Automated merge.

[wget] / src / retr.c

diff --git a/src/retr.c b/src/retr.c
index 021060819d3aa8e36b44c539d0aee46908a1ad3a..7a28ea32e76e825d452417d104fb538a178e79a8 100644 (file)
--- a/src/retr.c
+++ b/src/retr.c
@@ -52,6 +52,7 @@ as that of the covered work.  */
 #include "convert.h"
 #include "ptimer.h"
 #include "iri.h"
+#include "html-url.h"
 
 /* Total size of downloaded files.  Used to enforce quota.  */
 SUM_SIZE_INT total_downloaded_bytes;
@@ -613,8 +614,6 @@ retrieve_url (const char *origurl, char **file, char **newloc,
   char *saved_post_data = NULL;
   char *saved_post_file_name = NULL;
 
-  bool utf8_encoded = opt.enable_iri;
-
   /* If dt is NULL, use local storage.  */
   if (!dt)
     {
@@ -627,8 +626,10 @@ retrieve_url (const char *origurl, char **file, char **newloc,
   if (file)
     *file = NULL;
 
+  reset_utf8_encode ();
+
  second_try:
-  u = url_parse (url, &up_error_code, &utf8_encoded);
+  u = url_parse (url, &up_error_code);
   if (!u)
     {
       logprintf (LOG_NOTQUIET, "%s: %s.\n", url, url_error (up_error_code));
@@ -652,9 +653,10 @@ retrieve_url (const char *origurl, char **file, char **newloc,
   if (proxy)
     {
       /* sXXXav : support IRI for proxy */
-      bool proxy_utf8_encode = false;
       /* Parse the proxy URL.  */
-      proxy_url = url_parse (proxy, &up_error_code, &proxy_utf8_encode);
+      set_ugly_no_encode (true);
+      proxy_url = url_parse (proxy, &up_error_code);
+      set_ugly_no_encode (false);
       if (!proxy_url)
         {
           logprintf (LOG_NOTQUIET, _("Error parsing proxy URL %s: %s.\n"),
@@ -729,10 +731,10 @@ retrieve_url (const char *origurl, char **file, char **newloc,
       xfree (mynewloc);
       mynewloc = construced_newloc;
 
-      utf8_encoded = opt.enable_iri;
+      reset_utf8_encode ();
 
       /* Now, see if this new location makes sense. */
-      newloc_parsed = url_parse (mynewloc, &up_error_code, &utf8_encoded);
+      newloc_parsed = url_parse (mynewloc, &up_error_code);
       if (!newloc_parsed)
         {
           logprintf (LOG_NOTQUIET, "%s: %s.\n", escnonprint_uri (mynewloc),
@@ -780,9 +782,9 @@ retrieve_url (const char *origurl, char **file, char **newloc,
     }
 
   /* Try to not encode in UTF-8 if fetching failed */
-  if (!(*dt & RETROKF) && utf8_encoded)
+  if (!(*dt & RETROKF) && get_utf8_encode ())
     {
-      utf8_encoded = false;
+      set_utf8_encode (false);
       /*printf ("[Fallbacking to non-utf8 for `%s'\n", url);*/
       goto second_try;
     }
@@ -794,6 +796,16 @@ retrieve_url (const char *origurl, char **file, char **newloc,
         register_redirection (origurl, u->url);
       if (*dt & TEXTHTML)
         register_html (u->url, local_file);
+      if (*dt & RETROKF)
+        {
+          register_download (u->url, local_file);
+          if (redirection_count && 0 != strcmp (origurl, u->url))
+            register_redirection (origurl, u->url);
+          if (*dt & TEXTHTML)
+            register_html (u->url, local_file);
+          if (*dt & TEXTCSS)
+            register_css (u->url, local_file);
+        }
     }
 
   if (file)
@@ -834,10 +846,24 @@ retrieve_from_file (const char *file, bool html, int *count)
   uerr_t status;
   struct urlpos *url_list, *cur_url;
 
-  url_list = (html ? get_urls_html (file, NULL, NULL)
-              : get_urls_file (file));
+  char *input_file = NULL;
+  const char *url = file;
+
   status = RETROK;             /* Suppose everything is OK.  */
   *count = 0;                  /* Reset the URL count.  */
+  
+  if (url_has_scheme (url))
+    {
+      uerr_t status;
+      status = retrieve_url (url, &input_file, NULL, NULL, NULL, false);
+      if (status != RETROK)
+        return status;
+    }
+  else
+    input_file = (char *) file;
+
+  url_list = (html ? get_urls_html (input_file, NULL, NULL)
+              : get_urls_file (input_file));
 
   for (cur_url = url_list; cur_url; cur_url = cur_url->next, ++*count)
     {
@@ -1036,8 +1062,11 @@ getproxy (struct url *u)
 bool
 url_uses_proxy (const char *url)
 {
-  bool ret, utf8_encode = false;
-  struct url *u = url_parse (url, NULL, &utf8_encode);
+  bool ret;
+  struct url *u;
+  set_ugly_no_encode(true);
+  u= url_parse (url, NULL);
+  set_ugly_no_encode(false);
   if (!u)
     return false;
   ret = getproxy (u) != NULL;