1 files changed, 65 insertions, 59 deletions

diff --git a/lib/consumer.cc b/lib/consumer.cc
index ff5da91..df95b64 100644
--- a/lib/consumer.cc
+++ b/lib/consumer.cc
@@ -1,73 +1,111 @@
 #include <algorithm>
 #include <cassert>
 #include <opkele/util.h>
 #include <opkele/exception.h>
 #include <opkele/data.h>
 #include <opkele/consumer.h>
 #include <openssl/sha.h>
 #include <openssl/hmac.h>
 #include <curl/curl.h>
 
 #include <iostream>
 
 #include "config.h"
 
-#if defined(USE_LIBPCRECPP)
-# include <pcrecpp.h>
-#elif defined(USE_PCREPP)
-# include <pcre++.h>
-#else
- /* internal implementation won't be built */
-#endif
+#include <pcre.h>
 
 namespace opkele {
     using namespace std;
 
+    class pcre_matches_t {
+	public:
+	    int *_ov;
+	    int _s;
+
+	    pcre_matches_t() : _ov(0), _s(0) { }
+	    pcre_matches_t(int s) : _ov(0), _s(s) {
+		if(_s&1) ++_s;
+		_s += _s>>1;
+		_ov = new int[_s];
+	    }
+	    ~pcre_matches_t() throw() { if(_ov) delete[] _ov; }
+
+	    int begin(int i) const { return _ov[i<<1]; }
+	    int end(int i) const { return _ov[(i<<1)+1]; }
+	    int length(int i) const { int t=i<<1; return _ov[t+1]-_ov[t]; }
+    };
+
+    class pcre_t {
+	public:
+	    pcre *_p;
+
+	    pcre_t() : _p(0) { }
+	    pcre_t(pcre *p) : _p(p) { }
+	    pcre_t(const char *re,int opts) : _p(0) {
+		static const char *errptr; static int erroffset;
+		_p = pcre_compile(re,opts,&errptr,&erroffset,NULL);
+		if(!_p)
+		    throw internal_error(OPKELE_CP_ string("Failed to compile regexp: ")+errptr);
+	    }
+	    ~pcre_t() throw() { if(_p) (*pcre_free)(_p); }
+
+	    pcre_t& operator=(pcre *p) { if(_p) (*pcre_free)(_p); _p=p; return *this; }
+
+	    operator const pcre*(void) const { return _p; }
+	    operator pcre*(void) { return _p; }
+
+	    int exec(const string& s,pcre_matches_t& m) {
+		if(!_p)
+		    throw internal_error(OPKELE_CP_ "Trying to execute absent regexp");
+		return pcre_exec(_p,NULL,s.c_str(),s.length(),0,0,m._ov,m._s);
+	    }
+    };
+
     class curl_t {
 	public:
 	    CURL *_c;
 
 	    curl_t() : _c(0) { }
 	    curl_t(CURL *c) : _c(c) { }
 	    ~curl_t() throw() { if(_c) curl_easy_cleanup(_c); }
 
 	    curl_t& operator=(CURL *c) { if(_c) curl_easy_cleanup(_c); _c=c; return *this; }
 
 	    operator const CURL*(void) const { return _c; }
 	    operator CURL*(void) { return _c; }
     };
 
     static CURLcode curl_misc_sets(CURL* c) {
 	CURLcode r;
 	(r=curl_easy_setopt(c,CURLOPT_FOLLOWLOCATION,1))
 	|| (r=curl_easy_setopt(c,CURLOPT_MAXREDIRS,5))
 	|| (r=curl_easy_setopt(c,CURLOPT_DNS_CACHE_TIMEOUT,120))
 	|| (r=curl_easy_setopt(c,CURLOPT_DNS_USE_GLOBAL_CACHE,1))
 	|| (r=curl_easy_setopt(c,CURLOPT_USERAGENT,PACKAGE_NAME"/"PACKAGE_VERSION))
 	|| (r=curl_easy_setopt(c,CURLOPT_TIMEOUT,20))
 #ifdef	DISABLE_CURL_SSL_VERIFYHOST
 	|| (r=curl_easy_setopt(c,CURLOPT_SSL_VERIFYHOST,0))
 #endif
 #ifdef	DISABLE_CURL_SSL_VERIFYPEER
 	|| (r=curl_easy_setopt(c,CURLOPT_SSL_VERIFYPEER,0))
 #endif
 	;
 	return r;
     }
 
     static size_t _curl_tostring(void *ptr,size_t size,size_t nmemb,void *stream) {
 	string *str = (string*)stream;
 	size_t bytes = size*nmemb;
 	size_t get = min(16384-str->length(),bytes);
 	str->append((const char*)ptr,get);
 	return get;
     }
 
