Migrate to pcre2 (#249)

* [regex] Update pcre to pcre2 * [regex] Throw if accessing match after match error * [regex] Prevent null character breaking patterns e.g. `new EReg("abc\x00def", "")` would erroneously match "abc" HaxeFoundation/haxe#10592 * [regex] Add TODO regarding pcre2_substitute * [cmake] Fix static Apache build We can no longer link pcre as we use pcre2 * [cmake] Fix windows pcre static build * [cmake] Download old PCRE for Apache * [cmake] Remove unused flag for pcre windows build * [cmake] Reenable PCRE_STATIC cmake flag for Win * [cmake] Use https for pcre download for Apache Co-authored-by: Simon Krajewski <[email protected]>
HaxeFoundation · Feb 16, 2022 · 4a7b090 · 4a7b090
1 parent 58e8ca7
commit 4a7b090
Show file tree

Hide file tree

Showing 9 changed files with 200 additions and 150 deletions.
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -103,7 +103,7 @@ set(external_deps
 	Zlib
 	OpenSSL
 	MariaDBConnector
-	PCRE
+	pcre2
 	Sqlite3
 	APR
 	Apache

diff --git a/Earthfile b/Earthfile
@@ -8,7 +8,7 @@ ARG USER_UID=1000
 ARG USER_GID=$USER_UID
 
 ARG LINK_TYPE_DEFAULT=static # static or dynamic
-ARG LINK_DYNAMIC_PACKAGES="libgc-dev libpcre3-dev zlib1g-dev apache2-dev libmysqlclient-dev libsqlite3-dev libmbedtls-dev"
+ARG LINK_DYNAMIC_PACKAGES="libgc-dev libpcre2-dev zlib1g-dev apache2-dev libmysqlclient-dev libsqlite3-dev libmbedtls-dev"
 
 vscode-dev-containers-scripts:
     FROM curlimages/curl:7.80.0

diff --git a/README.md b/README.md
@@ -41,7 +41,7 @@ Neko needs to link with various third-party libraries, which are summarized as f
 |-----------------------------------------|-------------|-----------------------------------------------------------|
 | Boehm GC                                | all         | libgc-dev                                                 |
 | OpenSSL                                 | all         | libssl-dev                                                |
-| PCRE                                    | all         | libpcre3-dev                                              |
+| pcre2                                   | all         | libpcre2-dev                                              |
 | zlib                                    | all         | zlib1g-dev                                                |
 | Apache 2.2 / 2.4, with apr and apr-util | all         | apache2-dev                                               |
 | MariaDB / MySQL (Connector/C)           | all         | libmariadb-client-lgpl-dev-compat (or libmysqlclient-dev) |
@@ -115,7 +115,7 @@ Settings that allow to exclude libraries and their dependencies from the build;
 
 Default value: `all` for Windows, `none` otherwise
 
-It defines the dependencies that should be linked statically. Can be `all`, `none`, or a list of library names (e.g. `BoehmGC;Zlib;OpenSSL;MariaDBConnector;PCRE;Sqlite3;APR;APRutil;Apache;MbedTLS`). NOTE: On MacOS, APRutil cannot be added here as it has become part of the APR version used on MacOS builds.
+It defines the dependencies that should be linked statically. Can be `all`, `none`, or a list of library names (e.g. `BoehmGC;Zlib;OpenSSL;MariaDBConnector;pcre2;Sqlite3;APR;APRutil;Apache;MbedTLS`). NOTE: On MacOS, APRutil cannot be added here as it has become part of the APR version used on MacOS builds.
 
 CMake will automatically download and build the specified dependencies into the build folder. If a library is not present in this list, it should be installed manually, and it will be linked dynamically.
 

