X-Git-Url: http://sjero.net/git/?p=wget;a=blobdiff_plain;f=src%2Fretr.c;h=2a9dec7bbb658aa6bdb12365813c0121142e7cdb;hp=58e00d2fe74102909f78cfdaa57ae5cc7caa304e;hb=50e12521d63b8b42370a07d9c9971f56ca6dc58a;hpb=b28a6abfe66e03dae1f749d8215f4ba2b7303e5a

diff --git a/src/retr.c b/src/retr.c
index 58e00d2f..2a9dec7b 100644
--- a/src/retr.c
+++ b/src/retr.c
@@ -226,7 +226,8 @@ fd_read_body (int fd, FILE *out, wgint toread, wgint startpos,
       /* If we're skipping STARTPOS bytes, pass 0 as the INITIAL
          argument to progress_create because the indicator doesn't
          (yet) know about "skipping" data.  */
-      progress = progress_create (skip ? 0 : startpos, startpos + toread);
+      wgint start = skip ? 0 : startpos;
+      progress = progress_create (start, start + toread);
       progress_interactive = progress_interactive_p (progress);
     }
 
@@ -393,7 +394,7 @@ fd_read_hunk (int fd, hunk_terminator_t terminator, long sizehint, long maxsize)
   char *hunk = xmalloc (bufsize);
   int tail = 0;                 /* tail position in HUNK */
 
-  assert (maxsize >= bufsize);
+  assert (!maxsize || maxsize >= bufsize);
 
   while (1)
     {
@@ -596,15 +597,16 @@ static char *getproxy (struct url *);
    multiple points. */
 
 uerr_t
-retrieve_url (const char *origurl, char **file, char **newloc,
-              const char *refurl, int *dt, bool recursive)
+retrieve_url (struct url * orig_parsed, const char *origurl, char **file,
+              char **newloc, const char *refurl, int *dt, bool recursive,
+              struct iri *iri)
 {
   uerr_t result;
   char *url;
   bool location_changed;
   int dummy;
   char *mynewloc, *proxy;
-  struct url *u, *proxy_url;
+  struct url *u = orig_parsed, *proxy_url;
   int up_error_code;            /* url parse error code */
   char *local_file;
   int redirection_count = 0;
@@ -625,13 +627,10 @@ retrieve_url (const char *origurl, char **file, char **newloc,
   if (file)
     *file = NULL;
 
-  u = url_parse (url, &up_error_code);
-  if (!u)
-    {
-      logprintf (LOG_NOTQUIET, "%s: %s.\n", url, url_error (up_error_code));
-      xfree (url);
-      return URLERROR;
-    }
+ second_try:
+  DEBUGP (("[IRI Retrieving %s with %s (UTF-8=%d)\n", quote_n (0, url),
+           iri->uri_encoding ? quote_n (1, iri->uri_encoding) : "None",
+           iri->utf8_encode));
 
   if (!refurl)
     refurl = opt.referer;
@@ -646,13 +645,19 @@ retrieve_url (const char *origurl, char **file, char **newloc,
   proxy = getproxy (u);
   if (proxy)
     {
+      struct iri *pi = iri_new ();
+      set_uri_encoding (pi, opt.locale, true);
+      pi->utf8_encode = false;
+
       /* Parse the proxy URL.  */
-      proxy_url = url_parse (proxy, &up_error_code);
+      proxy_url = url_parse (proxy, &up_error_code, NULL, true);
       if (!proxy_url)
         {
+          char *error = url_error (proxy, up_error_code);
           logprintf (LOG_NOTQUIET, _("Error parsing proxy URL %s: %s.\n"),
-                     proxy, url_error (up_error_code));
+                     proxy, error);
           xfree (url);
+          xfree (error);
           RESTORE_POST_DATA;
           return PROXERR;
         }
@@ -672,7 +677,7 @@ retrieve_url (const char *origurl, char **file, char **newloc,
 #endif
       || (proxy_url && proxy_url->scheme == SCHEME_HTTP))
     {
-      result = http_loop (u, &mynewloc, &local_file, refurl, dt, proxy_url);
+      result = http_loop (u, &mynewloc, &local_file, refurl, dt, proxy_url, iri);
     }
   else if (u->scheme == SCHEME_FTP)
     {
@@ -722,15 +727,26 @@ retrieve_url (const char *origurl, char **file, char **newloc,
       xfree (mynewloc);
       mynewloc = construced_newloc;
 