     assoc_t consumer_t::associate(const string& server) {
 	util::dh_t dh = DH_new();
 	if(!dh)
 	    throw exception_openssl(OPKELE_CP_ "failed to DH_new()");
 	dh->p = util::dec_to_bignum(data::_default_p);
 	dh->g = util::dec_to_bignum(data::_default_g);
 	if(!DH_generate_key(dh))
 	    throw exception_openssl(OPKELE_CP_ "failed to DH_generate_key()");
@@ -223,175 +261,143 @@ namespace opkele {
 	    }catch(failed_check_authentication& fca) {
 		throw id_res_failed(OPKELE_CP_ "failed to check_authentication()");
 	    }
 	}
 	if(ext) ext->id_res_hook(pin,ps,identity);
     }
 
     void consumer_t::check_authentication(const string& server,const params_t& p) {
 	string request = "openid.mode=check_authentication";
 	for(params_t::const_iterator i=p.begin();i!=p.end();++i) {
 	    if(i->first!="openid.mode") {
 		request += '&';
 		request += i->first;
 		request += '=';
 		request += util::url_encode(i->second);
 	    }
 	}
 	curl_t curl = curl_easy_init();
 	if(!curl)
 	    throw exception_curl(OPKELE_CP_ "failed to curl_easy_init()");
 	string response;
 	CURLcode r;
 	(r=curl_misc_sets(curl))
 	|| (r=curl_easy_setopt(curl,CURLOPT_URL,server.c_str()))
 	|| (r=curl_easy_setopt(curl,CURLOPT_POST,1))
 	|| (r=curl_easy_setopt(curl,CURLOPT_POSTFIELDS,request.data()))
 	|| (r=curl_easy_setopt(curl,CURLOPT_POSTFIELDSIZE,request.length()))
 	|| (r=curl_easy_setopt(curl,CURLOPT_WRITEFUNCTION,_curl_tostring))
 	|| (r=curl_easy_setopt(curl,CURLOPT_WRITEDATA,&response))
 	;
 	if(r)
 	    throw exception_curl(OPKELE_CP_ "failed to curl_easy_setopt()",r);
 	if(r=curl_easy_perform(curl))
 	    throw exception_curl(OPKELE_CP_ "failed to curl_easy_perform()",r);
 	params_t pp; pp.parse_keyvalues(response);
 	if(pp.has_param("invalidate_handle"))
 	    invalidate_assoc(server,pp.get_param("invalidate_handle"));
 	if(pp.has_param("is_valid")) {
 	    if(pp.get_param("is_valid")=="true")
 		return;
 	}else if(pp.has_param("lifetime")) {
 	    if(util::string_to_long(pp.get_param("lifetime")))
 		return;
 	}
 	throw failed_check_authentication(OPKELE_CP_ "failed to verify response");
     }
 