diff --git a/cmake/FindPCRE.cmake b/cmake/FindPCRE.cmake
diff --git a/cmake/FindPCRE2.cmake b/cmake/FindPCRE2.cmake
@@ -0,0 +1,19 @@
+# https://github.com/refu-lang/rfbase/blob/master/cmake/FindPCRE2.cmake
+
+# This CMake file tries to find the Perl regular expression libraries
+# The following variables are set:
+# PCRE2_FOUND - System has the PCRE library
+# PCRE2_LIBRARIES - The PCRE library file
+# PCRE2_INCLUDE_DIRS - The folder with the PCRE headers
+
+find_library(PCRE2_LIBRARIES NAMES pcre2 pcre2-8)
+find_path(PCRE2_INCLUDE_DIRS pcre2.h)
+if(PCRE2_LIBRARIES AND PCRE2_INCLUDE_DIRS)
+  message(STATUS "PCRE2 libs: ${PCRE2_LIBRARIES}")
+  message(STATUS "PCRE2 include directory: ${PCRE2_INCLUDE_DIRS}")
+  set(PCRE2_FOUND TRUE CACHE BOOL "Found PCRE2 libraries" FORCE)
+  add_custom_target(pcre2)
+else()
+  set(PCRE2_FOUND FALSE CACHE BOOL "Found PCRE2 libraries" FORCE)
+  message(STATUS "PCRE2 library not found.")
+endif()
diff --git a/extra/azure-pipelines/Brewfile-STATIC_DEPS_NONE b/extra/azure-pipelines/Brewfile-STATIC_DEPS_NONE
@@ -4,4 +4,4 @@ brew "pkg-config"
 brew "bdw-gc"
 brew "mariadb-connector-c"
 brew "mbedtls@2", link: true
-brew "pcre"
+brew "pcre2"
diff --git a/libs/CMakeLists.txt b/libs/CMakeLists.txt
@@ -41,13 +41,14 @@ if (WITH_APACHE)
 			set(APRUTIL_CONF "")
 			set(APRUTIL_DEP "")
 		endif()
-		if (STATIC_PCRE)
+		# APACHE still uses old pcre currently
+		# if (STATIC_PCRE2)
 			set(PCRE_CONF --with-pcre=${CMAKE_BINARY_DIR}/libs/src/install-prefix)
 			set(PCRE_DEP PCRE)
-		elseif()
-			set(PCRE_CONF "")
-			set(PCRE_DEP "")
-		endif()
+		# elseif()
+		# 	set(PCRE_CONF "")
+		# 	set(PCRE_DEP "")
+		# endif()
 		if (STATIC_ZLIB)
 			set(ZLIB_CONF --with-z=${CMAKE_BINARY_DIR}/libs/src/install-prefix)
 			set(ZLIB_DEP Zlib)
@@ -84,6 +85,68 @@ if (WITH_APACHE)
 			)
 		endif()
 
+		# Apache still uses old PCRE. Remove this once they migrate
+		## Start of PCRE download info
+		set(PCRE_URL_DATA
+			URL "https://downloads.sourceforge.net/project/pcre/pcre/8.45/pcre-8.45.tar.gz"
+			URL_HASH SHA1=a19402ce56d770da1557cf331b109d33adb74062
+		)
+		if (WIN32)
+			ExternalProject_Add(PCRE
+				${EP_CONFIGS}
+				${PCRE_URL_DATA}
+				CMAKE_ARGS
+					-G ${CMAKE_GENERATOR}
+					-DCMAKE_INSTALL_PREFIX=${CMAKE_BINARY_DIR}/libs/src/install-prefix
+					-Wno-dev
+					-DPCRE_BUILD_PCRECPP=OFF
+					-DPCRE_BUILD_PCREGREP=OFF
+					-DPCRE_BUILD_TESTS=OFF
+					-DPCRE_SUPPORT_JIT=ON
+					-DPCRE_SUPPORT_UNICODE_PROPERTIES=ON
+			)
+			set(PCRE_LIBRARIES
+				optimized ${CMAKE_BINARY_DIR}/libs/src/install-prefix/lib/pcre.lib
+				debug ${CMAKE_BINARY_DIR}/libs/src/install-prefix/lib/pcred.lib
+			)
+		else()
+			if (APPLE)
+				set(PCRE_CFLAGS "-w -mmacosx-version-min=${CMAKE_OSX_DEPLOYMENT_TARGET}")
+			else()
+				set(PCRE_CFLAGS "-w")
+			endif()
+			set(PCRE_LIBRARIES
+				${CMAKE_BINARY_DIR}/libs/src/install-prefix/lib/libpcre.a
+			)
+			ExternalProject_Add(PCRE
+				${EP_CONFIGS}
+				${PCRE_URL_DATA}
+				CONFIGURE_COMMAND cd ${CMAKE_BINARY_DIR}/libs/src/PCRE &&
+					./configure
+						--prefix=${CMAKE_BINARY_DIR}/libs/src/install-prefix
+						--with-pic
+						--enable-unicode-properties
+						--enable-silent-rules
+						--enable-jit
+						--disable-cpp
+						--enable-shared=no
+						--enable-static=yes
+						--silent
+				BUILD_COMMAND cd ${CMAKE_BINARY_DIR}/libs/src/PCRE &&
+					make "CFLAGS=${PCRE_CFLAGS}"
+				INSTALL_COMMAND cd ${CMAKE_BINARY_DIR}/libs/src/PCRE &&
+					make install
+				BYPRODUCTS ${PCRE_LIBRARIES}
+			)
+		endif()
+
+		set_target_properties(PCRE PROPERTIES ${EP_PROPS})
+		# don't want to add PCRE 1 to external_deps,
+		# but want download_static_deps depends on it
+		add_dependencies(download_static_deps PCRE-download)
+
+		## End of PCRE download info
+
 		if (APPLE) # We need to use the most up to date version on MacOS
 			set(APR_SOURCE
 				SVN_REPOSITORY https://svn.apache.org/repos/asf/apr/apr/trunk/