+      /* Reset UTF-8 encoding state, keep the URI encoding and reset
+         the content encoding. */
+      iri->utf8_encode = opt.enable_iri;
+      set_content_encoding (iri, NULL);
+      xfree_null (iri->orig_url);
+
       /* Now, see if this new location makes sense. */
-      newloc_parsed = url_parse (mynewloc, &up_error_code);
+      newloc_parsed = url_parse (mynewloc, &up_error_code, iri, true);
       if (!newloc_parsed)
         {
+          char *error = url_error (mynewloc, up_error_code);
           logprintf (LOG_NOTQUIET, "%s: %s.\n", escnonprint_uri (mynewloc),
-                     url_error (up_error_code));
-          url_free (u);
+                     error);
+          if (orig_parsed != u)
+            {
+              url_free (u);
+            }
           xfree (url);
           xfree (mynewloc);
+          xfree (error);
           RESTORE_POST_DATA;
           return result;
         }
@@ -747,7 +763,10 @@ retrieve_url (const char *origurl, char **file, char **newloc,
           logprintf (LOG_NOTQUIET, _("%d redirections exceeded.\n"),
                      opt.max_redirect);
           url_free (newloc_parsed);
-          url_free (u);
+          if (orig_parsed != u)
+            {
+              url_free (u);
+            }
           xfree (url);
           xfree (mynewloc);
           RESTORE_POST_DATA;
@@ -756,7 +775,10 @@ retrieve_url (const char *origurl, char **file, char **newloc,
 
       xfree (url);
       url = mynewloc;
-      url_free (u);
+      if (orig_parsed != u)
+        {
+          url_free (u);
+        }
       u = newloc_parsed;
 
       /* If we're being redirected from POST, we don't want to POST
@@ -770,8 +792,31 @@ retrieve_url (const char *origurl, char **file, char **newloc,
       goto redirected;
     }
 
-  if (local_file)
+  /* Try to not encode in UTF-8 if fetching failed */
+  if (!(*dt & RETROKF) && iri->utf8_encode)
+    {
+      iri->utf8_encode = false;
+      if (orig_parsed != u)
+        {
+          url_free (u);
+        }
+      u = url_parse (origurl, NULL, iri, true);
+      if (u)
+        {
+          DEBUGP (("[IRI fallbacking to non-utf8 for %s\n", quote (url)));
+          goto second_try;
+        }
+      else
+          DEBUGP (("[Couldn't fallback to non-utf8 for %s\n", quote (url)));
+    }
+
+  if (local_file && *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 & RETROKF)
         {
           register_download (u->url, local_file);
@@ -789,7 +834,10 @@ retrieve_url (const char *origurl, char **file, char **newloc,
   else
     xfree_null (local_file);
 
-  url_free (u);
+  if (orig_parsed != u)
+    {
+      url_free (u);
+    }
 
   if (redirection_count)
     {
@@ -821,24 +869,51 @@ retrieve_from_file (const char *file, bool html, int *count)
 {
   uerr_t status;
   struct urlpos *url_list, *cur_url;
+  struct iri *iri = iri_new();
 
   char *input_file = NULL;
   const char *url = file;
 
