reworked identity resolution and service discovery

 The discovery, which does both XRDS-based (Yadis, XRI, for XRI, using proxy)
 and HTML-based search, now returns results in opkele:idiscovery_t structure.
 It uses expat-based parser idigger_t, which itself is not exposed via any
 header files, but hidden in lib/discovery.cc, the discovery testing program is
 renamed from openid_resolve to idiscover.

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

15 files changed, 575 insertions, 472 deletions

diff --git a/configure.ac b/configure.ac
index 2b94b41..1bd1873 100644
--- a/configure.ac
+++ b/configure.ac
@@ -130,6 +130,15 @@ if $nitpick ; then
  CXXFLAGS="$CXXFLAGS $CXX_NITPICK"
 fi
 
+xri_proxy_url="http://beta.xri.net/"
+AC_MSG_CHECKING([for XRI resolver proxy])
+AC_ARG_ENABLE([xri-proxy],
+ AC_HELP_STRING([--with-xri-proxy=url],[set xri proxy for use when resolving xri identities, default is http://xr_proxy_url]),
+ [ xri_proxy_url="$withval" ]
+)
+AC_MSG_RESULT([$xri_proxy_url])
+AC_DEFINE_UNQUOTED([XRI_PROXY_URL],["$xri_proxy_url"],[XRI proxy resolver URL])
+
 AC_CONFIG_FILES([
  Makefile
   libopkele.pc
diff --git a/include/Makefile.am b/include/Makefile.am
index 23c7e0d..0c2928d 100644
--- a/include/Makefile.am
+++ b/include/Makefile.am
@@ -12,7 +12,7 @@ nobase_include_HEADERS = \
 	opkele/xconsumer.h \
 	opkele/xserver.h \
 	opkele/curl.h opkele/expat.h \
-	opkele/openid_service_resolver.h \
+	opkele/discovery.h \
 	opkele/uris.h
 EXTRA_DIST = \
 	opkele/data.h \
diff --git a/include/opkele/discovery.h b/include/opkele/discovery.h
new file mode 100644
index 0000000..5d7129b
--- a/dev/null
+++ b/include/opkele/discovery.h
@@ -0,0 +1,33 @@
+#ifndef __OPKELE_DISCOVERY_H
+#define __OPKELE_DISCOVERY_H
+
+#include <string>
+#include <opkele/types.h>
+
+namespace opkele {
+    using std::string;
+
+    struct idiscovery_t;
+
+    void idiscover(idiscovery_t& result,const string& identity);
+
+    struct idiscovery_t {
+	string normalized_id;
+	string canonicalized_id;
+	xrd::XRD_t xrd;
+
+	idiscovery_t(const string& i) {
+	    idiscover(*this,i);
+	}
+	idiscovery_t(const char *i) {
+	    idiscover(*this,i);
+	}
+
+	void clear() {
+	    normalized_id.clear(); canonicalized_id.clear();
+	    xrd.clear();
+	}
+    };
+}
+
+#endif /* __OPKELE_DISCOVERY_H */
diff --git a/include/opkele/exception.h b/include/opkele/exception.h
index 753a818..a654d59 100644
--- a/include/opkele/exception.h
+++ b/include/opkele/exception.h
@@ -203,6 +203,25 @@ namespace opkele {
     };
 