diff --git a/libs/regexp/CMakeLists.txt b/libs/regexp/CMakeLists.txt
@@ -3,52 +3,52 @@
 # regexp.ndll
 
 add_library(regexp.ndll MODULE regexp.c)
-if (STATIC_PCRE)
+if (STATIC_PCRE2)
+	add_compile_definitions(PCRE2_STATIC_LINK)
 	if (WIN32)
-		set(PCRE_URL "https://ftp.pcre.org/pub/pcre/pcre-8.42.tar.gz")
+		set(PCRE2_URL "https://github.com/PhilipHazel/pcre2/releases/download/pcre2-10.37/pcre2-10.37.tar.gz")
 		if (NOT ${CMAKE_VERSION} VERSION_LESS 3.7)
-			list(APPEND PCRE_URL
-				"http://downloads.sourceforge.net/project/pcre/pcre/8.42/pcre-8.42.tar.gz"
+			list(APPEND PCRE2_URL
+				"https://sourceforge.net/projects/pcre/files/pcre2/10.37/pcre2-10.37.tar.gz/download"
 			)
 		endif()
-		ExternalProject_Add(PCRE
+		ExternalProject_Add(pcre2
 			${EP_CONFIGS}
-			URL ${PCRE_URL}
-			URL_MD5 fc18afa0f14a25475cf097ee102a3e4f
+			URL ${PCRE2_URL}
+			URL_MD5 a0b59d89828f62d2e1caac04f7c51e0b
 			CMAKE_ARGS
 				-G ${CMAKE_GENERATOR}
 				-DCMAKE_INSTALL_PREFIX=${CMAKE_BINARY_DIR}/libs/src/install-prefix
 				-Wno-dev
-				-DPCRE_BUILD_PCRECPP=OFF
-				-DPCRE_BUILD_PCREGREP=OFF
-				-DPCRE_BUILD_TESTS=OFF
-				-DPCRE_SUPPORT_JIT=ON
-				-DPCRE_SUPPORT_UNICODE_PROPERTIES=ON
+				-DPCRE2_BUILD_PCRE2GREP=OFF
+				-DPCRE2_BUILD_TESTS=OFF
+				-DPCRE2_SUPPORT_JIT=ON
+				-DPCRE2_SUPPORT_UNICODE=ON
 		)
-		set(PCRE_LIBRARIES
-			optimized ${CMAKE_BINARY_DIR}/libs/src/install-prefix/lib/pcre.lib
-			debug ${CMAKE_BINARY_DIR}/libs/src/install-prefix/lib/pcred.lib
+		set(PCRE2_LIBRARIES
+			optimized ${CMAKE_BINARY_DIR}/libs/src/install-prefix/lib/pcre2-8.lib
+			debug ${CMAKE_BINARY_DIR}/libs/src/install-prefix/lib/pcre2-8d.lib
 		)
 	else()
 		if (APPLE)
-			set(PCRE_CFLAGS "-w -mmacosx-version-min=${CMAKE_OSX_DEPLOYMENT_TARGET}")
+			set(PCRE2_CFLAGS "-w -mmacosx-version-min=${CMAKE_OSX_DEPLOYMENT_TARGET}")
 		else()
-			set(PCRE_CFLAGS "-w")
+			set(PCRE2_CFLAGS "-w")
 		endif()
-		set(PCRE_LIBRARIES
-			${CMAKE_BINARY_DIR}/libs/src/install-prefix/lib/libpcre.a
+		set(PCRE2_LIBRARIES
+			${CMAKE_BINARY_DIR}/libs/src/install-prefix/lib/libpcre2-8.a
 		)
-		set(PCRE_URL "https://ftp.pcre.org/pub/pcre/pcre-8.40.tar.gz")
