nexmon – Blame information for rev 1

Subversion Repositories:
Rev:
Rev Author Line No. Line
1 office 1 /* GLIB - Library of useful routines for C programming
2 *
3 * gconvert.c: Convert between character sets using iconv
4 * Copyright Red Hat Inc., 2000
5 * Authors: Havoc Pennington <hp@redhat.com>, Owen Taylor <otaylor@redhat.com>
6 *
7 * This library is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2 of the License, or (at your option) any later version.
11 *
12 * This library is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with this library; if not, see <http://www.gnu.org/licenses/>.
19 */
20  
21 #include "config.h"
22 #include "glibconfig.h"
23  
24 #ifndef G_OS_WIN32
25 #include <iconv.h>
26 #endif
27 #include <errno.h>
28 #include <stdio.h>
29 #include <string.h>
30 #include <stdlib.h>
31  
32 #ifdef G_OS_WIN32
33 #include "win_iconv.c"
34 #endif
35  
36 #ifdef G_PLATFORM_WIN32
37 #define STRICT
38 #include <windows.h>
39 #undef STRICT
40 #endif
41  
42 #include "gconvert.h"
43  
44 #include "gcharsetprivate.h"
45 #include "gslist.h"
46 #include "gstrfuncs.h"
47 #include "gtestutils.h"
48 #include "gthread.h"
49 #include "gunicode.h"
50 #include "gfileutils.h"
51  
52 #include "glibintl.h"
53  
54 #if defined(USE_LIBICONV_GNU) && !defined (_LIBICONV_H)
55 #error GNU libiconv in use but included iconv.h not from libiconv
56 #endif
57 #if !defined(USE_LIBICONV_GNU) && defined (_LIBICONV_H) \
58 && !defined (__APPLE_CC__) && !defined (__LP_64__)
59 #error GNU libiconv not in use but included iconv.h is from libiconv
60 #endif
61  
62  
63 /**
64 * SECTION:conversions
65 * @title: Character Set Conversion
66 * @short_description: convert strings between different character sets
67 *
68 * The g_convert() family of function wraps the functionality of iconv().
69 * In addition to pure character set conversions, GLib has functions to
70 * deal with the extra complications of encodings for file names.
71 *
72 * ## File Name Encodings
73 *
74 * Historically, UNIX has not had a defined encoding for file names:
75 * a file name is valid as long as it does not have path separators
76 * in it ("/"). However, displaying file names may require conversion:
77 * from the character set in which they were created, to the character
78 * set in which the application operates. Consider the Spanish file name
79 * "Presentaci&oacute;n.sxi". If the application which created it uses
80 * ISO-8859-1 for its encoding,
81 * |[
82 * Character: P r e s e n t a c i ó n . s x i
83 * Hex code: 50 72 65 73 65 6e 74 61 63 69 f3 6e 2e 73 78 69
84 * ]|
85 * However, if the application use UTF-8, the actual file name on
86 * disk would look like this:
87 * |[
88 * Character: P r e s e n t a c i ó n . s x i
89 * Hex code: 50 72 65 73 65 6e 74 61 63 69 c3 b3 6e 2e 73 78 69
90 * ]|
91 * Glib uses UTF-8 for its strings, and GUI toolkits like GTK+ that use
92 * Glib do the same thing. If you get a file name from the file system,
93 * for example, from readdir() or from g_dir_read_name(), and you wish
94 * to display the file name to the user, you will need to convert it
95 * into UTF-8. The opposite case is when the user types the name of a
96 * file he wishes to save: the toolkit will give you that string in
97 * UTF-8 encoding, and you will need to convert it to the character
98 * set used for file names before you can create the file with open()
99 * or fopen().
100 *
101 * By default, Glib assumes that file names on disk are in UTF-8
102 * encoding. This is a valid assumption for file systems which
103 * were created relatively recently: most applications use UTF-8
104 * encoding for their strings, and that is also what they use for
105 * the file names they create. However, older file systems may
106 * still contain file names created in "older" encodings, such as
107 * ISO-8859-1. In this case, for compatibility reasons, you may want
108 * to instruct Glib to use that particular encoding for file names
109 * rather than UTF-8. You can do this by specifying the encoding for
110 * file names in the [`G_FILENAME_ENCODING`][G_FILENAME_ENCODING]
111 * environment variable. For example, if your installation uses
112 * ISO-8859-1 for file names, you can put this in your `~/.profile`
113 * |[
114 * export G_FILENAME_ENCODING=ISO-8859-1
115 * ]|
116 * Glib provides the functions g_filename_to_utf8() and
117 * g_filename_from_utf8() to perform the necessary conversions.
118 * These functions convert file names from the encoding specified
119 * in `G_FILENAME_ENCODING` to UTF-8 and vice-versa. This
120 * [diagram][file-name-encodings-diagram] illustrates how
121 * these functions are used to convert between UTF-8 and the
122 * encoding for file names in the file system.
123 *
124 * ## Conversion between file name encodings # {#file-name-encodings-diagram)
125 *
126 * ![](file-name-encodings.png)
127 *
128 * ## Checklist for Application Writers
129 *
130 * This section is a practical summary of the detailed
131  
132 * things to do to make sure your applications process file
133 * name encodings correctly.
134 *
135 * 1. If you get a file name from the file system from a function
136 * such as readdir() or gtk_file_chooser_get_filename(), you do
137 * not need to do any conversion to pass that file name to
138 * functions like open(), rename(), or fopen() -- those are "raw"
139 * file names which the file system understands.
140 *
141 * 2. If you need to display a file name, convert it to UTF-8 first
142 * by using g_filename_to_utf8(). If conversion fails, display a
143 * string like "Unknown file name". Do not convert this string back
144 * into the encoding used for file names if you wish to pass it to
145 * the file system; use the original file name instead.
146 *
147 * For example, the document window of a word processor could display
148 * "Unknown file name" in its title bar but still let the user save
149 * the file, as it would keep the raw file name internally. This
150 * can happen if the user has not set the `G_FILENAME_ENCODING`
151 * environment variable even though he has files whose names are
152 * not encoded in UTF-8.
153 *
154 * 3. If your user interface lets the user type a file name for saving
155 * or renaming, convert it to the encoding used for file names in
156 * the file system by using g_filename_from_utf8(). Pass the converted
157 * file name to functions like fopen(). If conversion fails, ask the
158 * user to enter a different file name. This can happen if the user
159 * types Japanese characters when `G_FILENAME_ENCODING` is set to
160 * `ISO-8859-1`, for example.
161 */
162  
163 /* We try to terminate strings in unknown charsets with this many zero bytes
164 * to ensure that multibyte strings really are nul-terminated when we return
165 * them from g_convert() and friends.
166 */
167 #define NUL_TERMINATOR_LENGTH 4
168  
169 G_DEFINE_QUARK (g_convert_error, g_convert_error)
170  
171 static gboolean
172 try_conversion (const char *to_codeset,
173 const char *from_codeset,
174 iconv_t *cd)
175 {
176 *cd = iconv_open (to_codeset, from_codeset);
177  
178 if (*cd == (iconv_t)-1 && errno == EINVAL)
179 return FALSE;
180 else
181 return TRUE;
182 }
183  
184 static gboolean
185 try_to_aliases (const char **to_aliases,
186 const char *from_codeset,
187 iconv_t *cd)
188 {
189 if (to_aliases)
190 {
191 const char **p = to_aliases;
192 while (*p)
193 {
194 if (try_conversion (*p, from_codeset, cd))
195 return TRUE;
196  
197 p++;
198 }
199 }
200  
201 return FALSE;
202 }
203  
204 /**
205 * g_iconv_open:
206 * @to_codeset: destination codeset
207 * @from_codeset: source codeset
208 *
209 * Same as the standard UNIX routine iconv_open(), but
210 * may be implemented via libiconv on UNIX flavors that lack
211 * a native implementation.
212 *
213 * GLib provides g_convert() and g_locale_to_utf8() which are likely
214 * more convenient than the raw iconv wrappers.
215 *
216 * Returns: a "conversion descriptor", or (GIConv)-1 if
217 * opening the converter failed.
