1 /* Conversion of links to local files.
2 Copyright (C) 1996, 1997, 2000, 2001 Free Software Foundation, Inc.
4 This file is part of GNU Wget.
6 GNU Wget is free software; you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation; either version 2 of the License, or
9 (at your option) any later version.
11 GNU Wget is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with Wget; if not, write to the Free Software
18 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
20 In addition, as a special exception, the Free Software Foundation
21 gives permission to link the code of its release of Wget with the
22 OpenSSL project's "OpenSSL" library (or with modified versions of it
23 that use the same license as the "OpenSSL" library), and distribute
24 the linked executables. You must obey the GNU General Public License
25 in all respects for all of the code used other than "OpenSSL". If you
26 modify this file, you may extend this exception to your version of the
27 file, but you are not obligated to do so. If you do not wish to do
28 so, delete this exception statement from your version. */
38 #endif /* HAVE_STRING_H */
41 #endif /* HAVE_UNISTD_H */
44 #include <sys/types.h>
53 static struct hash_table *dl_file_url_map;
54 struct hash_table *dl_url_file_map;
56 /* List of HTML files downloaded in this Wget run, used for link
57 conversion after Wget is done. The list and the set contain the
58 same information, except the list maintains the order. Perhaps I
59 should get rid of the list, it's there for historical reasons. */
60 static slist *downloaded_html_list;
61 struct hash_table *downloaded_html_set;
63 static void convert_links PARAMS ((const char *, struct urlpos *));
65 /* This function is called when the retrieval is done to convert the
66 links that have been downloaded. It has to be called at the end of
67 the retrieval, because only then does Wget know conclusively which
68 URLs have been downloaded, and which not, so it can tell which
69 direction to convert to.
71 The "direction" means that the URLs to the files that have been
72 downloaded get converted to the relative URL which will point to
73 that file. And the other URLs get converted to the remote URL on
76 All the downloaded HTMLs are kept in downloaded_html_files, and
77 downloaded URLs in urls_downloaded. All the information is
78 extracted from these two lists. */
81 convert_all_links (void)
87 struct wget_timer *timer = wtimer_new ();
89 /* Destructively reverse downloaded_html_files to get it in the right order.
90 recursive_retrieve() used slist_prepend() consistently. */
91 downloaded_html_list = slist_nreverse (downloaded_html_list);
93 for (html = downloaded_html_list; html; html = html->next)
95 struct urlpos *urls, *cur_url;
97 char *file = html->string;
99 /* Determine the URL of the HTML file. get_urls_html will need
101 url = hash_table_get (dl_file_url_map, file);
104 DEBUGP (("Apparently %s has been removed.\n", file));
108 DEBUGP (("Scanning %s (from %s)\n", file, url));
110 /* Parse the HTML file... */
111 urls = get_urls_html (file, url, NULL);
113 /* We don't respect meta_disallow_follow here because, even if
114 the file is not followed, we might still want to convert the
115 links that have been followed from other files. */
117 for (cur_url = urls; cur_url; cur_url = cur_url->next)
120 struct url *u = cur_url->url;
122 if (cur_url->link_base_p)
124 /* Base references have been resolved by our parser, so
125 we turn the base URL into an empty string. (Perhaps
126 we should remove the tag entirely?) */
127 cur_url->convert = CO_NULLIFY_BASE;
131 /* We decide the direction of conversion according to whether
132 a URL was downloaded. Downloaded URLs will be converted
133 ABS2REL, whereas non-downloaded will be converted REL2ABS. */
134 local_name = hash_table_get (dl_url_file_map, u->url);
136 /* Decide on the conversion type. */
139 /* We've downloaded this URL. Convert it to relative
140 form. We do this even if the URL already is in
141 relative form, because our directory structure may
142 not be identical to that on the server (think `-nd',
143 `--cut-dirs', etc.) */
144 cur_url->convert = CO_CONVERT_TO_RELATIVE;
145 cur_url->local_name = xstrdup (local_name);
146 DEBUGP (("will convert url %s to local %s\n", u->url, local_name));
150 /* We haven't downloaded this URL. If it's not already
151 complete (including a full host name), convert it to
