]>
Commit | Line | Data |
---|---|---|
fb5663ca | 1 | /* Convert text in given files from the specified from-set to the to-set. |
0ecb606c | 2 | Copyright (C) 1998-2004, 2005, 2006 Free Software Foundation, Inc. |
fb5663ca UD |
3 | This file is part of the GNU C Library. |
4 | Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998. | |
5 | ||
0ecb606c JJ |
6 | This program is free software; you can redistribute it and/or modify |
7 | it under the terms of the GNU General Public License version 2 as | |
8 | published by the Free Software Foundation. | |
fb5663ca | 9 | |
0ecb606c | 10 | This program is distributed in the hope that it will be useful, |
fb5663ca | 11 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
0ecb606c JJ |
12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
13 | GNU General Public License for more details. | |
fb5663ca | 14 | |
0ecb606c JJ |
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. */ | |
fb5663ca UD |
18 | |
19 | #include <argp.h> | |
e4060100 | 20 | #include <assert.h> |
8fe0fd03 | 21 | #include <ctype.h> |
fb5663ca UD |
22 | #include <errno.h> |
23 | #include <error.h> | |
24 | #include <fcntl.h> | |
25 | #include <iconv.h> | |
ac3d553b | 26 | #include <langinfo.h> |
fb5663ca | 27 | #include <locale.h> |
8fe0fd03 | 28 | #include <search.h> |
fa00744e | 29 | #include <stdbool.h> |
fb5663ca UD |
30 | #include <stdio.h> |
31 | #include <stdlib.h> | |
32 | #include <string.h> | |
33 | #include <unistd.h> | |
4360eafd | 34 | #include <libintl.h> |
7cabd57c UD |
35 | #ifdef _POSIX_MAPPED_FILES |
36 | # include <sys/mman.h> | |
37 | #endif | |
93693c4d | 38 | #include <charmap.h> |
e62c19f1 | 39 | #include <gconv_int.h> |
93693c4d | 40 | #include "iconv_prog.h" |
9a1f71a7 | 41 | #include "iconvconfig.h" |
fb5663ca UD |
42 | |
43 | /* Get libc version number. */ | |
44 | #include "../version.h" | |
45 | ||
46 | #define PACKAGE _libc_intl_domainname | |
47 | ||
48 | ||
49 | /* Name and version of program. */ | |
50 | static void print_version (FILE *stream, struct argp_state *state); | |
51 | void (*argp_program_version_hook) (FILE *, struct argp_state *) = print_version; | |
52 | ||
53 | #define OPT_VERBOSE 1000 | |
adcf0e4a | 54 | #define OPT_LIST 'l' |
fb5663ca UD |
55 | |
56 | /* Definitions of arguments for argp functions. */ | |
57 | static const struct argp_option options[] = | |
58 | { | |
59 | { NULL, 0, NULL, 0, N_("Input/Output format specification:") }, | |
60 | { "from-code", 'f', "NAME", 0, N_("encoding of original text") }, | |
61 | { "to-code", 't', "NAME", 0, N_("encoding for output") }, | |
8fe0fd03 | 62 | { NULL, 0, NULL, 0, N_("Information:") }, |
adcf0e4a | 63 | { "list", 'l', NULL, 0, N_("list all known coded character sets") }, |
fb5663ca | 64 | { NULL, 0, NULL, 0, N_("Output control:") }, |
adcf0e4a | 65 | { NULL, 'c', NULL, 0, N_("omit invalid characters from output") }, |
fb5663ca | 66 | { "output", 'o', "FILE", 0, N_("output file") }, |
6d77214d | 67 | { "silent", 's', NULL, 0, N_("suppress warnings") }, |
fb5663ca UD |
68 | { "verbose", OPT_VERBOSE, NULL, 0, N_("print progress information") }, |
69 | { NULL, 0, NULL, 0, NULL } | |
70 | }; | |
71 | ||
72 | /* Short description of program. */ | |
73 | static const char doc[] = N_("\ | |
74 | Convert encoding of given files from one encoding to another."); | |
75 | ||
76 | /* Strings for arguments in help texts. */ | |
77 | static const char args_doc[] = N_("[FILE...]"); | |
