1 /* Copyright (C) 1996-2008, 2009, 2010, 2011 Free Software Foundation, Inc.
2 This file is part of the GNU C Library.
3 Contributed by Ulrich Drepper <drepper@redhat.com>, 1996.
5 This program is free software; you can redistribute it and/or modify
6 it under the terms of the GNU General Public License as published
7 by the Free Software Foundation; version 2 of the License, or
8 (at your option) any later version.
10 This program is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 GNU General Public License for more details.
15 You should have received a copy of the GNU General Public License
16 along with this program; if not, write to the Free Software Foundation,
17 Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
46 #include "catgetsinfo.h"
50 (((w) << 24) | (((w) & 0xff00) << 8) | (((w) >> 8) & 0xff00) | ((w) >> 24))
61 struct message_list
*next
;
69 struct message_list
*messages
;
76 struct set_list
*next
;
82 struct set_list
*all_sets
;
83 struct set_list
*current_set
;
84 size_t total_messages
;
88 struct obstack mem_pool
;
92 /* If non-zero force creation of new file, not using existing one. */
95 /* Name of output file. */
96 static const char *output_name
;
98 /* Name of generated C header file. */
99 static const char *header_name
;
101 /* Name and version of program. */
102 static void print_version (FILE *stream
, struct argp_state
*state
);
103 void (*argp_program_version_hook
) (FILE *, struct argp_state
*) = print_version
;
107 /* Definitions of arguments for argp functions. */
108 static const struct argp_option options
[] =
110 { "header", 'H', N_("NAME"), 0,
111 N_("Create C header file NAME containing symbol definitions") },
112 { "new", OPT_NEW
, NULL
, 0,
113 N_("Do not use existing catalog, force new output file") },
114 { "output", 'o', N_("NAME"), 0, N_("Write output to file NAME") },
115 { NULL
, 0, NULL
, 0, NULL
}
118 /* Short description of program. */
119 static const char doc
[] = N_("Generate message catalog.\
120 \vIf INPUT-FILE is -, input is read from standard input. If OUTPUT-FILE\n\
121 is -, output is written to standard output.\n");
123 /* Strings for arguments in help texts. */
124 static const char args_doc
[] = N_("\
125 -o OUTPUT-FILE [INPUT-FILE]...\n[OUTPUT-FILE [INPUT-FILE]...]");
127 /* Prototype for option handler. */
128 static error_t
parse_opt (int key
, char *arg
, struct argp_state
*state
);
130 /* Function to print some extra text in the help message. */
131 static char *more_help (int key
, const char *text
, void *input
);
133 /* Data structure to communicate with argp functions. */
134 static struct argp argp
=
136 options
, parse_opt
, args_doc
, doc
, NULL
, more_help
140 /* Wrapper functions with error checking for standard functions. */
141 extern void *xmalloc (size_t n
);
142 extern void *xcalloc (size_t n
, size_t s
);
143 extern void *xrealloc (void *o
, size_t n
);
144 extern char *xstrdup (const char *);
146 /* Prototypes for local functions. */
147 static void error_print (void);
148 static struct catalog
*read_input_file (struct catalog
*current
,
150 static void write_out (struct catalog
*result
, const char *output_name
,
151 const char *header_name
);
152 static struct set_list
*find_set (struct catalog
*current
, int number
);
153 static void normalize_line (const char *fname
, size_t line
, iconv_t cd
,
154 wchar_t *string
, wchar_t quote_char
,
155 wchar_t escape_char
);
156 static void read_old (struct catalog
*catalog
, const char *file_name
);
157 static int open_conversion (const char *codesetp
, iconv_t
*cd_towcp
,
158 iconv_t