   status = RETROK;             /* Suppose everything is OK.  */
   *count = 0;                  /* Reset the URL count.  */
-  
+
+  /* sXXXav : Assume filename and links in the file are in the locale */
+  set_uri_encoding (iri, opt.locale, true);
+  set_content_encoding (iri, opt.locale);
+
   if (url_has_scheme (url))
     {
+      int dt,url_err;
       uerr_t status;
-      status = retrieve_url (url, &input_file, NULL, NULL, NULL, false);
+      struct url * url_parsed = url_parse(url, &url_err, NULL, true);
+
+      if (!url_parsed)
+        {
+          char *error = url_error (url, url_err);
+          logprintf (LOG_NOTQUIET, "%s: %s.\n", url, error);
+          xfree (error);
+          return URLERROR;
+        }
+
+      if (!opt.base_href)
+        opt.base_href = xstrdup (url);
+
+      status = retrieve_url (url_parsed, url, &input_file, NULL, NULL, &dt,
+                             false, iri);
       if (status != RETROK)
         return status;
+
+      if (dt & TEXTHTML)
+        html = true;
+
+      /* If we have a found a content encoding, use it */
+      if (iri->content_encoding)
+	  set_uri_encoding (iri, iri->content_encoding, false);
     }
   else
     input_file = (char *) file;
 
-  url_list = (html ? get_urls_html (input_file, NULL, NULL)
+  url_list = (html ? get_urls_html (input_file, NULL, NULL, iri)
               : get_urls_file (input_file));
 
   for (cur_url = url_list; cur_url; cur_url = cur_url->next, ++*count)
@@ -854,21 +929,28 @@ retrieve_from_file (const char *file, bool html, int *count)
           status = QUOTEXC;
           break;
         }
+
+      /* Reset UTF-8 encode status */
+      iri->utf8_encode = opt.enable_iri;
+      xfree_null (iri->orig_url);
+      iri->orig_url = NULL;
+
       if ((opt.recursive || opt.page_requisites)
           && (cur_url->url->scheme != SCHEME_FTP || getproxy (cur_url->url)))
         {
           int old_follow_ftp = opt.follow_ftp;
 
           /* Turn opt.follow_ftp on in case of recursive FTP retrieval */
-          if (cur_url->url->scheme == SCHEME_FTP) 
+          if (cur_url->url->scheme == SCHEME_FTP)
             opt.follow_ftp = 1;
-          
-          status = retrieve_tree (cur_url->url->url);
+
+          status = retrieve_tree (cur_url->url, iri);
 
           opt.follow_ftp = old_follow_ftp;
         }
       else
-        status = retrieve_url (cur_url->url->url, &filename, &new_file, NULL, &dt, opt.recursive);
+        status = retrieve_url (cur_url->url, cur_url->url->url, &filename,
+                               &new_file, NULL, &dt, opt.recursive, iri);
 
       if (filename && opt.delete_after && file_exists_p (filename))
         {
@@ -887,6 +969,8 @@ Removing file due to --delete-after in retrieve_from_file():\n"));
   /* Free the linked list of URL-s.  */
   free_urlpos (url_list);
 
+  iri_free (iri);
+
   return status;
 }
 
@@ -1036,14 +1120,12 @@ getproxy (struct url *u)
 /* Returns true if URL would be downloaded through a proxy. */
 
 bool
-url_uses_proxy (const char *url)
+url_uses_proxy (struct url * u)
 {
   bool ret;
-  struct url *u = url_parse (url, NULL);
   if (!u)
     return false;
   ret = getproxy (u) != NULL;
-  url_free (u);
   return ret;
 }
 
@@ -1056,3 +1138,16 @@ no_proxy_match (const char *host, const char **no_proxy)
   else
     return sufmatch (no_proxy, host);
 }
+
+/* Set the file parameter to point to the local file string.  */
+void
+set_local_file (const char **file, const char *default_file)
+{
+  if (opt.output_document)
+    {
+      if (output_stream_regular)
+        *file = opt.output_document;
+    }
+  else
+    *file = default_file;
+}