78 | ||
79 | /* Prototype for option handler. */ | |
adcf0e4a | 80 | static error_t parse_opt (int key, char *arg, struct argp_state *state); |
fb5663ca UD |
81 | |
82 | /* Function to print some extra text in the help message. */ | |
adcf0e4a | 83 | static char *more_help (int key, const char *text, void *input); |
fb5663ca UD |
84 | |
85 | /* Data structure to communicate with argp functions. */ | |
86 | static struct argp argp = | |
87 | { | |
88 | options, parse_opt, args_doc, doc, NULL, more_help | |
89 | }; | |
90 | ||
e0e86ccb UD |
91 | /* Code sets to convert from and to respectively. An empty string as the |
92 | default causes the 'iconv_open' function to look up the charset of the | |
93 | currently selected locale and use it. */ | |
94 | static const char *from_code = ""; | |
95 | static const char *to_code = ""; | |
fb5663ca UD |
96 | |
97 | /* File to write output to. If NULL write to stdout. */ | |
98 | static const char *output_file; | |
99 | ||
100 | /* Nonzero if verbose ouput is wanted. */ | |
93693c4d | 101 | int verbose; |
fb5663ca | 102 | |
8fe0fd03 UD |
103 | /* Nonzero if list of all coded character sets is wanted. */ |
104 | static int list; | |
105 | ||
85830c4c | 106 | /* If nonzero omit invalid character from output. */ |
93693c4d | 107 | int omit_invalid; |
85830c4c | 108 | |
fb5663ca | 109 | /* Prototypes for the functions doing the actual work. */ |
3b434940 | 110 | static int process_block (iconv_t cd, char *addr, size_t len, FILE *output); |
fb5663ca UD |
111 | static int process_fd (iconv_t cd, int fd, FILE *output); |
112 | static int process_file (iconv_t cd, FILE *input, FILE *output); | |
2bd60880 | 113 | static void print_known_names (void) internal_function; |
fb5663ca UD |
114 | |
115 | ||
116 | int | |
117 | main (int argc, char *argv[]) | |
118 | { | |
119 | int status = EXIT_SUCCESS; | |
120 | int remaining; | |
121 | FILE *output; | |
122 | iconv_t cd; | |
85830c4c | 123 | const char *orig_to_code; |
93693c4d UD |
124 | struct charmap_t *from_charmap = NULL; |
125 | struct charmap_t *to_charmap = NULL; | |
fb5663ca UD |
126 | |
127 | /* Set locale via LC_ALL. */ | |
128 | setlocale (LC_ALL, ""); | |
129 | ||
130 | /* Set the text message domain. */ | |
131 | textdomain (_libc_intl_domainname); | |
132 | ||
133 | /* Parse and process arguments. */ | |
134 | argp_parse (&argp, argc, argv, 0, &remaining, NULL); | |
135 | ||
8fe0fd03 UD |
136 | /* List all coded character sets if wanted. */ |
137 | if (list) | |
138 | { | |
139 | print_known_names (); | |
140 | exit (EXIT_SUCCESS); | |
141 | } | |
fb5663ca | 142 | |
85830c4c UD |
143 | /* If we have to ignore errors make sure we use the appropriate name for |
144 | the to-character-set. */ | |
145 | orig_to_code = to_code; | |
146 | if (omit_invalid) | |
147 | { | |
148 | const char *errhand = strchrnul (to_code, '/'); | |
149 | int nslash = 2; | |
150 | char *newp; | |
151 | char *cp; | |
152 | ||
153 | if (*errhand == '/') | |
154 | { | |
155 | --nslash; | |
156 | errhand = strchrnul (errhand, '/'); | |
157 | ||
158 | if (*errhand == '/') | |
159 | { | |
160 | --nslash; | |
b8750711 | 161 | errhand = strchr (errhand, '\0'); |
85830c4c UD |
162 | } |
163 | } | |
164 | ||
b8750711 | 165 | newp = (char *) alloca (errhand - to_code + nslash + 7 + 1); |
85830c4c | 166 | cp = mempcpy (newp, to_code, errhand - to_code); |
23054ade | 167 | while (nslash-- > 0) |
85830c4c | 168 | *cp++ = '/'; |
b8750711 UD |