*cd_tombp
, wchar_t *escape_charp
);
162 main (int argc
, char *argv
[])
164 struct catalog
*result
;
167 /* Set program name for messages. */
168 error_print_progname
= error_print
;
170 /* Set locale via LC_ALL. */
171 setlocale (LC_ALL
, "");
173 /* Set the text message domain. */
174 textdomain (PACKAGE
);
176 /* Initialize local variables. */
179 /* Parse and process arguments. */
180 argp_parse (&argp
, argc
, argv
, 0, &remaining
, NULL
);
182 /* Determine output file. */
183 if (output_name
== NULL
)
184 output_name
= remaining
< argc
? argv
[remaining
++] : "-";
186 /* Process all input files. */
187 setlocale (LC_CTYPE
, "C");
188 if (remaining
< argc
)
190 result
= read_input_file (result
, argv
[remaining
]);
191 while (++remaining
< argc
);
193 result
= read_input_file (NULL
, "-");
195 /* Write out the result. */
197 write_out (result
, output_name
, header_name
);
199 return error_message_count
!= 0;
203 /* Handle program arguments. */
205 parse_opt (int key
, char *arg
, struct argp_state
*state
)
219 return ARGP_ERR_UNKNOWN
;
226 more_help (int key
, const char *text
, void *input
)
230 case ARGP_KEY_HELP_EXTRA
:
231 /* We print some extra information. */
232 return strdup (gettext ("\
233 For bug reporting instructions, please see:\n\
234 <http://www.gnu.org/software/libc/bugs.html>.\n"));
238 return (char *) text
;
241 /* Print the version information. */
243 print_version (FILE *stream
, struct argp_state
*state
)
245 fprintf (stream
, "gencat (GNU %s) %s\n", PACKAGE
, VERSION
);
246 fprintf (stream
, gettext ("\
247 Copyright (C) %s Free Software Foundation, Inc.\n\
248 This is free software; see the source for copying conditions. There is NO\n\
249 warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.\n\
251 fprintf (stream
, gettext ("Written by %s.\n"), "Ulrich Drepper");
255 /* The address of this function will be assigned to the hook in the
260 /* We don't want the program name to be printed in messages. Emacs'
261 compile.el does not like this. */
265 static struct catalog
*
266 read_input_file (struct catalog
*current
, const char *fname
)
274 iconv_t cd_towc
= (iconv_t
) -1;
275 iconv_t cd_tomb
= (iconv_t
) -1;
276 wchar_t escape_char
= L
'\\';
277 char *codeset
= NULL
;
279 if (strcmp (fname
, "-") == 0 || strcmp (fname
, "/dev/stdin") == 0)
282 fname
= gettext ("*standard input*");
285 fp
= fopen (fname
, "r");
288 error (0, errno
, gettext ("cannot open input file `%s'"), fname
);
292 /* If we haven't seen anything yet, allocate result structure. */
295 current
= (struct catalog
*) xcalloc (1, sizeof (*current
));
297 #define obstack_chunk_alloc malloc
298 #define obstack_chunk_free free
299 obstack_init (¤t
->mem_pool
);
301 current
->current_set
= find_set (current
, NL_SETD
);
309 wbuf
= (wchar_t *) xmalloc (wbufsize
);
315 size_t start_line
= line_number
+ 1;
322 act_len
= getline (&buf
, &len
, fp
);
327 /* It the line continued? */
329 if (buf
[act_len
- 1] == '\n')
333 /* There might be more than one backslash at the end of
334 the line. Only if there is an odd number of them is
335 the line continued. */
336 if (act_len
> 0 && buf
[act_len
- 1] == '\\')
338 int temp_act_len
= act_len
;
343 continued
= !continued
;
345 while (temp_act_len
> 0 && buf
[temp_act_len
- 1] == '\\');
352 /* Append to currently selected line. */
353 obstack_grow (¤t
->mem_pool
, buf
, act_len
);
357 obstack_1grow (¤t
->mem_pool
, '\0');
358 this_line
= (char *) obstack_finish (¤t
->mem_pool
);
361 if (this_line
[0] == '$')
363 if (isblank (this_line
[1]))
366 while (isblank (this_line
[cnt
]))