152 that form, so it can be reached while browsing this
154 if (!cur_url->link_complete_p)
155 cur_url->convert = CO_CONVERT_TO_COMPLETE;
156 cur_url->local_name = NULL;
157 DEBUGP (("will convert url %s to complete\n", u->url));
161 /* Convert the links in the file. */
162 convert_links (file, urls);
169 msecs = wtimer_elapsed (timer);
170 wtimer_delete (timer);
171 logprintf (LOG_VERBOSE, _("Converted %d files in %.2f seconds.\n"),
172 file_count, (double)msecs / 1000);
175 static void write_backup_file PARAMS ((const char *, downloaded_file_t));
176 static const char *replace_attr PARAMS ((const char *, int, FILE *,
178 static const char *replace_attr_refresh_hack PARAMS ((const char *, int, FILE *,
180 static char *local_quote_string PARAMS ((const char *));
181 static char *construct_relative PARAMS ((const char *, const char *));
183 /* Change the links in one HTML file. LINKS is a list of links in the
184 document, along with their positions and the desired direction of
187 convert_links (const char *file, struct urlpos *links)
189 struct file_memory *fm;
192 downloaded_file_t downloaded_file_return;
195 int to_url_count = 0, to_file_count = 0;
197 logprintf (LOG_VERBOSE, _("Converting %s... "), file);
200 /* First we do a "dry run": go through the list L and see whether
201 any URL needs to be converted in the first place. If not, just
202 leave the file alone. */
204 struct urlpos *dry = links;
205 for (dry = links; dry; dry = dry->next)
206 if (dry->convert != CO_NOCONVERT)
210 logputs (LOG_VERBOSE, _("nothing to do.\n"));
215 fm = read_file (file);
218 logprintf (LOG_NOTQUIET, _("Cannot convert links in %s: %s\n"),
219 file, strerror (errno));
223 downloaded_file_return = downloaded_file (CHECK_FOR_FILE, file);
224 if (opt.backup_converted && downloaded_file_return)
225 write_backup_file (file, downloaded_file_return);
227 /* Before opening the file for writing, unlink the file. This is
228 important if the data in FM is mmaped. In such case, nulling the
229 file, which is what fopen() below does, would make us read all
230 zeroes from the mmaped region. */
231 if (unlink (file) < 0 && errno != ENOENT)
233 logprintf (LOG_NOTQUIET, _("Unable to delete `%s': %s\n"),
234 file, strerror (errno));
238 /* Now open the file for writing. */
239 fp = fopen (file, "wb");
242 logprintf (LOG_NOTQUIET, _("Cannot convert links in %s: %s\n"),
243 file, strerror (errno));
248 /* Here we loop through all the URLs in file, replacing those of
249 them that are downloaded with relative references. */
251 for (link = links; link; link = link->next)
253 char *url_start = fm->content + link->pos;
255 if (link->pos >= fm->length)
257 DEBUGP (("Something strange is going on. Please investigate."));
260 /* If the URL is not to be converted, skip it. */
261 if (link->convert == CO_NOCONVERT)
263 DEBUGP (("Skipping %s at position %d.\n", link->url->url, link->pos));
267 /* Echo the file contents, up to the offending URL's opening
268 quote, to the outfile. */
269 fwrite (p, 1, url_start - p, fp);
272 switch (link->convert)
274 case CO_CONVERT_TO_RELATIVE:
275 /* Convert absolute URL to relative. */
277 char *newname = construct_relative (file, link->local_name);
278 char *quoted_newname = local_quote_string (newname);
280 if (!link->link_refresh_p)
281 p = replace_attr (p, link->size, fp, quoted_newname);
283 p = replace_attr_refresh_hack (p, link->size, fp, quoted_newname,
284 link->refresh_timeout);
286 DEBUGP (("TO_RELATIVE: %s to %s at position %d in %s.\n",
287 link->url->url, newname, link->pos, file));
289 xfree (quoted_newname);
293 case CO_CONVERT_TO_COMPLETE:
294 /* Convert the link to absolute URL. */
296 char *newlink = link->url->url;
297 char *quoted_newlink = html_quote_string (newlink);
299 if (!link->link_refresh_p)
300 p = replace_attr (p, link->size, fp, quoted_newlink);
302 p = replace_attr_refresh_hack (p, link->size, fp, quoted_newlink,
303 link->refresh_timeout);
305 DEBUGP (("TO_COMPLETE: <something> to %s at position %d in %s.\n",
306 newlink, link->pos, file));
307 xfree (quoted_newlink);
311 case CO_NULLIFY_BASE:
312 /* Change the base href to "". */
313 p = replace_attr (p, link->size, fp, "");
321 /* Output the rest of the file. */
322 if (p - fm->content < fm->length)
323 fwrite (p, 1, fm->length - (p - fm->content), fp);
327 logprintf (LOG_VERBOSE, "%d-%d\n", to_file_count, to_url_count);
330 /* Construct and return a link that points from BASEFILE to LINKFILE.
