-rw-r--r-- | include/opkele/consumer.h | 11 | ||||
-rw-r--r-- | lib/consumer.cc | 26 |
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 | |||
@@ -131,3 +131,10 @@ namespace opkele { | |||
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 |
@@ -135,3 +142,3 @@ namespace opkele { | |||
135 | */ | 142 | */ |
136 | static string canonicalize(const string& url); | 143 | virtual string canonicalize(const string& url); |
137 | 144 | ||
diff --git a/lib/consumer.cc b/lib/consumer.cc index 282f0cc..dd8e150 100644 --- a/lib/consumer.cc +++ b/lib/consumer.cc | |||
@@ -312,3 +312,3 @@ namespace opkele { | |||
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; |
@@ -344,2 +344,26 @@ namespace opkele { | |||
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 | } |