     /**
+     * exception thrown in case of failed discovery
+     */
+    class failed_discovery : public exception {
+	public:
+	    failed_discovery(OPKELE_E_PARS)
+		: exception(OPKELE_E_CONS) { }
+    };
+
+    /**
+     * unsuccessfull xri resolution
+     */
+    class failed_xri_resolution : public failed_discovery {
+	public:
+	    long _code;
+	    failed_xri_resolution(OPKELE_E_PARS,long _c=-1)
+		: failed_discovery(OPKELE_E_CONS), _code(_c) { }
+    };
+
+    /**
      * not implemented (think pure virtual) member function executed, signfies
      * programmer error
      */
diff --git a/include/opkele/openid_service_resolver.h b/include/opkele/openid_service_resolver.h
deleted file mode 100644
index 64edd28..0000000
--- a/include/opkele/openid_service_resolver.h
+++ b/dev/null
@@ -1,118 +0,0 @@
-#ifndef __OPKELE_OPENID_SERVICE_RESOLVER_H
-#define __OPKELE_OPENID_SERVICE_RESOLVER_H
-
-#include <climits>
-#include <string>
-#include <list>
-#include <set>
-#include <map>
-#include <opkele/curl.h>
-#include <opkele/expat.h>
-
-namespace opkele {
-    using std::list;
-    using std::string;
-    using std::set;
-    using std::map;
-
-    struct openid_auth_SEP_t {
-	    long priority;
-	    set<string> xrd_Type;
-	    string xrd_URI;
-	    string openid_Delegate;
-
-	    openid_auth_SEP_t() : priority(LONG_MAX) { }
-    };
-
-    struct openid_auth_info_t {
-	string canonical_id;
-	openid_auth_SEP_t auth_SEP;
-    };
-
-
-    class openid_service_resolver_t : public util::curl_t, public util::expat_t {
-	public:
-	    string xri_proxy;
-
-	    openid_service_resolver_t(const string& xp="");
-	    ~openid_service_resolver_t() throw() { }
-
-	    const openid_auth_info_t& resolve(const string& id);
-
-	    enum state_t {
-		state_parse = 0,
-		state_stopping_head, state_stopping_body,
-		state_stopping_size
-	    };
-	    state_t state;
-
-	    struct parser_node_t {
-		string element;
-		string content;
-		typedef map<string,string> attrs_t;
-		attrs_t attrs;
-		bool skip_text, skip_tags;
-		openid_auth_info_t auth_info;
-
-		parser_node_t(const XML_Char *n,const XML_Char **a)
-		    : skip_text(true), skip_tags(true)
-		{
-		    element = n;
-		    for(;*a;a+=2)
-			attrs[a[0]] = a[1];
-		}
-
-	    };
-
-	    class parser_tree_t : public list<parser_node_t> {
-		public:
-		    const_reference top() const { return back(); }
-		    reference top()  { return back(); }
-
-		    const_reference parent() const {
-			const_reverse_iterator rv = rbegin();
-			return *(++rv); }
-		    reference parent() {
-			reverse_iterator rv = rbegin();
-			return *(++rv); }
-
-		    inline void pop() { pop_back(); }
-		    inline void push(const_reference e) { push_back(e); }
-
-		    void push(const XML_Char *n,const XML_Char **a) {
-			parser_node_t nn(n,a);
-			if(empty())
-			    nn.skip_text = nn.skip_tags = true;
-			else{
-			    const_reference t = top();
-			    nn.skip_text = t.skip_text; nn.skip_tags = t.skip_tags;
-			}
-			push(nn);
-		    }
-	    };
-	    parser_tree_t tree;
-
-	    void start_element(const XML_Char *n,const XML_Char **a);
-	    void end_element(const XML_Char *n);
-	    void character_data(const XML_Char *s,int l);
-
-	    string xrds_location;
-	    openid_auth_SEP_t html_SEP;
-	    openid_auth_info_t auth_info;
-
-	    void pop_tag();
-
-	    size_t write(void *p,size_t s,size_t nm);
-
-	    string http_content_type;
-
-	    size_t header(void *p,size_t s,size_t nm);
-	    
-	    bool xri_mode;
-
-	    void discover_service(const string& url,bool xri=false);
-    };
-
-}
-
-#endif /* __OPKELE_OPENID_SERVICE_RESOLVER_H */
diff --git a/include/opkele/types.h b/include/opkele/types.h
index f732a1e..520618d 100644
--- a/include/opkele/types.h
+++ b/include/opkele/types.h
@@ -11,6 +11,7 @@
 #include <string>
 #include <map>
 #include <memory>
+#include <set>
 
 namespace opkele {
     using std::vector;
@@ -18,6 +19,8 @@ namespace opkele {
     using std::map;
     using std::ostream;
     using std::auto_ptr;
+    using std::multimap;
+    using std::set;
 
