Do not bomb out on unusual xml

In particular, properly handle directy Type child of XRD element.

As usually, thanks to Joseph Smarr of Plaxo for spotting it ;-)

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

1 files changed, 26 insertions, 15 deletions

diff --git a/lib/discovery.cc b/lib/discovery.cc
index cbdceb3..f8a3988 100644
--- a/lib/discovery.cc
+++ b/lib/discovery.cc
@@ -265,331 +265,342 @@ namespace opkele {
                         td.opt_set(TidyDoctypeMode,TidyDoctypeOmit);
                         td.opt_set(TidyMark,false);
                         td.opt_set(TidyNumEntities,true);
                         if(td.parse_string(save_html)<=0)
                             throw exception_tidy(OPKELE_CP_ "tidy failed to parse document");
                         if(td.clean_and_repair()<=0)
                             throw exception_tidy(OPKELE_CP_ "tidy failed to clean and repair");
                         util::tidy_buf_t tide;
                         if(td.save_buffer(tide)<=0)
                             throw exception_tidy(OPKELE_CP_ "tidy failed to save buffer");
                         prepare_to_parse();
                         parse(tide.c_str(),tide.size(),true);
                     }catch(exception_tidy& et) { }
                 }
                 save_html.clear();
             }
 
             void prepare_to_parse() {
                 (*(expat_t*)this) = parser_create_ns();
                 set_user_data(); set_element_handler();
                 set_character_data_handler();
                 set_unknown_encoding_handler();
 
                 if(xmode&xmode_html) {
                     html_openid1.clear(); html_openid2.clear();
                     parser_choked = false;
                 }
 
                 cdata = 0; xrd_service = 0; skipping = 0;
                 pt_stack.clear();
                 status_code = 100; status_string.clear();
             }
 
             void html2xrd(endpoint_discovery_iterator& oi,idiscovery_t& id) {
                 XRD_t& x = id.xrd;
                 if(!html_openid2.uris.empty()) {
                     html_openid2.types.insert(STURI_OPENID20);
                     x.services.add(-1,html_openid2);
                     queue_endpoints(oi,id,&op_service_types[st_index_2]);
                 }
                 if(!html_openid1.uris.empty()) {
                     html_openid1.types.insert(STURI_OPENID11);
                     x.services.add(-1,html_openid1);
                     queue_endpoints(oi,id,&op_service_types[st_index_1]);
                 }
             }
 
             size_t write(void *p,size_t s,size_t nm) {
                 /* TODO: limit total size */
                 size_t bytes = s*nm;
                 const char *inbuf = (const char*)p;
                 if(xmode&xmode_html) {
                     size_t mbts = save_html.capacity()-save_html.size();
                     size_t bts = 0;
                     if(mbts>0) {
                         bts = (bytes>mbts)?mbts:bytes;
                         save_html.append(inbuf,bts);
                     }
                     if(skipping<0) return bts;
                 }
                 if(skipping<0) return 0;
                 bool rp = parse(inbuf,bytes,false);
                 if(!rp) {
                     parser_choked = true;
                     skipping = -1;
                     if(!(xmode&xmode_html))
                         bytes = 0;
                 }
                 return bytes;
             }
             size_t header(void *p,size_t s,size_t nm) {
                 size_t bytes = s*nm;
                 const char *h = (const char*)p;
                 const char *colon = (const char*)memchr(p,':',bytes);
                 const char *space = (const char*)memchr(p,' ',bytes);
                 if(space && ( (!colon) || space<colon ) ) {
                     xrds_location.clear(); http_content_type.clear();
                 }else if(colon) {
                     const char *hv = ++colon;
                     size_t hnl = colon-h;
                     int rb;
                     for(rb = bytes-hnl-1;rb>0 && isspace(*hv);++hv,--rb) ;
                     while(rb>0 && isspace(hv[rb-1])) --rb;
                     if(rb) {
                         if( (hnl>=sizeof(XRDS_HEADER))
                                 && !strncasecmp(h,XRDS_HEADER":",
                                     sizeof(XRDS_HEADER)) ) {
                             xrds_location.assign(hv,rb);
                         }else if( (hnl>=sizeof(CT_HEADER))
                                 && !strncasecmp(h,CT_HEADER":",
                                     sizeof(CT_HEADER)) ) {
                             const char *sc = (const char*)memchr(
                                     hv,';',rb);
                             http_content_type.assign(hv,sc?(sc-hv):rb);
                         }
                     }
                 }
                 return curl_t::header(p,s,nm);
             }
 
