Automated merge.

[wget] / src / recur.c

diff --git a/src/recur.c b/src/recur.c
index e2f58d1cb6cffc3bbd5027c345ca6a0f6a9f918d..95581486b28e5e06a9d43ff023e73b7cf3e4a8ba 100644 (file)
--- a/src/recur.c
+++ b/src/recur.c
@@ -51,7 +51,6 @@ as that of the covered work.  */
 #include "html-url.h"
 #include "css-url.h"
 #include "spider.h"
-#include "iri.h"
 \f
 /* Functions for maintaining the URL queue.  */
 
@@ -116,7 +115,8 @@ url_enqueue (struct url_queue *queue, struct iri *i,
   DEBUGP (("Queue count %d, maxcount %d.\n", queue->count, queue->maxcount));
 
   if (i)
-    printf ("[Enqueuing %s with %s\n", url, i->uri_encoding);
+    DEBUGP (("[IRI Enqueuing %s with %s\n", quote_n (0, url),
+             i->uri_encoding ? quote_n (1, i->uri_encoding) : "None"));
 
   if (queue->tail)
     queue->tail->next = qel;
@@ -187,7 +187,7 @@ static bool descend_redirect_p (const char *, const char *, int,
           options, add it to the queue. */
 
 uerr_t
-retrieve_tree (const char *start_url)
+retrieve_tree (const char *start_url, struct iri *pi)
 {
   uerr_t status = RETROK;
 
@@ -201,13 +201,25 @@ retrieve_tree (const char *start_url)
   int up_error_code;
   struct url *start_url_parsed;
   struct iri *i = iri_new ();
-  set_uri_encoding (i, opt.locale);
 
-  start_url_parsed = url_parse (start_url, &up_error_code, i);
+#define COPYSTR(x)  (x) ? xstrdup(x) : NULL;
+  /* Duplicate pi struct if not NULL */
+  if (pi)
+    {
+      i->uri_encoding = COPYSTR (pi->uri_encoding);
+      i->content_encoding = COPYSTR (pi->content_encoding);
+      i->utf8_encode = pi->utf8_encode;
+    }
+  else
+    set_uri_encoding (i, opt.locale, true);
+#undef COPYSTR
+
+  start_url_parsed = url_parse (start_url, &up_error_code, i, true);
   if (!start_url_parsed)
     {
-      logprintf (LOG_NOTQUIET, "%s: %s.\n", start_url,
-                 url_error (up_error_code));
+      char *error = url_error (start_url, up_error_code);
+      logprintf (LOG_NOTQUIET, "%s: %s.\n", start_url, error);
+      xfree (error);
       return URLERROR;
     }
 
@@ -369,7 +381,7 @@ retrieve_tree (const char *start_url)
           if (children)
             {
               struct urlpos *child = children;
-              struct url *url_parsed = url_parse (url, NULL, i);
+              struct url *url_parsed = url_parse (url, NULL, i, false);
               struct iri *ci;
               char *referer_url = url;
               bool strip_auth = (url_parsed != NULL
@@ -390,7 +402,7 @@ retrieve_tree (const char *start_url)
                                         blacklist, i))
                     {
                       ci = iri_new ();
-                      set_uri_encoding (ci, i->content_encoding);
+                      set_uri_encoding (ci, i->content_encoding, false);
                       url_enqueue (queue, ci, xstrdup (child->url->url),
                                    xstrdup (referer_url), depth + 1,
                                    child->link_expect_html,
@@ -682,10 +694,10 @@ descend_redirect_p (const char *redirected, const char *original, int depth,
   struct urlpos *upos;
   bool success;
 
-  orig_parsed = url_parse (original, NULL, NULL);
+  orig_parsed = url_parse (original, NULL, NULL, false);
   assert (orig_parsed != NULL);
 
-  new_parsed = url_parse (redirected, NULL, NULL);
+  new_parsed = url_parse (redirected, NULL, NULL, false);
   assert (new_parsed != NULL);
 
   upos = xnew0 (struct urlpos);