218 **/
219 GIConv
220 g_iconv_open (const gchar *to_codeset,
221 const gchar *from_codeset)
222 {
223 iconv_t cd;
224  
225 if (!try_conversion (to_codeset, from_codeset, &cd))
226 {
227 const char **to_aliases = _g_charset_get_aliases (to_codeset);
228 const char **from_aliases = _g_charset_get_aliases (from_codeset);
229  
230 if (from_aliases)
231 {
232 const char **p = from_aliases;
233 while (*p)
234 {
235 if (try_conversion (to_codeset, *p, &cd))
236 goto out;
237  
238 if (try_to_aliases (to_aliases, *p, &cd))
239 goto out;
240  
241 p++;
242 }
243 }
244  
245 if (try_to_aliases (to_aliases, from_codeset, &cd))
246 goto out;
247 }
248  
249 out:
250 return (cd == (iconv_t)-1) ? (GIConv)-1 : (GIConv)cd;
251 }
252  
253 /**
254 * g_iconv:
255 * @converter: conversion descriptor from g_iconv_open()
256 * @inbuf: bytes to convert
257 * @inbytes_left: inout parameter, bytes remaining to convert in @inbuf
258 * @outbuf: converted output bytes
259 * @outbytes_left: inout parameter, bytes available to fill in @outbuf
260 *
261 * Same as the standard UNIX routine iconv(), but
262 * may be implemented via libiconv on UNIX flavors that lack
263 * a native implementation.
264 *
265 * GLib provides g_convert() and g_locale_to_utf8() which are likely
266 * more convenient than the raw iconv wrappers.
267 *
268 * Returns: count of non-reversible conversions, or -1 on error
269 **/
270 gsize
271 g_iconv (GIConv converter,
272 gchar **inbuf,
273 gsize *inbytes_left,
274 gchar **outbuf,
275 gsize *outbytes_left)
276 {
277 iconv_t cd = (iconv_t)converter;
278  
279 return iconv (cd, inbuf, inbytes_left, outbuf, outbytes_left);
280 }
281  
282 /**
283 * g_iconv_close:
284 * @converter: a conversion descriptor from g_iconv_open()
285 *
286 * Same as the standard UNIX routine iconv_close(), but
287 * may be implemented via libiconv on UNIX flavors that lack
288 * a native implementation. Should be called to clean up
289 * the conversion descriptor from g_iconv_open() when
290 * you are done converting things.
291 *
292 * GLib provides g_convert() and g_locale_to_utf8() which are likely
293 * more convenient than the raw iconv wrappers.
294 *
295 * Returns: -1 on error, 0 on success
296 **/
297 gint
298 g_iconv_close (GIConv converter)
299 {
300 iconv_t cd = (iconv_t)converter;
301  
302 return iconv_close (cd);
303 }
304  
305 static GIConv
306 open_converter (const gchar *to_codeset,
307 const gchar *from_codeset,
308 GError **error)
309 {
310 GIConv cd;
311  
312 cd = g_iconv_open (to_codeset, from_codeset);
313  
314 if (cd == (GIConv) -1)
315 {
316 /* Something went wrong. */
317 if (error)
318 {
319 if (errno == EINVAL)
320 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_NO_CONVERSION,
321 _("Conversion from character set '%s' to '%s' is not supported"),
322 from_codeset, to_codeset);
323 else
324 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
325 _("Could not open converter from '%s' to '%s'"),
326 from_codeset, to_codeset);
327 }
328 }
329  
330 return cd;
331 }
332  
333 static int
334 close_converter (GIConv cd)
335 {
336 if (cd == (GIConv) -1)
337 return 0;
338  
339 return g_iconv_close (cd);
340 }
341  
342 /**
343 * g_convert_with_iconv:
344 * @str: the string to convert
345 * @len: the length of the string in bytes, or -1 if the string is
346 * nul-terminated (Note that some encodings may allow nul
347 * bytes to occur inside strings. In that case, using -1
348 * for the @len parameter is unsafe)
349 * @converter: conversion descriptor from g_iconv_open()
350 * @bytes_read: location to store the number of bytes in the
351 * input string that were successfully converted, or %NULL.
352 * Even if the conversion was successful, this may be
353 * less than @len if there were partial characters
354 * at the end of the input. If the error
355 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
356 * stored will the byte offset after the last valid
357 * input sequence.
358 * @bytes_written: the number of bytes stored in the output buffer (not
359 * including the terminating nul).
360 * @error: location to store the error occurring, or %NULL to ignore
361 * errors. Any of the errors in #GConvertError may occur.
362 *
363 * Converts a string from one character set to another.
364 *
365 * Note that you should use g_iconv() for streaming conversions.
366 * Despite the fact that @byes_read can return information about partial
367 * characters, the g_convert_... functions are not generally suitable
368 * for streaming. If the underlying converter maintains internal state,
369 * then this won't be preserved across successive calls to g_convert(),
370 * g_convert_with_iconv() or g_convert_with_fallback(). (An example of
371 * this is the GNU C converter for CP1255 which does not emit a base
372 * character until it knows that the next character is not a mark that
373 * could combine with the base character.)
374 *
375 * Returns: If the conversion was successful, a newly allocated
376 * nul-terminated string, which must be freed with
377 * g_free(). Otherwise %NULL and @error will be set.
378 **/
379 gchar*
380 g_convert_with_iconv (const gchar *str,
381 gssize len,
382 GIConv converter,
383 gsize *bytes_read,
384 gsize *bytes_written,
385 GError **error)
386 {
387 gchar *dest;
388 gchar *outp;
389 const gchar *p;
390 gsize inbytes_remaining;
391 gsize outbytes_remaining;
392 gsize err;
393 gsize outbuf_size;
394 gboolean have_error = FALSE;
395 gboolean done = FALSE;
396 gboolean reset = FALSE;
397  
398 g_return_val_if_fail (converter != (GIConv) -1, NULL);
399  
400 if (len < 0)
401 len = strlen (str);
402  
403 p = str;
404 inbytes_remaining = len;
405 outbuf_size = len + NUL_TERMINATOR_LENGTH;
406  
407 outbytes_remaining = outbuf_size - NUL_TERMINATOR_LENGTH;
408 outp = dest = g_malloc (outbuf_size);
409  
410 while (!done && !have_error)
411 {
412 if (reset)
413 err = g_iconv (converter, NULL, &inbytes_remaining, &outp, &outbytes_remaining);
414 else
415 err = g_iconv (converter, (char **)&p, &inbytes_remaining, &outp, &outbytes_remaining);
416  
417 if (err == (gsize) -1)
418 {
419 switch (errno)
420 {
421 case EINVAL:
422 /* Incomplete text, do not report an error */
423 done = TRUE;
424 break;
425 case E2BIG:
426 {
427 gsize used = outp - dest;
428  
429 outbuf_size *= 2;
430 dest = g_realloc (dest, outbuf_size);
431  
432 outp = dest + used;
433 outbytes_remaining = outbuf_size - used - NUL_TERMINATOR_LENGTH;
434 }
435 break;
436 case EILSEQ:
437 g_set_error_literal (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
438 _("Invalid byte sequence in conversion input"));
439 have_error = TRUE;
440 break;
441 default:
442 {
443 int errsv = errno;
444  
445 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
446 _("Error during conversion: %s"),
447 g_strerror (errsv));
448 }
449 have_error = TRUE;
450 break;
451 }
452 }
453 else
454 {
455 if (!reset)
456 {
457 /* call g_iconv with NULL inbuf to cleanup shift state */
458 reset = TRUE;
459 inbytes_remaining = 0;
460 }
461 else
462 done = TRUE;
463 }
464 }
465  
466 memset (outp, 0, NUL_TERMINATOR_LENGTH);
467  
468 if (bytes_read)
469 *bytes_read = p - str;
470 else
471 {
472 if ((p - str) != len)
473 {
474 if (!have_error)
475 {
476 g_set_error_literal (error, G_CONVERT_ERROR, G_CONVERT_ERROR_PARTIAL_INPUT,
477 _("Partial character sequence at end of input"));
478 have_error = TRUE;
479 }
480 }
481 }
482  
483 if (bytes_written)
484 *bytes_written = outp - dest; /* Doesn't include '\0' */
485  
486 if (have_error)
487 {
488 g_free (dest);
489 return NULL;
490 }
491 else
492 return dest;
493 }
494  
495 /**
496 * g_convert:
497 * @str: the string to convert
498 * @len: the length of the string in bytes, or -1 if the string is
499 * nul-terminated (Note that some encodings may allow nul
500 * bytes to occur inside strings. In that case, using -1
501 * for the @len parameter is unsafe)
502 * @to_codeset: name of character set into which to convert @str
503 * @from_codeset: character set of @str.
