/cp-mv-backup \ + cp/cp-parents \ + cp/deref-slink \ + cp/dir-rm-dest \ + cp/dir-slash \ + cp/dir-vs-file \ + cp/existing-perm-race \ + cp/fail-perm \ + cp/file-perm-race \ + cp/into-self \ + cp/link \ + cp/link-no-deref \ + cp/link-preserve \ + cp/no-deref-link1 \ + cp/no-deref-link2 \ + cp/no-deref-link3 \ + cp/parent-perm \ + cp/parent-perm-race \ + cp/perm \ + cp/preserve-2 \ + cp/proc-zero-len \ + cp/r-vs-symlink \ + cp/same-file \ + cp/slink-2-slink \ + cp/sparse \ + cp/special-f \ + cp/src-base-dot \ + cp/symlink-slash \ + cp/thru-dangling \ + dd/misc \ + dd/not-rewound \ + dd/reblock \ + dd/skip-seek \ + dd/skip-seek2 \ + dd/skip-seek-past-file \ + dd/unblock-sync \ + df/total-verify \ + du/2g \ + du/8gb \ + du/basic \ + du/deref \ + du/deref-args \ + du/exclude \ + du/fd-leak \ + du/files0-from \ + du/hard-link \ + du/inacc-dest \ + du/inacc-dir \ + du/inaccessible-cwd \ + du/long-from-unreadable \ + du/long-sloop \ + du/no-deref \ + du/no-x \ + du/one-file-system \ + du/restore-wd \ + du/slash \ + du/slink \ + du/trailing-slash \ + du/two-args \ + install/basic-1 \ + install/create-leading \ + install/d-slashdot \ + install/install-C \ + install/install-C-selinux \ + install/strip-program \ + install/trap \ + ln/backup-1 \ + ln/hard-backup \ + ln/misc \ + ln/sf-1 \ + ln/target-1 \ + ls/color-clear-to-eol \ + ls/color-dtype-dir \ + ls/dangle \ + ls/dired \ + ls/file-type \ + ls/follow-slink \ + ls/hardlink \ + ls/infloop \ + ls/inode \ + ls/m-option \ + ls/no-arg \ + ls/no-cap \ + ls/proc-selinux-segfault \ + ls/recursive \ + ls/rt-1 \ + ls/stat-dtype \ + ls/stat-failed \ + ls/stat-free-symlinks \ + ls/stat-vs-dirent \ + ls/symlink-slash \ + ls/x-option \ + mkdir/p-1 \ + mkdir/p-2 \ + mkdir/p-3 \ + mkdir/p-slashdot \ + mkdir/p-thru-slink \ + mkdir/p-v \ + mkdir/parents \ + mkdir/perm \ + mkdir/selinux \ + mkdir/special-1 \ + mkdir/t-slash \ + mv/acl \ + mv/atomic \ + mv/atomic2 \ + mv/backup-dir \ + mv/backup-is-src \ + mv/childproof \ + mv/diag \ + mv/dir-file \ + mv/dir2dir \ + mv/dup-source \ + mv/force \ + mv/hard-2 \ + mv/hard-3 \ + mv/hard-4 \ + mv/hard-link-1 \ + mv/hard-verbose \ + mv/i-1 \ + mv/i-2 \ + mv/i-3 \ + mv/i-4 \ + mv/i-5 \ + mv/i-link-no \ + mv/into-self \ + mv/into-self-2 \ + mv/into-self-3 \ + mv/into-self-4 \ + mv/leak-fd \ + mv/mv-n \ + mv/mv-special-1 \ + mv/no-target-dir \ + mv/part-fail \ + mv/part-hardlink \ + mv/part-rename \ + mv/part-symlink \ + mv/partition-perm \ + mv/perm-1 \ + mv/to-symlink \ + mv/trailing-slash \ + mv/update \ + readlink/can-e \ + readlink/can-f \ + readlink/can-m \ + readlink/rl-1 \ + rmdir/fail-perm \ + rmdir/ignore \ + rmdir/t-slash \ + tail-2/assert-2 \ + tail-2/big-4gb \ + tail-2/proc-ksyms \ + tail-2/start-middle \ + touch/dangling-symlink \ + touch/dir-1 \ + touch/fail-diag \ + touch/fifo \ + touch/no-create-missing \ + touch/no-rights \ + touch/not-owner \ + touch/obsolescent \ + touch/read-only \ + touch/relative \ + $(root_tests) + +pr_data = \ + pr/0F \ + pr/0FF \ + pr/0FFnt \ + pr/0FFt \ + pr/0FnFnt \ + pr/0FnFt \ + pr/0Fnt \ + pr/0Ft \ + pr/2-S_f-t_notab \ + pr/2-Sf-t_notab \ + pr/2f-t_notab \ + pr/2s_f-t_notab \ + pr/2s_w60f-t_nota \ + pr/2sf-t_notab \ + pr/2sw60f-t_notab \ + pr/2w60f-t_notab \ + pr/3-0F \ + pr/3-5l24f-t \ + pr/3-FF \ + pr/3a2l17-FF \ + pr/3a3f-0F \ + pr/3a3l15-t \ + pr/3a3l15f-t \ + pr/3b2l17-FF \ + pr/3b3f-0F \ + pr/3b3f-0FF \ + pr/3b3f-FF \ + pr/3b3l15-t \ + pr/3b3l15f-t \ + pr/3f-0F \ + pr/3f-FF \ + pr/3l24-t \ + pr/3l24f-t \ + pr/3ml24-FF \ + pr/3ml24-t \ + pr/3ml24-t-FF \ + pr/3ml24f-t \ + pr/4-7l24-FF \ + pr/4l24-FF \ + pr/FF \ + pr/FFn \ + pr/FFtn \ + pr/FnFn \ + pr/Ja3l24f-lm \ + pr/Jb3l24f-lm \ + pr/Jml24f-lm-lo \ + pr/W-72l24f-ll \ + pr/W20l24f-ll \ + pr/W26l24f-ll \ + pr/W27l24f-ll \ + pr/W28l24f-ll \ + pr/W35Ja3l24f-lm \ + pr/W35Jb3l24f-lm \ + pr/W35Jml24f-lmlo \ + pr/W35a3l24f-lm \ + pr/W35b3l24f-lm \ + pr/W35ml24f-lm-lo \ + pr/W72Jl24f-ll \ + pr/a2l15-FF \ + pr/a2l17-FF \ + pr/a3-0F \ + pr/a3f-0F \ + pr/a3f-0FF \ + pr/a3f-FF \ + pr/a3l15-t \ + pr/a3l15f-t \ + pr/a3l24f-lm \ + pr/b2l15-FF \ + pr/b2l17-FF \ + pr/b3-0F \ + pr/b3f-0F \ + pr/b3f-0FF \ + pr/b3f-FF \ + pr/b3l15-t \ + pr/b3l15f-t \ + pr/b3l24f-lm \ + pr/l24-FF \ + pr/l24-t \ + pr/l24f-t \ + pr/loli \ + pr/ml20-FF-t \ + pr/ml24-FF \ + pr/ml24-t \ + pr/ml24-t-FF \ + pr/ml24f-0F \ + pr/ml24f-lm-lo \ + pr/ml24f-t \ + pr/ml24f-t-0F \ + pr/n+2-5l24f-0FF \ + pr/n+2l24f-0FF \ + pr/n+2l24f-bl \ + pr/n+3-7l24-FF \ + pr/n+3l24f-0FF \ + pr/n+3l24f-bl \ + pr/n+3ml20f-bl-FF \ + pr/n+3ml24f-bl-tn \ + pr/n+3ml24f-tn-bl \ + pr/n+4-8a2l17-FF \ + pr/n+4b2l17f-0FF \ + pr/n+5-8b3l17f-FF \ + pr/n+5a3l13f-0FF \ + pr/n+6a2l17-FF \ + pr/n+6b3l13f-FF \ + pr/n+7l24-FF \ + pr/n+8l20-FF \ + pr/nJml24f-lmlmlo \ + pr/nJml24f-lmlolm \ + pr/nN1+3l24f-bl \ + pr/nN15l24f-bl \ + pr/nSml20-bl-FF \ + pr/nSml20-t-t-FF \ + pr/nSml20-t-tFFFF \ + pr/nSml24-bl-FF \ + pr/nSml24-t-t-FF \ + pr/nSml24-t-tFFFF \ + pr/nl24f-bl \ + pr/o3Jml24f-lm-lo \ + pr/o3a3Sl24f-tn \ + pr/o3a3Snl24f-tn \ + pr/o3a3l24f-tn \ + pr/o3b3Sl24f-tn \ + pr/o3b3Snl24f-tn \ + pr/o3b3l24f-tn \ + pr/o3mSl24f-bl-tn \ + pr/o3mSnl24fbltn \ + pr/o3ml24f-bl-tn \ + pr/t-0FF \ + pr/t-FF \ + pr/t-bl \ + pr/t-t \ + pr/tFFn \ + pr/tFFt \ + pr/tFFt-bl \ + pr/tFFt-ll \ + pr/tFFt-lm \ + pr/tFnFt \ + pr/t_notab \ + pr/t_tab \ + pr/t_tab_ \ + pr/ta3-0FF \ + pr/ta3-FF \ + pr/tb3-0FF \ + pr/tb3-FF \ + pr/tn \ + pr/tn2e5o3-t_tab \ + pr/tn2e8-t_tab \ + pr/tn2e8o3-t_tab \ + pr/tn_2e8-t_tab \ + pr/tn_2e8S-t_tab \ + pr/tne8-t_tab \ + pr/tne8o3-t_tab \ + pr/tt-0FF \ + pr/tt-FF \ + pr/tt-bl \ + pr/tt-t \ + pr/tta3-0FF \ + pr/tta3-FF \ + pr/ttb3-0FF \ + pr/ttb3-FF \ + pr/w72l24f-ll + +AUTOMAKE_OPTIONS = +include $(srcdir)/check.mk diff -Naur coreutils-7.2.orig/tests/misc/cut coreutils-7.2/tests/misc/cut --- coreutils-7.2.orig/tests/misc/cut 2009-02-27 16:36:00.000000000 +0000 +++ coreutils-7.2/tests/misc/cut 2009-04-18 10:41:11.000000000 +0100 @@ -26,7 +26,7 @@ my $prog = 'cut'; my $try = "Try \`$prog --help' for more information.\n"; my $from_1 = "$prog: fields and positions are numbered from 1\n$try"; -my $inval = "$prog: invalid byte or field list\n$try"; +my $inval = "$prog: invalid byte, character or field list\n$try"; my $no_endpoint = "$prog: invalid range with no endpoint: -\n$try"; my @Tests = @@ -141,7 +141,7 @@ # None of the following invalid ranges provoked an error up to coreutils-6.9. ['inval1', qw(-f 2-0), {IN=>''}, {OUT=>''}, {EXIT=>1}, - {ERR=>"$prog: invalid decreasing range\n$try"}], + {ERR=>"$prog: invalid byte, character or field list\n$try"}], ['inval2', qw(-f -), {IN=>''}, {OUT=>''}, {EXIT=>1}, {ERR=>$no_endpoint}], ['inval3', '-f', '4,-', {IN=>''}, {OUT=>''}, {EXIT=>1}, {ERR=>$no_endpoint}], ['inval4', '-f', '1-2,-', {IN=>''}, {OUT=>''}, {EXIT=>1}, {ERR=>$no_endpoint}], diff -Naur coreutils-7.2.orig/tests/misc/mb1.I coreutils-7.2/tests/misc/mb1.I --- coreutils-7.2.orig/tests/misc/mb1.I 1970-01-01 01:00:00.000000000 +0100 +++ coreutils-7.2/tests/misc/mb1.I 2009-04-18 10:41:11.000000000 +0100 @@ -0,0 +1,4 @@ +Apple＠10 +Banana＠5 +Citrus＠20 +Cherry＠30 diff -Naur coreutils-7.2.orig/tests/misc/mb1.X coreutils-7.2/tests/misc/mb1.X --- coreutils-7.2.orig/tests/misc/mb1.X 1970-01-01 01:00:00.000000000 +0100 +++ coreutils-7.2/tests/misc/mb1.X 2009-04-18 10:41:11.000000000 +0100 @@ -0,0 +1,4 @@ +Banana＠5 +Apple＠10 +Citrus＠20 +Cherry＠30 diff -Naur coreutils-7.2.orig/tests/misc/mb2.I coreutils-7.2/tests/misc/mb2.I --- coreutils-7.2.orig/tests/misc/mb2.I 1970-01-01 01:00:00.000000000 +0100 +++ coreutils-7.2/tests/misc/mb2.I 2009-04-18 10:41:11.000000000 +0100 @@ -0,0 +1,4 @@ +Apple＠ＡＡ10＠＠20 +Banana＠ＡＡ5＠＠30 +Citrus＠ＡＡ20＠＠5 +Cherry＠ＡＡ30＠＠10 diff -Naur coreutils-7.2.orig/tests/misc/mb2.X coreutils-7.2/tests/misc/mb2.X --- coreutils-7.2.orig/tests/misc/mb2.X 1970-01-01 01:00:00.000000000 +0100 +++ coreutils-7.2/tests/misc/mb2.X 2009-04-18 10:41:11.000000000 +0100 @@ -0,0 +1,4 @@ +Citrus＠ＡＡ20＠＠5 +Cherry＠ＡＡ30＠＠10 +Apple＠ＡＡ10＠＠20 +Banana＠ＡＡ5＠＠30 diff -Naur coreutils-7.2.orig/tests/misc/sort-mb-tests coreutils-7.2/tests/misc/sort-mb-tests --- coreutils-7.2.orig/tests/misc/sort-mb-tests 1970-01-01 01:00:00.000000000 +0100 +++ coreutils-7.2/tests/misc/sort-mb-tests 2009-04-18 10:41:11.000000000 +0100 @@ -0,0 +1,58 @@ +#! /bin/sh +case $# in + 0) xx='../src/sort';; + *) xx="$1";; +esac +test "$VERBOSE" && echo=echo || echo=: +$echo testing program: $xx +errors=0 +test "$srcdir" || srcdir=. +test "$VERBOSE" && $xx --version 2> /dev/null + +export LC_ALL=en_US.UTF-8 +locale -k LC_CTYPE 2>&1 | grep -q charmap.*UTF-8 || exit 77 +errors=0 + +$xx -t ＠ -k2 -n misc/mb1.I > misc/mb1.O +code=$? +if test $code != 0; then + $echo "Test mb1 failed: $xx return code $code differs from expected value 0" 1>&2 + errors=`expr $errors + 1` +else + cmp misc/mb1.O $srcdir/misc/mb1.X > /dev/null 2>&1 + case $? in + 0) if test "$VERBOSE"; then $echo "passed mb1"; fi;; + 1) $echo "Test mb1 failed: files misc/mb1.O and $srcdir/misc/mb1.X differ" 1>&2 + (diff -c misc/mb1.O $srcdir/misc/mb1.X) 2> /dev/null + errors=`expr $errors + 1`;; + 2) $echo "Test mb1 may have failed." 1>&2 + $echo The command "cmp misc/mb1.O $srcdir/misc/mb1.X" failed. 1>&2 + errors=`expr $errors + 1`;; + esac +fi + +$xx -t ＠ -k4 -n misc/mb2.I > misc/mb2.O +code=$? +if test $code != 0; then + $echo "Test mb2 failed: $xx return code $code differs from expected value 0" 1>&2 + errors=`expr $errors + 1` +else + cmp misc/mb2.O $srcdir/misc/mb2.X > /dev/null 2>&1 + case $? in + 0) if test "$VERBOSE"; then $echo "passed mb2"; fi;; + 1) $echo "Test mb2 failed: files misc/mb2.O and $srcdir/misc/mb2.X differ" 1>&2 + (diff -c misc/mb2.O $srcdir/misc/mb2.X) 2> /dev/null + errors=`expr $errors + 1`;; + 2) $echo "Test mb2 may have failed." 1>&2 + $echo The command "cmp misc/mb2.O $srcdir/misc/mb2.X" failed. 1>&2 + errors=`expr $errors + 1`;; + esac +fi + +if test $errors = 0; then + $echo Passed all 113 tests. 1>&2 +else + $echo Failed $errors tests. 1>&2 +fi +test $errors = 0 || errors=1 +exit $errors