     /**
      * the OpenID operation mode
@@ -167,6 +170,66 @@ namespace opkele {
      */
     ostream& operator << (ostream& o,const params_t& p);
 
+    namespace xrd {
+
+	struct priority_compare {
+	    inline bool operator()(long a,long b) const {
+		return (a<0) ? false : (b<0) ? false : (a<b);
+	    }
+	};
+
+	template <typename _DT>
+	    class priority_map : public multimap<long,_DT,priority_compare> {
+		typedef multimap<long,_DT,priority_compare> map_type;
+		public:
+
+		    inline _DT& add(long priority,const _DT& d) {
+			return insert(typename map_type::value_type(priority,d))->second;
+		    }
+	    };
+
+	typedef priority_map<string> canonical_ids_t;
+	typedef priority_map<string> local_ids_t;
+	typedef set<string> types_t;
+	typedef priority_map<string> uris_t;
+
+	class service_t {
+	    public:
+		types_t types;
+		uris_t uris;
+		local_ids_t local_ids;
+
+		void clear() {
+		    types.clear();
+		    uris.clear(); local_ids.clear();
+		}
+	};
+	typedef priority_map<service_t> services_t;
+
+	class XRD_t {
+	    public:
+		time_t expires;
+
+		canonical_ids_t canonical_ids;
+		local_ids_t local_ids;
+		services_t services;
+
+		void clear() {
+		    expires = 0;
+		    canonical_ids.clear(); local_ids.clear();
+		    services.clear();
+		}
+		bool empty() const {
+		    return
+			canonical_ids.empty()
+			&& local_ids.empty()
+			&& services.empty();
+		}
+
+	};
+
+    }
+
 }
 
 #endif /* __OPKELE_TYPES_H */
diff --git a/include/opkele/uris.h b/include/opkele/uris.h
index 9a6a3cd..a432b13 100644
--- a/include/opkele/uris.h
+++ b/include/opkele/uris.h
@@ -1,10 +1,15 @@
 #ifndef __OPKELE_URIS_H
 #define __OPKELE_URIS_H
 
-#define NSURI_XRDS     "xri://$xrds"
-#define NSURI_XRD      "xri://$xrd*($v*2.0)"
-#define NSURI_OPENID10 "http://openid.net/xmlns/1.0"
+#define NSURI_XRDS		"xri://$xrds"
+#define NSURI_XRD		"xri://$xrd*($v*2.0)"
+#define NSURI_OPENID10		"http://openid.net/xmlns/1.0"
 
-#define STURI_OPENID10 "http://openid.net/signon/1.0"
+#define STURI_OPENID10		"http://openid.net/signon/1.0"
+#define STURI_OPENID11		"http://openid.net/signon/1.1"
+#define STURI_OPENID20		"http://specs.openid.net/auth/2.0/signon"
+#define STURI_OPENID20_OP	"http://specs.openid.net/auth/2.0/server"
+
+#define IDURI_SELECT20		"http://specs.openid.net/auth/2.0/identifier_select"
 
