added passthrough unknown encoding handler to discovery parser

1 files changed, 8 insertions, 0 deletions

diff --git a/lib/discovery.cc b/lib/discovery.cc
index bd1f917..cbdceb3 100644
--- a/lib/discovery.cc
+++ b/lib/discovery.cc
@@ -222,128 +222,129 @@ namespace opkele {
 		    }
 		}
 		return rv;
 	    }
 
 	    void discover_at(idiscovery_t& idis,const string& url,int xm) {
 		CURLcode r = easy_setopt(CURLOPT_MAXREDIRS, (xm&xmode_noredirs)?0:5);
 		if(r)
 		    throw exception_curl(OPKELE_CP_ "failed to set curly maxredirs option");
 		if( (r=easy_setopt(CURLOPT_URL,url.c_str())) )
 		    throw exception_curl(OPKELE_CP_ "failed to set curly 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 if(xmode&xmode_html){
 		    /* 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");
 #ifdef NDEBUG
 			td.opt_set(TidyQuiet,true);
 			td.opt_set(TidyShowWarnings,false);
 #else /* NDEBUG */
 			td.opt_set(TidyQuiet,false);
 			td.opt_set(TidyShowWarnings,true);
 #endif /* NDEBUG */
 			td.opt_set(TidyForceOutput,true);
 			td.opt_set(TidyXhtmlOut,true);
 			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":",
@@ -508,80 +509,87 @@ namespace opkele {
 			    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;
     }
 
 }