169 | if (cp[-1] != '/') |
170 | *cp++ = ','; | |
c7fb46a9 | 171 | memcpy (cp, "IGNORE", sizeof ("IGNORE")); |
85830c4c UD |
172 | |
173 | to_code = newp; | |
174 | } | |
175 | ||
93693c4d UD |
176 | /* POSIX 1003.2b introduces a silly thing: the arguments to -t anf -f |
177 | can be file names of charmaps. In this case iconv will have to read | |
178 | those charmaps and use them to do the conversion. But there are | |
179 | holes in the specification. There is nothing said that if -f is a | |
180 | charmap filename that -t must be, too. And vice versa. There is | |
181 | also no word about the symbolic names used. What if they don't | |
182 | match? */ | |
183 | if (strchr (from_code, '/') != NULL) | |
184 | /* The from-name might be a charmap file name. Try reading the | |
185 | file. */ | |
0ecb606c | 186 | from_charmap = charmap_read (from_code, /*0, 1*/1, 0, 0, 0); |
93693c4d UD |
187 | |
188 | if (strchr (orig_to_code, '/') != NULL) | |
189 | /* The to-name might be a charmap file name. Try reading the | |
190 | file. */ | |
0ecb606c | 191 | to_charmap = charmap_read (orig_to_code, /*0, 1,*/1, 0, 0, 0); |
93693c4d | 192 | |
fb5663ca UD |
193 | |
194 | /* Determine output file. */ | |
377c725f | 195 | if (output_file != NULL && strcmp (output_file, "-") != 0) |
fb5663ca UD |
196 | { |
197 | output = fopen (output_file, "w"); | |
198 | if (output == NULL) | |
199 | error (EXIT_FAILURE, errno, _("cannot open output file")); | |
200 | } | |
201 | else | |
202 | output = stdout; | |
203 | ||
93693c4d UD |
204 | /* At this point we have to handle two cases. The first one is |
205 | where a charmap is used for the from- or to-charset, or both. We | |
206 | handle this special since it is very different from the sane way of | |
207 | doing things. The other case allows converting using the iconv() | |
208 | function. */ | |
209 | if (from_charmap != NULL || to_charmap != NULL) | |
210 | /* Construct the conversion table and do the conversion. */ | |
211 | status = charmap_conversion (from_code, from_charmap, to_code, to_charmap, | |
212 | argc, remaining, argv, output); | |
fb5663ca | 213 | else |
93693c4d UD |
214 | { |
215 | /* Let's see whether we have these coded character sets. */ | |
216 | cd = iconv_open (to_code, from_code); | |
217 | if (cd == (iconv_t) -1) | |
218 | { | |
219 | if (errno == EINVAL) | |
fa00744e UD |
220 | { |
221 | /* Try to be nice with the user and tell her which of the | |
222 | two encoding names is wrong. This is possible because | |
223 | all supported encodings can be converted from/to Unicode, | |
224 | in other words, because the graph of encodings is | |
225 | connected. */ | |
226 | bool from_wrong = | |
227 | (iconv_open ("UTF-8", from_code) == (iconv_t) -1 | |
228 | && errno == EINVAL); | |
229 | bool to_wrong = | |
230 | (iconv_open (to_code, "UTF-8") == (iconv_t) -1 | |
231 | && errno == EINVAL); | |
232 | const char *from_pretty = | |
233 | (from_code[0] ? from_code : nl_langinfo (CODESET)); | |
234 | const char *to_pretty = | |
235 | (orig_to_code[0] ? orig_to_code : nl_langinfo (CODESET)); | |
236 | ||
237 | if (from_wrong) | |
238 | { | |
239 | if (to_wrong) | |
0ecb606c | 240 | error (0, 0, |
fa00744e | 241 | _("\ |
0ecb606c | 242 | conversions from `%s' and to `%s' are not supported"), |
fa00744e UD |
243 | from_pretty, to_pretty); |
244 | else | |
0ecb606c | 245 | error (0, 0, |
fa00744e UD |
246 | _("conversion from `%s' is not supported"), |