368 if (strncmp (&this_line
[cnt
], "codeset=", 8) != 0)
369 /* This is a comment line. Do nothing. */;
370 else if (codeset
!= NULL
)
371 /* Ignore multiple codeset. */;
376 while (this_line
[cnt
] != '\0' && !isspace (this_line
[cnt
]))
380 int len
= cnt
- start
;
381 codeset
= xmalloc (len
+ 1);
382 *((char *) mempcpy (codeset
, &this_line
[start
], len
))
387 else if (strncmp (&this_line
[1], "set", 3) == 0)
389 int cnt
= sizeof ("set");
391 const char *symbol
= NULL
;
392 while (isspace (this_line
[cnt
]))
395 if (isdigit (this_line
[cnt
]))
397 set_number
= atol (&this_line
[cnt
]);
399 /* If the given number for the character set is
400 higher than any we used for symbolic set names
401 avoid clashing by using only higher numbers for
402 the following symbolic definitions. */
403 if (set_number
> current
->last_set
)
404 current
->last_set
= set_number
;
408 /* See whether it is a reasonable identifier. */
410 while (isalnum (this_line
[cnt
]) || this_line
[cnt
] == '_')
415 /* No correct character found. */
416 error_at_line (0, 0, fname
, start_line
,
417 gettext ("illegal set number"));
422 /* We have found seomthing that looks like a
423 correct identifier. */
424 struct set_list
*runp
;
426 this_line
[cnt
] = '\0';
428 symbol
= &this_line
[start
];
430 /* Test whether the identifier was already used. */
431 runp
= current
->all_sets
;
433 if (runp
->symbol
!= NULL
434 && strcmp (runp
->symbol
, symbol
) == 0)
441 /* We cannot allow duplicate identifiers for
443 error_at_line (0, 0, fname
, start_line
,
444 gettext ("duplicate set definition"));
445 error_at_line (0, 0, runp
->fname
, runp
->line
,
447 this is the first definition"));
451 /* Allocate next free message set for identifier. */
452 set_number
= ++current
->last_set
;
458 /* We found a legal set number. */
459 current
->current_set
= find_set (current
, set_number
);
462 current
->current_set
->symbol
= symbol
;
463 current
->current_set
->fname
= fname
;
464 current
->current_set
->line
= start_line
;
467 else if (strncmp (&this_line
[1], "delset", 6) == 0)
469 int cnt
= sizeof ("delset");
470 while (isspace (this_line
[cnt
]))
473 if (isdigit (this_line
[cnt
]))
475 size_t set_number
= atol (&this_line
[cnt
]);
476 struct set_list
*set
;
478 /* Mark the message set with the given number as
480 set
= find_set (current
, set_number
);
485 /* See whether it is a reasonable identifier. */
487 while (isalnum (this_line
[cnt
]) || this_line
[cnt
] == '_')
491 error_at_line (0, 0, fname
, start_line
,
492 gettext ("illegal set number"));
496 struct set_list
*runp
;
498 this_line
[cnt
] = '\0';
500 symbol
= &this_line
[start
];
502 /* We have a symbolic set name. This name must
503 appear somewhere else in the catalogs read so
505 for (runp
= current
->all_sets
; runp
!= NULL
;
508 if (strcmp (runp
->symbol
, symbol
) == 0)
515 /* Name does not exist before. */
516 error_at_line (0, 0, fname
, start_line
,
517 gettext ("unknown set `%s'"), symbol
);
521 else if (strncmp (&this_line
[1], "quote", 5) == 0)
530 cnt
= sizeof ("quote");
531 while (isspace (this_line
[cnt
]))
534 /* We need the conversion. */
535 if (cd_towc
== (iconv_t
) -1
536 && open_conversion (codeset
, &cd_towc
, &cd_tomb
,
538 /* Something is wrong. */
541 /* Yes, the quote char can be '\0'; this means no quote
542 char. The function using the information works on
543 wide characters so we have to convert it here. */
544 buf
[0] = this_line
[cnt
];
549 wbufptr
= (char *) wbuf
;
552 /* Flush the state. */
553 iconv (cd_towc
, NULL
, NULL
, NULL
, NULL
);
555 iconv (cd_towc
, &bufptr
