X-Git-Url: http://erislabs.net/gitweb/?a=blobdiff_plain;ds=inline;f=lib%2Funilbrk%2Fu8-possible-linebreaks.c;h=58a15a7521f87757f9d0f2478d54bfae1de8fd2a;hb=e16561484b19d960cbb07bd82033155b2979c4f0;hp=fa1fbc136c816e35ab4f221029df0771a9fe0d03;hpb=9abda809d44be61025d67d5980a4e780010c7b5b;p=gnulib.git diff --git a/lib/unilbrk/u8-possible-linebreaks.c b/lib/unilbrk/u8-possible-linebreaks.c index fa1fbc136..58a15a752 100644 --- a/lib/unilbrk/u8-possible-linebreaks.c +++ b/lib/unilbrk/u8-possible-linebreaks.c @@ -1,5 +1,5 @@ /* Line breaking of UTF-8 strings. - Copyright (C) 2001-2003, 2006-2008 Free Software Foundation, Inc. + Copyright (C) 2001-2003, 2006-2012 Free Software Foundation, Inc. Written by Bruno Haible , 2001. This program is free software: you can redistribute it and/or modify it @@ -23,7 +23,7 @@ #include #include -#include "unilbrk/tables.h" +#include "unilbrk/lbrktables.h" #include "uniwidth/cjk.h" #include "unistr.h" @@ -46,93 +46,117 @@ u8_possible_linebreaks (const uint8_t *s, size_t n, const char *encoding, char * int prop = unilbrkprop_lookup (uc); if (prop == LBP_BK) - { - /* Mandatory break. */ - *p = UC_BREAK_MANDATORY; - last_prop = LBP_BK; - seen_space = NULL; - seen_space2 = NULL; - } + { + /* Mandatory break. */ + *p = UC_BREAK_MANDATORY; + last_prop = LBP_BK; + seen_space = NULL; + seen_space2 = NULL; + } else - { - char *q; - - /* Resolve property values whose behaviour is not fixed. */ - switch (prop) - { - case LBP_AI: - /* Resolve ambiguous. */ - prop = LBP_AI_REPLACEMENT; - break; - case LBP_CB: - /* This is arbitrary. */ - prop = LBP_ID; - break; - case LBP_SA: - /* We don't handle complex scripts yet. - Treat LBP_SA like LBP_XX. */ - case LBP_XX: - /* This is arbitrary. */ - prop = LBP_AL; - break; - } - - /* Deal with combining characters. */ - q = p; - if (prop == LBP_CM) - { - /* Don't break just before a combining character. */ - *p = UC_BREAK_PROHIBITED; - /* A combining character turns a preceding space into LBP_AL. */ - if (seen_space != NULL) - { - q = seen_space; - seen_space = seen_space2; - prop = LBP_AL; - goto lookup_via_table; - } - } - else if (prop == LBP_SP) - { - /* Don't break just before a space. */ - *p = UC_BREAK_PROHIBITED; - seen_space2 = seen_space; - seen_space = p; - } - else - { - lookup_via_table: - /* prop must be usable as an index for table 7.3 of UTR #14. */ - if (!(prop >= 1 && prop <= sizeof (unilbrk_table) / sizeof (unilbrk_table[0]))) - abort (); - - if (last_prop == LBP_BK) - { - /* Don't break at the beginning of a line. */ - *q = UC_BREAK_PROHIBITED; - } - else - { - switch (unilbrk_table [last_prop-1] [prop-1]) - { - case D: - *q = UC_BREAK_POSSIBLE; - break; - case I: - *q = (seen_space != NULL ? UC_BREAK_POSSIBLE : UC_BREAK_PROHIBITED); - break; - case P: - *q = UC_BREAK_PROHIBITED; - break; - default: - abort (); - } - } - last_prop = prop; - seen_space = NULL; - seen_space2 = NULL; - } - } + { + char *q; + + /* Resolve property values whose behaviour is not fixed. */ + switch (prop) + { + case LBP_AI: + /* Resolve ambiguous. */ + prop = LBP_AI_REPLACEMENT; + break; + case LBP_CB: + /* This is arbitrary. */ + prop = LBP_ID; + break; + case LBP_SA: + /* We don't handle complex scripts yet. + Treat LBP_SA like LBP_XX. */ + case LBP_XX: + /* This is arbitrary. */ + prop = LBP_AL; + break; + } + + /* Deal with spaces and combining characters. */ + q = p; + if (prop == LBP_SP) + { + /* Don't break just before a space. */ + *p = UC_BREAK_PROHIBITED; + seen_space2 = seen_space; + seen_space = p; + } + else if (prop == LBP_ZW) + { + /* Don't break just before a zero-width space. */ + *p = UC_BREAK_PROHIBITED; + last_prop = LBP_ZW; + seen_space = NULL; + seen_space2 = NULL; + } + else if (prop == LBP_CM) + { + /* Don't break just before a combining character, except immediately after a + zero-width space. */ + if (last_prop == LBP_ZW) + { + /* Break after zero-width space. */ + *p = UC_BREAK_POSSIBLE; + /* A combining character turns a preceding space into LBP_ID. */ + last_prop = LBP_ID; + } + else + { + *p = UC_BREAK_PROHIBITED; + /* A combining character turns a preceding space into LBP_ID. */ + if (seen_space != NULL) + { + q = seen_space; + seen_space = seen_space2; + prop = LBP_ID; + goto lookup_via_table; + } + } + } + else + { + lookup_via_table: + /* prop must be usable as an index for table 7.3 of UTR #14. */ + if (!(prop >= 0 && prop < sizeof (unilbrk_table) / sizeof (unilbrk_table[0]))) + abort (); + + if (last_prop == LBP_BK) + { + /* Don't break at the beginning of a line. */ + *q = UC_BREAK_PROHIBITED; + } + else if (last_prop == LBP_ZW) + { + /* Break after zero-width space. */ + *q = UC_BREAK_POSSIBLE; + } + else + { + switch (unilbrk_table [last_prop] [prop]) + { + case D: + *q = UC_BREAK_POSSIBLE; + break; + case I: + *q = (seen_space != NULL ? UC_BREAK_POSSIBLE : UC_BREAK_PROHIBITED); + break; + case P: + *q = UC_BREAK_PROHIBITED; + break; + default: + abort (); + } + } + last_prop = prop; + seen_space = NULL; + seen_space2 = NULL; + } + } s += count; p += count; @@ -159,28 +183,28 @@ read_file (FILE *stream) while (! feof (stream)) { if (size + BUFSIZE > alloc) - { - alloc = alloc + alloc / 2; - if (alloc < size + BUFSIZE) - alloc = size + BUFSIZE; - buf = realloc (buf, alloc); - if (buf == NULL) - { - fprintf (stderr, "out of memory\n"); - exit (1); - } - } + { + alloc = alloc + alloc / 2; + if (alloc < size + BUFSIZE) + alloc = size + BUFSIZE; + buf = realloc (buf, alloc); + if (buf == NULL) + { + fprintf (stderr, "out of memory\n"); + exit (1); + } + } count = fread (buf + size, 1, BUFSIZE, stream); if (count == 0) - { - if (ferror (stream)) - { - perror ("fread"); - exit (1); - } - } + { + if (ferror (stream)) + { + perror ("fread"); + exit (1); + } + } else - size += count; + size += count; } buf = realloc (buf, size + 1); if (buf == NULL) @@ -207,24 +231,24 @@ main (int argc, char * argv[]) u8_possible_linebreaks ((uint8_t *) input, length, "UTF-8", breaks); for (i = 0; i < length; i++) - { - switch (breaks[i]) - { - case UC_BREAK_POSSIBLE: - /* U+2027 in UTF-8 encoding */ - putc (0xe2, stdout); putc (0x80, stdout); putc (0xa7, stdout); - break; - case UC_BREAK_MANDATORY: - /* U+21B2 (or U+21B5) in UTF-8 encoding */ - putc (0xe2, stdout); putc (0x86, stdout); putc (0xb2, stdout); - break; - case UC_BREAK_PROHIBITED: - break; - default: - abort (); - } - putc (input[i], stdout); - } + { + switch (breaks[i]) + { + case UC_BREAK_POSSIBLE: + /* U+2027 in UTF-8 encoding */ + putc (0xe2, stdout); putc (0x80, stdout); putc (0xa7, stdout); + break; + case UC_BREAK_MANDATORY: + /* U+21B2 (or U+21B5) in UTF-8 encoding */ + putc (0xe2, stdout); putc (0x86, stdout); putc (0xb2, stdout); + break; + case UC_BREAK_PROHIBITED: + break; + default: + abort (); + } + putc (input[i], stdout); + } free (breaks);