added support for crappy libpcre c++ bindings

Added optional support for crappy google's c++ bindings. The bindings are
ridiculously googley, so I think, taking into account the amount of pcre
involved, I'll soon drop requirement for any bindings and will use libpcre
directly.

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

3 files changed, 92 insertions, 16 deletions

diff --git a/acinclude.m4 b/acinclude.m4
index 349d3cf..532f978 100644
--- a/acinclude.m4
+++ b/acinclude.m4
@@ -63,50 +63,50 @@ AC_DEFUN([AC_WITH_DOXYGEN],[
 
 dnl AC_WITH_DOT([ACTION-IF-FOUND[,ACTION-IF-NOT-FOUND]])
 dnl Outputs:
 dnl AC_SUBST: DOT HAVE_DOT
 dnl AM_CONDITIONAL: HAVE_DOT
 AC_DEFUN([AC_WITH_DOT],[
  HAVE_DOT="no"
  AC_PATH_PROG([DOT],[dot],[false])
  if test "${DOT}" = "false" ; then
   ifelse([$2], , :, [$2])
  else
   HAVE_DOT="yes"
   AC_SUBST([DOT])
   $1
  fi
 AC_SUBST([HAVE_DOT])
  AM_CONDITIONAL([HAVE_DOT],[test "${HAVE_DOT}" = "yes"])
 ])
 
 dnl AC_WITH_PCRE([ACTION-IF-FOUND[,ACTION-IF-NOT-FOUND]])
 dnl Outputs:
 dnl AC_SUBST: PCRE_CONFIG PCRE_PREFIX PCRE_EXEC_PREFIX
 dnl           PCRE_VERSION PCRE_CFLAGS PCRE_LIBS
 dnl           PCRE_LIBS_POSIX PCRE_CFLAGS_POSIX
-dnl AM_CONDITIONAL: HAVE_PCRE
 dnl AC_DEFINE: HAVE_PCRE PCRE_VERSION
+dnl env: HAVE_PCRE=yes|no
 AC_DEFUN([AC_WITH_PCRE],[
  HAVE_PCRE="no"
  PCRE_CONFIG=""
  PCRE_PREFIX=""
  PCRE_EXEC_PREFIX=""
  PCRE_VERSION=""
  PCRE_CFLAGS=""
  PCRE_LIBS=""
  PCRE_LOCATIONS="${PATH}:/usr/local/bin:/usr/bin"
  test -z "$WANT_PCRE" && WANT_PCRE=""
  AC_ARG_WITH([pcre],
   AC_HELP_STRING([--with-pcre=location],[Look for pcre in specified locations]),
   [
    if test "${withval}" = "no" ; then
     WANT_PCRE="no"
    else
     if test -x "${withval}" ; then
      PCRE_CONFIG="${withval}"
     elif test -x "${withval}/pcre-config" ; then
      PCRE_CONFIG="${withval}/pcre-config"
     elif test -x "${withval}/bin/pcre-config" ; then
      PCRE_CONFIG="${withval}/bin/pcre-config"
     fi
    fi
@@ -121,57 +121,56 @@ AC_DEFUN([AC_WITH_PCRE],[
     ifelse([$2], , :, [$2])
    else
     HAVE_PCRE="yes"
     PCRE_PREFIX="`${PCRE_CONFIG} --prefix`"
     PCRE_EXEC_PREFIX="`${PCRE_CONFIG} --exec-prefix`"
     PCRE_VERSION="`${PCRE_CONFIG} --version`"
     PCRE_CFLAGS="`${PCRE_CONFIG} --cflags`"
     PCRE_LIBS="`${PCRE_CONFIG} --libs`"
     PCRE_CFLAGS_POSIX="`${PCRE_CONFIG} --cflags-posix`"
     PCRE_LIBS_POSIX="`${PCRE_CONFIG} --libs-posix`"
     AC_SUBST([PCRE_CONFIG])
     AC_SUBST([PCRE_PREFIX])
     AC_SUBST([PCRE_EXEC_PREFIX])
     AC_SUBST([PCRE_VERSION])
     AC_SUBST([PCRE_CFLAGS])
     AC_SUBST([PCRE_LIBS])
     AC_SUBST([PCRE_CFLAGS_POSIX])
     AC_SUBST([PCRE_LIBS_POSIX])
     AC_DEFINE([HAVE_PCRE],,[pcre support])
     AC_DEFINE_UNQUOTED([PCRE_VERSION],["${PCRE_VERSION}"],[pcre version])
     $1
    fi
   fi
  fi
- AM_CONDITIONAL([HAVE_PCRE],[test "${HAVE_PCRE}" = "yes"])
 ])
 