, &buflen
, &wbufptr
, &wbuflen
);
556 if (buflen
!= 0 || (wchar_t *) wbufptr
!= &wbuf
[2])
557 error_at_line (0, 0, fname
, start_line
,
558 gettext ("invalid quote character"));
560 /* Use the converted wide character. */
561 current
->quote_char
= wbuf
[0];
567 while (this_line
[cnt
] != '\0' && !isspace (this_line
[cnt
]))
569 this_line
[cnt
] = '\0';
570 error_at_line (0, 0, fname
, start_line
,
571 gettext ("unknown directive `%s': line ignored"),
575 else if (isalnum (this_line
[0]) || this_line
[0] == '_')
577 const char *ident
= this_line
;
578 char *line
= this_line
;
583 while (line
[0] != '\0' && !isspace (line
[0]));
585 *line
++ = '\0'; /* Terminate the identifier. */
587 /* Now we found the beginning of the message itself. */
589 if (isdigit (ident
[0]))
591 struct message_list
*runp
;
592 struct message_list
*lastp
;
594 message_number
= atoi (ident
);
596 /* Find location to insert the new message. */
597 runp
= current
->current_set
->messages
;
600 if (runp
->number
== message_number
)
609 /* Oh, oh. There is already a message with this
610 number in the message set. */
611 if (runp
->symbol
== NULL
)
613 /* The existing message had its number specified
614 by the user. Fatal collision type uh, oh. */
615 error_at_line (0, 0, fname
, start_line
,
616 gettext ("duplicated message number"));
617 error_at_line (0, 0, runp
->fname
, runp
->line
,
618 gettext ("this is the first definition"));
623 /* Collision was with number auto-assigned to a
624 symbolic. Change existing symbolic number
625 and move to end the list (if not already there). */
626 runp
->number
= ++current
->current_set
->last_message
;
628 if (runp
->next
!= NULL
)
630 struct message_list
*endp
;
633 current
->current_set
->messages
=runp
->next
;
635 lastp
->next
=runp
->next
;
638 while (endp
->next
!= NULL
)
646 ident
= NULL
; /* We don't have a symbol. */
648 if (message_number
!= 0
649 && message_number
> current
->current_set
->last_message
)
650 current
->current_set
->last_message
= message_number
;
652 else if (ident
[0] != '\0')
654 struct message_list
*runp
;
656 /* Test whether the symbolic name was not used for
657 another message in this message set. */
658 runp
= current
->current_set
->messages
;
660 if (runp
->symbol
!= NULL
&& strcmp (ident
, runp
->symbol
) == 0)
666 /* The name is already used. */
667 error_at_line (0, 0, fname
, start_line
, gettext ("\
668 duplicated message identifier"));
669 error_at_line (0, 0, runp
->fname
, runp
->line
,
670 gettext ("this is the first definition"));
674 /* Give the message the next unused number. */
675 message_number
= ++current
->current_set
->last_message
;
680 if (message_number
!= 0)
686 struct message_list
*newp
;
687 size_t line_len
= strlen (line
) + 1;
688 size_t ident_len
= 0;
690 /* We need the conversion. */
691 if (cd_towc
== (iconv_t
) -1
692 && open_conversion (codeset
, &cd_towc
, &cd_tomb
,
694 /* Something is wrong. */
697 /* Convert to a wide character string. We have to
698 interpret escape sequences which will be impossible
699 without doing the conversion if the codeset of the
700 message is stateful. */
705 outbuf
= (char *) wbuf
;
708 /* Flush the state. */
709 iconv (cd_towc
, NULL
, NULL
, NULL
, NULL
);
711 iconv (cd_towc
, &inbuf
, &inlen
, &outbuf
, &outlen
);
714 /* The string is converted. */
715 assert (outlen
< wbufsize
);
716 assert (wbuf
[(wbufsize
- outlen
) / sizeof (wchar_t) - 1]
723 /* Something is wrong with this string, we ignore it. */
724 error_at_line (0, 0, fname
, start_line
, gettext ("\
725 invalid character: message ignored"));