504 * @bytes_read: (out): location to store the number of bytes in the
505 * input string that were successfully converted, or %NULL.
506 * Even if the conversion was successful, this may be
507 * less than @len if there were partial characters
508 * at the end of the input. If the error
509 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
510 * stored will the byte offset after the last valid
511 * input sequence.
512 * @bytes_written: (out): the number of bytes stored in the output buffer (not
513 * including the terminating nul).
514 * @error: location to store the error occurring, or %NULL to ignore
515 * errors. Any of the errors in #GConvertError may occur.
516 *
517 * Converts a string from one character set to another.
518 *
519 * Note that you should use g_iconv() for streaming conversions.
520 * Despite the fact that @byes_read can return information about partial
521 * characters, the g_convert_... functions are not generally suitable
522 * for streaming. If the underlying converter maintains internal state,
523 * then this won't be preserved across successive calls to g_convert(),
524 * g_convert_with_iconv() or g_convert_with_fallback(). (An example of
525 * this is the GNU C converter for CP1255 which does not emit a base
526 * character until it knows that the next character is not a mark that
527 * could combine with the base character.)
528 *
529 * Using extensions such as "//TRANSLIT" may not work (or may not work
530 * well) on many platforms. Consider using g_str_to_ascii() instead.
531 *
532 * Returns: If the conversion was successful, a newly allocated
533 * nul-terminated string, which must be freed with
534 * g_free(). Otherwise %NULL and @error will be set.
535 **/
536 gchar*
537 g_convert (const gchar *str,
538 gssize len,
539 const gchar *to_codeset,
540 const gchar *from_codeset,
541 gsize *bytes_read,
542 gsize *bytes_written,
543 GError **error)
544 {
545 gchar *res;
546 GIConv cd;
547  
548 g_return_val_if_fail (str != NULL, NULL);
549 g_return_val_if_fail (to_codeset != NULL, NULL);
550 g_return_val_if_fail (from_codeset != NULL, NULL);
551  
552 cd = open_converter (to_codeset, from_codeset, error);
553  
554 if (cd == (GIConv) -1)
555 {
556 if (bytes_read)
557 *bytes_read = 0;
558  
559 if (bytes_written)
560 *bytes_written = 0;
561  
562 return NULL;
563 }
564  
565 res = g_convert_with_iconv (str, len, cd,
566 bytes_read, bytes_written,
567 error);
568  
569 close_converter (cd);
570  
571 return res;
572 }
573  
574 /**
575 * g_convert_with_fallback:
576 * @str: the string to convert
577 * @len: the length of the string in bytes, or -1 if the string is
578 * nul-terminated (Note that some encodings may allow nul
579 * bytes to occur inside strings. In that case, using -1
580 * for the @len parameter is unsafe)
581 * @to_codeset: name of character set into which to convert @str
582 * @from_codeset: character set of @str.
583 * @fallback: UTF-8 string to use in place of character not
584 * present in the target encoding. (The string must be
585 * representable in the target encoding).
586 If %NULL, characters not in the target encoding will
587 be represented as Unicode escapes \uxxxx or \Uxxxxyyyy.
588 * @bytes_read: location to store the number of bytes in the
589 * input string that were successfully converted, or %NULL.
590 * Even if the conversion was successful, this may be
591 * less than @len if there were partial characters
592 * at the end of the input.
593 * @bytes_written: the number of bytes stored in the output buffer (not
594 * including the terminating nul).
595 * @error: location to store the error occurring, or %NULL to ignore
596 * errors. Any of the errors in #GConvertError may occur.
597 *
598 * Converts a string from one character set to another, possibly
599 * including fallback sequences for characters not representable
600 * in the output. Note that it is not guaranteed that the specification
601 * for the fallback sequences in @fallback will be honored. Some
602 * systems may do an approximate conversion from @from_codeset
603 * to @to_codeset in their iconv() functions,
604 * in which case GLib will simply return that approximate conversion.
605 *
606 * Note that you should use g_iconv() for streaming conversions.
607 * Despite the fact that @byes_read can return information about partial
608 * characters, the g_convert_... functions are not generally suitable
609 * for streaming. If the underlying converter maintains internal state,
610 * then this won't be preserved across successive calls to g_convert(),
611 * g_convert_with_iconv() or g_convert_with_fallback(). (An example of
612 * this is the GNU C converter for CP1255 which does not emit a base
613 * character until it knows that the next character is not a mark that
614 * could combine with the base character.)
615 *
616 * Returns: If the conversion was successful, a newly allocated
617 * nul-terminated string, which must be freed with
618 * g_free(). Otherwise %NULL and @error will be set.
619 **/
620 gchar*
621 g_convert_with_fallback (const gchar *str,
622 gssize len,
623 const gchar *to_codeset,
624 const gchar *from_codeset,
625 const gchar *fallback,
626 gsize *bytes_read,
627 gsize *bytes_written,
628 GError **error)
629 {
630 gchar *utf8;
631 gchar *dest;
632 gchar *outp;
633 const gchar *insert_str = NULL;
634 const gchar *p;
635 gsize inbytes_remaining;
636 const gchar *save_p = NULL;
637 gsize save_inbytes = 0;
638 gsize outbytes_remaining;
639 gsize err;
640 GIConv cd;
641 gsize outbuf_size;
642 gboolean have_error = FALSE;
643 gboolean done = FALSE;
644  
645 GError *local_error = NULL;
646  
647 g_return_val_if_fail (str != NULL, NULL);
648 g_return_val_if_fail (to_codeset != NULL, NULL);
649 g_return_val_if_fail (from_codeset != NULL, NULL);
650  
651 if (len < 0)
652 len = strlen (str);
653  
654 /* Try an exact conversion; we only proceed if this fails
655 * due to an illegal sequence in the input string.
656 */
657 dest = g_convert (str, len, to_codeset, from_codeset,
658 bytes_read, bytes_written, &local_error);
659 if (!local_error)
660 return dest;
661  
662 if (!g_error_matches (local_error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE))
663 {
664 g_propagate_error (error, local_error);
665 return NULL;
666 }
667 else
668 g_error_free (local_error);
669  
670 local_error = NULL;
671  
672 /* No go; to proceed, we need a converter from "UTF-8" to
673 * to_codeset, and the string as UTF-8.
674 */
675 cd = open_converter (to_codeset, "UTF-8", error);
676 if (cd == (GIConv) -1)
677 {
678 if (bytes_read)
679 *bytes_read = 0;
680  
681 if (bytes_written)
682 *bytes_written = 0;
683  
684 return NULL;
685 }
686  
687 utf8 = g_convert (str, len, "UTF-8", from_codeset,
688 bytes_read, &inbytes_remaining, error);
689 if (!utf8)
690 {
691 close_converter (cd);
692 if (bytes_written)
693 *bytes_written = 0;
694 return NULL;
695 }
696  
697 /* Now the heart of the code. We loop through the UTF-8 string, and
698 * whenever we hit an offending character, we form fallback, convert
699 * the fallback to the target codeset, and then go back to
700 * converting the original string after finishing with the fallback.
