got rid of all pcre wrappers altogether

although all claims against pcre++ make absolutely no sense. But for now I'll
live with no wrappers at all. I may reconsider it when implementing more
sophisticated discovery mechanisms for upcoming extensions.

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

5 files changed, 79 insertions, 105 deletions

diff --git a/configure.ac b/configure.ac
index 00c6bc4..53e22ba 100644
--- a/configure.ac
+++ b/configure.ac
@@ -8,125 +8,84 @@ AC_PROG_CXX
 AC_PROG_CC
 AC_PROG_LIBTOOL
 
 AC_HEADER_STDC
 
 AC_PATH_PROG([XSLTPROC],[xsltproc],[true])
 AC_WITH_PKGCONFIG
 
 PKG_CHECK_MODULES([OPENSSL],[openssl],,[
  AC_MSG_ERROR([no openssl library found. get one from http://www.openssl.org/])
 ])
 
 WANT_KONFORKA="yes"
 AC_ARG_ENABLE([konforka],
  AC_HELP_STRING([--disable-konforka],[do not use konforka library (default: use if found)]),
  [
   test "${enableval}" = "no" && WANT_KONFORKA="no"
  ]
 )
 if test "${WANT_KONFORKA}" = "yes" ; then
  PKG_CHECK_MODULES([KONFORKA],[konforka],[
   AC_SUBST([KONFORKA_CFLAGS])
   AC_SUBST([KONFORKA_LIBS])
   AC_DEFINE([HAVE_KONFORKA],,[defined in presence of konforka library])
   AC_DEFINE([OPKELE_HAVE_KONFORKA],,[defined in presence of konforka library])
   AC_SUBST([KONFORKA_KONFORKA],[konforka])
  ],[true])
 fi
 
 WANT_DOXYGEN="yes"
 AC_ARG_ENABLE([doxygen],
  AC_HELP_STRING([--disable-doxygen],[do not generate documentation]),
  [
   test "${enableval}" = "no" && WANT_DOXYGEN="no"
  ]
 )
 if test "${WANT_DOXYGEN}" = "yes" ; then
  AC_WITH_DOXYGEN
  AC_WITH_DOT
 else
  AM_CONDITIONAL([HAVE_DOXYGEN],[false])
  AM_CONDITIONAL([HAVE_DOT],[false])
 fi
 
 LIBCURL_CHECK_CONFIG(,,,[
  AC_MSG_ERROR([no required libcurl library. get one from http://curl.haxx.se/])
 ])
 
-want_pcre_impl=""
-AC_ARG_WITH([pcre-bindings],
- AC_HELP_STRING([--with-pcre-bindings=(pcrepp|none|libpcrecpp)],[Specify which pcre c++ bindings to use. 'pcrepp' stands for quite sensible library, found at http://www.daemon.de/PCRE/, 'libcrecpp' makes use of crappy bindings by google and 'none' disables internal implementation of OP discovery]),
- [
-  case "$withval" in
-   pcrepp) want_pcre_impl="pcrepp" ;;
-   libpcrecpp) want_pcre_impl="libpcrecpp" ;;
-   none) want_pcre_impl="none";;
-   *) AC_MSG_ERROR([I'm not sure I understand what do you want for a pcre c++ bindings]) ;;
-  esac
- ]
-)
-
-found_pcre_impl=""
-
-if test -z "$want_pcre_impl" -o "$want_pcre_impl" = "pcrepp" ; then
- AC_WITH_PCRE([
-   AC_WITH_PCREPP([
-    found_pcre_impl=pcrepp
+PKG_CHECK_MODULES([PCRE],[libpcre],,[
+ AC_MSG_ERROR([no libpcre found, go get it at http://www.pcre.org/])
    ])
-  ],[
-   AC_MSG_ERROR([no pcre library found. get one at http://www.pcre.org/])
-  ]
- )
- test "$want_pcre_impl,$found_pcre_impl" = "pcrepp," && AC_MSG_ERROR([no pcre++ library found. get one at http://www.daemon.de/PCRE])
-fi
-
-if test "$found_pcre_impl,$want_pcre_impl" = "," -o "$want_pcre_impl" = "libpcrecpp" ; then
- test -z "$want_pcre_impl" || AC_MSG_NOTICE([You want to use crappy libpcre c++ bindings])
- PKG_CHECK_MODULES([LIBPCRECPP],[libpcrecpp],[
-   found_pcre_impl=libpcrecpp
-   CXXFLAGS="$CXXFLAGS $LIBPCRECPP_CFLAGS"
-   LIBS="$LIBS $LIBPCRECPP_LIBS"
-  ],[
-   test -z "$want_pcre_impl" || AC_MSG_ERROR([no libpcre c++ bindings found. why would you want it if you don't have it installed?])
-  ]
- )
-fi
-test "$want_pcre_impl,$found_pcre_impl" = "," && AC_MSG_ERROR([no pcre c++ bindings found, use --with-pcre-bindings=none to disable code that makes use of it])
-
-case "$found_pcre_impl" in
- pcrepp) AC_DEFINE([USE_PCREPP],,[defined if pcre++ is to be used]) ;;
- libpcrecpp) AC_DEFINE([USE_LIBPCRECPP],,[defined if crappy google bindings are to be used]) ;;
-esac
 