247 | from_pretty); | |
248 | } | |
249 | else | |
250 | { | |
251 | if (to_wrong) | |
0ecb606c | 252 | error (0, 0, |
fa00744e UD |
253 | _("conversion to `%s' is not supported"), |
254 | to_pretty); | |
255 | else | |
0ecb606c | 256 | error (0, 0, |
fa00744e UD |
257 | _("conversion from `%s' to `%s' is not supported"), |
258 | from_pretty, to_pretty); | |
259 | } | |
0ecb606c JJ |
260 | |
261 | argp_help (&argp, stderr, ARGP_HELP_SEE, | |
262 | program_invocation_short_name); | |
263 | exit (1); | |
fa00744e | 264 | } |
93693c4d UD |
265 | else |
266 | error (EXIT_FAILURE, errno, | |
267 | _("failed to start conversion processing")); | |
268 | } | |
adcf0e4a | 269 | |
93693c4d UD |
270 | /* Now process the remaining files. Write them to stdout or the file |
271 | specified with the `-o' parameter. If we have no file given as | |
272 | the parameter process all from stdin. */ | |
273 | if (remaining == argc) | |
274 | { | |
275 | if (process_file (cd, stdin, output) != 0) | |
276 | status = EXIT_FAILURE; | |
277 | } | |
278 | else | |
279 | do | |
280 | { | |
d3f8be6d | 281 | #ifdef _POSIX_MAPPED_FILES |
93693c4d UD |
282 | struct stat st; |
283 | char *addr; | |
d3f8be6d | 284 | #endif |
f2fcb018 | 285 | int fd, ret; |
93693c4d UD |
286 | |
287 | if (verbose) | |
c4f4ef87 | 288 | fprintf (stderr, "%s:\n", argv[remaining]); |
93693c4d UD |
289 | if (strcmp (argv[remaining], "-") == 0) |
290 | fd = 0; | |
291 | else | |
adcf0e4a | 292 | { |
93693c4d UD |
293 | fd = open (argv[remaining], O_RDONLY); |
294 | ||
295 | if (fd == -1) | |
296 | { | |
297 | error (0, errno, _("cannot open input file `%s'"), | |
298 | argv[remaining]); | |
299 | status = EXIT_FAILURE; | |
300 | continue; | |
301 | } | |
adcf0e4a | 302 | } |
fb5663ca | 303 | |
7cabd57c | 304 | #ifdef _POSIX_MAPPED_FILES |
93693c4d UD |
305 | /* We have possibilities for reading the input file. First try |
306 | to mmap() it since this will provide the fastest solution. */ | |
307 | if (fstat (fd, &st) == 0 | |
308 | && ((addr = mmap (NULL, st.st_size, PROT_READ, MAP_PRIVATE, | |
309 | fd, 0)) != MAP_FAILED)) | |
fb5663ca | 310 | { |
93693c4d UD |
311 | /* Yes, we can use mmap(). The descriptor is not needed |
312 | anymore. */ | |
313 | if (close (fd) != 0) | |
314 | error (EXIT_FAILURE, errno, | |
315 | _("error while closing input `%s'"), | |
316 | argv[remaining]); | |
317 | ||
f2fcb018 UD |
318 | ret = process_block (cd, addr, st.st_size, output); |
319 | ||
320 | /* We don't need the input data anymore. */ | |
321 | munmap ((void *) addr, st.st_size); | |
322 | ||
323 | if (ret != 0) | |
93693c4d | 324 | { |
93693c4d UD |
325 | status = EXIT_FAILURE; |
326 | ||
f2fcb018 UD |
327 | if (ret < 0) |
328 | /* We cannot go on with producing output since it might | |
329 | lead to problem because the last output might leave | |
330 | the output stream in an undefined state. */ | |
331 | break; | |
93693c4d | 332 | } |
fb5663ca | 333 | } |
93693c4d | 334 | else |
7cabd57c | 335 | #endif /* _POSIX_MAPPED_FILES */ |
fb5663ca | 336 | { |
93693c4d | 337 | /* Read the file in pieces. */ |
f2fcb018 UD |
338 | ret = process_fd (cd, fd, output); |
339 | ||
340 | /* Now close the file. */ | |
341 | close (fd); | |
342 | ||
343 | if (ret != 0) | |
93693c4d UD |
344 | { |
345 | /* Something went wrong. */ | |
346 | status = EXIT_FAILURE; | |
347 | ||
f2fcb018 UD |
348 | if (ret < 0) |
349 | /* We cannot go on with producing output since it might | |