701 *
702 * The variables save_p and save_inbytes store the input state
703 * for the original string while we are converting the fallback
704 */
705 p = utf8;
706  
707 outbuf_size = len + NUL_TERMINATOR_LENGTH;
708 outbytes_remaining = outbuf_size - NUL_TERMINATOR_LENGTH;
709 outp = dest = g_malloc (outbuf_size);
710  
711 while (!done && !have_error)
712 {
713 gsize inbytes_tmp = inbytes_remaining;
714 err = g_iconv (cd, (char **)&p, &inbytes_tmp, &outp, &outbytes_remaining);
715 inbytes_remaining = inbytes_tmp;
716  
717 if (err == (gsize) -1)
718 {
719 switch (errno)
720 {
721 case EINVAL:
722 g_assert_not_reached();
723 break;
724 case E2BIG:
725 {
726 gsize used = outp - dest;
727  
728 outbuf_size *= 2;
729 dest = g_realloc (dest, outbuf_size);
730  
731 outp = dest + used;
732 outbytes_remaining = outbuf_size - used - NUL_TERMINATOR_LENGTH;
733  
734 break;
735 }
736 case EILSEQ:
737 if (save_p)
738 {
739 /* Error converting fallback string - fatal
740 */
741 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
742 _("Cannot convert fallback '%s' to codeset '%s'"),
743 insert_str, to_codeset);
744 have_error = TRUE;
745 break;
746 }
747 else if (p)
748 {
749 if (!fallback)
750 {
751 gunichar ch = g_utf8_get_char (p);
752 insert_str = g_strdup_printf (ch < 0x10000 ? "\\u%04x" : "\\U%08x",
753 ch);
754 }
755 else
756 insert_str = fallback;
757  
758 save_p = g_utf8_next_char (p);
759 save_inbytes = inbytes_remaining - (save_p - p);
760 p = insert_str;
761 inbytes_remaining = strlen (p);
762 break;
763 }
764 /* fall thru if p is NULL */
765 default:
766 {
767 int errsv = errno;
768  
769 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_FAILED,
770 _("Error during conversion: %s"),
771 g_strerror (errsv));
772 }
773  
774 have_error = TRUE;
775 break;
776 }
777 }
778 else
779 {
780 if (save_p)
781 {
782 if (!fallback)
783 g_free ((gchar *)insert_str);
784 p = save_p;
785 inbytes_remaining = save_inbytes;
786 save_p = NULL;
787 }
788 else if (p)
789 {
790 /* call g_iconv with NULL inbuf to cleanup shift state */
791 p = NULL;
792 inbytes_remaining = 0;
793 }
794 else
795 done = TRUE;
796 }
797 }
798  
799 /* Cleanup
800 */
801 memset (outp, 0, NUL_TERMINATOR_LENGTH);
802  
803 close_converter (cd);
804  
805 if (bytes_written)
806 *bytes_written = outp - dest; /* Doesn't include '\0' */
807  
808 g_free (utf8);
809  
810 if (have_error)
811 {
812 if (save_p && !fallback)
813 g_free ((gchar *)insert_str);
814 g_free (dest);
815 return NULL;
816 }
817 else
818 return dest;
819 }
820  
821 /*
822 * g_locale_to_utf8
823 *
824 *
825 */
826  
827 static gchar *
828 strdup_len (const gchar *string,
829 gssize len,
830 gsize *bytes_written,
831 gsize *bytes_read,
832 GError **error)
833  
834 {
835 gsize real_len;
836  
837 if (!g_utf8_validate (string, len, NULL))
838 {
839 if (bytes_read)
840 *bytes_read = 0;
841 if (bytes_written)
842 *bytes_written = 0;
843  
844 g_set_error_literal (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
845 _("Invalid byte sequence in conversion input"));
846 return NULL;
847 }
848  
849 if (len < 0)
850 real_len = strlen (string);
851 else
852 {
853 real_len = 0;
854  
855 while (real_len < len && string[real_len])
856 real_len++;
857 }
858  
859 if (bytes_read)
860 *bytes_read = real_len;
861 if (bytes_written)
862 *bytes_written = real_len;
863  
864 return g_strndup (string, real_len);
865 }
866  
867 /**
868 * g_locale_to_utf8:
869 * @opsysstring: a string in the encoding of the current locale. On Windows
870 * this means the system codepage.
871 * @len: the length of the string, or -1 if the string is
872 * nul-terminated (Note that some encodings may allow nul
873 * bytes to occur inside strings. In that case, using -1
874 * for the @len parameter is unsafe)
875 * @bytes_read: (out) (optional): location to store the number of bytes in the
876 * input string that were successfully converted, or %NULL.
877 * Even if the conversion was successful, this may be
878 * less than @len if there were partial characters
879 * at the end of the input. If the error
880 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
881 * stored will the byte offset after the last valid
882 * input sequence.
883 * @bytes_written: (out) (optional): the number of bytes stored in the output
884 * buffer (not including the terminating nul).
885 * @error: location to store the error occurring, or %NULL to ignore
886 * errors. Any of the errors in #GConvertError may occur.
887 *
888 * Converts a string which is in the encoding used for strings by
889 * the C runtime (usually the same as that used by the operating
890 * system) in the [current locale][setlocale] into a UTF-8 string.
891 *
892 * Returns: A newly-allocated buffer containing the converted string,
893 * or %NULL on an error, and error will be set.
894 **/
895 gchar *
896 g_locale_to_utf8 (const gchar *opsysstring,
897 gssize len,
898 gsize *bytes_read,
899 gsize *bytes_written,
900 GError **error)
901 {
902 const char *charset;
903  
904 if (g_get_charset (&charset))
905 return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
906 else
907 return g_convert (opsysstring, len,
908 "UTF-8", charset, bytes_read, bytes_written, error);
909 }
910  
911 /**
912 * g_locale_from_utf8:
913 * @utf8string: a UTF-8 encoded string
914 * @len: the length of the string, or -1 if the string is
915 * nul-terminated (Note that some encodings may allow nul
916 * bytes to occur inside strings. In that case, using -1
917 * for the @len parameter is unsafe)
918 * @bytes_read: (out) (optional): location to store the number of bytes in the
919 * input string that were successfully converted, or %NULL.
920 * Even if the conversion was successful, this may be
921 * less than @len if there were partial characters
922 * at the end of the input. If the error
923 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
924 * stored will the byte offset after the last valid
925 * input sequence.
926 * @bytes_written: (out) (optional): the number of bytes stored in the output
927 * buffer (not including the terminating nul).
928 * @error: location to store the error occurring, or %NULL to ignore
929 * errors. Any of the errors in #GConvertError may occur.
930 *
931 * Converts a string from UTF-8 to the encoding used for strings by
932 * the C runtime (usually the same as that used by the operating
933 * system) in the [current locale][setlocale]. On Windows this means
934 * the system codepage.
935 *
936 * Returns: A newly-allocated buffer containing the converted string,
937 * or %NULL on an error, and error will be set.
938 **/
939 gchar *
940 g_locale_from_utf8 (const gchar *utf8string,
941 gssize len,
942 gsize *bytes_read,
943 gsize *bytes_written,
944 GError **error)
945 {
946 const gchar *charset;
947  
948 if (g_get_charset (&charset))
949 return strdup_len (utf8string, len, bytes_read, bytes_written, error);
950 else
951 return g_convert (utf8string, len,
952 charset, "UTF-8", bytes_read, bytes_written, error);
953 }
954  
955 #ifndef G_PLATFORM_WIN32
956  
957 typedef struct _GFilenameCharsetCache GFilenameCharsetCache;
958  
959 struct _GFilenameCharsetCache {
960 gboolean is_utf8;
961 gchar *charset;
962 gchar **filename_charsets;
963 };
964  
965 static void
966 filename_charset_cache_free (gpointer data)
967 {
968 GFilenameCharsetCache *cache = data;
969 g_free (cache->charset);
970 g_strfreev (cache->filename_charsets);
971 g_free (cache);
972 }
973  
974 /**
975 * g_get_filename_charsets:
976 * @charsets: return location for the %NULL-terminated list of encoding names
977 *
978 * Determines the preferred character sets used for filenames.
979 * The first character set from the @charsets is the filename encoding, the
980 * subsequent character sets are used when trying to generate a displayable
981 * representation of a filename, see g_filename_display_name().