331 Both files should be local file names, BASEFILE of the referrering
332 file, and LINKFILE of the referred file.
336 cr("foo", "bar") -> "bar"
337 cr("A/foo", "A/bar") -> "bar"
338 cr("A/foo", "A/B/bar") -> "B/bar"
339 cr("A/X/foo", "A/Y/bar") -> "../Y/bar"
340 cr("X/", "Y/bar") -> "../Y/bar" (trailing slash does matter in BASE)
342 Both files should be absolute or relative, otherwise strange
343 results might ensue. The function makes no special efforts to
344 handle "." and ".." in links, so make sure they're not there
345 (e.g. using path_simplify). */
348 construct_relative (const char *basefile, const char *linkfile)
355 /* First, skip the initial directory components common to both
358 for (b = basefile, l = linkfile; *b == *l && *b != '\0'; ++b, ++l)
361 start = (b - basefile) + 1;
366 /* With common directories out of the way, the situation we have is
368 b - b1/b2/[...]/bfile
369 l - l1/l2/[...]/lfile
371 The link we're constructing needs to be:
372 lnk - ../../l1/l2/[...]/lfile
374 Where the number of ".."'s equals the number of bN directory
377 /* Count the directory components in B. */
379 for (b = basefile; *b; b++)
385 /* Construct LINK as explained above. */
386 link = (char *)xmalloc (3 * basedirs + strlen (linkfile) + 1);
387 for (i = 0; i < basedirs; i++)
388 memcpy (link + 3 * i, "../", 3);
389 strcpy (link + 3 * i, linkfile);
394 write_backup_file (const char *file, downloaded_file_t downloaded_file_return)
396 /* Rather than just writing over the original .html file with the
397 converted version, save the former to *.orig. Note we only do
398 this for files we've _successfully_ downloaded, so we don't
399 clobber .orig files sitting around from previous invocations. */
401 /* Construct the backup filename as the original name plus ".orig". */
402 size_t filename_len = strlen(file);
403 char* filename_plus_orig_suffix;
404 boolean already_wrote_backup_file = FALSE;
405 slist* converted_file_ptr;
406 static slist* converted_files = NULL;
408 if (downloaded_file_return == FILE_DOWNLOADED_AND_HTML_EXTENSION_ADDED)
410 /* Just write "orig" over "html". We need to do it this way
411 because when we're checking to see if we've downloaded the
412 file before (to see if we can skip downloading it), we don't
413 know if it's a text/html file. Therefore we don't know yet
414 at that stage that -E is going to cause us to tack on
415 ".html", so we need to compare vs. the original URL plus
416 ".orig", not the original URL plus ".html.orig". */
417 filename_plus_orig_suffix = alloca (filename_len + 1);
418 strcpy(filename_plus_orig_suffix, file);
419 strcpy((filename_plus_orig_suffix + filename_len) - 4, "orig");
421 else /* downloaded_file_return == FILE_DOWNLOADED_NORMALLY */
423 /* Append ".orig" to the name. */
424 filename_plus_orig_suffix = alloca (filename_len + sizeof(".orig"));
425 strcpy(filename_plus_orig_suffix, file);
426 strcpy(filename_plus_orig_suffix + filename_len, ".orig");
429 /* We can get called twice on the same URL thanks to the
430 convert_all_links() call in main(). If we write the .orig file
431 each time in such a case, it'll end up containing the first-pass
432 conversion, not the original file. So, see if we've already been
433 called on this file. */
434 converted_file_ptr = converted_files;
435 while (converted_file_ptr != NULL)
436 if (strcmp(converted_file_ptr->string, file) == 0)
438 already_wrote_backup_file = TRUE;
442 converted_file_ptr = converted_file_ptr->next;
444 if (!already_wrote_backup_file)
446 /* Rename <file> to <file>.orig before former gets written over. */
447 if (rename(file, filename_plus_orig_suffix) != 0)
448 logprintf (LOG_NOTQUIET, _("Cannot back up %s as %s: %s\n"),
449 file, filename_plus_orig_suffix, strerror (errno));
451 /* Remember that we've already written a .orig backup for this file.
