google url normalization works on Windows, test1 must be improved:

there are linking problems (/DSHARED in *.lib normalization libraries
produce clashing registry structures)

6 files changed, 31 insertions, 13 deletions

diff --git a/Makefile.W32 b/Makefile.W32
index a22e0e3..b1100cf 100755
--- a/Makefile.W32
+++ b/Makefile.W32
@@ -1,7 +1,7 @@
 TOPDIR = .
 
 #SUBDIRS = libfetch streamhtmlparser googleurl sqlite3 src tests
-SUBDIRS = utils sqlite3 src tests
+SUBDIRS = utils googleurl sqlite3 src tests
 
 PACKAGE_NAME = CrawlingWolf
 PACKAGE_VERSION = 0.0.1
diff --git a/googleurl/Makefile.W32 b/googleurl/Makefile.W32
index b2af6be..81bf240 100755
--- a/googleurl/Makefile.W32
+++ b/googleurl/Makefile.W32
@@ -9,12 +9,14 @@ INCLUDE_CXXFLAGS = \
 	/DNDEBUG /DWIN32
 
 INCLUDE_DIRS = \
-	/I.
+	/I. \
+	/I"$(ICU_DIR)\include"
 
 INCLUDE_LDFLAGS = \
-
+	/LIBPATH:"$(ICU_DIR)\lib"
+	
 INCLUDE_LIBS = \
-
+	
 CPP_OBJS = \
 	url_canon_etc.obj \
 	url_canon_filesystemurl.obj \
diff --git a/makefiles/nmake/config.mk b/makefiles/nmake/config.mk
index a3d0484..5593036 100755
--- a/makefiles/nmake/config.mk
+++ b/makefiles/nmake/config.mk
@@ -8,6 +8,11 @@
 
 # please customize
 
+# ICU
+#####
+
+ICU_DIR = C:\cygwin\home\Andreas Baumann\icu
+
 # OpenSSL (http://www.slproweb.com/products/Win32OpenSSL.html)
 ##############################################################
 
diff --git a/src/modules/urlnormalizer/googleurl/Makefile.W32 b/src/modules/urlnormalizer/googleurl/Makefile.W32
index f739b32..de4d644 100755
--- a/src/modules/urlnormalizer/googleurl/Makefile.W32
+++ b/src/modules/urlnormalizer/googleurl/Makefile.W32
@@ -16,7 +16,8 @@ INCLUDE_LDFLAGS = \
 
 INCLUDE_LIBS = \
 	$(TOPDIR)\googleurl\googleurl.lib \
-	$(TOPDIR)\src\crawlingwolf.lib
+	$(TOPDIR)\src\crawlingwolf.lib \
+	"$(ICU_DIR)\lib\icuuc.lib"
 
 DYNAMIC_MODULE = \
 	mod_urlnormalizer_googleurl.dll
diff --git a/tests/url/Makefile.W32 b/tests/url/Makefile.W32
index c65000f..cd545dd 100755
--- a/tests/url/Makefile.W32
+++ b/tests/url/Makefile.W32
@@ -11,13 +11,17 @@ INCLUDE_CXXFLAGS = \
 INCLUDE_DIRS = \
 	/I. \
 	/I$(TOPDIR)\src \
-	/I$(TOPDIR)\src\modules\urlnormalizer\simpleurl
+	/I$(TOPDIR)\src\modules\urlnormalizer\simpleurl \
+	/I$(TOPDIR)\src\modules\urlnormalizer\googleurl
 
 INCLUDE_LDFLAGS = \
 
 INCLUDE_LIBS = \
 	$(TOPDIR)\src\crawlingwolf.lib \
-	$(TOPDIR)\src\modules\urlnormalizer\simpleurl\simpleurlnormalizer.lib
+	$(TOPDIR)\src\modules\urlnormalizer\googleurl\googleurlnormalizer.lib \
+	$(TOPDIR)\googleurl\googleurl.lib \
+	"$(ICU_DIR)\lib\icuuc.lib"
+#	$(TOPDIR)\src\modules\urlnormalizer\simpleurl\simpleurlnormalizer.lib \
 
 TEST_CPP_BINS = \
 	test1.exe
@@ -36,6 +40,8 @@ local_clean:
 local_distclean:
 
 local_test:
+	@-copy "$(ICU_DIR)\bin\icuuc49.dll" .
+	@-copy "$(ICU_DIR)\bin\icudt49.dll" .
 	@-exec_test test1 test1 "parse illegal protocol" simple parse www.andreasbaumann.cc
 	@-exec_test test1 test2 "parse normal start URL without slash" simple parse http://www.andreasbaumann.cc
 	@-exec_test test1 test3 "parse normal start URL with slash" simple parse http://www.andreasbaumann.cc/
diff --git a/tests/url/test1.cpp b/tests/url/test1.cpp
index 2512e79..b46e690 100755
--- a/tests/url/test1.cpp
+++ b/tests/url/test1.cpp
@@ -4,7 +4,7 @@
 #include "ModuleLoader.hpp"
 #else
 #include "SimpleURLNormalizer.hpp"
-//#include "GoogleURLNormalizer.hpp"
+#include "GoogleURLNormalizer.hpp"
 #endif
 
 #include <vector>
@@ -29,18 +29,22 @@ int main( int argc, char *argv[] )
 
 #ifdef USE_MODULELOADER
 		vector<string> modules;
-	//	modules.push_back( "../../src/modules/urlnormalizer/simpleurl/mod_urlnormalizer_simple.so" );
+#ifndef _WIN32
+		modules.push_back( "../../src/modules/urlnormalizer/simpleurl/mod_urlnormalizer_simple.so" );
+		modules.push_back( "../../src/modules/urlnormalizer/googleurl/mod_urlnormalizer_googleurl.so" );
+#else
 		modules.push_back( "..\\..\\src\\modules\\urlnormalizer\\simpleurl\\mod_urlnormalizer_simple.dll" );
-	//	modules.push_back( "../../src/modules/urlnormalizer/googleurl/mod_urlnormalizer_googleurl.so" );
+		modules.push_back( "..\\..\\src\\modules\\urlnormalizer\\googleurl\\mod_urlnormalizer_googleurl.dll" );
+#endif
 		ModuleLoader<URLNormalizer> urlNormalizers( modules );
 
 		URLNormalizer *normalizer = urlNormalizers.create( method );
 #else
 		URLNormalizer *normalizer;
 		if( strcmp( method, "simple" ) == 0 ) {
-			normalizer = new SimpleURLNormalizer( );
-	//	} else if( strcmp( method, "google" ) == 0 ) {
-	//		normalizer = new GoogleURLNormalizer( );
+//			normalizer = new SimpleURLNormalizer( );
+		} else if( strcmp( method, "google" ) == 0 ) {
+			normalizer = new GoogleURLNormalizer( );
 		} else {
 			cerr << "Unknown normalization method '" << method << "'" << endl;
 			return 1;