350 | lead to problem because the last output might leave | |
351 | the output stream in an undefined state. */ | |
352 | break; | |
93693c4d | 353 | } |
fb5663ca | 354 | } |
fb5663ca | 355 | } |
93693c4d UD |
356 | while (++remaining < argc); |
357 | } | |
fb5663ca UD |
358 | |
359 | /* Close the output file now. */ | |
360 | if (fclose (output)) | |
361 | error (EXIT_FAILURE, errno, _("error while closing output file")); | |
362 | ||
363 | return status; | |
364 | } | |
365 | ||
366 | ||
367 | /* Handle program arguments. */ | |
368 | static error_t | |
369 | parse_opt (int key, char *arg, struct argp_state *state) | |
370 | { | |
371 | switch (key) | |
372 | { | |
373 | case 'f': | |
374 | from_code = arg; | |
375 | break; | |
376 | case 't': | |
377 | to_code = arg; | |
378 | break; | |
379 | case 'o': | |
380 | output_file = arg; | |
381 | break; | |
adcf0e4a UD |
382 | case 's': |
383 | /* Nothing, for now at least. We are not giving out any information | |
384 | about missing character or so. */ | |
385 | break; | |
386 | case 'c': | |
85830c4c UD |
387 | /* Omit invalid characters from output. */ |
388 | omit_invalid = 1; | |
adcf0e4a | 389 | break; |
fb5663ca UD |
390 | case OPT_VERBOSE: |
391 | verbose = 1; | |
392 | break; | |
8fe0fd03 UD |
393 | case OPT_LIST: |
394 | list = 1; | |
395 | break; | |
fb5663ca UD |
396 | default: |
397 | return ARGP_ERR_UNKNOWN; | |
398 | } | |
399 | return 0; | |
400 | } | |
401 | ||
402 | ||
403 | static char * | |
404 | more_help (int key, const char *text, void *input) | |
405 | { | |
406 | switch (key) | |
407 | { | |
408 | case ARGP_KEY_HELP_EXTRA: | |
409 | /* We print some extra information. */ | |
410 | return strdup (gettext ("\ | |
d40eb37a UD |
411 | For bug reporting instructions, please see:\n\ |
412 | <http://www.gnu.org/software/libc/bugs.html>.\n")); | |
fb5663ca UD |
413 | default: |
414 | break; | |
415 | } | |
416 | return (char *) text; | |
417 | } | |
418 | ||
419 | ||
420 | /* Print the version information. */ | |
421 | static void | |
422 | print_version (FILE *stream, struct argp_state *state) | |
423 | { | |
424 | fprintf (stream, "iconv (GNU %s) %s\n", PACKAGE, VERSION); | |
425 | fprintf (stream, gettext ("\ | |
426 | Copyright (C) %s Free Software Foundation, Inc.\n\ | |
427 | This is free software; see the source for copying conditions. There is NO\n\ | |
428 | warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.\n\ | |
0ecb606c | 429 | "), "2006"); |
fb5663ca UD |
430 | fprintf (stream, gettext ("Written by %s.\n"), "Ulrich Drepper"); |
431 | } | |
432 | ||
433 | ||
434 | static int | |
3b434940 | 435 | process_block (iconv_t cd, char *addr, size_t len, FILE *output) |
fb5663ca UD |
436 | { |
437 | #define OUTBUF_SIZE 32768 | |
9b26f5c4 | 438 | const char *start = addr; |
fb5663ca | 439 | char outbuf[OUTBUF_SIZE]; |
8619129f UD |
440 | char *outptr; |
441 | size_t outlen; | |
442 | size_t n; | |
f2fcb018 | 443 | int ret = 0; |
fb5663ca UD |
444 | |
445 | while (len > 0) | |
446 | { | |
8619129f UD |
447 | outptr = outbuf; |
448 | outlen = OUTBUF_SIZE; | |
449 | n = iconv (cd, &addr, &len, &outptr, &outlen); | |
fb5663ca | 450 | |
f2fcb018 UD |
451 | if (n == (size_t) -1 && omit_invalid && errno == EILSEQ) |
452 | { | |
453 | ret = 1; | |
454 | if (len == 0) | |
455 | n = 0; | |
456 | else | |
457 | errno = E2BIG; | |
458 | } | |
459 | ||
fb5663ca UD |
460 | if (outptr != outbuf) |
461 | { | |