 #endif /* __OPKELE_URIS_H */
diff --git a/lib/Makefile.am b/lib/Makefile.am
index b278faf..185411f 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -25,6 +25,6 @@ libopkele_la_SOURCES = \
 		       sreg.cc \
 		       extension_chain.cc \
 		       curl.cc expat.cc \
-		       openid_service_resolver.cc
+		       discovery.cc
 libopkele_la_LDFLAGS = \
 	-version-info 2:0:0
diff --git a/lib/discovery.cc b/lib/discovery.cc
new file mode 100644
index 0000000..a35ce32
--- a/dev/null
+++ b/lib/discovery.cc
@@ -0,0 +1,375 @@
+#include <iostream>
+using namespace std;
+#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 "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;
+    }
+
+    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;
+
+	    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;
+		    /* 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");
+		}else{
+		    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);
+
+		(*(expat_t*)this) = parser_create_ns();
+		set_user_data(); set_element_handler();
+		set_character_data_handler();
+
+		xrds_location.clear(); http_content_type.clear();
+		xmode = xm;
+		if(xmode&xmode_html) {
+		    xrds_location.clear();
+		    html_openid1.clear(); html_openid2.clear();
+		}
+		xrd = &result.xrd;
+		cdata = 0; xrd_service = 0; skipping = 0;
+		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);
+		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;
+		    }
+		}else{
+		    int pt_s = pt_stack.size();
+		    if(pt_s==1) {
+			/* TODO: xrd:XRD/xrd:Expires */
+			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 "\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(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->uris.add(element_priority(a),string()));
+			    }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;
+		    }
+		}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(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;
+		}
+	    }
+
+    };
+
+    void idiscover(idiscovery_t& result,const string& identity) {
+	idigger_t idigger;
+	idigger.discover(result,identity);
+    }
+
+}
diff --git a/lib/openid_service_resolver.cc b/lib/openid_service_resolver.cc
deleted file mode 100644
index 505e5b2..0000000
--- a/lib/openid_service_resolver.cc
+++ b/dev/null
@@ -1,298 +0,0 @@
-#include <cctype>
-#include <opkele/exception.h>
-#include <opkele/util.h>
-#include <opkele/openid_service_resolver.h>
-#include <opkele/uris.h>
-
-#define LOCATION_HEADER "X-XRDS-Location"
-
-namespace opkele {
-    static const char *whitespace = " \t\r\n";
-
-    openid_service_resolver_t::openid_service_resolver_t(const string& xp)
-	: util::curl_t(easy_init()),
-	util::expat_t(0),
-	xri_proxy(xp.empty()?"http://beta.xri.net/":xp)
-    {
-	CURLcode r;
-	(r=misc_sets())
-	|| (r=set_write())
-	|| (r==set_header())
-	;
-	if(r)
-	    throw opkele::exception_curl(OPKELE_CP_ "failed to set curly options",r);
-    }
-
-    static 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 inline bool is_qelement(const XML_Char *n,const char *qen) {
-	return !strcasecmp(n,qen);
-    }
-    static inline bool is_element(
-	    const openid_service_resolver_t::parser_node_t& n,
