canonicalization bug - do not add '/' after '?', add it right after hostname if needed.

1 files changed, 9 insertions, 2 deletions

diff --git a/lib/consumer.cc b/lib/consumer.cc
index cbe0769..bb6358c 100644
--- a/lib/consumer.cc
+++ b/lib/consumer.cc
@@ -272,70 +272,77 @@ namespace opkele {
         (r=curl_misc_sets(curl))
         || (r=curl_easy_setopt(curl,CURLOPT_URL,url.c_str()))
         || (r=curl_easy_setopt(curl,CURLOPT_WRITEFUNCTION,_curl_tostring))
         || (r=curl_easy_setopt(curl,CURLOPT_WRITEDATA,&html))
         ;
         if(r)
             throw exception_curl(OPKELE_CP_ "failed to curl_easy_setopt()",r);
         r = curl_easy_perform(curl);
         if(r && r!=CURLE_WRITE_ERROR)
             throw exception_curl(OPKELE_CP_ "failed to curl_easy_perform()",r);
         pcrepp::Pcre bre("<body\\b",PCRE_CASELESS);
         // strip out everything past body
         if(bre.search(html))
             html.erase(bre.get_match_start());
         pcrepp::Pcre hdre("<head[^>]*>",PCRE_CASELESS);
         if(!hdre.search(html))
             throw bad_input(OPKELE_CP_ "failed to find head");
         html.erase(0,hdre.get_match_end()+1);
         pcrepp::Pcre lre("<link\\b([^>]+)>",PCRE_CASELESS),
             rre("\\brel=['\"]([^'\"]+)['\"]",PCRE_CASELESS),
             hre("\\bhref=['\"]([^'\"]+)['\"]",PCRE_CASELESS);
         while(lre.search(html)) {
             string attrs = lre[0];
             html.erase(0,lre.get_match_end()+1);
             if(!(rre.search(attrs)&&hre.search(attrs)))
                 continue;
             if(rre[0]=="openid.server") {
                 server = hre[0];
                 if(!delegate.empty())
                     break;
             }else if(rre[0]=="openid.delegate") {
                 delegate = hre[0];
                 if(!server.empty())
                     break;
             }
         }
         if(server.empty())
             throw failed_assertion(OPKELE_CP_ "The location has no openid.server declaration");
     }
 
     assoc_t consumer_t::find_assoc(const string& server) {
         throw failed_lookup(OPKELE_CP_ "no find_assoc() provided");
     }
 
     string consumer_t::canonicalize(const string& url) {
         string rv = url;
         // strip leading and trailing spaces
         string::size_type i = rv.find_first_not_of(" \t\r\n");
         if(i==string::npos)
             throw bad_input(OPKELE_CP_ "empty URL");
         if(i)
             rv.erase(0,i);
         i = rv.find_last_not_of(" \t\r\n");
         assert(i!=string::npos);
         if(i<(rv.length()-1))
             rv.erase(i+1);
         // add missing http://
         i = rv.find("://");
         if(i==string::npos) { // primitive. but do we need more?
             rv.insert(0,"http://");
             i = sizeof("http://")-1;
         }else{
             i += sizeof("://")-1;
         }
-        if(rv.find('/',i)==string::npos)
-            rv += '/';
+        string::size_type qm = rv.find('?',i);
+        string::size_type sl = rv.find('/',i);
+        if(qm!=string::npos) {
+            if(sl==string::npos || sl>qm)
+                rv.insert(qm,1,'/');
+        }else{
+            if(sl==string::npos)
+                rv += '/';
+        }
         return rv;
     }
 
 }