452 Note that we never free this memory since we need it till the
453 convert_all_links() call, which is one of the last things the
454 program does before terminating. BTW, I'm not sure if it would be
455 safe to just set 'converted_file_ptr->string' to 'file' below,
456 rather than making a copy of the string... Another note is that I
457 thought I could just add a field to the urlpos structure saying
458 that we'd written a .orig file for this URL, but that didn't work,
459 so I had to make this separate list.
460 -- Dan Harkless <wget@harkless.org>
462 This [adding a field to the urlpos structure] didn't work
463 because convert_file() is called from convert_all_links at
464 the end of the retrieval with a freshly built new urlpos
466 -- Hrvoje Niksic <hniksic@xemacs.org>
468 converted_file_ptr = xmalloc (sizeof (*converted_file_ptr));
469 converted_file_ptr->string = xstrdup (file);
470 converted_file_ptr->next = converted_files;
471 converted_files = converted_file_ptr;
475 static int find_fragment PARAMS ((const char *, int, const char **,
478 /* Replace an attribute's original text with NEW_TEXT. */
481 replace_attr (const char *p, int size, FILE *fp, const char *new_text)
484 char quote_char = '\"'; /* use "..." for quoting, unless the
485 original value is quoted, in which
486 case reuse its quoting char. */
487 const char *frag_beg, *frag_end;
489 /* Structure of our string is:
491 <--- size ---> (with quotes)
494 <--- size --> (no quotes) */
496 if (*p == '\"' || *p == '\'')
501 size -= 2; /* disregard opening and closing quote */
503 putc (quote_char, fp);
504 fputs (new_text, fp);
506 /* Look for fragment identifier, if any. */
507 if (find_fragment (p, size, &frag_beg, &frag_end))
508 fwrite (frag_beg, 1, frag_end - frag_beg, fp);
512 putc (quote_char, fp);
517 /* The same as REPLACE_ATTR, but used when replacing
518 <meta http-equiv=refresh content="new_text"> because we need to
519 append "timeout_value; URL=" before the next_text. */
522 replace_attr_refresh_hack (const char *p, int size, FILE *fp,
523 const char *new_text, int timeout)
526 char *new_with_timeout = (char *)alloca (numdigit (timeout)
530 sprintf (new_with_timeout, "%d; URL=%s", timeout, new_text);
532 return replace_attr (p, size, fp, new_with_timeout);
535 /* Find the first occurrence of '#' in [BEG, BEG+SIZE) that is not
536 preceded by '&'. If the character is not found, return zero. If
537 the character is found, return 1 and set BP and EP to point to the
538 beginning and end of the region.
540 This is used for finding the fragment indentifiers in URLs. */
543 find_fragment (const char *beg, int size, const char **bp, const char **ep)
545 const char *end = beg + size;
547 for (; beg < end; beg++)
569 /* Quote FILE for use as local reference to an HTML file.
571 We quote ? as %3F to avoid passing part of the file name as the
572 parameter when browsing the converted file through HTTP. However,
573 it is safe to do this only when `--html-extension' is turned on.
574 This is because converting "index.html?foo=bar" to
575 "index.html%3Ffoo=bar" would break local browsing, as the latter
576 isn't even recognized as an HTML file! However, converting
577 "index.html?foo=bar.html" to "index.html%3Ffoo=bar.html" should be
578 safe for both local and HTTP-served browsing. */
581 local_quote_string (const char *file)
583 const char *file_sans_qmark;
586 if (!opt.html_extension)
587 return html_quote_string (file);
589 qm = count_char (file, '?');
593 const char *from = file;
596 /* qm * 2 because we replace each question mark with "%3F",
597 i.e. replace one char with three, hence two more. */
598 int fsqlen = strlen (file) + qm * 2;
600 to = newname = (char *)alloca (fsqlen + 1);
601 for (; *from; from++)
612 assert (to - newname == fsqlen);
615 file_sans_qmark = newname;
618 file_sans_qmark = file;
620 return html_quote_string (file_sans_qmark);