729 /* The output buffer is too small. */
731 wbuf
= (wchar_t *) xrealloc (wbuf
, wbufsize
);
734 /* Strip quote characters, change escape sequences into
735 correct characters etc. */
736 normalize_line (fname
, start_line
, cd_towc
, wbuf
,
737 current
->quote_char
, escape_char
);
740 ident_len
= line
- this_line
;
742 /* Now the string is free of escape sequences. Convert it
743 back into a multibyte character string. First free the
744 memory allocated for the original string. */
745 obstack_free (¤t
->mem_pool
, this_line
);
747 used
= 1; /* Yes, we use the line. */
749 /* Now fill in the new string. It should never happen that
750 the replaced string is longer than the original. */
751 inbuf
= (char *) wbuf
;
752 inlen
= (wcslen (wbuf
) + 1) * sizeof (wchar_t);
754 outlen
= obstack_room (¤t
->mem_pool
);
755 obstack_blank (¤t
->mem_pool
, outlen
);
756 this_line
= (char *) obstack_base (¤t
->mem_pool
);
757 outbuf
= this_line
+ ident_len
;
760 /* Flush the state. */
761 iconv (cd_tomb
, NULL
, NULL
, NULL
, NULL
);
763 iconv (cd_tomb
, &inbuf
, &inlen
, &outbuf
, &outlen
);
766 error_at_line (0, 0, fname
, start_line
,
767 gettext ("invalid line"));
770 assert (outbuf
[-1] == '\0');
772 /* Free the memory in the obstack we don't use. */
773 obstack_blank (¤t
->mem_pool
, -(int) outlen
);
774 line
= obstack_finish (¤t
->mem_pool
);
776 newp
= (struct message_list
*) xmalloc (sizeof (*newp
));
777 newp
->number
= message_number
;
778 newp
->message
= line
+ ident_len
;
779 /* Remember symbolic name; is NULL if no is given. */
780 newp
->symbol
= ident
? line
: NULL
;
781 /* Remember where we found the character. */
783 newp
->line
= start_line
;
785 /* Find place to insert to message. We keep them in a
786 sorted single linked list. */
787 if (current
->current_set
->messages
== NULL
788 || current
->current_set
->messages
->number
> message_number
)
790 newp
->next
= current
->current_set
->messages
;
791 current
->current_set
->messages
= newp
;
795 struct message_list
*runp
;
796 runp
= current
->current_set
->messages
;
797 while (runp
->next
!= NULL
)
798 if (runp
->next
->number
> message_number
)
802 newp
->next
= runp
->next
;
806 ++current
->total_messages
;
813 /* See whether we have any non-white space character in this
815 while (this_line
[cnt
] != '\0' && isspace (this_line
[cnt
]))
818 if (this_line
[cnt
] != '\0')
819 /* Yes, some unknown characters found. */
820 error_at_line (0, 0, fname
, start_line
,
821 gettext ("malformed line ignored"));
825 /* We can save the memory for the line if it was not used. */
827 obstack_free (¤t
->mem_pool
, this_line
);
830 /* Close the conversion modules. */
831 iconv_close (cd_towc
);
832 iconv_close (cd_tomb
);
845 write_out (struct catalog
*catalog
, const char *output_name
,
846 const char *header_name
)
848 /* Computing the "optimal" size. */
849 struct set_list
*set_run
;
850 size_t best_total
, best_size
, best_depth
;
851 size_t act_size
, act_depth
;
852 struct catalog_obj obj
;
853 struct obstack string_pool
;
856 uint32_t *array1
, *array2
;
860 /* If not otherwise told try to read file with existing
863 read_old (catalog
, output_name
);
865 /* Initialize best_size with a very high value. */
866 best_total
= best_size
= best_depth
= UINT_MAX
;
868 /* We need some start size for testing. Let's start with
869 TOTAL_MESSAGES / 5, which theoretically provides a mean depth of
871 act_size
= 1 + catalog
->total_messages
/ 5;
873 /* We determine the size of a hash table here. Because the message
874 numbers can be chosen arbitrary by the programmer we cannot use
