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
@@ -68,75 +68,82 @@ namespace opkele {
 
 	    /**
 	     * perform the associate request to OpenID server.
 	     * @param server the OpenID server
 	     * @return the auto_ptr<> for the newly allocated association_t
 	     * object, representing established association
 	     * @throw exception in case of error
 	     */
 	    assoc_t associate(const string& server);
 	    /**
 	     * prepare the parameters for the checkid_immediate
 	     * request.
 	     * @param identity the identity to verify
 	     * @param return_to the return_to url to pass with the request
 	     * @param trust_root the trust root to advertise with the request
 	     * @param ext pointer to an extension(s) hooks object
 	     * @return the location string
 	     * @throw exception in case of error
 	     */
 	    virtual string checkid_immediate(const string& identity,const string& return_to,const string& trust_root="",extension_t *ext=0);
 	    /**
 	     * prepare the parameters for the checkid_setup
 	     * request.
 	     * @param identity the identity to verify
 	     * @param return_to the return_to url to pass with the request
 	     * @param trust_root the trust root to advertise with the request
 	     * @param ext pointer to an extension(s) hooks object
 	     * @return the location string
 	     * @throw exception in case of error
 	     */
 	    virtual string checkid_setup(const string& identity,const string& return_to,const string& trust_root="",extension_t *ext=0);
 	    /**
 	     * the actual implementation behind checkid_immediate() and
 	     * checkid_setup() functions.
 	     * @param mode checkid_* mode - either mode_checkid_immediate or mode_checkid_setup
 	     * @param identity the identity to verify
 	     * @param return_to the return_to url to pass with the request
 	     * @param trust_root the trust root to advertise with the request
 	     * @param ext pointer to an extension(s) hooks object
 	     * @return the location string
 	     * @throw exception in case of error
 	     */
 	    virtual string checkid_(mode_t mode,const string& identity,const string& return_to,const string& trust_root="",extension_t *ext=0);
 	    /**
 	     * verify the id_res response
 	     * @param pin the response parameters
 	     * @param identity the identity being checked (if not specified,
 	     * @param ext pointer to an extension(s) hooks object
 	     * extracted from the openid.identity parameter
 	     * @throw id_res_mismatch in case of signature mismatch
 	     * @throw id_res_setup in case of openid.user_setup_url failure
 	     * (supposedly checkid_immediate only)
 	     * @throw id_res_failed in case of failure
 	     * @throw exception in case of other failures
 	     */
 	    virtual void id_res(const params_t& pin,const string& identity="",extension_t *ext=0);
 	    /**
 	     * perform a check_authentication request.
 	     * @param server the OpenID server
 	     * @param p request parameters
 	     */
 	    void check_authentication(const string& server,const params_t& p);
 
 	    /**
-	     * 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);
 
     };
 
 }
 
 #endif /* __OPKELE_CONSUMER_H */
diff --git a/lib/consumer.cc b/lib/consumer.cc
index 282f0cc..dd8e150 100644
--- a/lib/consumer.cc
+++ b/lib/consumer.cc
@@ -249,97 +249,121 @@ namespace opkele {
 	if(pp.has_param("invalidate_handle"))
 	    invalidate_assoc(server,pp.get_param("invalidate_handle"));
 	if(pp.has_param("is_valid")) {
 	    if(pp.get_param("is_valid")=="true")
 		return;
 	}else if(pp.has_param("lifetime")) {
 	    if(util::string_to_long(pp.get_param("lifetime")))
 		return;
 	}
 	throw failed_check_authentication(OPKELE_CP_ "failed to verify response");
     }
 
     void consumer_t::retrieve_links(const string& url,string& server,string& delegate) {
 	server.erase();
 	delegate.erase();
 	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,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 consumer_t::normalize(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;
 	}
 	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;
     }
 
+    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);
+    }
+
 }