discovery: stop parser with XML_StopParser()

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

2 files changed, 16 insertions, 8 deletions

diff --git a/include/opkele/expat.h b/include/opkele/expat.h
index 60c41ac..97ed61a 100644
--- a/include/opkele/expat.h
+++ b/include/opkele/expat.h
@@ -2,50 +2,52 @@
 #define __OPKELE_EXPAT_H
 
 #include <cassert>
 #include <expat.h>
 
 namespace opkele {
 
     namespace util {
 
 	class expat_t {
 	    public:
 		XML_Parser _x;
 
 		expat_t() : _x(0) { }
 		expat_t(XML_Parser x) : _x(x) { }
 		virtual ~expat_t() throw();
 
 		expat_t& operator=(XML_Parser x);
 
 		operator const XML_Parser(void) const { return _x; }
 		operator XML_Parser(void) { return _x; }
 
 		inline bool parse(const char *s,int len,bool final=false) {
 		    assert(_x);
-		    return XML_Parse(_x,s,len,final);
-		}
+		    return XML_Parse(_x,s,len,final); }
+		enum XML_Status stop_parser(bool resumable=false) {
+		    assert(_x);
+		    return XML_StopParser(_x,resumable); }
 
 		virtual void start_element(const XML_Char *n,const XML_Char **a) { }
 		virtual void end_element(const XML_Char *n) { }
 		void set_element_handler();
 
 		virtual void character_data(const XML_Char *s,int l) { }
 		void set_character_data_handler();
 
 		virtual void processing_instruction(const XML_Char *t,const XML_Char *d) { }
 		void set_processing_instruction_handler();
 
 		virtual void comment(const XML_Char *d) { }
 		void set_comment_handler();
 
 		virtual void start_cdata_section() { }
 		virtual void end_cdata_section() { }
 		void set_cdata_section_handler();
 
 		virtual void default_handler(const XML_Char *s,int l) { }
 		void set_default_handler();
 		void set_default_handler_expand();
 
 		virtual void start_namespace_decl(const XML_Char *p,const XML_Char *u) { }
 		virtual void end_namespace_decl(const XML_Char *p) { }
diff --git a/lib/discovery.cc b/lib/discovery.cc
index 81727c0..8729cfb 100644
--- a/lib/discovery.cc
+++ b/lib/discovery.cc
@@ -1,31 +1,32 @@
 #include <list>
 #include <opkele/curl.h>
 #include <opkele/expat.h>
 #include <opkele/uris.h>
 #include <opkele/discovery.h>
 #include <opkele/exception.h>
 #include <opkele/util.h>
+#include <opkele/debug.h>
 
 #include "config.h"
 
 #define XRDS_HEADER "X-XRDS-Location"
 #define CT_HEADER "Content-Type"
 
 namespace opkele {
     using std::list;
     using xrd::XRD_t;
     using xrd::service_t;
 
     static const char *whitespace = " \t\r\n";
     static const char *i_leaders = "=@+$!(";
 
     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;
@@ -154,49 +155,53 @@ namespace opkele {
 		status_code = 100; status_string.clear();
 
 		r = easy_perform();
 		if(r && r!=CURLE_WRITE_ERROR)
 		    throw exception_curl(OPKELE_CP_ "failed to perform curly request",r);
 
 		parse(0,0,true);
 	    }
 
 	    void html2xrd(XRD_t& x) {
 		if(!html_openid1.uris.empty()) {
 		    html_openid1.types.insert(STURI_OPENID11);
 		    x.services.add(-1,html_openid1);
 		}
 		if(!html_openid2.uris.empty()) {
 		    html_openid2.types.insert(STURI_OPENID20);
 		    x.services.add(-1,html_openid2);
 		}
 	    }
 
 	    size_t write(void *p,size_t s,size_t nm) {
 		if(skipping<0) return 0;
 		/* TODO: limit total size */
 		size_t bytes = s*nm;
-		parse((const char *)p,bytes,false);
+		bool rp = parse((const char *)p,bytes,false);
+		if(!rp) {
+		    skipping = -1;
+		    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;
 		    int 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(
@@ -204,49 +209,49 @@ namespace opkele {
 			    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;
+			skipping = -1; stop_parser();
 		    }
 		}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);
@@ -289,56 +294,57 @@ namespace opkele {
 		    }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();
 		    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);
 			pt_stack.pop_back();
-			if(status_code!=100)
-			    skipping = -1;
+			if(status_code!=100) {
+			    skipping = -1; stop_parser();
+			}
 		    }
 		}else if(is_qelement(n,NSURI_XRD "\tExpires")) {
 		    assert(xrd);
 		    xrd->expires = util::w3c_to_time(cdata_buf);
 		}else if((xmode&xmode_html) && is_element(n,"head")) {
-		    skipping = -1;
+		    skipping = -1; stop_parser();
 		}
 		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;
@@ -353,36 +359,36 @@ namespace opkele {
 			    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);
 			else if(rel=="openid.delegate")
 			    html_openid1.local_ids.add(-1,href);
 			else if(rel=="openid2.provider")
 			    html_openid2.uris.add(-1,href);
 			else if(rel=="openid2.local_id")
 			    html_openid2.local_ids.add(-1,href);
 		    }
 		}else if(is_element(n,"body")) {
-		    skipping = -1;
+		    skipping = -1; stop_parser();
 		}
 	    }
 
     };
 
     void idiscover(idiscovery_t& result,const string& identity) {
 	idigger_t idigger;
 	idigger.discover(result,identity);
     }
 
 }