first running lua code with URL normalization, cleanup needed..

6 files changed, 81 insertions, 4 deletions

diff --git a/src/crawl/GNUmakefile b/src/crawl/GNUmakefile
index f664f0d..5a9051a 100755
--- a/src/crawl/GNUmakefile
+++ b/src/crawl/GNUmakefile
@@ -26,13 +26,22 @@ INCLUDE_LIBS = \
 # openssl
 ifeq ($(WITH_SSL),1)
 
-INCLUDE_CFLAGS += \
+INCLUDE_CXXFLAGS += \
 	-DWITH_SSL
 
 INCLUDE_LIBS += \
 	$(OPENSSL_LIBS)
 endif
 
+ifeq ($(WITH_LUA),1)
+INCLUDE_CXXFLAGS += \
+	-DWITH_LUA
+
+INCLUDE_DIRS += \
+	-I$(TOPDIR)/src/libcrawler \
+	$(TOLUA_INCLUDES)
+endif
+
 CPP_OBJS = \
 
 CPP_BINS = \
diff --git a/src/crawl/crawl.conf b/src/crawl/crawl.conf
index 922ae42..5d1c380 100644
--- a/src/crawl/crawl.conf
+++ b/src/crawl/crawl.conf
@@ -1,8 +1,8 @@
 local normalizer = GoogleURLNormalizer:new( )
 local baseUrl = normalizer:parseUrl( "http://www.base.com" )
---io.write( "base URL is: " .. baseUrl.str( ) )
+io.write( "base URL is: " .. baseUrl:str( ) .. "\n" )
 local url = normalizer:normalize( baseUrl, "/relativedir/relativefile.html" )
---io.write( "URL is: " .. url.str( ) )
+io.write( "URL is: " .. url:str( ) .. "\n" )
 
 -- global setting
 
diff --git a/src/crawl/crawl.cpp b/src/crawl/crawl.cpp
index 08d3dbf..3a8fdff 100755
--- a/src/crawl/crawl.cpp
+++ b/src/crawl/crawl.cpp
@@ -55,6 +55,12 @@ BOOL WINAPI termHandler( DWORD ctrlType )
 	
 #endif
 
+#ifdef WITH_LUA
+// TODO: should be in the laoding function of libcrawl
+#include "tolua.h"
+#include "URLLua.hpp"
+#endif
+
 static int counter = 0;
 
 int main( int /* argc */, char *argv[] )
@@ -281,6 +287,11 @@ int main( int /* argc */, char *argv[] )
 
 		LOG( logNOTICE ) << "Crawler stopped.. normal shutdown..";
 
+#ifdef WITH_LUA
+		// TODO: should be in the laoding function of libcrawl
+		tolua_URL_open( luaVm.handle( ) );
+#endif
+
 		luaVm.executeMain( );
 		//luaVm.dumpState( );
 			
diff --git a/src/libcrawler/GNUmakefile b/src/libcrawler/GNUmakefile
index e28b916..d546058 100755
--- a/src/libcrawler/GNUmakefile
+++ b/src/libcrawler/GNUmakefile
@@ -8,14 +8,31 @@ INCLUDE_CPPFLAGS = \
 
 INCLUDE_LDFLAGS = \
 
+ifeq ($(WITH_LUA),1)
+INCLUDE_LDFLAGS += \
+	$(TOLUA_LDFLAGS)
+endif
+
 INCLUDE_DIRS = \
 	-I. \
 	-I$(TOPDIR)/include/logger \
 	-I$(TOPDIR)/include/util \
 	-I$(TOPDIR)/include/crawler
 
+ifeq ($(WITH_LUA),1)
+INCLUDE_DIRS += \
+	-I$(TOPDIR)/include/luaglue \
+	$(TOLUA_INCLUDES)
+endif
+
 INCLUDE_LIBS = \
 
+ifeq ($(WITH_LUA),1)
+INCLUDE_LIBS += \
+	-llua \
+	$(TOLUA_LIBS)
+endif
+
 STATIC_LIB = libcrawler.a
 
 DYNAMIC_LIB = libcrawler.so
@@ -28,8 +45,16 @@ CPP_OBJS = \
 	MIMEType.o \
 	SpoolRewindInputStream.o
 
+ifeq ($(WITH_LUA),1)
+CPP_OBJS += \
+	URLLua.o
+endif
+
 -include $(TOPDIR)/makefiles/gmake/sub.mk
 
+URLLua.cpp: URL.pkg
+	$(TOLUA) -H URLLua.hpp -o URLLua.cpp URL.pkg
+
 local_all:
 
 local_clean:
diff --git a/src/libcrawler/URL.pkg b/src/libcrawler/URL.pkg
new file mode 100644
index 0000000..d27288d
--- /dev/null
+++ b/src/libcrawler/URL.pkg
@@ -0,0 +1,32 @@
+$#include "URL.hpp"
+
+$using namespace std;
+
+class URL
+{
+	URL( ) {}
+		
+	URL( const URL& url ) {}
+		
+	URL( const std::string _protocol, const std::string _host, const unsigned short _port, const std::string _path, const std::string _query, const std::string _fragment ) {}
+				
+	const string protocol( ) const {}
+		
+	const string host( ) const {}
+				
+	unsigned short port( ) const {}
+		
+	const string path( ) const {}
+		
+	const string query( ) const {}
+		
+	std::string fragment( ) const {}
+		
+	std::string str( ) const {}
+		
+	static URL Null;
+	
+	bool operator==( const URL &other ) const {}
+	
+	bool operator<( const URL &other ) const {}
+};
diff --git a/src/modules/urlnormalizer/googleurl/GoogleURLNormalizer.pkg b/src/modules/urlnormalizer/googleurl/GoogleURLNormalizer.pkg
index ca62fe3..3af1f5f 100644
--- a/src/modules/urlnormalizer/googleurl/GoogleURLNormalizer.pkg
+++ b/src/modules/urlnormalizer/googleurl/GoogleURLNormalizer.pkg
@@ -8,7 +8,7 @@ class GoogleURLNormalizer : public URLNormalizer
 		
 	virtual ~GoogleURLNormalizer( ) { }
 		
-	virtual URL parseUrl( string s );
+	virtual URL parseUrl( const string s );
 	
 	virtual URL normalize( const URL url, const string s );
 };