875 the simple method of accessing the array using the message
876 number. The algorithm is based on the trivial hash function
877 NUMBER % TABLE_SIZE, where collisions are stored in a second
878 dimension up to TABLE_DEPTH. We here compute TABLE_SIZE so that
879 the needed space (= TABLE_SIZE * TABLE_DEPTH) is minimal. */
880 while (act_size
<= best_total
)
882 size_t deep
[act_size
];
885 memset (deep
, '\0', act_size
* sizeof (size_t));
886 set_run
= catalog
->all_sets
;
887 while (set_run
!= NULL
)
889 struct message_list
*message_run
;
891 message_run
= set_run
->messages
;
892 while (message_run
!= NULL
)
894 size_t idx
= (message_run
->number
* set_run
->number
) % act_size
;
897 if (deep
[idx
] > act_depth
)
899 act_depth
= deep
[idx
];
900 if (act_depth
* act_size
> best_total
)
903 message_run
= message_run
->next
;
905 set_run
= set_run
->next
;
908 if (act_depth
* act_size
<= best_total
)
910 /* We have found a better solution. */
911 best_total
= act_depth
* act_size
;
912 best_size
= act_size
;
913 best_depth
= act_depth
;
919 /* let's be prepared for an empty message file. */
920 if (best_size
== UINT_MAX
)
926 /* OK, now we have the size we will use. Fill in the header, build
927 the table and the second one with swapped byte order. */
928 obj
.magic
= CATGETS_MAGIC
;
929 obj
.plane_size
= best_size
;
930 obj
.plane_depth
= best_depth
;
932 /* Allocate room for all needed arrays. */
934 (uint32_t *) alloca (best_size
* best_depth
* sizeof (uint32_t) * 3);
935 memset (array1
, '\0', best_size
* best_depth
* sizeof (uint32_t) * 3);
937 = (uint32_t *) alloca (best_size
* best_depth
* sizeof (uint32_t) * 3);
938 obstack_init (&string_pool
);
940 set_run
= catalog
->all_sets
;
941 while (set_run
!= NULL
)
943 struct message_list
*message_run
;
945 message_run
= set_run
->messages
;
946 while (message_run
!= NULL
)
948 size_t idx
= (((message_run
->number
* set_run
->number
) % best_size
)
950 /* Determine collision depth. */
951 while (array1
[idx
] != 0)
952 idx
+= best_size
* 3;
954 /* Store set number, message number and pointer into string
955 space, relative to the first string. */
956 array1
[idx
+ 0] = set_run
->number
;
957 array1
[idx
+ 1] = message_run
->number
;
958 array1
[idx
+ 2] = obstack_object_size (&string_pool
);
960 /* Add current string to the continuous space containing all
962 obstack_grow0 (&string_pool
, message_run
->message
,
963 strlen (message_run
->message
));
965 message_run
= message_run
->next
;
968 set_run
= set_run
->next
;
970 strings_size
= obstack_object_size (&string_pool
);
971 strings
= obstack_finish (&string_pool
);
973 /* Compute ARRAY2 by changing the byte order. */
974 for (cnt
= 0; cnt
< best_size
* best_depth
* 3; ++cnt
)
975 array2
[cnt
] = SWAPU32 (array1
[cnt
]);
977 /* Now we can write out the whole data. */
978 if (strcmp (output_name
, "-") == 0
979 || strcmp (output_name
, "/dev/stdout") == 0)
983 fd
= creat (output_name
, 0666);
985 error (EXIT_FAILURE
, errno
, gettext ("cannot open output file `%s'"),
989 /* Write out header. */
990 write (fd
, &obj
, sizeof (obj
));
992 /* We always write out the little endian version of the index
994 #if __BYTE_ORDER == __LITTLE_ENDIAN
995 write (fd
, array1
, best_size
* best_depth
* sizeof (uint32_t) * 3);
996 write (fd
, array2
, best_size
* best_depth
* sizeof (uint32_t) * 3);
997 #elif __BYTE_ORDER == __BIG_ENDIAN
998 write (fd
, array2
, best_size
* best_depth
* sizeof (uint32_t) * 3);
999 write (fd
, array1
, best_size
* best_depth
* sizeof (uint32_t) * 3);
1001 # error Cannot handle __BYTE_ORDER byte order