462 | /* We have something to write out. */ | |
4d3a563f UD |
463 | int errno_save = errno; |
464 | ||
6dd67bd5 UD |
465 | if (fwrite (outbuf, 1, outptr - outbuf, output) |
466 | < (size_t) (outptr - outbuf) | |
fb5663ca UD |
467 | || ferror (output)) |
468 | { | |
469 | /* Error occurred while printing the result. */ | |
470 | error (0, 0, _("\ | |
471 | conversion stopped due to problem in writing the output")); | |
472 | return -1; | |
473 | } | |
4d3a563f UD |
474 | |
475 | errno = errno_save; | |
fb5663ca UD |
476 | } |
477 | ||
478 | if (n != (size_t) -1) | |
c559a3ca UD |
479 | { |
480 | /* All the input test is processed. For state-dependent | |
481 | character sets we have to flush the state now. */ | |
482 | outptr = outbuf; | |
483 | outlen = OUTBUF_SIZE; | |
f2fcb018 | 484 | n = iconv (cd, NULL, NULL, &outptr, &outlen); |
c559a3ca UD |
485 | |
486 | if (outptr != outbuf) | |
487 | { | |
488 | /* We have something to write out. */ | |
489 | int errno_save = errno; | |
490 | ||
6dd67bd5 UD |
491 | if (fwrite (outbuf, 1, outptr - outbuf, output) |
492 | < (size_t) (outptr - outbuf) | |
c559a3ca UD |
493 | || ferror (output)) |
494 | { | |
495 | /* Error occurred while printing the result. */ | |
496 | error (0, 0, _("\ | |
497 | conversion stopped due to problem in writing the output")); | |
498 | return -1; | |
499 | } | |
500 | ||
501 | errno = errno_save; | |
502 | } | |
503 | ||
f2fcb018 UD |
504 | if (n != (size_t) -1) |
505 | break; | |
506 | ||
507 | if (omit_invalid && errno == EILSEQ) | |
508 | { | |
509 | ret = 1; | |
510 | break; | |
511 | } | |
c559a3ca | 512 | } |
fb5663ca UD |
513 | |
514 | if (errno != E2BIG) | |
515 | { | |
516 | /* iconv() ran into a problem. */ | |
517 | switch (errno) | |
518 | { | |
519 | case EILSEQ: | |
b8750711 UD |
520 | if (! omit_invalid) |
521 | error (0, 0, _("illegal input sequence at position %ld"), | |
522 | (long int) (addr - start)); | |
fb5663ca UD |
523 | break; |
524 | case EINVAL: | |
525 | error (0, 0, _("\ | |
526 | incomplete character or shift sequence at end of buffer")); | |
527 | break; | |
528 | case EBADF: | |
529 | error (0, 0, _("internal error (illegal descriptor)")); | |
530 | break; | |
531 | default: | |
532 | error (0, 0, _("unknown iconv() error %d"), errno); | |
533 | break; | |
534 | } | |
535 | ||
536 | return -1; | |
537 | } | |
538 | } | |
539 | ||
f2fcb018 | 540 | return ret; |
fb5663ca UD |
541 | } |
542 | ||
543 | ||
544 | static int | |
545 | process_fd (iconv_t cd, int fd, FILE *output) | |
546 | { | |
547 | /* we have a problem with reading from a desriptor since we must not | |
548 | provide the iconv() function an incomplete character or shift | |
549 | sequence at the end of the buffer. Since we have to deal with | |
550 | arbitrary encodings we must read the whole text in a buffer and | |
551 | process it in one step. */ | |
552 | static char *inbuf = NULL; | |
553 | static size_t maxlen = 0; | |
554 | char *inptr = NULL; | |
555 | size_t actlen = 0; | |
556 | ||
557 | while (actlen < maxlen) | |
558 | { | |
2e47aff5 | 559 | ssize_t n = read (fd, inptr, maxlen - actlen); |
fb5663ca UD |
560 | |
561 | if (n == 0) | |
562 | /* No more text to read. */ | |
563 | break; | |
564 | ||
565 | if (n == -1) | |
566 | { | |
567 | /* Error while reading. */ | |
568 | error (0, errno, _("error while reading the input")); | |
569 | return -1; | |
570 | } | |
571 | ||
572 | inptr += n; | |
573 | actlen += n; | |
574 | } | |
