Fixes to restore Test-iri and Test-iri-forced-remote

[wget] / src / retr.c

diff --git a/src/retr.c b/src/retr.c
index e3d62978f0b1ee4ea38adb1e36358794f4f2f82b..2a9dec7bbb658aa6bdb12365813c0121142e7cdb 100644 (file)
--- 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);
     }
 
@@ -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, struct iri *iri)
+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;
@@ -626,16 +628,6 @@ retrieve_url (const char *origurl, char **file, char **newloc,
     *file = NULL;
 
  second_try:
-  u = url_parse (url, &up_error_code, iri);
-  if (!u)
-    {
-      char *error = url_error (url, up_error_code);
-      logprintf (LOG_NOTQUIET, "%s: %s.\n", url, error);
-      xfree (url);
-      xfree (error);
-      return URLERROR;
-    }
-
   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));
@@ -658,7 +650,7 @@ retrieve_url (const char *origurl, char **file, char **newloc,
       pi->utf8_encode = false;
 
       /* Parse the proxy URL.  */
-      proxy_url = url_parse (proxy, &up_error_code, NULL);
+      proxy_url = url_parse (proxy, &up_error_code, NULL, true);
       if (!proxy_url)
         {
           char *error = url_error (proxy, up_error_code);
@@ -739,15 +731,19 @@ retrieve_url (const char *origurl, char **file, char **newloc,
          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, iri);
+      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),
                      error);
-          url_free (u);
+          if (orig_parsed != u)
+            {
+              url_free (u);
+            }
           xfree (url);
           xfree (mynewloc);
           xfree (error);
@@ -767,7 +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;
@@ -776,7 +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
@@ -794,8 +796,18 @@ retrieve_url (const char *origurl, char **file, char **newloc,
   if (!(*dt & RETROKF) && iri->utf8_encode)
     {
       iri->utf8_encode = false;
-      DEBUGP (("[IRI Fallbacking to non-utf8 for %s\n", quote (url)));
-      goto second_try;
+      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)
@@ -822,7 +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)
     {
@@ -868,13 +883,23 @@ retrieve_from_file (const char *file, bool html, int *count)
 
   if (url_has_scheme (url))
     {
-      int dt;
+      int dt,url_err;
       uerr_t status;
+      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, &input_file, NULL, NULL, &dt, false, iri);
+      status = retrieve_url (url_parsed, url, &input_file, NULL, NULL, &dt,
+                             false, iri);
       if (status != RETROK)
         return status;
 
@@ -907,6 +932,8 @@ retrieve_from_file (const char *file, bool html, int *count)
 
       /* 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)))
@@ -917,13 +944,13 @@ retrieve_from_file (const char *file, bool html, int *count)
           if (cur_url->url->scheme == SCHEME_FTP)
             opt.follow_ftp = 1;
 
-          status = retrieve_tree (cur_url->url->url, iri);
+          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, iri);
+        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))
         {
@@ -1093,18 +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;
-  struct iri *i = iri_new();
-  /* url was given in the command line, so use locale as encoding */
-  set_uri_encoding (i, opt.locale, true);
-  u= url_parse (url, NULL, i);
   if (!u)
     return false;
   ret = getproxy (u) != NULL;
-  url_free (u);
   return ret;
 }
 
@@ -1117,3 +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;
+}