982 *
983 * On Unix, the character sets are determined by consulting the
984 * environment variables `G_FILENAME_ENCODING` and `G_BROKEN_FILENAMES`.
985 * On Windows, the character set used in the GLib API is always UTF-8
986 * and said environment variables have no effect.
987 *
988 * `G_FILENAME_ENCODING` may be set to a comma-separated list of
989 * character set names. The special token "\@locale" is taken
990 * to mean the character set for the [current locale][setlocale].
991 * If `G_FILENAME_ENCODING` is not set, but `G_BROKEN_FILENAMES` is,
992 * the character set of the current locale is taken as the filename
993 * encoding. If neither environment variable is set, UTF-8 is taken
994 * as the filename encoding, but the character set of the current locale
995 * is also put in the list of encodings.
996 *
997 * The returned @charsets belong to GLib and must not be freed.
998 *
999 * Note that on Unix, regardless of the locale character set or
1000 * `G_FILENAME_ENCODING` value, the actual file names present
1001 * on a system might be in any random encoding or just gibberish.
1002 *
1003 * Returns: %TRUE if the filename encoding is UTF-8.
1004 *
1005 * Since: 2.6
1006 */
1007 gboolean
1008 g_get_filename_charsets (const gchar ***filename_charsets)
1009 {
1010 static GPrivate cache_private = G_PRIVATE_INIT (filename_charset_cache_free);
1011 GFilenameCharsetCache *cache = g_private_get (&cache_private);
1012 const gchar *charset;
1013  
1014 if (!cache)
1015 {
1016 cache = g_new0 (GFilenameCharsetCache, 1);
1017 g_private_set (&cache_private, cache);
1018 }
1019  
1020 g_get_charset (&charset);
1021  
1022 if (!(cache->charset && strcmp (cache->charset, charset) == 0))
1023 {
1024 const gchar *new_charset;
1025 gchar *p;
1026 gint i;
1027  
1028 g_free (cache->charset);
1029 g_strfreev (cache->filename_charsets);
1030 cache->charset = g_strdup (charset);
1031  
1032 p = getenv ("G_FILENAME_ENCODING");
1033 if (p != NULL && p[0] != '\0')
1034 {
1035 cache->filename_charsets = g_strsplit (p, ",", 0);
1036 cache->is_utf8 = (strcmp (cache->filename_charsets[0], "UTF-8") == 0);
1037  
1038 for (i = 0; cache->filename_charsets[i]; i++)
1039 {
1040 if (strcmp ("@locale", cache->filename_charsets[i]) == 0)
1041 {
1042 g_get_charset (&new_charset);
1043 g_free (cache->filename_charsets[i]);
1044 cache->filename_charsets[i] = g_strdup (new_charset);
1045 }
1046 }
1047 }
1048 else if (getenv ("G_BROKEN_FILENAMES") != NULL)
1049 {
1050 cache->filename_charsets = g_new0 (gchar *, 2);
1051 cache->is_utf8 = g_get_charset (&new_charset);
1052 cache->filename_charsets[0] = g_strdup (new_charset);
1053 }
1054 else
1055 {
1056 cache->filename_charsets = g_new0 (gchar *, 3);
1057 cache->is_utf8 = TRUE;
1058 cache->filename_charsets[0] = g_strdup ("UTF-8");
1059 if (!g_get_charset (&new_charset))
1060 cache->filename_charsets[1] = g_strdup (new_charset);
1061 }
1062 }
1063  
1064 if (filename_charsets)
1065 *filename_charsets = (const gchar **)cache->filename_charsets;
1066  
1067 return cache->is_utf8;
1068 }
1069  
1070 #else /* G_PLATFORM_WIN32 */
1071  
1072 gboolean
1073 g_get_filename_charsets (const gchar ***filename_charsets)
1074 {
1075 static const gchar *charsets[] = {
1076 "UTF-8",
1077 NULL
1078 };
1079  
1080 #ifdef G_OS_WIN32
1081 /* On Windows GLib pretends that the filename charset is UTF-8 */
1082 if (filename_charsets)
1083 *filename_charsets = charsets;
1084  
1085 return TRUE;
1086 #else
1087 gboolean result;
1088  
1089 /* Cygwin works like before */
1090 result = g_get_charset (&(charsets[0]));
1091  
1092 if (filename_charsets)
1093 *filename_charsets = charsets;
1094  
1095 return result;
1096 #endif
1097 }
1098  
1099 #endif /* G_PLATFORM_WIN32 */
1100  
1101 static gboolean
1102 get_filename_charset (const gchar **filename_charset)
1103 {
1104 const gchar **charsets;
1105 gboolean is_utf8;
1106  
1107 is_utf8 = g_get_filename_charsets (&charsets);
1108  
1109 if (filename_charset)
1110 *filename_charset = charsets[0];
1111  
1112 return is_utf8;
1113 }
1114  
1115 /**
1116 * g_filename_to_utf8:
1117 * @opsysstring: a string in the encoding for filenames
1118 * @len: the length of the string, or -1 if the string is
1119 * nul-terminated (Note that some encodings may allow nul
1120 * bytes to occur inside strings. In that case, using -1
1121 * for the @len parameter is unsafe)
1122 * @bytes_read: (out) (optional): location to store the number of bytes in the
1123 * input string that were successfully converted, or %NULL.
1124 * Even if the conversion was successful, this may be
1125 * less than @len if there were partial characters
1126 * at the end of the input. If the error
1127 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
1128 * stored will the byte offset after the last valid
1129 * input sequence.
1130 * @bytes_written: (out) (optional): the number of bytes stored in the output
1131 * buffer (not including the terminating nul).
1132 * @error: location to store the error occurring, or %NULL to ignore
1133 * errors. Any of the errors in #GConvertError may occur.
1134 *
1135 * Converts a string which is in the encoding used by GLib for
1136 * filenames into a UTF-8 string. Note that on Windows GLib uses UTF-8
1137 * for filenames; on other platforms, this function indirectly depends on
1138 * the [current locale][setlocale].
1139 *
1140 * Returns: The converted string, or %NULL on an error.
1141 **/
1142 gchar*
1143 g_filename_to_utf8 (const gchar *opsysstring,
1144 gssize len,
1145 gsize *bytes_read,
1146 gsize *bytes_written,
1147 GError **error)
1148 {
1149 const gchar *charset;
1150  
1151 g_return_val_if_fail (opsysstring != NULL, NULL);
1152  
1153 if (get_filename_charset (&charset))
1154 return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
1155 else
1156 return g_convert (opsysstring, len,
1157 "UTF-8", charset, bytes_read, bytes_written, error);
1158 }
1159  
1160 #if defined (G_OS_WIN32) && !defined (_WIN64)
1161  
1162 #undef g_filename_to_utf8
1163  
1164 /* Binary compatibility version. Not for newly compiled code. Also not needed for
1165 * 64-bit versions as there should be no old deployed binaries that would use
1166 * the old versions.
1167 */
1168  
1169 gchar*
1170 g_filename_to_utf8 (const gchar *opsysstring,
1171 gssize len,
1172 gsize *bytes_read,
1173 gsize *bytes_written,
1174 GError **error)
1175 {
1176 const gchar *charset;
1177  
1178 g_return_val_if_fail (opsysstring != NULL, NULL);
1179  
1180 if (g_get_charset (&charset))
1181 return strdup_len (opsysstring, len, bytes_read, bytes_written, error);
1182 else
1183 return g_convert (opsysstring, len,
1184 "UTF-8", charset, bytes_read, bytes_written, error);
1185 }
1186  
1187 #endif
1188  
1189 /**
1190 * g_filename_from_utf8:
1191 * @utf8string: a UTF-8 encoded string.
1192 * @len: the length of the string, or -1 if the string is
1193 * nul-terminated.
1194 * @bytes_read: (out) (optional): location to store the number of bytes in
1195 * the input string that were successfully converted, or %NULL.
1196 * Even if the conversion was successful, this may be
1197 * less than @len if there were partial characters
1198 * at the end of the input. If the error
1199 * #G_CONVERT_ERROR_ILLEGAL_SEQUENCE occurs, the value
1200 * stored will the byte offset after the last valid
1201 * input sequence.
