2 files changed, 7 insertions, 10 deletions

diff --git a/include/opkele/expat.h b/include/opkele/expat.h
index 97ed61a..60c41ac 100644
--- a/include/opkele/expat.h
+++ b/include/opkele/expat.h
@@ -1,93 +1,91 @@
 #ifndef __OPKELE_EXPAT_H
 #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); }
-                enum XML_Status stop_parser(bool resumable=false) {
-                    assert(_x);
-                    return XML_StopParser(_x,resumable); }
+                    return XML_Parse(_x,s,len,final);
+                }
 
                 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) { }
                 void set_namespace_decl_handler();
 
                 inline enum XML_Error get_error_code() {
                     assert(_x); return XML_GetErrorCode(_x); }
                 static inline const XML_LChar *error_string(XML_Error c) {
                     return XML_ErrorString(c); }
 
                 inline long get_current_byte_index() {
                     assert(_x); return XML_GetCurrentByteIndex(_x); }
                 inline int get_current_line_number() {
                     assert(_x); return XML_GetCurrentLineNumber(_x); }
                 inline int get_current_column_number() {
                     assert(_x); return XML_GetCurrentColumnNumber(_x); }
 
                 inline void set_user_data() {
                     assert(_x); XML_SetUserData(_x,this); }
 
                 inline bool set_base(const XML_Char *b) {
                     assert(_x); return XML_SetBase(_x,b); }
                 inline const XML_Char *get_base() {
                     assert(_x); return XML_GetBase(_x); }
 
                 inline int get_specified_attribute_count() {
                     assert(_x); return XML_GetSpecifiedAttributeCount(_x); }
 
                 inline bool set_param_entity_parsing(enum XML_ParamEntityParsing c) {
                     assert(_x); return XML_SetParamEntityParsing(_x,c); }
 
                 inline static XML_Parser parser_create(const XML_Char *e=0) {
                     return XML_ParserCreate(e); }
                 inline static XML_Parser parser_create_ns(const XML_Char *e=0,XML_Char s='\t') {
                     return XML_ParserCreateNS(e,s); }
 
         };
 
     }
 
 }
 
 #endif /* __OPKELE_EXPAT_H */
diff --git a/lib/discovery.cc b/lib/discovery.cc
index af9686a..d868308 100644
--- a/lib/discovery.cc
+++ b/lib/discovery.cc
@@ -1,447 +1,446 @@
 #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/tidy.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 const size_t max_html = 16384;
 
     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;
     }
 
     class idigger_t : public util::curl_t, public util::expat_t {
         public:
             string xri_proxy;
 
             enum {
                 xmode_html = 1, xmode_xrd = 2
             };
             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 discover(idiscovery_t& result,const string& identity) {
                 result.clear();
                 string::size_type fsc = identity.find_first_not_of(whitespace);
                 if(fsc==string::npos)
                     throw bad_input(OPKELE_CP_ "whtiespace-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);
                 if(strchr(i_leaders,id[0])) {
                     result.normalized_id = id;
                     result.xri_identity = true;
                     /* TODO: further canonicalize xri identity? Like folding case  or whatever... */
                     discover_at(
                             result,
                             xri_proxy + util::url_encode(id)+
                             "?_xrd_r=application/xrd+xml;sep=false", xmode_xrd);
                     if(status_code!=100)
                         throw failed_xri_resolution(OPKELE_CP_
                                 "XRI resolution failed with '"+status_string+"' message",status_code);
                     if(result.xrd.canonical_ids.empty())
                         throw opkele::failed_discovery(OPKELE_CP_ "No CanonicalID for XRI identity found");
                     result.canonicalized_id = result.xrd.canonical_ids.begin()->second;
                 }else{
                     result.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);
                     }
                     result.normalized_id = util::rfc_3986_normalize_uri(id);
                     discover_at(result,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);
                     result.canonicalized_id = util::rfc_3986_normalize_uri(eu); /* XXX: strip fragment part? */
                     if(xrds_location.empty()) {
                         html2xrd(result.xrd);
                     }else{
                         discover_at(result,xrds_location,xmode_xrd);
                         if(result.xrd.empty())
                             html2xrd(result.xrd);
                     }
                 }
             }
 
             void discover_at(idiscovery_t& result,const string& url,int xm) {
                 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 = &result.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);
                         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;
                 status_code = 100; status_string.clear();
             }
 
             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) {
                 /* 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;
                     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(
                                     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; stop_parser();
+                        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 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()));
                             }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();
                     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; stop_parser();
-                        }
+                        if(status_code!=100)
+                            skipping = -1;
                     }
                 }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; stop_parser();
+                    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(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);
                         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; stop_parser();
+                    skipping = -1;
                 }
             }
 
     };
 
     void idiscover(idiscovery_t& result,const string& identity) {
         idigger_t idigger;
         idigger.discover(result,identity);
     }
 
 }