1004 /* Finally write the strings. */
1005 write (fd
, strings
, strings_size
);
1007 if (fd
!= STDOUT_FILENO
)
1010 /* If requested now write out the header file. */
1011 if (header_name
!= NULL
)
1016 /* Open output file. "-" or "/dev/stdout" means write to
1018 if (strcmp (header_name
, "-") == 0
1019 || strcmp (header_name
, "/dev/stdout") == 0)
1023 fp
= fopen (header_name
, "w");
1025 error (EXIT_FAILURE
, errno
,
1026 gettext ("cannot open output file `%s'"), header_name
);
1029 /* Iterate over all sets and all messages. */
1030 set_run
= catalog
->all_sets
;
1031 while (set_run
!= NULL
)
1033 struct message_list
*message_run
;
1035 /* If the current message set has a symbolic name write this
1037 if (set_run
->symbol
!= NULL
)
1038 fprintf (fp
, "%s#define %sSet %#x\t/* %s:%Zu */\n",
1039 first
? "" : "\n", set_run
->symbol
, set_run
->number
- 1,
1040 set_run
->fname
, set_run
->line
);
1043 message_run
= set_run
->messages
;
1044 while (message_run
!= NULL
)
1046 /* If the current message has a symbolic name write
1047 #define out. But we have to take care for the set
1048 not having a symbolic name. */
1049 if (message_run
->symbol
!= NULL
)
1051 if (set_run
->symbol
== NULL
)
1052 fprintf (fp
, "#define AutomaticSet%d%s %#x\t/* %s:%Zu */\n",
1053 set_run
->number
, message_run
->symbol
,
1054 message_run
->number
, message_run
->fname
,
1057 fprintf (fp
, "#define %s%s %#x\t/* %s:%Zu */\n",
1058 set_run
->symbol
, message_run
->symbol
,
1059 message_run
->number
, message_run
->fname
,
1063 message_run
= message_run
->next
;
1066 set_run
= set_run
->next
;
1075 static struct set_list
*
1076 find_set (struct catalog
*current
, int number
)
1078 struct set_list
*result
= current
->all_sets
;
1080 /* We must avoid set number 0 because a set of this number signals
1081 in the tables that the entry is not occupied. */
1084 while (result
!= NULL
)
1085 if (result
->number
== number
)
1088 result
= result
->next
;
1090 /* Prepare new message set. */
1091 result
= (struct set_list
*) xcalloc (1, sizeof (*result
));
1092 result
->number
= number
;
1093 result
->next
= current
->all_sets
;
1094 current
->all_sets
= result
;
1100 /* Normalize given string *in*place* by processing escape sequences
1101 and quote characters. */
1103 normalize_line (const char *fname
, size_t line
, iconv_t cd
, wchar_t *string
,
1104 wchar_t quote_char
, wchar_t escape_char
)
1107 wchar_t *rp
= string
;
1108 wchar_t *wp
= string
;
1110 if (quote_char
!= L
'\0' && *rp
== quote_char
)
1118 while (*rp
!= L
'\0')
1119 if (*rp
== quote_char
)
1120 /* We simply end the string when we find the first time an
1121 not-escaped quote character. */
1123 else if (*rp
== escape_char
)
1126 if (quote_char
!= L
'\0' && *rp
== quote_char
)
1127 /* This is an extension to XPG. */
1130 /* Recognize escape sequences. */
1167 number
= *rp
++ - L
'0';
1168 while (number
<= (255 / 8) && *rp
>= L
'0' && *rp
<= L
'7')
1171 number
+= *rp
++ - L
'0';
1174 cbuf
[0] = (char) number
;
1179 wcbufptr
= (char *) wcbuf
;
1180 wcbufin
= sizeof (wcbuf
);
1182 /* Flush the state. */
1183 iconv (cd
, NULL
, NULL
, NULL
, NULL
);
1185 iconv (cd
, &cbufptr
, &cbufin
, &wcbufptr
, &wcbufin
);
1186 if (cbufptr
!= &cbuf
[2] || (wchar_t *) wcbufptr
!= &wcbuf
[2])
1187 error_at_line (0, 0, fname
, line
,
1188 gettext ("invalid escape sequence"));
1194 if (*rp
== escape_char
)
1196 *wp
++ = escape_char
;
1200 /* Simply ignore the backslash character. */;
1207 /* If we saw a quote character at the beginning we expect another
1209 if (is_quoted
&& *rp
!= quote_char
)
