Implement URI construction handling append='qxri' xrd:URI attribute.

This is the right thing to do and may come in handy assisting OPs in improving
UI.

Signed-off-by: Michael Krelin <hacker@klever.net>

1 files changed, 22 insertions, 7 deletions

diff --git a/lib/discovery.cc b/lib/discovery.cc
index 93409f4..6e20654 100644
--- a/lib/discovery.cc
+++ b/lib/discovery.cc
@@ -30,201 +30,209 @@ namespace opkele {
         const char *forceid;
     } service_types[] = {
         { STURI_OPENID20_OP, IDURI_SELECT20 },
         { STURI_OPENID20, 0 },
         { STURI_OPENID11, 0 },
         { STURI_OPENID10, 0 }
     };
     enum {
         st_index_1 = 2, st_index_2 = 1
     };
 
 
     static inline bool is_qelement(const XML_Char *n,const char *qen) {
         return !strcasecmp(n,qen);
     }
     static inline bool is_element(const XML_Char *n,const char *en) {
         if(!strcasecmp(n,en)) return true;
         int nl = strlen(n), enl = strlen(en);
         if( (nl>=(enl+1)) && n[nl-enl-1]=='\t'
                 && !strcasecmp(&n[nl-enl],en) )
             return true;
         return false;
     }
 
     static long element_priority(const XML_Char **a) {
         for(;*a;++a)
             if(!strcasecmp(*(a++),"priority")) {
                 long rv;
                 return (sscanf(*a,"%ld",&rv)==1)?rv:-1;
             }
         return -1;
     }
+    /* TODO: ideally all attributes should be
+     * retrieved in one run */
+    static const char *element_attr(const XML_Char **a, const char *at) {
+        for(;*a;++a)
+            if(!strcasecmp(*(a++),at)) {
+                return *a;
+            }
+        return 0;
+    }
 
     class idigger_t : public util::curl_t, public util::expat_t {
         public:
             string xri_proxy;
 
             enum {
                 xmode_html = 1, xmode_xrd = 2, xmode_cid = 4
             };
             int xmode;
 
             string xrds_location;
             string http_content_type;
             service_t html_openid1;
             service_t html_openid2;
             string cdata_buf;
             long status_code;
             string status_string;
 
             typedef list<string> pt_stack_t;
             pt_stack_t pt_stack;
             int skipping;
             bool parser_choked;
             string save_html;
 
             XRD_t *xrd;
             service_t *xrd_service;
             string* cdata;
 
             idigger_t()
                 : util::curl_t(easy_init()),
                 util::expat_t(0),
                 xri_proxy(XRI_PROXY_URL) {
                     CURLcode r;
                     (r=misc_sets())
                         || (r=set_write())
                         || (r=set_header())
                         ;
                     if(r)
                         throw exception_curl(OPKELE_CP_ "failed to set curly options",r);
                 }
             ~idigger_t() throw() { }
 