-	    const char *en) {
-	return is_element(n.element.c_str(),en);
-    }
-    static inline bool is_qelement(
-	    const openid_service_resolver_t::parser_node_t& n,
-	    const char *qen) {
-	return is_qelement(n.element.c_str(),qen);
-    }
-
-    void openid_service_resolver_t::start_element(const XML_Char *n,const XML_Char **a) {
-	if(state!=state_parse) return;
-	tree.push(n,a);
-	parser_node_t& t = tree.top();
-	if(is_element(n,"html") || is_element(n,"head")
-		|| is_qelement(n,NSURI_XRDS "\tXRDS")
-		|| is_qelement(n,NSURI_XRD "\tXRD") )
-	    t.skip_tags = false;
-	else if(is_qelement(n,NSURI_XRD "\tService")
-		|| is_qelement(n,NSURI_XRD "\tType")
-		|| is_qelement(n,NSURI_XRD "\tURI")
-		|| is_qelement(n,NSURI_OPENID10 "\tDelegate")
-		|| is_qelement(n,NSURI_XRD "\tCanonicalID") )
-	    t.skip_tags = t.skip_text = false;
-	else if(is_element(n,"body"))
-	    state = state_stopping_body;
-    }
-    void openid_service_resolver_t::end_element(const XML_Char *n) {
-	if(state!=state_parse) return;
-	assert(tree.top().element == n);
-	pop_tag();
-    }
-    void openid_service_resolver_t::character_data(const XML_Char *s,int l) {
-	if(state!=state_parse) return;
-	if( !( tree.empty() || tree.top().skip_text ) )
-	    tree.top().content.append(s,l);
-    }
-
-    static void copy_trim_whitespace(string& to,const string& from) {
-	string::size_type ns0 = from.find_first_not_of(whitespace);
-	if(ns0==string::npos) {
-	    to.clear(); return;
-	}
-	string::size_type ns1 = from.find_last_not_of(whitespace);
-	assert(ns1!=string::npos);
-	to.assign(from,ns0,ns1-ns0+1);
-    }
-
-    void openid_service_resolver_t::pop_tag() {
-	assert(!tree.empty());
-	parser_node_t& t = tree.top();
-	if( is_element(t,"meta")
-		&& !strcasecmp(t.attrs["http-equiv"].c_str(),LOCATION_HEADER) ) {
-	    xrds_location = t.attrs["content"];
-	}else if( is_element(t,"link") ) {
-	    parser_node_t::attrs_t::const_iterator ir = t.attrs.find("rel");
-	    if(ir!=t.attrs.end()) {
-		const string& rels = ir->second;
-		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")
-			copy_trim_whitespace(html_SEP.xrd_URI,t.attrs["href"]);
-		    else if(rel=="openid.delegate")
-			copy_trim_whitespace(html_SEP.openid_Delegate,t.attrs["href"]);
-		}
-	    }
-	}else if( is_element(t,"head") )
-	    state = state_stopping_head;
-	else if( is_qelement(t,NSURI_XRD "\tXRD")) {
-	    if( !(
-			(
-			 xri_mode
-			 && t.auth_info.canonical_id.empty()
-			) ||
-			t.auth_info.auth_SEP.xrd_Type.empty()
-		 ) )
-		auth_info = t.auth_info;
-	}else if( tree.size()>1 ) {
-	    parser_node_t& p = tree.parent();
-	    if( is_qelement(p,NSURI_XRD "\tService") ) {
-		if( is_qelement(t,NSURI_XRD "\tType") ) {
-		    if(t.content==STURI_OPENID10) {
-			string tmp; copy_trim_whitespace(tmp,t.content);
-			p.auth_info.auth_SEP.xrd_Type.insert(tmp);
-		    }
-		}else if( is_qelement(t,NSURI_XRD "\tURI") )
-		    copy_trim_whitespace(p.auth_info.auth_SEP.xrd_URI,t.content);
-		else if( is_qelement(t,NSURI_OPENID10 "\tDelegate") )
-		    copy_trim_whitespace(p.auth_info.auth_SEP.openid_Delegate,t.content);
-	    }else if( is_qelement(p,NSURI_XRD "\tXRD") ) {
