summaryrefslogtreecommitdiff
path: root/lib/unilbrk/u8-possible-linebreaks.c
diff options
context:
space:
mode:
authorAndreas Rottmann <a.rottmann@gmx.at>2010-05-27 18:23:15 +0200
committerAndreas Rottmann <a.rottmann@gmx.at>2010-05-27 18:23:15 +0200
commit3e0814cd9862b89c7a39672672937477bd87ddfb (patch)
tree159134a624e51509f40ed8823249f09a70d1dda3 /lib/unilbrk/u8-possible-linebreaks.c
parent3bebb94360f1c2333feb8d504a9fa4f96984d8b7 (diff)
Imported Upstream version 0.9.3upstream/0.9.3
Diffstat (limited to 'lib/unilbrk/u8-possible-linebreaks.c')
-rw-r--r--lib/unilbrk/u8-possible-linebreaks.c290
1 files changed, 145 insertions, 145 deletions
diff --git a/lib/unilbrk/u8-possible-linebreaks.c b/lib/unilbrk/u8-possible-linebreaks.c
index 94e14f40..d65335ea 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, 2006-2009 Free Software Foundation, Inc.
+ Copyright (C) 2001-2003, 2006-2010 Free Software Foundation, Inc.
Written by Bruno Haible <bruno@clisp.org>, 2001.
This program is free software: you can redistribute it and/or modify it
@@ -46,117 +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;
+ {
+ 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;
- }
+ /* 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 ();
+ /* 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;
- }
- }
+ 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;
@@ -183,28 +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)
@@ -231,24 +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);