1202 * @bytes_written: (out): the number of bytes stored in the output buffer (not
1203 * including the terminating nul).
1204 * @error: location to store the error occurring, or %NULL to ignore
1205 * errors. Any of the errors in #GConvertError may occur.
1206 *
1207 * Converts a string from UTF-8 to the encoding GLib uses for
1208 * filenames. Note that on Windows GLib uses UTF-8 for filenames;
1209 * on other platforms, this function indirectly depends on the
1210 * [current locale][setlocale].
1211 *
1212 * Returns: (array length=bytes_written) (element-type guint8) (transfer full):
1213 * The converted string, or %NULL on an error.
1214 **/
1215 gchar*
1216 g_filename_from_utf8 (const gchar *utf8string,
1217 gssize len,
1218 gsize *bytes_read,
1219 gsize *bytes_written,
1220 GError **error)
1221 {
1222 const gchar *charset;
1223  
1224 if (get_filename_charset (&charset))
1225 return strdup_len (utf8string, len, bytes_read, bytes_written, error);
1226 else
1227 return g_convert (utf8string, len,
1228 charset, "UTF-8", bytes_read, bytes_written, error);
1229 }
1230  
1231 #if defined (G_OS_WIN32) && !defined (_WIN64)
1232  
1233 #undef g_filename_from_utf8
1234  
1235 /* Binary compatibility version. Not for newly compiled code. */
1236  
1237 gchar*
1238 g_filename_from_utf8 (const gchar *utf8string,
1239 gssize len,
1240 gsize *bytes_read,
1241 gsize *bytes_written,
1242 GError **error)
1243 {
1244 const gchar *charset;
1245  
1246 if (g_get_charset (&charset))
1247 return strdup_len (utf8string, len, bytes_read, bytes_written, error);
1248 else
1249 return g_convert (utf8string, len,
1250 charset, "UTF-8", bytes_read, bytes_written, error);
1251 }
1252  
1253 #endif
1254  
1255 /* Test of haystack has the needle prefix, comparing case
1256 * insensitive. haystack may be UTF-8, but needle must
1257 * contain only ascii. */
1258 static gboolean
1259 has_case_prefix (const gchar *haystack, const gchar *needle)
1260 {
1261 const gchar *h, *n;
1262  
1263 /* Eat one character at a time. */
1264 h = haystack;
1265 n = needle;
1266  
1267 while (*n && *h &&
1268 g_ascii_tolower (*n) == g_ascii_tolower (*h))
1269 {
1270 n++;
1271 h++;
1272 }
1273  
1274 return *n == '\0';
1275 }
1276  
1277 typedef enum {
1278 UNSAFE_ALL = 0x1, /* Escape all unsafe characters */
1279 UNSAFE_ALLOW_PLUS = 0x2, /* Allows '+' */
1280 UNSAFE_PATH = 0x8, /* Allows '/', '&', '=', ':', '@', '+', '$' and ',' */
1281 UNSAFE_HOST = 0x10, /* Allows '/' and ':' and '@' */
1282 UNSAFE_SLASHES = 0x20 /* Allows all characters except for '/' and '%' */
1283 } UnsafeCharacterSet;
1284  
1285 static const guchar acceptable[96] = {
1286 /* A table of the ASCII chars from space (32) to DEL (127) */
1287 /* ! " # $ % & ' ( ) * + , - . / */
1288 0x00,0x3F,0x20,0x20,0x28,0x00,0x2C,0x3F,0x3F,0x3F,0x3F,0x2A,0x28,0x3F,0x3F,0x1C,
1289 /* 0 1 2 3 4 5 6 7 8 9 : ; < = > ? */
1290 0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x38,0x20,0x20,0x2C,0x20,0x20,
1291 /* @ A B C D E F G H I J K L M N O */
1292 0x38,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,
1293 /* P Q R S T U V W X Y Z [ \ ] ^ _ */
1294 0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x20,0x20,0x20,0x20,0x3F,
1295 /* ` a b c d e f g h i j k l m n o */
1296 0x20,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,
1297 /* p q r s t u v w x y z { | } ~ DEL */
1298 0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x3F,0x20,0x20,0x20,0x3F,0x20
1299 };
1300  
1301 static const gchar hex[16] = "0123456789ABCDEF";
1302  
1303 /* Note: This escape function works on file: URIs, but if you want to
1304 * escape something else, please read RFC-2396 */
1305 static gchar *
1306 g_escape_uri_string (const gchar *string,
1307 UnsafeCharacterSet mask)
1308 {
1309 #define ACCEPTABLE(a) ((a)>=32 && (a)<128 && (acceptable[(a)-32] & use_mask))
1310  
1311 const gchar *p;
1312 gchar *q;
1313 gchar *result;
1314 int c;
1315 gint unacceptable;
1316 UnsafeCharacterSet use_mask;
1317  
1318 g_return_val_if_fail (mask == UNSAFE_ALL
1319 || mask == UNSAFE_ALLOW_PLUS
1320 || mask == UNSAFE_PATH
1321 || mask == UNSAFE_HOST
1322 || mask == UNSAFE_SLASHES, NULL);
1323  
1324 unacceptable = 0;
1325 use_mask = mask;
1326 for (p = string; *p != '\0'; p++)
1327 {
1328 c = (guchar) *p;
1329 if (!ACCEPTABLE (c))
1330 unacceptable++;
1331 }
1332  
1333 result = g_malloc (p - string + unacceptable * 2 + 1);
1334  
1335 use_mask = mask;
1336 for (q = result, p = string; *p != '\0'; p++)
1337 {
1338 c = (guchar) *p;
1339  
1340 if (!ACCEPTABLE (c))
1341 {
1342 *q++ = '%'; /* means hex coming */
1343 *q++ = hex[c >> 4];
1344 *q++ = hex[c & 15];
1345 }
1346 else
1347 *q++ = *p;
1348 }
1349  
1350 *q = '\0';
1351  
1352 return result;
1353 }
1354  
1355  
1356 static gchar *
1357 g_escape_file_uri (const gchar *hostname,
1358 const gchar *pathname)
1359 {
1360 char *escaped_hostname = NULL;
1361 char *escaped_path;
1362 char *res;
1363  
1364 #ifdef G_OS_WIN32
1365 char *p, *backslash;
1366  
1367 /* Turn backslashes into forward slashes. That's what Netscape
1368 * does, and they are actually more or less equivalent in Windows.