-		if(is_qelement(t,NSURI_XRD "\tService") ) {
-		    if( !t.auth_info.auth_SEP.xrd_Type.empty() ) {
-			parser_node_t::attrs_t::const_iterator ip
-			    = t.attrs.find("priority");
-			if(ip!=t.attrs.end()) {
-			    const char *nptr = ip->second.c_str();
-			    char *eptr = 0;
-			    t.auth_info.auth_SEP.priority = strtol(nptr,&eptr,10);
-			    if(nptr==eptr)
-				t.auth_info.auth_SEP.priority = LONG_MAX;
-			}
-			if( (t.auth_info.auth_SEP.priority < p.auth_info.auth_SEP.priority)
-				|| p.auth_info.auth_SEP.xrd_Type.empty() )
-			    p.auth_info.auth_SEP = t.auth_info.auth_SEP;
-		    }
-		}else if( is_qelement(t,NSURI_XRD "\tCanonicalID") )
-		    copy_trim_whitespace(p.auth_info.canonical_id,t.content);
-	    }
-	}
-	
-	tree.pop();
-    }
-
-    size_t openid_service_resolver_t::write(void *p,size_t s,size_t nm) {
-	if(state != state_parse)
-	    return 0;
-	/* TODO: limit total size */
-	size_t bytes = s*nm;
-	parse((const char *)p,bytes,false);
-	return bytes;
-    }
-
-    size_t openid_service_resolver_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(LOCATION_HEADER))
-			&& !strncasecmp(h,LOCATION_HEADER ":",
-			    sizeof(LOCATION_HEADER)) ) {
-		    xrds_location.assign(hv,rb);
-		}else if( (hnl >= sizeof("Content-Type"))
-			&& !strncasecmp(h,"Content-Type:",
-			    sizeof("Content-Type")) ) {
-		    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 openid_service_resolver_t::discover_service(const string& url,bool xri) {
-	CURLcode r = easy_setopt(CURLOPT_URL,url.c_str());
-	if(r)
-	    throw opkele::exception_curl(OPKELE_CP_ "failed to set curly urlie",r);
-
-	(*(expat_t*)this) = parser_create_ns();
-	set_user_data(); set_element_handler();
-	set_character_data_handler();
-	tree.clear();
-	state = state_parse;
-
-	r = easy_perform();
-	if(r && r!=CURLE_WRITE_ERROR)
-	    throw exception_curl(OPKELE_CP_ "failed to perform curly request",r);
-
-	parse(0,0,true);
-	while(!tree.empty()) pop_tag();
-    }
-
-    const openid_auth_info_t& openid_service_resolver_t::resolve(const string& id) {
-	auth_info = openid_auth_info_t();
-	html_SEP = openid_auth_SEP_t();
-
-	string::size_type fns = id.find_first_not_of(whitespace);
-	if(fns==string::npos)
-	    throw opkele::bad_input(OPKELE_CP_ "whitespace-only identity");
-	string::size_type lns = id.find_last_not_of(whitespace);
-	assert(lns!=string::npos);
-	if(!strncasecmp(
-		    id.c_str()+fns,"xri://",
-		    sizeof("xri://")-1))
-	    fns+=sizeof("xri://")-1;
-	string nid(id,fns,lns-fns+1);
-	if(nid.empty())
-	    throw opkele::bad_input(OPKELE_CP_ "nothing significant in identity");
-	if(strchr("=@+$!(",*nid.c_str())) {
-	    discover_service(
-		    xri_proxy + util::url_encode(nid) +
-		    "?_xrd_t=" STURI_OPENID10 "&_xrd_r=application/xrd+xml;sep=true",
-		    true );
-	    if(auth_info.canonical_id.empty()
-		    || auth_info.auth_SEP.xrd_Type.empty() )
-		throw opkele::failed_lookup(OPKELE_CP_ "no OpenID service for XRI found");
-	}else{
-	    const char *np = nid.c_str();
-	    if( (strncasecmp(np,"http",4) || strncmp(
-			    tolower(*(np+4))=='s'? np+5 : np+4, "://", 3))
-#ifndef NDEBUG
