make canonicalize follow redirects

 - renamed old consumer_t::canonicalize to consumer_t::normalize
 - made the real canonicalize virtual to allow caching layer

2 files changed, 34 insertions, 3 deletions

diff --git a/include/opkele/consumer.h b/include/opkele/consumer.h
index fdb6119..042e2d1 100644
--- a/include/opkele/consumer.h
+++ b/include/opkele/consumer.h
@@ -130,9 +130,16 @@ namespace opkele {
 
 	    /**
-	     * make URL canonical, by adding http:// and trailing slash, if needed.
+	     * normalize URL by adding http:// and trailing slash if needed.
+	     * @param url
+	     * @return normalized url
+	     */
+	    static string normalize(const string& url);
+
+	    /**
+	     * Canonicalize URL, by normalizing its appearance and following redirects.
 	     * @param url
 	     * @return canonicalized url
 	     */
-	    static string canonicalize(const string& url);
+	    virtual string canonicalize(const string& url);
 
     };
diff --git a/lib/consumer.cc b/lib/consumer.cc
index 282f0cc..dd8e150 100644
--- a/lib/consumer.cc
+++ b/lib/consumer.cc
@@ -311,5 +311,5 @@ namespace opkele {
     }
 
-    string consumer_t::canonicalize(const string& url) {
+    string consumer_t::normalize(const string& url) {
 	string rv = url;
 	// strip leading and trailing spaces
@@ -343,3 +343,27 @@ namespace opkele {
     }
 
+    string consumer_t::canonicalize(const string& url) {
+	string rv = normalize(url);
+	curl_t curl = curl_easy_init();
+	if(!curl)
+	    throw exception_curl(OPKELE_CP_ "failed to curl_easy_init()");
+	string html;
+	CURLcode r;
+	(r=curl_misc_sets(curl))
+	|| (r=curl_easy_setopt(curl,CURLOPT_URL,rv.c_str()))
+	|| (r=curl_easy_setopt(curl,CURLOPT_NOBODY,1))
+	;
+	if(r)
+	    throw exception_curl(OPKELE_CP_ "failed to curl_easy_setopt()",r);
+	r = curl_easy_perform(curl);
+	if(r)
+	    throw exception_curl(OPKELE_CP_ "failed to curl_easy_perform()",r);
+	const char *eu = 0;
+	r = curl_easy_getinfo(curl,CURLINFO_EFFECTIVE_URL,&eu);
+	if(r)
+	    throw exception_curl(OPKELE_CP_ "failed to curl_easy_getinfo(..CURLINFO_EFFECTIVE_URL..)",r);
+	rv = eu;
+	return normalize(rv);
+    }
+
 }