summaryrefslogtreecommitdiffabout
Unidiff
Diffstat (more/less context) (ignore whitespace changes)
-rw-r--r--include/opkele/consumer.h11
-rw-r--r--lib/consumer.cc26
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
@@ -129,11 +129,18 @@ namespace opkele {
129 void check_authentication(const string& server,const params_t& p); 129 void check_authentication(const string& server,const params_t& p);
130 130
131 /** 131 /**
132 * make URL canonical, by adding http:// and trailing slash, if needed. 132 * normalize URL by adding http:// and trailing slash if needed.
133 * @param url
134 * @return normalized url
135 */
136 static string normalize(const string& url);
137
138 /**
139 * Canonicalize URL, by normalizing its appearance and following redirects.
133 * @param url 140 * @param url
134 * @return canonicalized url 141 * @return canonicalized url
135 */ 142 */
136 static string canonicalize(const string& url); 143 virtual string canonicalize(const string& url);
137 144
138 }; 145 };
139 146
diff --git a/lib/consumer.cc b/lib/consumer.cc
index 282f0cc..dd8e150 100644
--- a/lib/consumer.cc
+++ b/lib/consumer.cc
@@ -310,7 +310,7 @@ namespace opkele {
310 throw failed_lookup(OPKELE_CP_ "no find_assoc() provided"); 310 throw failed_lookup(OPKELE_CP_ "no find_assoc() provided");
311 } 311 }
312 312
313 string consumer_t::canonicalize(const string& url) { 313 string consumer_t::normalize(const string& url) {
314 string rv = url; 314 string rv = url;
315 // strip leading and trailing spaces 315 // strip leading and trailing spaces
316 string::size_type i = rv.find_first_not_of(" \t\r\n"); 316 string::size_type i = rv.find_first_not_of(" \t\r\n");
@@ -342,4 +342,28 @@ namespace opkele {
342 return rv; 342 return rv;
343 } 343 }
344 344
345 string consumer_t::canonicalize(const string& url) {
346 string rv = normalize(url);
347 curl_t curl = curl_easy_init();
348 if(!curl)
349 throw exception_curl(OPKELE_CP_ "failed to curl_easy_init()");
350 string html;
351 CURLcode r;
352 (r=curl_misc_sets(curl))
353 || (r=curl_easy_setopt(curl,CURLOPT_URL,rv.c_str()))
354 || (r=curl_easy_setopt(curl,CURLOPT_NOBODY,1))
355 ;
356 if(r)
357 throw exception_curl(OPKELE_CP_ "failed to curl_easy_setopt()",r);
358 r = curl_easy_perform(curl);
359 if(r)
360 throw exception_curl(OPKELE_CP_ "failed to curl_easy_perform()",r);
361 const char *eu = 0;
362 r = curl_easy_getinfo(curl,CURLINFO_EFFECTIVE_URL,&eu);
363 if(r)
364 throw exception_curl(OPKELE_CP_ "failed to curl_easy_getinfo(..CURLINFO_EFFECTIVE_URL..)",r);
365 rv = eu;
366 return normalize(rv);
367 }
368
345} 369}