-		    && strncasecmp(np,"file:///",sizeof("file:///")-1)
-#endif		/* XXX: or how do I let tests work? */
-		    )
-		nid.insert(0,"http://");
-	    string::size_type fp = nid.find('#');
-	    if(fp!=string::npos) {
-		string::size_type qp = nid.find('?');
-		if(qp==string::npos || qp<fp) {
-		    nid.erase(fp);
-		}else if(qp>fp)
-		    nid.erase(fp,qp-fp);
-	    }
-	    discover_service(nid);
-	    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 canonicalized_id = util::rfc_3986_normalize_uri(eu);
-	    if(xrds_location.empty()) {
-		if(auth_info.auth_SEP.xrd_Type.empty()) {
-		    if(html_SEP.xrd_URI.empty())
-			throw opkele::failed_lookup(OPKELE_CP_ "no OpenID 1.0 service discovered");
-		    auth_info.auth_SEP = html_SEP;
-		    auth_info.auth_SEP.xrd_Type.clear(); auth_info.auth_SEP.xrd_Type.insert( STURI_OPENID10 );
-		    auth_info.canonical_id = canonicalized_id;
-		}else{
-		    if(auth_info.canonical_id.empty())
-			auth_info.canonical_id = canonicalized_id;
-		}
-	    }else{
-		discover_service(xrds_location);
-		if(auth_info.auth_SEP.xrd_Type.empty()) {
-		    if(html_SEP.xrd_URI.empty())
-			throw opkele::failed_lookup(OPKELE_CP_ "no OpenID 1.0 service discovered");
-		    auth_info.auth_SEP = html_SEP;
-		    auth_info.auth_SEP.xrd_Type.clear(); auth_info.auth_SEP.xrd_Type.insert( STURI_OPENID10 );
-		    auth_info.canonical_id = canonicalized_id;
-		}else{
-		    if(auth_info.canonical_id.empty())
-			auth_info.canonical_id = canonicalized_id;
-		}
-	    }
-	}
-	return auth_info;
-    }
-
-}
diff --git a/lib/util.cc b/lib/util.cc
index 69d37b5..83f0eef 100644
--- a/lib/util.cc
+++ b/lib/util.cc
@@ -185,21 +185,23 @@ namespace opkele {
 		     uri.begin()+ns, uri.begin()+colon+1,
 		     back_inserter(rv), ::tolower );
 	     bool s;
-	     if(rv=="http:")
-		 s = false;
-	     else if(rv=="https:")
-		 s = true;
-#ifndef NDEBUG
-	     else if(rv=="file:")
-		 s = false;
-#endif /* XXX: or try to make tests work some other way */
-	     else
-		 throw not_implemented(OPKELE_CP_ "Only http(s) URIs can be normalized here");
 	     string::size_type ul = uri.find_last_not_of(whitespace)+1;
 	     if(ul <= (colon+3))
 		 throw bad_input(OPKELE_CP_ "Unexpected end of URI being normalized encountered");
 	     if(uri[colon+1]!='/' || uri[colon+2]!='/')
 		 throw bad_input(OPKELE_CP_ "Unexpected input in URI being normalized after scheme component");
+	     if(rv=="http:")
+		 s = false;
+	     else if(rv=="https:")
+		 s = true;
+	     else{
+		 /* TODO: support more schemes.
+		  * e.g. xri. How do we normalize
+		  * xri?
+		  */
+		 rv.append(uri,colon+1,ul-colon-1);
+		 return rv;
+	     }
 	     rv += "//";
 	     string::size_type interesting = uri.find_first_of(":/#?",colon+3);
 	     if(interesting==string::npos) {
diff --git a/test/.gitignore b/test/.gitignore
index 5ce4dc9..31ae686 100644
--- a/test/.gitignore
+++ b/test/.gitignore
@@ -2,4 +2,4 @@
 /.libs
 /test
 *.o
-/openid_resolve
+/idiscover
diff --git a/test/Makefile.am b/test/Makefile.am
index 5aa87b3..13c4cd2 100644
--- a/test/Makefile.am
+++ b/test/Makefile.am
@@ -1,4 +1,4 @@
-noinst_PROGRAMS = test openid_resolve
+noinst_PROGRAMS = test idiscover
 
