do not tidy XRD documents

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

1 files changed, 2 insertions, 2 deletions

diff --git a/lib/discovery.cc b/lib/discovery.cc
index 6f9926c..b7f2db6 100644
--- a/lib/discovery.cc
+++ b/lib/discovery.cc
@@ -66,275 +66,275 @@ namespace opkele {
 	    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,
 		xmode_noredirs = 8
 	    };
 	    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() { }
 
 	    void yadiscover(endpoint_discovery_iterator oi,const string& yurl,const char **types,bool redirs) {
 		idiscovery_t idis;
 		idis.xri_identity = false;
 		discover_at(idis,yurl,xmode_html|xmode_xrd|(redirs?0:xmode_noredirs));
 		if(!xrds_location.empty()) {
 		    idis.clear();
 		    discover_at(idis,xrds_location,xmode_xrd);
 		}
 		idis.normalized_id = idis.canonicalized_id = yurl;
 		service_type_t st;
 		for(st.uri=*types;*types;st.uri=*(++types))
 		    queue_endpoints(oi,idis,&st);
 	    }
 
 	    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 = id;
 		    set<string> cids;
 		    for(const struct service_type_t *st=op_service_types;
 			    st<&op_service_types[sizeof(op_service_types)/sizeof(*op_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");
+			    throw opkele::failed_discovery(OPKELE_CP_ "No CanonicalID 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=op_service_types;
 				    st<&op_service_types[sizeof(op_service_types)/sizeof(*op_service_types)];++st)
 				queue_endpoints(oi,idis,st);
 			}
 		    }
 		}
 		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{
+		}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");
 #ifndef NDEBUG
 			td.opt_set(TidyQuiet,false);
 			td.opt_set(TidyShowWarnings,false);
 #endif /* NDEBUG */
 			td.opt_set(TidyForceOutput,true);
 			td.opt_set(TidyXhtmlOut,true);
 			td.opt_set(TidyDoctypeMode,TidyDoctypeOmit);
 			td.opt_set(TidyMark,false);
 			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();
 
 		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)) ) {