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
@@ -84,8 +84,8 @@ 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=""
@@ -142,15 +142,14 @@ AC_DEFUN([AC_WITH_PCRE],[
    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=""
@@ -207,7 +206,6 @@ AC_DEFUN([AC_WITH_PCREPP],[
    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
@@ -52,15 +52,52 @@ 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]),
diff --git a/lib/consumer.cc b/lib/consumer.cc
index 8f66688..299b3bc 100644
--- a/lib/consumer.cc
+++ b/lib/consumer.cc
@@ -7,12 +7,19 @@
 #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;
 
@@ -261,6 +268,7 @@ namespace opkele {
     }
 
     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();
@@ -278,17 +286,44 @@ namespace opkele {
         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);
@@ -304,8 +339,14 @@ namespace opkele {
                     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) {