Updated config.guess, config.sub, install.sh.

[wget] / src / html-url.c

diff --git a/src/html-url.c b/src/html-url.c
index ef93a7e49c5c61ded555c2b6c022be3346bbec2c..218659d454d4a23b34c0288f1b848585a14b1e4b 100644 (file)
--- a/src/html-url.c
+++ b/src/html-url.c
@@ -44,7 +44,6 @@ as that of the covered work.  */
 #include "recur.h"
 #include "html-url.h"
 #include "css-url.h"
-#include "iri.h"
 
 typedef void (*tag_handler_t) (int, struct taginfo *, struct map_context *);
 
@@ -175,6 +174,10 @@ static const char *additional_attributes[] = {
 static struct hash_table *interesting_tags;
 static struct hash_table *interesting_attributes;
 
+/* Will contains the (last) charset found in 'http-equiv=content-type'
+   meta tags  */
+static char *meta_charset;
+
 static void
 init_interesting (void)
 {
@@ -285,9 +288,7 @@ append_url (const char *link_uri, int position, int size,
           return NULL;
         }
 
-      set_ugly_no_encode (true);
-      url = url_parse (link_uri, NULL);
-      set_ugly_no_encode (false);
+      url = url_parse (link_uri, NULL, NULL, false);
       if (!url)
         {
           DEBUGP (("%s: link \"%s\" doesn't parse.\n",
@@ -303,12 +304,13 @@ append_url (const char *link_uri, int position, int size,
 
       char *complete_uri = uri_merge (base, link_uri);
 
-      DEBUGP (("%s: merge(\"%s\", \"%s\") -> %s\n",
-               ctx->document_file, base, link_uri, complete_uri));
+      DEBUGP (("%s: merge(%s, %s) -> %s\n",
+               quotearg_n_style (0, escape_quoting_style, ctx->document_file),
+               quote_n (1, base),
+               quote_n (2, link_uri),
+               quotearg_n_style (3, escape_quoting_style, complete_uri)));
 
-      set_ugly_no_encode (true);
-      url = url_parse (complete_uri, NULL);
-      set_ugly_no_encode (false);
+      url = url_parse (complete_uri, NULL, NULL, false);
       if (!url)
         {
           DEBUGP (("%s: merged link \"%s\" doesn't parse.\n",
@@ -319,7 +321,7 @@ append_url (const char *link_uri, int position, int size,
       xfree (complete_uri);
     }
 
-  DEBUGP (("appending \"%s\" to urlpos.\n", url->url));
+  DEBUGP (("appending %s to urlpos.\n", quote (url->url)));
 
   newel = xnew0 (struct urlpos);
   newel->url = url;
@@ -572,10 +574,8 @@ tag_handle_meta (int tagid, struct taginfo *tag, struct map_context *ctx)
       if (!mcharset)
         return;
 
-      /*logprintf (LOG_VERBOSE, "Meta tag charset : %s\n", quote (mcharset));*/
-
-      set_current_charset (mcharset);
-      xfree (mcharset);
+      xfree_null (meta_charset);
+      meta_charset = mcharset;
     }
   else if (name && 0 == strcasecmp (name, "robots"))
     {
@@ -590,15 +590,25 @@ tag_handle_meta (int tagid, struct taginfo *tag, struct map_context *ctx)
         {
           while (*content)
             {
-              /* Find the next occurrence of ',' or the end of
-                 the string.  */
-              char *end = strchr (content, ',');
-              if (end)
-                ++end;
-              else
-                end = content + strlen (content);
+              char *end;
+              /* Skip any initial whitespace. */
+              content += strspn (content, " \f\n\r\t\v");
+              /* Find the next occurrence of ',' or whitespace,
+               * or the end of the string.  */
+              end = content + strcspn (content, ", \f\n\r\t\v");
               if (!strncasecmp (content, "nofollow", end - content))
                 ctx->nofollow = true;
+              /* Skip past the next comma, if any. */
+              if (*end == ',')
+                ++end;
+              else
+                {
+                  end = strchr (end, ',');
+                  if (end)
+                    ++end;
+                  else
+                    end = content + strlen (content);
+                }
               content = end;
             }
         }
@@ -641,7 +651,8 @@ collect_tags_mapper (struct taginfo *tag, void *arg)
    <base href=...> and does the right thing.  */
 
 struct urlpos *
-get_urls_html (const char *file, const char *url, bool *meta_disallow_follow)
+get_urls_html (const char *file, const char *url, bool *meta_disallow_follow,
+               struct iri *iri)
 {
   struct file_memory *fm;
   struct map_context ctx;
@@ -681,6 +692,10 @@ get_urls_html (const char *file, const char *url, bool *meta_disallow_follow)
   map_html_tags (fm->content, fm->length, collect_tags_mapper, &ctx, flags,
                  NULL, interesting_attributes);
 
+  /* If meta charset isn't null, override content encoding */
+  if (iri && meta_charset)
+    set_content_encoding (iri, meta_charset);
+
   DEBUGP (("no-follow in %s: %d\n", file, ctx.nofollow));
   if (meta_disallow_follow)
     *meta_disallow_follow = ctx.nofollow;
@@ -750,14 +765,14 @@ get_urls_file (const char *file)
           url_text = merged;
         }
 
-      set_ugly_no_encode (true);
-      url = url_parse (url_text, &up_error_code);
-      set_ugly_no_encode (false);
+      url = url_parse (url_text, &up_error_code, NULL, false);
       if (!url)
         {
+          char *error = url_error (url_text, up_error_code);
           logprintf (LOG_NOTQUIET, _("%s: Invalid URL %s: %s\n"),
-                     file, url_text, url_error (up_error_code));
+                     file, url_text, error);
           xfree (url_text);
+          xfree (error);
           continue;
         }
       xfree (url_text);