 DEFAULT_INCLUDES = -I${top_builddir}
 INCLUDES = -I${top_srcdir}/include/ ${KONFORKA_CFLAGS}
@@ -11,5 +11,5 @@ EXTRA_DIST=$(addsuffix .html,$(addprefix html/, \
 	unclosed-head spaced-links spaced-link-attrs 2rels \
 	))
 
-openid_resolve_SOURCES = openid_resolve.cc
-openid_resolve_LDADD = ${top_builddir}/lib/libopkele.la
+idiscover_SOURCES = idiscover.cc
+idiscover_LDADD = ${top_builddir}/lib/libopkele.la
diff --git a/test/idiscover.cc b/test/idiscover.cc
new file mode 100644
index 0000000..2abedc9
--- a/dev/null
+++ b/test/idiscover.cc
@@ -0,0 +1,49 @@
+#include <iostream>
+#include <stdexcept>
+#include <iterator>
+#include <algorithm>
+using namespace std;
+#include <opkele/exception.h>
+#include <opkele/discovery.h>
+
+template<typename _PDT>
+    ostream& operator<<(ostream& o,const opkele::xrd::priority_map<_PDT>& pm) {
+	for(typename opkele::xrd::priority_map<_PDT>::const_iterator i=pm.begin();
+		i!=pm.end();++i)
+	    o << ' ' << i->second << '[' << i->first << ']';
+	return o;
+    }
+
+ostream& operator<<(ostream& o,const opkele::xrd::service_t s) {
+    o << "{" << endl
+	<< " Type: ";
+    copy(s.types.begin(),s.types.end(),
+	    ostream_iterator<string>(o," "));
+    o << endl
+	<< " URI: " << s.uris << endl
+	<< " LocalID: " << s.local_ids << endl;
+    o << "}";
+}
+
+int main(int argc,char **argv) {
+    try {
+	if(argc<2)
+	    throw opkele::exception(OPKELE_CP_ "Please, give me something to resolve");
+	for(int a=1;a<argc;++a) {
+	    opkele::idiscovery_t discovery(argv[a]);
+	    clog
+		<< "===============================================================" << endl
+		<< "User-supplied ID: " << argv[a] << endl
+		<< "Normalized ID: " << discovery.normalized_id << endl
+		<< "Canonicalized ID: " << discovery.canonicalized_id << endl
+		<< endl
+		<< "CanonicalID: " << discovery.xrd.canonical_ids << endl
+		<< "LocalID: " << discovery.xrd.local_ids << endl
+		<< "Services: " << discovery.xrd.services << endl;
+	}
+    }catch(exception& e) {
+	cerr << "oops: " << e.what() << endl;
+	_exit(1);
+    }
+    _exit(0);
+}
diff --git a/test/openid_resolve.cc b/test/openid_resolve.cc
deleted file mode 100644
index 31c7a6a..0000000
--- a/test/openid_resolve.cc
+++ b/dev/null
@@ -1,36 +0,0 @@
-#include <iostream>
-#include <stdexcept>
-#include <iterator>
-#include <algorithm>
-using namespace std;
-#include <opkele/exception.h>
-#include <opkele/openid_service_resolver.h>
-
-int main(int argc,char **argv) {
-    try {
-	if(argc<2)
-	    throw opkele::exception(OPKELE_CP_ "Please, give me something to resolve");
-	opkele::openid_service_resolver_t resolver;
-	for(int a=1;a<argc;++a) {
-	    const opkele::openid_auth_info_t& iai = resolver.resolve(argv[a]);
-	    clog
-		<< "====================" << endl
-		<< "canonical id is " << iai.canonical_id << endl
-		<< endl
-		<< "service priority is " << iai.auth_SEP.priority << endl
-		<< "service types are " ;
-	    copy(
-		    iai.auth_SEP.xrd_Type.begin(), iai.auth_SEP.xrd_Type.end(),
-		    ostream_iterator<string>(clog," ") );
-	    clog << endl
-		<< "service URI is " << iai.auth_SEP.xrd_URI  << endl;
-	    if(!iai.auth_SEP.openid_Delegate.empty())
-		clog << "openid:Delegate is " << iai.auth_SEP.openid_Delegate << endl;
-	    clog << endl;
-	}
-    }catch(exception& e) {
-	cerr << "oops: " << e.what() << endl;
-	_exit(1);
-    }
-    _exit(0);
-}