575 | ||
576 | if (actlen == maxlen) | |
577 | while (1) | |
578 | { | |
2e47aff5 | 579 | ssize_t n; |
d1dddedf | 580 | char *new_inbuf; |
fb5663ca UD |
581 | |
582 | /* Increase the buffer. */ | |
d1dddedf UD |
583 | new_inbuf = (char *) realloc (inbuf, maxlen + 32768); |
584 | if (new_inbuf == NULL) | |
585 | { | |
586 | error (0, errno, _("unable to allocate buffer for input")); | |
587 | return -1; | |
588 | } | |
589 | inbuf = new_inbuf; | |
fb5663ca | 590 | maxlen += 32768; |
fb5663ca UD |
591 | inptr = inbuf + actlen; |
592 | ||
593 | do | |
594 | { | |
595 | n = read (fd, inptr, maxlen - actlen); | |
596 | ||
597 | if (n == 0) | |
598 | /* No more text to read. */ | |
599 | break; | |
600 | ||
601 | if (n == -1) | |
602 | { | |
603 | /* Error while reading. */ | |
604 | error (0, errno, _("error while reading the input")); | |
605 | return -1; | |
606 | } | |
607 | ||
608 | inptr += n; | |
609 | actlen += n; | |
610 | } | |
611 | while (actlen < maxlen); | |
612 | ||
613 | if (n == 0) | |
614 | /* Break again so we leave both loops. */ | |
615 | break; | |
616 | } | |
617 | ||
618 | /* Now we have all the input in the buffer. Process it in one run. */ | |
619 | return process_block (cd, inbuf, actlen, output); | |
620 | } | |
621 | ||
622 | ||
623 | static int | |
624 | process_file (iconv_t cd, FILE *input, FILE *output) | |
625 | { | |
626 | /* This should be safe since we use this function only for `stdin' and | |
627 | we haven't read anything so far. */ | |
628 | return process_fd (cd, fileno (input), output); | |
629 | } | |
8fe0fd03 UD |
630 | |
631 | ||
632 | /* Print all known character sets/encodings. */ | |
633 | static void *printlist; | |
634 | static size_t column; | |
635 | static int not_first; | |
636 | ||
637 | static void | |
638 | insert_print_list (const void *nodep, VISIT value, int level) | |
639 | { | |
640 | if (value == leaf || value == postorder) | |
641 | { | |
642 | const struct gconv_alias *s = *(const struct gconv_alias **) nodep; | |
9b26f5c4 | 643 | tsearch (s->fromname, &printlist, (__compar_fn_t) strverscmp); |
8fe0fd03 UD |
644 | } |
645 | } | |
646 | ||
647 | static void | |
b17c0a8e | 648 | do_print_human (const void *nodep, VISIT value, int level) |
8fe0fd03 UD |
649 | { |
650 | if (value == leaf || value == postorder) | |
651 | { | |
652 | const char *s = *(const char **) nodep; | |
653 | size_t len = strlen (s); | |
654 | size_t cnt; | |
655 | ||
656 | while (len > 0 && s[len - 1] == '/') | |
657 | --len; | |
658 | ||
659 | for (cnt = 0; cnt < len; ++cnt) | |
660 | if (isalnum (s[cnt])) | |
661 | break; | |
662 | if (cnt == len) | |
663 | return; | |
664 | ||
665 | if (not_first) | |
666 | { | |
667 | putchar (','); | |
668 | ++column; | |
669 | ||
670 | if (column > 2 && column + len > 77) | |
671 | { | |
672 | fputs ("\n ", stdout); | |
673 | column = 2; | |
674 | } | |
675 | else | |
676 | { | |
677 | putchar (' '); | |
678 | ++column; | |
679 | } | |
680 | } | |
681 | else | |
9b26f5c4 | 682 | not_first = 1; |
8fe0fd03 UD |
683 | |
684 | fwrite (s, len, 1, stdout); | |
685 | column += len; | |
686 | } | |
687 | } | |
688 | ||
b17c0a8e UD |
689 | static void |
690 | do_print (const void *nodep, VISIT value, int level) | |
691 | { | |
692 | if (value == leaf || value == postorder) | |
693 | { | |
694 | const char *s = *(const char **) nodep; | |
695 | ||
696 | puts (s); | |
697 | } | |
698 | } | |
699 | ||
8fe0fd03 | 700 | static void |