623 /* Book-keeping code for dl_file_url_map, dl_url_file_map,
624 downloaded_html_list, and downloaded_html_set. Other code calls
625 these functions to let us know that a file has been downloaded. */
627 #define ENSURE_TABLES_EXIST do { \
628 if (!dl_file_url_map) \
629 dl_file_url_map = make_string_hash_table (0); \
630 if (!dl_url_file_map) \
631 dl_url_file_map = make_string_hash_table (0); \
634 /* Return 1 if S1 and S2 are the same, except for "/index.html". The
635 three cases in which it returns one are (substitute any substring
638 m("foo/index.html", "foo/") ==> 1
639 m("foo/", "foo/index.html") ==> 1
640 m("foo", "foo/index.html") ==> 1
641 m("foo", "foo/" ==> 1
642 m("foo", "foo") ==> 1 */
645 match_except_index (const char *s1, const char *s2)
650 /* Skip common substring. */
651 for (i = 0; *s1 && *s2 && *s1 == *s2; s1++, s2++, i++)
654 /* Strings differ at the very beginning -- bail out. We need to
655 check this explicitly to avoid `lng - 1' reading outside the
660 /* Both strings hit EOF -- strings are equal. */
663 /* Strings are randomly different, e.g. "/foo/bar" and "/foo/qux". */
666 /* S1 is the longer one. */
669 /* S2 is the longer one. */
673 /* foo/index.html */ /* or */ /* foo/index.html */
677 /* The right-hand case. */
680 if (*lng == '/' && *(lng + 1) == '\0')
685 return 0 == strcmp (lng, "/index.html");
689 dissociate_urls_from_file_mapper (void *key, void *value, void *arg)
691 char *mapping_url = (char *)key;
692 char *mapping_file = (char *)value;
693 char *file = (char *)arg;
695 if (0 == strcmp (mapping_file, file))
697 hash_table_remove (dl_url_file_map, mapping_url);
699 xfree (mapping_file);
702 /* Continue mapping. */
706 /* Remove all associations from various URLs to FILE from dl_url_file_map. */
709 dissociate_urls_from_file (const char *file)
711 hash_table_map (dl_url_file_map, dissociate_urls_from_file_mapper,
715 /* Register that URL has been successfully downloaded to FILE. This
716 is used by the link conversion code to convert references to URLs
717 to references to local files. It is also being used to check if a
718 URL has already been downloaded. */
721 register_download (const char *url, const char *file)
723 char *old_file, *old_url;
727 /* With some forms of retrieval, it is possible, although not likely
728 or particularly desirable. If both are downloaded, the second
729 download will override the first one. When that happens,
730 dissociate the old file name from the URL. */
732 if (hash_table_get_pair (dl_file_url_map, file, &old_file, &old_url))
734 if (0 == strcmp (url, old_url))
735 /* We have somehow managed to download the same URL twice.
739 if (match_except_index (url, old_url)
740 && !hash_table_contains (dl_url_file_map, url))
741 /* The two URLs differ only in the "index.html" ending. For
742 example, one is "http://www.server.com/", and the other is
743 "http://www.server.com/index.html". Don't remove the old
744 one, just add the new one as a non-canonical entry. */
747 hash_table_remove (dl_file_url_map, file);
751 /* Remove all the URLs that point to this file. Yes, there can
752 be more than one such URL, because we store redirections as
753 multiple entries in dl_url_file_map. For example, if URL1
754 redirects to URL2 which gets downloaded to FILE, we map both
755 URL1 and URL2 to FILE in dl_url_file_map. (dl_file_url_map
756 only points to URL2.) When another URL gets loaded to FILE,
757 we want both URL1 and URL2 dissociated from it.
759 This is a relatively expensive operation because it performs
760 a linear search of the whole hash table, but it should be
761 called very rarely, only when two URLs resolve to the same
762 file name, *and* the "<file>.1" extensions are turned off.
763 In other words, almost never. */
764 dissociate_urls_from_file (file);
767 hash_table_put (dl_file_url_map, xstrdup (file), xstrdup (url));
770 /* A URL->FILE mapping is not possible without a FILE->URL mapping.