             string discover(endpoint_discovery_iterator& oi,const string& identity) {
                 string rv;
                 idiscovery_t idis;
                 string::size_type fsc = identity.find_first_not_of(whitespace);
                 if(fsc==string::npos)
                     throw bad_input(OPKELE_CP_ "whitespace-only identity");
                 string::size_type lsc = identity.find_last_not_of(whitespace);
                 assert(lsc!=string::npos);
                 if(!strncasecmp(identity.c_str()+fsc,"xri://",sizeof("xri://")-1))
                     fsc += sizeof("xri://")-1;
                 if((fsc+1)>=lsc)
                     throw bad_input(OPKELE_CP_ "not a character of importance in identity");
                 string id(identity,fsc,lsc-fsc+1);
                 idis.clear();
                 if(strchr(i_leaders,id[0])) {
                     /* TODO: further normalize xri identity? Like folding case
                      * or whatever... */
-                    rv = idis.normalized_id = id;
-                    idis.xri_identity = true;
+                    rv = id;
                     set<string> cids;
                     for(const struct service_type_t *st=service_types;
                             st<&service_types[sizeof(service_types)/sizeof(*service_types)];++st) {
                         idis.clear();
                         discover_at( idis,
                                 xri_proxy + util::url_encode(id)+
                                 "?_xrd_t="+util::url_encode(st->uri)+
                                 "&_xrd_r=application/xrd%2Bxml"
                                 ";sep=true;refs=true",
                                 xmode_xrd );
                         if(status_code==241) continue;
                         if(status_code!=100)
                             throw failed_xri_resolution(OPKELE_CP_
                                     "XRI resolution failed with '"+status_string+"' message"
                                     ", while looking for SEP with type '"+st->uri+"'", status_code);
                         if(idis.xrd.canonical_ids.empty())
                             throw opkele::failed_discovery(OPKELE_CP_ "No CanonicalID found for XRI identity found");
                         string cid = idis.xrd.canonical_ids.begin()->second;
                         if(cids.find(cid)==cids.end()) {
                             cids.insert(cid);
                             idis.clear();
                             discover_at( idis,
                                     xri_proxy + util::url_encode(id)+
                                     "?_xrd_t="+util::url_encode(st->uri)+
                                     "&_xrd_r=application/xrd%2Bxml"
                                     ";sep=true;refs=true",
                                     xmode_xrd );
                             if(status_code==241) continue;
                             if(status_code!=100)
                                 throw failed_xri_resolution(OPKELE_CP_
                                         "XRI resolution failed with '"+status_string+"' message"
                                         ", while looking for SEP with type '"+st->uri+"'"
                                         " on canonical id", status_code);
                         }
                         idis.canonicalized_id = cid;
+                        idis.normalized_id = rv; idis.xri_identity = true;
                         queue_endpoints(oi,idis,st);
                     }
                 }else{
                     idis.xri_identity = false;
                     if(id.find("://")==string::npos)
                         id.insert(0,"http://");
                     string::size_type fp = id.find('#');
                     if(fp!=string::npos) {
                         string::size_type qp = id.find('?');
                         if(qp==string::npos || qp<fp)
                             id.erase(fp);
                         else if(qp>fp)
                             id.erase(fp,qp-fp);
                     }
                     rv = idis.normalized_id = util::rfc_3986_normalize_uri(id);
                     discover_at(idis,id,xmode_html|xmode_xrd);
                     const char * eu = 0;
                     CURLcode r = easy_getinfo(CURLINFO_EFFECTIVE_URL,&eu);
                     if(r)
                         throw exception_curl(OPKELE_CP_ "failed to get CURLINFO_EFFECTIVE_URL",r);
                     string cid = util::strip_uri_fragment_part( idis.canonicalized_id = util::rfc_3986_normalize_uri(eu) );
                     if(xrds_location.empty()) {
                         html2xrd(oi,idis);
                     }else{
                         idis.clear();
                         idis.canonicalized_id = cid;
                         discover_at(idis,xrds_location,xmode_xrd);
                         if(idis.xrd.empty())
                             html2xrd(oi,idis);
                         else{
                             for(const service_type_t *st=service_types;
                                     st<&service_types[sizeof(service_types)/sizeof(*service_types)];++st)
                                 queue_endpoints(oi,idis,st);
                         }
                     }
                 }
                 return rv;
             }
 
             void discover_at(idiscovery_t& idis,const string& url,int xm) {
-                DOUT_("Doing discovery at " << url);
                 CURLcode r = easy_setopt(CURLOPT_URL,url.c_str());
                 if(r)
                     throw exception_curl(OPKELE_CP_ "failed to set culry urlie",r);
 
                 http_content_type.clear();
                 xmode = xm;
                 prepare_to_parse();
                 if(xmode&xmode_html) {
                     xrds_location.clear();
                     save_html.clear();
                     save_html.reserve(max_html);
                 }
                 xrd = &idis.xrd;
 
                 r = easy_perform();
                 if(r && r!=CURLE_WRITE_ERROR)
                     throw exception_curl(OPKELE_CP_ "failed to perform curly request",r);
 
