conversion after Wget is done. */
struct hash_table *downloaded_html_set;
-static struct hash_table *nonexisting_urls_hash;
-
static void convert_links (const char *, struct urlpos *);
/* This function is called when the retrieval is done to convert the
}
static void downloaded_files_free (void);
-static void nonexisting_urls_free (void);
/* Cleanup the data structures associated with this file. */
if (downloaded_html_set)
string_set_free (downloaded_html_set);
downloaded_files_free ();
- nonexisting_urls_free ();
if (converted_files)
string_set_free (converted_files);
}
downloaded_files_hash = NULL;
}
}
-\f
-/* Remembers broken links. */
-
-struct broken_urls_list
-{
- char *url;
- struct broken_urls_list *next;
-};
-
-static bool
-in_list (const struct broken_urls_list *list, const char *url)
-{
- const struct broken_urls_list *ptr;
-
- for (ptr = list; ptr; ptr = ptr->next)
- {
- /* str[case]cmp is inadequate for URL comparison */
- if (are_urls_equal (url, ptr->url) == 0) return true;
- }
-
- return false;
-}
-
-void
-nonexisting_url (const char *url, const char *referrer)
-{
- struct broken_urls_list *list;
-
- /* Ignore robots.txt URLs */
- if (is_robots_txt_url (url))
- return;
-
- if (!nonexisting_urls_hash)
- nonexisting_urls_hash = make_string_hash_table (0);
-
- list = hash_table_get (nonexisting_urls_hash, url);
- if (!list)
- {
- list = (struct broken_urls_list *) xnew0 (struct broken_urls_list);
- list->url = referrer ? xstrdup (referrer) : NULL;
- hash_table_put (nonexisting_urls_hash, xstrdup (url), list);
- }
- else if (list && !in_list (list, referrer))
- {
- /* Append referrer at the end of the list */
- struct broken_urls_list *newnode;
-
- while (list->next) list = list->next;
-
- newnode = xnew0 (struct broken_urls_list);
- newnode->url = xstrdup (referrer);
- list->next = newnode;
- }
-}
-
-static void
-nonexisting_urls_free (void)
-{
- if (nonexisting_urls_hash)
- {
- hash_table_iterator iter;
- for (hash_table_iterate (nonexisting_urls_hash, &iter);
- hash_table_iter_next (&iter);
- )
- {
- xfree (iter.key);
- xfree (iter.value);
- }
- hash_table_destroy (nonexisting_urls_hash);
- nonexisting_urls_hash = NULL;
- }
-}
-
-void
-print_broken_links (void)
-{
- hash_table_iterator iter;
- int num_elems;
-
- if (!nonexisting_urls_hash)
- {
- logprintf (LOG_NOTQUIET, _("Found no broken links.\n\n"));
- return;
- }
-
- num_elems = hash_table_count (nonexisting_urls_hash);
- assert (num_elems > 0);
-
- if (num_elems > 1)
- {
- logprintf (LOG_NOTQUIET, _("Found %d broken links.\n\n"),
- num_elems);
- }
- else
- {
- logprintf (LOG_NOTQUIET, _("Found 1 broken link.\n\n"));
- }
-
- for (hash_table_iterate (nonexisting_urls_hash, &iter);
- hash_table_iter_next (&iter);
- )
- {
- struct broken_urls_list *list;
-
- logprintf (LOG_NOTQUIET, _("%s referred by:\n"), (const char *)iter.key);
-
- for (list = (struct broken_urls_list *) iter.value;
- list;
- list = list->next)
- {
- logprintf (LOG_NOTQUIET, _(" %s\n"), list->url);
- }
- }
- logputs (LOG_NOTQUIET, "\n");
-}
-
\f
/* The function returns the pointer to the malloc-ed quoted version of
string s. It will recognize and quote numeric and special graphic