1210 error_at_line (0, 0, fname
, line
, gettext ("unterminated message"));
1212 /* Terminate string. */
1219 read_old (struct catalog
*catalog
, const char *file_name
)
1221 struct catalog_info old_cat_obj
;
1222 struct set_list
*set
= NULL
;
1226 /* Try to open catalog, but don't look through the NLSPATH. */
1227 if (__open_catalog (file_name
, NULL
, NULL
, &old_cat_obj
) != 0)
1229 if (errno
== ENOENT
)
1230 /* No problem, the catalog simply does not exist. */
1233 error (EXIT_FAILURE
, errno
,
1234 gettext ("while opening old catalog file"));
1237 /* OK, we have the catalog loaded. Now read all messages and merge
1238 them. When set and message number clash for any message the new
1239 one is used. If the new one is empty it indicates that the
1240 message should be deleted. */
1241 for (cnt
= 0; cnt
< old_cat_obj
.plane_size
* old_cat_obj
.plane_depth
; ++cnt
)
1243 struct message_list
*message
, *last
;
1245 if (old_cat_obj
.name_ptr
[cnt
* 3 + 0] == 0)
1246 /* No message in this slot. */
1249 if (old_cat_obj
.name_ptr
[cnt
* 3 + 0] - 1 != (uint32_t) last_set
)
1251 last_set
= old_cat_obj
.name_ptr
[cnt
* 3 + 0] - 1;
1252 set
= find_set (catalog
, old_cat_obj
.name_ptr
[cnt
* 3 + 0] - 1);
1256 message
= set
->messages
;
1257 while (message
!= NULL
)
1259 if ((uint32_t) message
->number
>= old_cat_obj
.name_ptr
[cnt
* 3 + 1])
1262 message
= message
->next
;
1266 || (uint32_t) message
->number
> old_cat_obj
.name_ptr
[cnt
* 3 + 1])
1268 /* We have found a message which is not yet in the catalog.
1269 Insert it at the right position. */
1270 struct message_list
*newp
;
1272 newp
= (struct message_list
*) xmalloc (sizeof(*newp
));
1273 newp
->number
= old_cat_obj
.name_ptr
[cnt
* 3 + 1];
1275 &old_cat_obj
.strings
[old_cat_obj
.name_ptr
[cnt
* 3 + 2]];
1278 newp
->symbol
= NULL
;
1279 newp
->next
= message
;
1282 set
->messages
= newp
;
1286 ++catalog
->total_messages
;
1288 else if (*message
->message
== '\0')
1290 /* The new empty message has overridden the old one thus
1291 "deleting" it as required. Now remove the empty remains. */
1293 set
->messages
= message
->next
;
1295 last
->next
= message
->next
;
1302 open_conversion (const char *codeset
, iconv_t
*cd_towcp
, iconv_t
*cd_tombp
,
1303 wchar_t *escape_charp
)
1312 /* If the input file does not specify the codeset use the locale's. */
1313 if (codeset
== NULL
)
1315 setlocale (LC_ALL
, "");
1316 codeset
= nl_langinfo (CODESET
);
1317 setlocale (LC_ALL
, "C");
1320 /* Get the conversion modules. */
1321 *cd_towcp
= iconv_open ("WCHAR_T", codeset
);
1322 *cd_tombp
= iconv_open (codeset
, "WCHAR_T");
1323 if (*cd_towcp
== (iconv_t
) -1 || *cd_tombp
== (iconv_t
) -1)
1325 error (0, 0, gettext ("conversion modules not available"));
1326 if (*cd_towcp
!= (iconv_t
) -1)
1327 iconv_close (*cd_towcp
);
1332 /* One special case for historical reasons is the backslash
1333 character. In some codesets the byte value 0x5c is not mapped to
1334 U005c in Unicode. These charsets then don't have a backslash
1335 character at all. Therefore we have to live with whatever the
1336 codeset provides and recognize, instead of the U005c, the character
1337 the byte value 0x5c is mapped to. */
1343 wbufptr
= (char *) wbuf
;
1344 wbufsize
= sizeof (wbuf
);
1346 iconv (*cd_towcp
, &bufptr
, &bufsize
, &wbufptr
, &wbufsize
);
1347 if (bufsize
!= 0 || wbufsize
!= 0)
1349 /* Something went wrong, we couldn't convert the byte 0x5c. Go
1350 on with using U005c. */
1351 error (0, 0, gettext ("cannot determine escape character"));
1352 *escape_charp
= L
'\\';
1355 *escape_charp
= wbuf
[0];