                 if(!parser_choked) {
                     parse(0,0,true);
                 }else{
                     /* TODO: do not bother if we've seen xml */
                     try {
                         util::tidy_doc_t td = util::tidy_doc_t::create();
                         if(!td)
                             throw exception_tidy(OPKELE_CP_ "failed to create htmltidy document");
 #ifndef NDEBUG
                         td.opt_set(TidyQuiet,false);
                         td.opt_set(TidyShowWarnings,false);
 #endif /* NDEBUG */
                         td.opt_set(TidyForceOutput,true);
                         td.opt_set(TidyXhtmlOut,true);
@@ -361,65 +369,67 @@ namespace opkele {
                             assert(xrd);
                             xrd_service = &(xrd->services.add(element_priority(a),
                                         service_t()));
                             pt_stack.push_back(n);
                         }else if(is_qelement(n,NSURI_XRD "\tStatus")) {
                             for(;*a;) {
                                 if(!strcasecmp(*(a++),"code")) {
                                     if(sscanf(*(a++),"%ld",&status_code)==1 && status_code!=100) {
                                         cdata = &status_string;
                                         pt_stack.push_back(n);
                                         break;
                                     }
                                 }else
                                     ++a;
                             }
                         }else if(is_qelement(n,NSURI_XRD "\tExpires")) {
                             assert(xrd);
                             cdata_buf.clear();
                             cdata = &cdata_buf;
                         }else if(xmode&xmode_html) {
                             html_start_element(n,a);
                         }else{
                             skipping = 1;
                         }
                     }else if(pt_s==2) {
                         if(is_qelement(pt_stack.back().c_str(), NSURI_XRD "\tService")) {
                             if(is_qelement(n,NSURI_XRD "\tType")) {
                                 assert(xrd); assert(xrd_service);
                                 cdata_buf.clear();
                                 cdata = &cdata_buf;
                             }else if(is_qelement(n,NSURI_XRD "\tURI")) {
                                 assert(xrd); assert(xrd_service);
-                                cdata = &(xrd_service->uris.add(element_priority(a),string()));
+                                const char *append = element_attr(a,"append");
+                                xrd::uri_t& uri = xrd_service->uris.add(element_priority(a),xrd::uri_t("",append?append:""));
+                                cdata = &uri.uri;
                             }else if(is_qelement(n,NSURI_XRD "\tLocalID")
                                     || is_qelement(n,NSURI_OPENID10 "\tDelegate") ) {
                                 assert(xrd); assert(xrd_service);
                                 cdata = &(xrd_service->local_ids.add(element_priority(a),string()));
                             }else if(is_qelement(n,NSURI_XRD "\tProviderID")) {
                                     assert(xrd); assert(xrd_service);
                                     cdata = &(xrd_service->provider_id);
                             }else{
                                 skipping = 1;
                             }
                         }else
                             skipping = 1;
                     }else if(xmode&xmode_html) {
                         html_start_element(n,a);
                     }else{
                         skipping = 1;
                     }
                 }
             }
             void end_element(const XML_Char *n) {
                 if(skipping<0) return;
                 if(skipping) {
                     --skipping; return;
                 }
                 if(is_qelement(n,NSURI_XRD "\tType")) {
                     assert(xrd); assert(xrd_service); assert(cdata==&cdata_buf);
                     xrd_service->types.insert(cdata_buf);
                 }else if(is_qelement(n,NSURI_XRD "\tService")) {
                     assert(xrd); assert(xrd_service);
                     assert(!pt_stack.empty());
                     assert(pt_stack.back()==(NSURI_XRD "\tService"));
                     pt_stack.pop_back();
@@ -455,83 +465,88 @@ namespace opkele {
                             heq = true;
                         else if(!strcasecmp(a[0],"content"))
                             l.assign(a[1]);
                     }
                     if(heq)
                         xrds_location = l;
                 }else if(is_element(n,"link")) {
                     string rels;
                     string href;
                     for(;*a;a+=2) {
                         if( !strcasecmp(a[0],"rel") ) {
                             rels.assign(a[1]);
                         }else if( !strcasecmp(a[0],"href") ) {
                             const char *ns = a[1];
                             for(;*ns && isspace(*ns);++ns);
                             href.assign(ns);
                             string::size_type lns=href.find_last_not_of(whitespace);
                             href.erase(lns+1);
                         }
                     }
                     for(string::size_type ns=rels.find_first_not_of(whitespace);
                             ns!=string::npos; ns=rels.find_first_not_of(whitespace,ns)) {
                         string::size_type s = rels.find_first_of(whitespace,ns);
                         string rel;
                         if(s==string::npos) {
                             rel.assign(rels,ns,string::npos);
                             ns = string::npos;
                         }else{
                             rel.assign(rels,ns,s-ns);
                             ns = s;
                         }
                         if(rel=="openid.server")
-                            html_openid1.uris.add(-1,href);
+                            html_openid1.uris.add(-1,xrd::uri_t(href));
                         else if(rel=="openid.delegate")
                             html_openid1.local_ids.add(-1,href);
                         else if(rel=="openid2.provider")
-                            html_openid2.uris.add(-1,href);
+                            html_openid2.uris.add(-1,xrd::uri_t(href));
                         else if(rel=="openid2.local_id")
                             html_openid2.local_ids.add(-1,href);
                     }
                 }else if(is_element(n,"body")) {
                     skipping = -1;
                 }
             }
 
             void queue_endpoints(endpoint_discovery_iterator& oi,
                     const idiscovery_t &id,
                     const service_type_t *st) {
                 openid_endpoint_t ep;
                 ep.claimed_id = id.canonicalized_id;
                 for(xrd::services_t::const_iterator isvc=id.xrd.services.begin();
                         isvc!=id.xrd.services.end(); ++isvc) {
                     const xrd::service_t svc = isvc->second;
                     if(svc.types.find(st->uri)==svc.types.end()) continue;
                     for(xrd::uris_t::const_iterator iu=svc.uris.begin();iu!=svc.uris.end();++iu) {
-                        ep.uri = iu->second;
+                        ep.uri = iu->second.uri;
+                        if(id.xri_identity) {
+                            if(iu->second.append=="qxri") {
+                                ep.uri += id.normalized_id;
+                            } /* TODO: else handle other append attribute values */
+                        }
                         if(st->forceid) {
                             ep.local_id = ep.claimed_id = st->forceid;
                             *(oi++) = ep;
                         }else{
                             if(svc.local_ids.empty()) {
                                 ep.local_id = ep.claimed_id;
                                 *(oi++) = ep;
                             }else{
                                 for(xrd::local_ids_t::const_iterator ilid=svc.local_ids.begin();
                                         ilid!=svc.local_ids.end(); ++ilid) {
                                     ep.local_id = ilid->second;
                                     *(oi++) = ep;
                                 }
                             }
                         }
                     }
                 }
             }
 
     };
 
     string idiscover(endpoint_discovery_iterator oi,const string& identity) {
         idigger_t idigger;
         return idigger.discover(oi,identity);
     }
 
 }