     void consumer_t::retrieve_links(const string& url,string& server,string& delegate) {
-#if defined(USE_LIBPCRECPP) || defined(USE_PCREPP)
 	server.erase();
 	delegate.erase();
 	curl_t curl = curl_easy_init();
 	if(!curl)
 	    throw exception_curl(OPKELE_CP_ "failed to curl_easy_init()");
 	string html;
 	CURLcode r;
 	(r=curl_misc_sets(curl))
 	|| (r=curl_easy_setopt(curl,CURLOPT_URL,url.c_str()))
 	|| (r=curl_easy_setopt(curl,CURLOPT_WRITEFUNCTION,_curl_tostring))
 	|| (r=curl_easy_setopt(curl,CURLOPT_WRITEDATA,&html))
 	;
 	if(r)
 	    throw exception_curl(OPKELE_CP_ "failed to curl_easy_setopt()",r);
 	r = curl_easy_perform(curl);
 	if(r && r!=CURLE_WRITE_ERROR)
 	    throw exception_curl(OPKELE_CP_ "failed to curl_easy_perform()",r);
-	// strip out everything past body
-	static const char *re_hdre = "<\\s*head[^>]*>",
+	static const char *re_bre = "<\\s*body\\b", *re_hdre = "<\\s*head[^>]*>",
 		     *re_lre = "<\\s*link\\b([^>]+)>",
 		     *re_rre = "\\brel\\s*=\\s*['\"]\\s*([^'\"\\s]+)\\s*['\"]",
 		     *re_hre = "\\bhref\\s*=\\s*['\"]\\s*([^'\"\\s]+)\\s*['\"]";
-#if defined(USE_LIBPCRECPP)
-	static pcrecpp::RE_Options ro(PCRE_CASELESS|PCRE_DOTALL);
-	static pcrecpp::RE
-	    bre("<body\\b.*",ro), hdre(re_hdre,ro),
-	    lre(re_lre,ro), rre(re_rre), hre(re_hre,ro);
-	bre.Replace("",&html);
-	pcrecpp::StringPiece hpiece(html);
-	if(!hdre.FindAndConsume(&hpiece))
-	    throw bad_input(OPKELE_CP_ "failed to find head");
-	string attrs;
-	while(lre.FindAndConsume(&hpiece,&attrs)) {
-	    pcrecpp::StringPiece rel, href;
-	    if(!(rre.PartialMatch(attrs,&rel) && hre.PartialMatch(attrs,&href)))
+	pcre_matches_t m1(3), m2(3);
+	pcre_t bre(re_bre,PCRE_CASELESS);
+	if(bre.exec(html,m1)>0)
+	    html.erase(m1.begin(0));
+	pcre_t hdre(re_hdre,PCRE_CASELESS);
+	if(hdre.exec(html,m1)<=0)
+	    throw bad_input(OPKELE_CP_ "failed to find <head>");
+	html.erase(0,m1.end(0)+1);
+	pcre_t lre(re_lre,PCRE_CASELESS), rre(re_rre,PCRE_CASELESS), hre(re_hre,PCRE_CASELESS);
+	while(lre.exec(html,m1)>=2) {
+	    string attrs(html,m1.begin(1),m1.length(1));
+	    html.erase(0,m1.end(0)+1);
+	    if(!( rre.exec(attrs,m1)>=2 && hre.exec(attrs,m2)>=2 ))
 		continue;
+	    string rel(attrs,m1.begin(1),m1.length(1));
+	    string href(attrs,m2.begin(1),m2.length(1));
 	    if(rel=="openid.server") {
-		href.CopyToString(&server);
-		if(!delegate.empty())
-		    break;
+		server = href;
+		if(!delegate.empty()) break;
 	    }else if(rel=="openid.delegate") {
-		href.CopyToString(&delegate);
-		if(!server.empty())
-		    break;
-	    }
+		delegate = href;
+		if(!server.empty()) break;
 	}
-#elif defined(USE_PCREPP)
-	pcrepp::Pcre bre("<body\\b",PCRE_CASELESS);
-	if(bre.search(html))
-	    html.erase(bre.get_match_start());
-	pcrepp::Pcre hdre(re_hdre,PCRE_CASELESS);
-	if(!hdre.search(html))
-	    throw bad_input(OPKELE_CP_ "failed to find head");
-	html.erase(0,hdre.get_match_end()+1);
-	pcrepp::Pcre lre(re_lre,PCRE_CASELESS), rre(re_rre,PCRE_CASELESS), hre(re_hre,PCRE_CASELESS);
-	while(lre.search(html)) {
-	    string attrs = lre[0];
-	    html.erase(0,lre.get_match_end()+1);
-	    if(!(rre.search(attrs)&&hre.search(attrs)))
-		continue;
-	    if(rre[0]=="openid.server") {
-		server = hre[0];
-		if(!delegate.empty())
-		    break;
-	    }else if(rre[0]=="openid.delegate") {
-		delegate = hre[0];
-		if(!server.empty())
-		    break;
 	    }
-	}
-#else
-	#error "I must have gone crazy"
-#endif
 	if(server.empty())
 	    throw failed_assertion(OPKELE_CP_ "The location has no openid.server declaration");
-#else /* none of the RE bindings enabled */
-	throw not_implemented(OPKELE_CP_ "No internal implementation of retrieve_links were provided at compile-time");
-#endif
     }
 
     assoc_t consumer_t::find_assoc(const string& server) {
 	throw failed_lookup(OPKELE_CP_ "no find_assoc() provided");
     }
 
     string consumer_t::normalize(const string& url) {
 	string rv = url;
 	// strip leading and trailing spaces
 	string::size_type i = rv.find_first_not_of(" \t\r\n");
 	if(i==string::npos)
 	    throw bad_input(OPKELE_CP_ "empty URL");
 	if(i)
 	    rv.erase(0,i);
 	i = rv.find_last_not_of(" \t\r\n");
 	assert(i!=string::npos);
 	if(i<(rv.length()-1))
 	    rv.erase(i+1);
 	// add missing http://
 	i = rv.find("://");
 	if(i==string::npos) { // primitive. but do we need more?
 	    rv.insert(0,"http://");
 	    i = sizeof("http://")-1;
 	}else{
 	    i += sizeof("://")-1;
 	}
 	string::size_type qm = rv.find('?',i);
 	string::size_type sl = rv.find('/',i);
 	if(qm!=string::npos) {
 	    if(sl==string::npos || sl>qm)
 		rv.insert(qm,1,'/');
 	}else{
 	    if(sl==string::npos)
 		rv += '/';
 	}
 	return rv;
     }
 
     string consumer_t::canonicalize(const string& url) {
 	string rv = normalize(url);
 	curl_t curl = curl_easy_init();
 	if(!curl)
 	    throw exception_curl(OPKELE_CP_ "failed to curl_easy_init()");
 	string html;
 	CURLcode r;
 	(r=curl_misc_sets(curl))
 	|| (r=curl_easy_setopt(curl,CURLOPT_URL,rv.c_str()))
 	|| (r=curl_easy_setopt(curl,CURLOPT_NOBODY,1))