 dnl AC_WITH_PCREPP([ACTION-IF-FOUND[,ACTION-IF-NOT-FOUND]])
 dnl Outputs:
 dnl AC_SUBST: PCREPP_CONFIG PCREPP_PREFIX PCREPP_EXEC_PREFIX
 dnl           PCREPP_VERSION PCREPP_CFLAGS PCREPP_LIBS
-dnl AM_CONDITIONAL: HAVE_PCREPP
 dnl AC_DEFINE: HAVE_PCREPP PCREPP_VERSION
+dnl env: HAVE_PCREPP=yes|no
 AC_DEFUN([AC_WITH_PCREPP],[
  HAVE_PCREPP="no"
  PCREPP_CONFIG=""
  PCREPP_PREFIX=""
  PCREPP_EXEC_PREFIX=""
  PCREPP_VERSION=""
  PCREPP_CFLAGS=""
  PCREPP_LIBS=""
  PCREPP_LOCATIONS="${PATH}:/usr/local/bin:/usr/bin"
  test -z "$WANT_PCREPP" && WANT_PCREPP=""
  AC_ARG_WITH([pcrepp],
   AC_HELP_STRING([--with-pcrepp=location],[Look for pcre++ in specified locations]),
   [
    if test "${withval}" = "no" ; then
     WANT_PCREPP="no"
    else
     if test -x "${withval}" ; then
      PCREPP_CONFIG="${withval}"
     elif test -x "${withval}/pcre++-config" ; then
      PCREPP_CONFIG="${withval}/pcre++-config"
     elif test -x "${withval}/bin/pcre++-config" ; then
      PCREPP_CONFIG="${withval}/bin/pcre++-config"
     fi
    fi
@@ -186,28 +185,27 @@ AC_DEFUN([AC_WITH_PCREPP],[
    if test -z "${PCREPP_CONFIG}" ; then
     AC_PATH_PROG([PCREPP_CONFIG],[pcre++-config],false,[${PCREPP_LOCATIONS}])
     if test "${PCREPP_CONFIG}" = "false" ; then
      ifelse([$2], , :, [$2])
     else
      HAVE_PCREPP="yes"
      PCREPP_PREFIX="`${PCREPP_CONFIG} --prefix`"
      PCREPP_EXEC_PREFIX="`${PCREPP_CONFIG} --exec-prefix`"
      PCREPP_VERSION="`${PCREPP_CONFIG} --version`"
      PCREPP_CFLAGS="`${PCREPP_CONFIG} --cflags` ${PCRE_CFLAGS}"
      PCREPP_LIBS="`${PCREPP_CONFIG} --libs` ${PCRE_LIBS}"
      AC_SUBST([PCREPP_CONFIG])
      AC_SUBST([PCREPP_PREFIX])
      AC_SUBST([PCREPP_EXEC_PREFIX])
      AC_SUBST([PCREPP_VERSION])
      AC_SUBST([PCREPP_CFLAGS])
      AC_SUBST([PCREPP_LIBS])
      AC_DEFINE([HAVE_PCREPP],,[pcre++ support])
      AC_DEFINE_UNQUOTED([PCREPP_VERSION],["${PCREPP_VERSION}"],[pcre++ version])
      $1
     fi
    fi
   fi
  fi
- AM_CONDITIONAL([HAVE_PCREPP],[test "${HAVE_PCREPP}" = "yes"])
 ])
 
 m4_include([acinclude.d/libcurl.m4])
diff --git a/configure.ac b/configure.ac
index 2094273..00c6bc4 100644
--- a/configure.ac
+++ b/configure.ac
@@ -31,57 +31,94 @@ if test "${WANT_KONFORKA}" = "yes" ; then
   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/])
 ])
-AC_WITH_PCRE([
-  AC_WITH_PCREPP(,[
-   AC_MSG_ERROR([no pcre++ library found. get one at http://www.daemon.de/PCRE])
-  ])
- ],[
-  AC_MSG_ERROR([no pcre library found. get one at http://www.pcre.org/])
+
+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
+   ])
+  ],[
+   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([
diff --git a/lib/consumer.cc b/lib/consumer.cc
index 8f66688..299b3bc 100644
--- a/lib/consumer.cc
+++ b/lib/consumer.cc
@@ -1,39 +1,46 @@
 #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 <pcre++.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
+
 namespace opkele {
     using namespace std;
 
     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))
@@ -240,93 +247,127 @@ namespace opkele {
 	|| (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);
-	pcrepp::Pcre bre("<body\\b",PCRE_CASELESS);
 	// strip out everything past body
+	static const char *re_hdre = "<head[^>]*>",
+		     *re_lre = "<link\\b([^>]+)>",
+		     *re_rre = "\\brel=['\"]([^'\"]+)['\"]",
+		     *re_hre = "\\bhref=['\"]([^'\"]+)['\"]";
+#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)))
+		continue;
+	    if(rel=="openid.server") {
+		href.CopyToString(&server);
+		if(!delegate.empty())
+		    break;
+	    }else if(rel=="openid.delegate") {
+		href.CopyToString(&delegate);
+		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("<head[^>]*>",PCRE_CASELESS);
+	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("<link\\b([^>]+)>",PCRE_CASELESS),
-	    rre("\\brel=['\"]([^'\"]+)['\"]",PCRE_CASELESS),
-	    hre("\\bhref=['\"]([^'\"]+)['\"]",PCRE_CASELESS);
+	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{