2bd60880 UD |
701 | internal_function |
702 | add_known_names (struct gconv_module *node) | |
703 | { | |
704 | if (node->left != NULL) | |
705 | add_known_names (node->left); | |
706 | if (node->right != NULL) | |
707 | add_known_names (node->right); | |
2bd60880 UD |
708 | do |
709 | { | |
d2dfc5de UD |
710 | if (strcmp (node->from_string, "INTERNAL")) |
711 | tsearch (node->from_string, &printlist, | |
712 | (__compar_fn_t) strverscmp); | |
9a1f71a7 | 713 | if (strcmp (node->to_string, "INTERNAL") != 0) |
d2dfc5de | 714 | tsearch (node->to_string, &printlist, (__compar_fn_t) strverscmp); |
2bd60880 | 715 | |
d2dfc5de | 716 | node = node->same; |
2bd60880 UD |
717 | } |
718 | while (node != NULL); | |
719 | } | |
720 | ||
9a1f71a7 UD |
721 | |
722 | static void | |
723 | insert_cache (void) | |
724 | { | |
725 | const struct gconvcache_header *header; | |
726 | const char *strtab; | |
727 | const struct hash_entry *hashtab; | |
728 | size_t cnt; | |
729 | ||
230491f0 UD |
730 | header = (const struct gconvcache_header *) __gconv_get_cache (); |
731 | strtab = (char *) header + header->string_offset; | |
732 | hashtab = (struct hash_entry *) ((char *) header + header->hash_offset); | |
9a1f71a7 UD |
733 | |
734 | for (cnt = 0; cnt < header->hash_size; ++cnt) | |
735 | if (hashtab[cnt].string_offset != 0) | |
736 | { | |
737 | const char *str = strtab + hashtab[cnt].string_offset; | |
738 | ||
739 | if (strcmp (str, "INTERNAL") != 0) | |
740 | tsearch (str, &printlist, (__compar_fn_t) strverscmp); | |
741 | } | |
742 | } | |
743 | ||
744 | ||
2bd60880 UD |
745 | static void |
746 | internal_function | |
8fe0fd03 UD |
747 | print_known_names (void) |
748 | { | |
8fe0fd03 | 749 | iconv_t h; |
230491f0 | 750 | void *cache; |
8fe0fd03 UD |
751 | |
752 | /* We must initialize the internal databases first. */ | |
753 | h = iconv_open ("L1", "L1"); | |
754 | iconv_close (h); | |
755 | ||
9a1f71a7 | 756 | /* See whether we have a cache. */ |
230491f0 UD |
757 | cache = __gconv_get_cache (); |
758 | if (cache != NULL) | |
9a1f71a7 UD |
759 | /* Yep, use only this information. */ |
760 | insert_cache (); | |
761 | else | |
762 | { | |
230491f0 UD |
763 | struct gconv_module *modules; |
764 | ||
9a1f71a7 UD |
765 | /* No, then use the information read from the gconv-modules file. |
766 | First add the aliases. */ | |
230491f0 | 767 | twalk (__gconv_get_alias_db (), insert_print_list); |
8fe0fd03 | 768 | |
9a1f71a7 | 769 | /* Add the from- and to-names from the known modules. */ |
230491f0 UD |
770 | modules = __gconv_get_modules_db (); |
771 | if (modules != NULL) | |
772 | add_known_names (modules); | |
9a1f71a7 | 773 | } |
8fe0fd03 | 774 | |
1b6840e5 UD |
775 | bool human_readable = isatty (fileno (stdout)); |
776 | ||
777 | if (human_readable) | |
778 | fputs (_("\ | |
8fe0fd03 UD |
779 | The following list contain all the coded character sets known. This does\n\ |
780 | not necessarily mean that all combinations of these names can be used for\n\ | |
781 | the FROM and TO command line parameters. One coded character set can be\n\ | |
d2dfc5de | 782 | listed with several different names (aliases).\n\n "), stdout); |
8fe0fd03 UD |
783 | |
784 | /* Now print the collected names. */ | |
785 | column = 2; | |
1b6840e5 | 786 | twalk (printlist, human_readable ? do_print_human : do_print); |
8fe0fd03 | 787 | |
1b6840e5 UD |
788 | if (human_readable && column != 0) |
789 | puts (""); | |
8fe0fd03 | 790 | } |