771 If the latter were present, it should have been removed by the
772 above `if'. So we could write:
774 assert (!hash_table_contains (dl_url_file_map, url));
776 The above is correct when running in recursive mode where the
777 same URL always resolves to the same file. But if you do
782 then the first URL will resolve to "FILE", and the other to
783 "FILE.1". In that case, FILE.1 will not be found in
784 dl_file_url_map, but URL will still point to FILE in
786 if (hash_table_get_pair (dl_url_file_map, url, &old_url, &old_file))
788 hash_table_remove (dl_url_file_map, url);
793 hash_table_put (dl_url_file_map, xstrdup (url), xstrdup (file));
796 /* Register that FROM has been redirected to TO. This assumes that TO
797 is successfully downloaded and already registered using
798 register_download() above. */
801 register_redirection (const char *from, const char *to)
807 file = hash_table_get (dl_url_file_map, to);
808 assert (file != NULL);
809 if (!hash_table_contains (dl_url_file_map, from))
810 hash_table_put (dl_url_file_map, xstrdup (from), xstrdup (file));
813 /* Register that the file has been deleted. */
816 register_delete_file (const char *file)
818 char *old_url, *old_file;
822 if (!hash_table_get_pair (dl_file_url_map, file, &old_file, &old_url))
825 hash_table_remove (dl_file_url_map, file);
828 dissociate_urls_from_file (file);
831 /* Register that FILE is an HTML file that has been downloaded. */
834 register_html (const char *url, const char *file)
836 if (!downloaded_html_set)
837 downloaded_html_set = make_string_hash_table (0);
838 else if (hash_table_contains (downloaded_html_set, file))
841 /* The set and the list should use the same copy of FILE, but the
842 slist interface insists on strduping the string it gets. Oh
844 string_set_add (downloaded_html_set, file);
845 downloaded_html_list = slist_prepend (downloaded_html_list, file);
848 /* Cleanup the data structures associated with recursive retrieving
849 (the variables above). */
851 convert_cleanup (void)
855 free_keys_and_values (dl_file_url_map);
856 hash_table_destroy (dl_file_url_map);
857 dl_file_url_map = NULL;
861 free_keys_and_values (dl_url_file_map);
862 hash_table_destroy (dl_url_file_map);
863 dl_url_file_map = NULL;
865 if (downloaded_html_set)
866 string_set_free (downloaded_html_set);
867 slist_free (downloaded_html_list);
868 downloaded_html_list = NULL;
871 /* Book-keeping code for downloaded files that enables extension
874 /* This table should really be merged with dl_file_url_map and
875 downloaded_html_files. This was originally a list, but I changed
876 it to a hash table beause it was actually taking a lot of time to
877 find things in it. */
879 static struct hash_table *downloaded_files_hash;
881 /* We're storing "modes" of type downloaded_file_t in the hash table.
882 However, our hash tables only accept pointers for keys and values.
883 So when we need a pointer, we use the address of a
884 downloaded_file_t variable of static storage. */
886 static downloaded_file_t *
887 downloaded_mode_to_ptr (downloaded_file_t mode)
889 static downloaded_file_t
890 v1 = FILE_NOT_ALREADY_DOWNLOADED,
891 v2 = FILE_DOWNLOADED_NORMALLY,
892 v3 = FILE_DOWNLOADED_AND_HTML_EXTENSION_ADDED,
897 case FILE_NOT_ALREADY_DOWNLOADED:
899 case FILE_DOWNLOADED_NORMALLY:
901 case FILE_DOWNLOADED_AND_HTML_EXTENSION_ADDED:
909 /* Remembers which files have been downloaded. In the standard case,
910 should be called with mode == FILE_DOWNLOADED_NORMALLY for each
911 file we actually download successfully (i.e. not for ones we have
912 failures on or that we skip due to -N).
914 When we've downloaded a file and tacked on a ".html" extension due
915 to -E, call this function with
916 FILE_DOWNLOADED_AND_HTML_EXTENSION_ADDED rather than
917 FILE_DOWNLOADED_NORMALLY.
919 If you just want to check if a file has been previously added
920 without adding it, call with mode == CHECK_FOR_FILE. Please be
921 sure to call this function with local filenames, not remote
925 downloaded_file (downloaded_file_t mode, const char *file)
927 downloaded_file_t *ptr;
929 if (mode == CHECK_FOR_FILE)
931 if (!downloaded_files_hash)
932 return FILE_NOT_ALREADY_DOWNLOADED;
933 ptr = hash_table_get (downloaded_files_hash, file);
935 return FILE_NOT_ALREADY_DOWNLOADED;
939 if (!downloaded_files_hash)
940 downloaded_files_hash = make_string_hash_table (0);
942 ptr = hash_table_get (downloaded_files_hash, file);
946 ptr = downloaded_mode_to_ptr (mode);
947 hash_table_put (downloaded_files_hash, xstrdup (file), &ptr);
949 return FILE_NOT_ALREADY_DOWNLOADED;
953 df_free_mapper (void *key, void *value, void *ignored)
960 downloaded_files_free (void)
962 if (downloaded_files_hash)
964 hash_table_map (downloaded_files_hash, df_free_mapper, NULL);
965 hash_table_destroy (downloaded_files_hash);
966 downloaded_files_hash = NULL;