+		set(PCRE2_URL "https://github.com/PhilipHazel/pcre2/releases/download/pcre2-10.37/pcre2-10.37.tar.gz")
 		if (NOT ${CMAKE_VERSION} VERSION_LESS 3.7)
-			list(APPEND PCRE_URL
-				"http://downloads.sourceforge.net/project/pcre/pcre/8.40/pcre-8.40.tar.gz"
+			list(APPEND PCRE2_URL
+				"https://sourceforge.net/projects/pcre/files/pcre2/10.37/pcre2-10.37.tar.gz/download"
 			)
 		endif()
-		ExternalProject_Add(PCRE
+		ExternalProject_Add(pcre2
 			${EP_CONFIGS}
-			URL ${PCRE_URL}
-			URL_MD5 890c808122bd90f398e6bc40ec862102
-			CONFIGURE_COMMAND cd ${CMAKE_BINARY_DIR}/libs/src/PCRE &&
+			URL ${PCRE2_URL}
+			URL_MD5 a0b59d89828f62d2e1caac04f7c51e0b
+			CONFIGURE_COMMAND cd ${CMAKE_BINARY_DIR}/libs/src/pcre2 &&
 				./configure
 					--prefix=${CMAKE_BINARY_DIR}/libs/src/install-prefix
 					--with-pic
@@ -59,24 +59,24 @@ if (STATIC_PCRE)
 					--enable-shared=no
 					--enable-static=yes
 					--silent
-			BUILD_COMMAND cd ${CMAKE_BINARY_DIR}/libs/src/PCRE &&
+			BUILD_COMMAND cd ${CMAKE_BINARY_DIR}/libs/src/pcre2 &&
 				make "CFLAGS=${PCRE_CFLAGS}"
-			INSTALL_COMMAND cd ${CMAKE_BINARY_DIR}/libs/src/PCRE &&
+			INSTALL_COMMAND cd ${CMAKE_BINARY_DIR}/libs/src/pcre2 &&
 				make install
-			BYPRODUCTS ${PCRE_LIBRARIES}
+			BYPRODUCTS ${PCRE2_LIBRARIES}
 		)
 	endif()
-	set_target_properties(PCRE PROPERTIES ${EP_PROPS})
-	set(PCRE_INCLUDE_DIRS ${CMAKE_BINARY_DIR}/libs/src/install-prefix/include)
-	add_dependencies(regexp.ndll PCRE)
+	set_target_properties(pcre2 PROPERTIES ${EP_PROPS})
+	set(PCRE2_INCLUDE_DIRS ${CMAKE_BINARY_DIR}/libs/src/install-prefix/include)
+	add_dependencies(regexp.ndll pcre2)
 	# Download project for fat source archive
-	add_dependencies(download_static_deps PCRE-download)
+	add_dependencies(download_static_deps pcre2-download)
 else()
-	find_package(PCRE REQUIRED)
+	find_package(PCRE2 REQUIRED)
 endif()
 
-target_include_directories(regexp.ndll PRIVATE ${PCRE_INCLUDE_DIRS})
-target_link_libraries(regexp.ndll libneko ${PCRE_LIBRARIES})
+target_include_directories(regexp.ndll PRIVATE ${PCRE2_INCLUDE_DIRS})
+target_link_libraries(regexp.ndll libneko ${PCRE2_LIBRARIES})
 
 set_target_properties(regexp.ndll
 	PROPERTIES