             void start_element(const XML_Char *n,const XML_Char **a) {
                 if(skipping<0) return;
                 if(skipping) {
                     if(xmode&xmode_html)
                         html_start_element(n,a);
                     ++skipping; return;
                 }
                 if(pt_stack.empty()) {
                     if(is_qelement(n,NSURI_XRDS "\tXRDS"))
                         return;
                     if(is_qelement(n,NSURI_XRD "\tXRD")) {
                         assert(xrd);
                         xrd->clear();
                         pt_stack.push_back(n);
                     }else if(xmode&xmode_html) {
                         html_start_element(n,a);
                     }else{
                         skipping = -1;
                     }
                 }else{
                     int pt_s = pt_stack.size();
                     if(pt_s==1) {
                         if(is_qelement(n,NSURI_XRD "\tCanonicalID")) {
                             assert(xrd);
                             cdata = &(xrd->canonical_ids.add(element_priority(a),string()));
                         }else if(is_qelement(n,NSURI_XRD "\tLocalID")) {
                             assert(xrd);
                             cdata = &(xrd->local_ids.add(element_priority(a),string()));
                         }else if(is_qelement(n,NSURI_XRD "\tProviderID")) {
                             assert(xrd);
                             cdata = &(xrd->provider_id);
                         }else if(is_qelement(n,NSURI_XRD "\tService")) {
                             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);
                                 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);
+                    if(xrd && 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();
-                    xrd_service = 0;
-                }else if(is_qelement(n,NSURI_XRD "\tStatus")) {
-                    assert(xrd);
-                    if(is_qelement(pt_stack.back().c_str(),n)) {
-                        assert(cdata==&status_string);
+                    if(!(xrd && xrd_service)) {
+                        skipping = -1;
+                    }else{
+                        assert(!pt_stack.empty());
+                        assert(pt_stack.back()==(NSURI_XRD "\tService"));
                         pt_stack.pop_back();
-                        if(status_code!=100)
-                            skipping = -1;
+                        xrd_service = 0;
+                    }
+                }else if(is_qelement(n,NSURI_XRD "\tStatus")) {
+                    if(!xrd) {
+                        skipping=-1;
+                    }else{
+                        if(is_qelement(pt_stack.back().c_str(),n)) {
+                            assert(cdata==&status_string);
+                            pt_stack.pop_back();
+                            if(status_code!=100)
+                                skipping = -1;
+                        }
                     }
                 }else if(is_qelement(n,NSURI_XRD "\tExpires")) {
-                    assert(xrd);
-                    xrd->expires = util::w3c_to_time(cdata_buf);
+                    if(!xrd) {
+                        skipping=-1;
+                    }else{
+                        xrd->expires = util::w3c_to_time(cdata_buf);
+                    }
                 }else if((xmode&xmode_html) && is_element(n,"head")) {
                     skipping = -1;
                 }
                 cdata = 0;
             }
             void character_data(const XML_Char *s,int l) {
                 if(skipping) return;
                 if(cdata) cdata->append(s,l);
             }
 
             void html_start_element(const XML_Char *n,const XML_Char **a) {
                 if(is_element(n,"meta")) {
                     bool heq = false;
                     string l;
                     for(;*a;a+=2) {
                         if(!( strcasecmp(a[0],"http-equiv")
                                 || strcasecmp(a[1],XRDS_HEADER) ))
                             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(data::_whitespace_chars);
                             href.erase(lns+1);
                         }
                     }
                     for(string::size_type ns=rels.find_first_not_of(data::_whitespace_chars);
                             ns!=string::npos; ns=rels.find_first_not_of(data::_whitespace_chars,ns)) {
                         string::size_type s = rels.find_first_of(data::_whitespace_chars,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,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,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.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;
                                 }
                             }
                         }
                     }
                 }
             }
 
             int unknown_encoding(const XML_Char *n,XML_Encoding *i) {
                 for(int ii=0;ii < sizeof(i->map)/sizeof(i->map[0]);++ii)
                     i->map[ii] = ii;
                 i->convert = 0; i->release = 0;
                 return XML_STATUS_OK;
             }
 
     };
 
     string idiscover(endpoint_discovery_iterator oi,const string& identity) {
         idigger_t idigger;
         return idigger.discover(oi,identity);
     }
 
     void yadiscover(endpoint_discovery_iterator oi,const string& yurl,const char **types,bool redirs) try {
         idigger_t idigger;
         idigger.yadiscover(oi,yurl,types,redirs);
     }catch(exception_curl& ec) {
         if(redirs || ec._error!=CURLE_TOO_MANY_REDIRECTS)
             throw;
     }
 
 }