 curl_ssl_verify_host="true"
 AC_ARG_ENABLE([ssl-verify-host],
  AC_HELP_STRING([--disable-ssl-verify-host],[disable cURL cert/host relationships verification]),
  [ test "${enableval}" = "no" && curl_ssl_verify_host="false" ]
 )
 ${curl_ssl_verify_host} || AC_DEFINE([DISABLE_CURL_SSL_VERIFYHOST],,[defined if cURL is not to verify cert/host])
 
 curl_ssl_verify_peer="true"
 AC_ARG_ENABLE([ssl-verify-peer],
  AC_HELP_STRING([--disable-ssl-verify-peer],[disable cURL cert validity verification]),
  [ test "${enableval}" = "no" && curl_ssl_verify_peer="false" ]
 )
 ${curl_ssl_verify_peer} || AC_DEFINE([DISABLE_CURL_SSL_VERIFYPEER],,[defined if cURL is not to verify cert validity])
 
 postels_law=true
 AC_ARG_ENABLE([postels-law],
  AC_HELP_STRING([--disable-postels-law],[Be strict, do not adhere to Postel's Law ("be conservative in what you do, be liberal in what you accept from others", RFC 793)]),
  [ test "${enableval}" = "no" && postels_law=false ]
 )
 $postels_law && AC_DEFINE([POSTELS_LAW],,[defined if we want to adhere to Postel's Law])
 
 AC_DEFINE_UNQUOTED([OPKELE_SRC_DIR],["$PWD"],[source directory])
 
 AC_CONFIG_FILES([
  Makefile
   libopkele.pc
   Doxyfile
  include/Makefile
  lib/Makefile
  test/Makefile
 ])
 AC_OUTPUT
diff --git a/include/opkele/exception.h b/include/opkele/exception.h
index 0150e6b..753a818 100644
--- a/include/opkele/exception.h
+++ b/include/opkele/exception.h
@@ -167,51 +167,60 @@ namespace opkele {
 	public:
 	    id_res_mismatch(OPKELE_E_PARS)
 		: id_res_failed(OPKELE_E_CONS) { }
     };
 
     /**
      * openssl malfunction occured
      */
     class exception_openssl : public exception {
 	public:
 	    unsigned long _error;
 	    string _ssl_string;
 	    exception_openssl(OPKELE_E_PARS);
 	    ~exception_openssl() throw() { }
     };
 
     /**
      * network operation related error occured
      */
     class exception_network : public exception {
 	public:
 	    exception_network(OPKELE_E_PARS)
 		: exception(OPKELE_E_CONS) { }
     };
 
     /**
      * network operation related error occured, specifically, related to
      * libcurl
      */
     class exception_curl : public exception_network {
 	public:
 	    CURLcode _error;
 	    string _curl_string;
 	    exception_curl(OPKELE_E_PARS);
 	    exception_curl(OPKELE_E_PARS,CURLcode e);
 	    ~exception_curl() throw() { }
     };
 
     /**
      * not implemented (think pure virtual) member function executed, signfies
      * programmer error
      */
     class not_implemented : public exception {
 	public:
 	    not_implemented(OPKELE_E_PARS)
 		: exception(OPKELE_E_CONS) { }
     };
 
+    /**
+     * internal error, indicates internal libopkele problem
+     */
+    class internal_error : public exception {
+	public:
+	    internal_error(OPKELE_E_PARS)
+		: exception(OPKELE_E_CONS) { }
+    };
+
 }
 
 #endif /* __OPKELE_EXCEPTION_H */
diff --git a/lib/Makefile.am b/lib/Makefile.am
index a68faf6..b008a52 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -1,27 +1,27 @@
 lib_LTLIBRARIES = libopkele.la
 
 INCLUDES = \
 	   -I${top_srcdir}/include/ \
 	   ${KONFORKA_CFLAGS} \
 	   ${OPENSSL_CFLAGS} \
 	   ${LIBCURL_CPPFLAGS} \
-	   ${PCREPP_CFLAGS}
+	   ${PCRE_CFLAGS}
 libopkele_la_LIBADD = \
 	${LIBCURL} \
-	${PCREPP_LIBS} \
+	${PCRE_LIBS} \
 	${OPENSSL_LIBS} \
 	${KONFORKA_LIBS}
 
 libopkele_la_SOURCES = \
 		       params.cc \
 		       util.cc \
 		       server.cc \
 		       secret.cc \
 		       data.cc \
 		       consumer.cc \
 		       exception.cc \
 		       extension.cc \
 		       sreg.cc \
 		       extension_chain.cc
 libopkele_la_LDFLAGS = \
 	-version-info 2:0:0
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))
diff --git a/libopkele.pc.in b/libopkele.pc.in
index 286720e..5f8684b 100644
--- a/libopkele.pc.in
+++ b/libopkele.pc.in
@@ -1,11 +1,11 @@
 prefix=@prefix@
 exec_prefix=@exec_prefix@
 libdir=@libdir@
 includedir=@includedir@
 
 Name: libopkele
 Description: C++ implementation of OpenID protocol
 Version: @VERSION@
-Requires: openssl @KONFORKA_KONFORKA@
+Requires: openssl libpcre @KONFORKA_KONFORKA@
 Cflags: -I${includedir} @LIBCURL_CPPFLAGS@ @PCREPP_CFLAGS@
 Libs: -L${libdir} -lopkele @LIBCURL@ @PCREPP_LIBS@