1369 */
1370  
1371 pathname = g_strdup (pathname);
1372 p = (char *) pathname;
1373  
1374 while ((backslash = strchr (p, '\\')) != NULL)
1375 {
1376 *backslash = '/';
1377 p = backslash + 1;
1378 }
1379 #endif
1380  
1381 if (hostname && *hostname != '\0')
1382 {
1383 escaped_hostname = g_escape_uri_string (hostname, UNSAFE_HOST);
1384 }
1385  
1386 escaped_path = g_escape_uri_string (pathname, UNSAFE_PATH);
1387  
1388 res = g_strconcat ("file://",
1389 (escaped_hostname) ? escaped_hostname : "",
1390 (*escaped_path != '/') ? "/" : "",
1391 escaped_path,
1392 NULL);
1393  
1394 #ifdef G_OS_WIN32
1395 g_free ((char *) pathname);
1396 #endif
1397  
1398 g_free (escaped_hostname);
1399 g_free (escaped_path);
1400  
1401 return res;
1402 }
1403  
1404 static int
1405 unescape_character (const char *scanner)
1406 {
1407 int first_digit;
1408 int second_digit;
1409  
1410 first_digit = g_ascii_xdigit_value (scanner[0]);
1411 if (first_digit < 0)
1412 return -1;
1413  
1414 second_digit = g_ascii_xdigit_value (scanner[1]);
1415 if (second_digit < 0)
1416 return -1;
1417  
1418 return (first_digit << 4) | second_digit;
1419 }
1420  
1421 static gchar *
1422 g_unescape_uri_string (const char *escaped,
1423 int len,
1424 const char *illegal_escaped_characters,
1425 gboolean ascii_must_not_be_escaped)
1426 {
1427 const gchar *in, *in_end;
1428 gchar *out, *result;
1429 int c;
1430  
1431 if (escaped == NULL)
1432 return NULL;
1433  
1434 if (len < 0)
1435 len = strlen (escaped);
1436  
1437 result = g_malloc (len + 1);
1438  
1439 out = result;
1440 for (in = escaped, in_end = escaped + len; in < in_end; in++)
1441 {
1442 c = *in;
1443  
1444 if (c == '%')
1445 {
1446 /* catch partial escape sequences past the end of the substring */
1447 if (in + 3 > in_end)
1448 break;
1449  
1450 c = unescape_character (in + 1);
1451  
1452 /* catch bad escape sequences and NUL characters */
1453 if (c <= 0)
1454 break;
1455  
1456 /* catch escaped ASCII */
1457 if (ascii_must_not_be_escaped && c <= 0x7F)
1458 break;
1459  
1460 /* catch other illegal escaped characters */
1461 if (strchr (illegal_escaped_characters, c) != NULL)
1462 break;
1463  
1464 in += 2;
1465 }
1466  
1467 *out++ = c;
1468 }
1469  
1470 g_assert (out - result <= len);
1471 *out = '\0';
1472  
1473 if (in != in_end)
1474 {
1475 g_free (result);
1476 return NULL;
1477 }
1478  
1479 return result;
1480 }
1481  
1482 static gboolean
1483 is_asciialphanum (gunichar c)
1484 {
1485 return c <= 0x7F && g_ascii_isalnum (c);
1486 }
1487  
1488 static gboolean
1489 is_asciialpha (gunichar c)
1490 {
1491 return c <= 0x7F && g_ascii_isalpha (c);
1492 }
1493  
1494 /* allows an empty string */
1495 static gboolean
1496 hostname_validate (const char *hostname)
1497 {
1498 const char *p;
1499 gunichar c, first_char, last_char;
1500  
1501 p = hostname;
1502 if (*p == '\0')
1503 return TRUE;
1504 do
1505 {
1506 /* read in a label */
1507 c = g_utf8_get_char (p);
1508 p = g_utf8_next_char (p);
1509 if (!is_asciialphanum (c))
1510 return FALSE;
1511 first_char = c;
1512 do
1513 {
1514 last_char = c;
1515 c = g_utf8_get_char (p);
1516 p = g_utf8_next_char (p);
1517 }
1518 while (is_asciialphanum (c) || c == '-');
1519 if (last_char == '-')
1520 return FALSE;
1521  
1522 /* if that was the last label, check that it was a toplabel */
1523 if (c == '\0' || (c == '.' && *p == '\0'))
1524 return is_asciialpha (first_char);
1525 }
1526 while (c == '.');
1527 return FALSE;
1528 }
1529  
1530 /**
1531 * g_filename_from_uri:
1532 * @uri: a uri describing a filename (escaped, encoded in ASCII).
1533 * @hostname: (out) (optional) (nullable): Location to store hostname for the
1534 * URI.
1535 * If there is no hostname in the URI, %NULL will be
1536 * stored in this location.
1537 * @error: location to store the error occurring, or %NULL to ignore
1538 * errors. Any of the errors in #GConvertError may occur.
1539 *
1540 * Converts an escaped ASCII-encoded URI to a local filename in the
1541 * encoding used for filenames.
1542 *
1543 * Returns: (type filename): a newly-allocated string holding
1544 * the resulting filename, or %NULL on an error.
1545 **/
1546 gchar *
1547 g_filename_from_uri (const gchar *uri,
1548 gchar **hostname,
1549 GError **error)
1550 {
1551 const char *path_part;
1552 const char *host_part;
1553 char *unescaped_hostname;
1554 char *result;
1555 char *filename;
1556 int offs;
1557 #ifdef G_OS_WIN32
1558 char *p, *slash;
1559 #endif
1560  
1561 if (hostname)
1562 *hostname = NULL;
1563  
1564 if (!has_case_prefix (uri, "file:/"))
1565 {
1566 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1567 _("The URI '%s' is not an absolute URI using the \"file\" scheme"),
1568 uri);
1569 return NULL;
1570 }
1571  
1572 path_part = uri + strlen ("file:");
1573  
1574 if (strchr (path_part, '#') != NULL)
1575 {
1576 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1577 _("The local file URI '%s' may not include a '#'"),
1578 uri);
1579 return NULL;
1580 }
1581  
1582 if (has_case_prefix (path_part, "///"))
1583 path_part += 2;
1584 else if (has_case_prefix (path_part, "//"))
1585 {
1586 path_part += 2;
1587 host_part = path_part;
1588  
1589 path_part = strchr (path_part, '/');
1590  
1591 if (path_part == NULL)
1592 {
1593 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1594 _("The URI '%s' is invalid"),
1595 uri);
1596 return NULL;
1597 }
1598  
1599 unescaped_hostname = g_unescape_uri_string (host_part, path_part - host_part, "", TRUE);
1600  
1601 if (unescaped_hostname == NULL ||
1602 !hostname_validate (unescaped_hostname))
1603 {
1604 g_free (unescaped_hostname);
1605 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1606 _("The hostname of the URI '%s' is invalid"),
1607 uri);
1608 return NULL;
1609 }
1610  
1611 if (hostname)
1612 *hostname = unescaped_hostname;
1613 else
1614 g_free (unescaped_hostname);
1615 }
1616  
1617 filename = g_unescape_uri_string (path_part, -1, "/", FALSE);
1618  
1619 if (filename == NULL)
1620 {
1621 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_BAD_URI,
1622 _("The URI '%s' contains invalidly escaped characters"),
1623 uri);
1624 return NULL;
1625 }
1626  
1627 offs = 0;
1628 #ifdef G_OS_WIN32
1629 /* Drop localhost */
1630 if (hostname && *hostname != NULL &&
1631 g_ascii_strcasecmp (*hostname, "localhost") == 0)
1632 {
1633 g_free (*hostname);
1634 *hostname = NULL;
1635 }
1636  
1637 /* Turn slashes into backslashes, because that's the canonical spelling */
1638 p = filename;
1639 while ((slash = strchr (p, '/')) != NULL)
1640 {
1641 *slash = '\\';
1642 p = slash + 1;
1643 }
1644  
1645 /* Windows URIs with a drive letter can be like "file://host/c:/foo"
1646 * or "file://host/c|/foo" (some Netscape versions). In those cases, start
1647 * the filename from the drive letter.
1648 */
1649 if (g_ascii_isalpha (filename[1]))
1650 {
1651 if (filename[2] == ':')
1652 offs = 1;
1653 else if (filename[2] == '|')
1654 {
1655 filename[2] = ':';
1656 offs = 1;
1657 }
1658 }
1659 #endif
1660  
1661 result = g_strdup (filename + offs);
1662 g_free (filename);
1663  
1664 return result;
1665 }
1666  
1667 #if defined (G_OS_WIN32) && !defined (_WIN64)
1668  
1669 #undef g_filename_from_uri
1670  
1671 gchar *
1672 g_filename_from_uri (const gchar *uri,
1673 gchar **hostname,
1674 GError **error)
1675 {
1676 gchar *utf8_filename;
1677 gchar *retval = NULL;
1678  
1679 utf8_filename = g_filename_from_uri_utf8 (uri, hostname, error);
1680 if (utf8_filename)
1681 {
1682 retval = g_locale_from_utf8 (utf8_filename, -1, NULL, NULL, error);
1683 g_free (utf8_filename);
1684 }
1685 return retval;
1686 }
1687  
1688 #endif
1689  
1690 /**
1691 * g_filename_to_uri:
1692 * @filename: an absolute filename specified in the GLib file name encoding,
1693 * which is the on-disk file name bytes on Unix, and UTF-8 on
1694 * Windows
1695 * @hostname: (allow-none): A UTF-8 encoded hostname, or %NULL for none.
1696 * @error: location to store the error occurring, or %NULL to ignore
1697 * errors. Any of the errors in #GConvertError may occur.
1698 *
1699 * Converts an absolute filename to an escaped ASCII-encoded URI, with the path
1700 * component following Section 3.3. of RFC 2396.
1701 *
1702 * Returns: a newly-allocated string holding the resulting
1703 * URI, or %NULL on an error.
1704 **/
1705 gchar *
1706 g_filename_to_uri (const gchar *filename,
1707 const gchar *hostname,
1708 GError **error)
1709 {
1710 char *escaped_uri;
1711  
1712 g_return_val_if_fail (filename != NULL, NULL);
1713  
1714 if (!g_path_is_absolute (filename))
1715 {
1716 g_set_error (error, G_CONVERT_ERROR, G_CONVERT_ERROR_NOT_ABSOLUTE_PATH,
1717 _("The pathname '%s' is not an absolute path"),
1718 filename);
1719 return NULL;
1720 }
1721  
1722 if (hostname &&
1723 !(g_utf8_validate (hostname, -1, NULL)
1724 && hostname_validate (hostname)))
1725 {
1726 g_set_error_literal (error, G_CONVERT_ERROR, G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
1727 _("Invalid hostname"));
1728 return NULL;
1729 }
1730  
1731 #ifdef G_OS_WIN32
1732 /* Don't use localhost unnecessarily */
1733 if (hostname && g_ascii_strcasecmp (hostname, "localhost") == 0)
1734 hostname = NULL;
1735 #endif
1736  
1737 escaped_uri = g_escape_file_uri (hostname, filename);
1738  
1739 return escaped_uri;
1740 }
1741  
1742 #if defined (G_OS_WIN32) && !defined (_WIN64)
1743  
1744 #undef g_filename_to_uri
1745  
1746 gchar *
1747 g_filename_to_uri (const gchar *filename,
1748 const gchar *hostname,
1749 GError **error)
1750 {
1751 gchar *utf8_filename;
1752 gchar *retval = NULL;
1753  
1754 utf8_filename = g_locale_to_utf8 (filename, -1, NULL, NULL, error);
1755  
1756 if (utf8_filename)
1757 {
1758 retval = g_filename_to_uri_utf8 (utf8_filename, hostname, error);
1759 g_free (utf8_filename);
1760 }
1761  
1762 return retval;
1763 }
1764  
1765 #endif
1766  
1767 /**
1768 * g_uri_list_extract_uris:
1769 * @uri_list: an URI list
1770 *
1771 * Splits an URI list conforming to the text/uri-list
1772 * mime type defined in RFC 2483 into individual URIs,
1773 * discarding any comments. The URIs are not validated.
1774 *
1775 * Returns: (transfer full): a newly allocated %NULL-terminated list
1776 * of strings holding the individual URIs. The array should be freed
1777 * with g_strfreev().
1778 *
1779 * Since: 2.6
1780 */
1781 gchar **
1782 g_uri_list_extract_uris (const gchar *uri_list)
1783 {
1784 GSList *uris, *u;
1785 const gchar *p, *q;
1786 gchar **result;
1787 gint n_uris = 0;
1788  
1789 uris = NULL;
1790  
1791 p = uri_list;
1792  
1793 /* We don't actually try to validate the URI according to RFC
1794 * 2396, or even check for allowed characters - we just ignore
1795 * comments and trim whitespace off the ends. We also
1796 * allow LF delimination as well as the specified CRLF.
1797 *
1798 * We do allow comments like specified in RFC 2483.
1799 */
1800 while (p)
1801 {
1802 if (*p != '#')
1803 {
1804 while (g_ascii_isspace (*p))
1805 p++;
1806  
1807 q = p;
1808 while (*q && (*q != '\n') && (*q != '\r'))
1809 q++;
1810  
1811 if (q > p)
1812 {
1813 q--;
1814 while (q > p && g_ascii_isspace (*q))
1815 q--;
1816  
1817 if (q > p)
1818 {
1819 uris = g_slist_prepend (uris, g_strndup (p, q - p + 1));
1820 n_uris++;
1821 }
1822 }
1823 }
1824 p = strchr (p, '\n');
1825 if (p)
1826 p++;
1827 }
1828  
1829 result = g_new (gchar *, n_uris + 1);
1830  
1831 result[n_uris--] = NULL;
1832 for (u = uris; u; u = u->next)
1833 result[n_uris--] = u->data;
1834  
1835 g_slist_free (uris);
1836  
1837 return result;
1838 }
1839  
1840 /**
1841 * g_filename_display_basename:
1842 * @filename: an absolute pathname in the GLib file name encoding
1843 *
1844 * Returns the display basename for the particular filename, guaranteed
1845 * to be valid UTF-8. The display name might not be identical to the filename,
1846 * for instance there might be problems converting it to UTF-8, and some files
1847 * can be translated in the display.
1848 *
1849 * If GLib cannot make sense of the encoding of @filename, as a last resort it
1850 * replaces unknown characters with U+FFFD, the Unicode replacement character.
1851 * You can search the result for the UTF-8 encoding of this character (which is
1852 * "\357\277\275" in octal notation) to find out if @filename was in an invalid
1853 * encoding.
1854 *
1855 * You must pass the whole absolute pathname to this functions so that
1856 * translation of well known locations can be done.
1857 *
1858 * This function is preferred over g_filename_display_name() if you know the
1859 * whole path, as it allows translation.
1860 *
1861 * Returns: a newly allocated string containing
1862 * a rendition of the basename of the filename in valid UTF-8
1863 *
1864 * Since: 2.6
1865 **/
1866 gchar *
1867 g_filename_display_basename (const gchar *filename)
1868 {
1869 char *basename;
1870 char *display_name;
1871  
1872 g_return_val_if_fail (filename != NULL, NULL);
1873  
1874 basename = g_path_get_basename (filename);
1875 display_name = g_filename_display_name (basename);
1876 g_free (basename);
1877 return display_name;
1878 }
1879  
1880 /**
1881 * g_filename_display_name:
1882 * @filename: a pathname hopefully in the GLib file name encoding
1883 *
1884 * Converts a filename into a valid UTF-8 string. The conversion is
1885 * not necessarily reversible, so you should keep the original around
1886 * and use the return value of this function only for display purposes.
1887 * Unlike g_filename_to_utf8(), the result is guaranteed to be non-%NULL
1888 * even if the filename actually isn't in the GLib file name encoding.
1889 *
1890 * If GLib cannot make sense of the encoding of @filename, as a last resort it
1891 * replaces unknown characters with U+FFFD, the Unicode replacement character.
1892 * You can search the result for the UTF-8 encoding of this character (which is
1893 * "\357\277\275" in octal notation) to find out if @filename was in an invalid
1894 * encoding.
1895 *
1896 * If you know the whole pathname of the file you should use
1897 * g_filename_display_basename(), since that allows location-based
1898 * translation of filenames.
1899 *
1900 * Returns: a newly allocated string containing
1901 * a rendition of the filename in valid UTF-8
1902 *
1903 * Since: 2.6
1904 **/
1905 gchar *
1906 g_filename_display_name (const gchar *filename)
1907 {
1908 gint i;
1909 const gchar **charsets;
1910 gchar *display_name = NULL;
1911 gboolean is_utf8;
1912  
1913 is_utf8 = g_get_filename_charsets (&charsets);
1914  
1915 if (is_utf8)
1916 {
1917 if (g_utf8_validate (filename, -1, NULL))
1918 display_name = g_strdup (filename);
1919 }
1920  
1921 if (!display_name)
1922 {
1923 /* Try to convert from the filename charsets to UTF-8.
1924 * Skip the first charset if it is UTF-8.
1925 */
1926 for (i = is_utf8 ? 1 : 0; charsets[i]; i++)
1927 {
1928 display_name = g_convert (filename, -1, "UTF-8", charsets[i],
1929 NULL, NULL, NULL);
1930  
1931 if (display_name)
1932 break;
1933 }
1934 }
1935  
1936 /* if all conversions failed, we replace invalid UTF-8
1937 * by a question mark
1938 */
1939 if (!display_name)
1940 display_name = _g_utf8_make_valid (filename);
1941  
1942 return display_name;
1943 }