diff -uprN ghc-6.4.orig/libraries/base/Text/Regex/Posix.hsc base/Text/Regex/Posix.hsc --- ghc-6.4.orig/libraries/base/Text/Regex/Posix.hsc 2005-02-01 09:52:21.000000000 +0900 +++ ghc-6.4/libraries/base/Text/Regex/Posix.hsc 2005-05-05 01:04:41.000000000 +0900 @@ -44,6 +44,13 @@ module Text.Regex.Posix ( #include +#ifdef HAVE_ONIGURUMA +#include "onigposix.h" +#ifdef __HUGS__ +{-# CFILES cbits/onigUtils.c #-} +#endif +#else + #if HAVE_REGEX_H && HAVE_REGCOMP #include "regex.h" #else @@ -57,6 +64,8 @@ module Text.Regex.Posix ( {-# CFILES cbits/regex/regfree.c #-} #endif +#endif + import Prelude import Foreign @@ -77,7 +86,10 @@ regcomp -> IO Regex -- ^ Returns: the compiled regular expression regcomp pattern flags = do regex_fptr <- mallocForeignPtrBytes (#const sizeof(regex_t)) - r <- withCString pattern $ \cstr -> +#ifdef HAVE_ONIGURUMA + onigSetDefaultEncodingUTF32 +#endif + r <- withString pattern $ \cstr -> withForeignPtr regex_fptr $ \p -> c_regcomp p cstr (fromIntegral flags) if (r == 0) @@ -103,7 +115,7 @@ regexec :: Regex -- ^ Compiled regular -- @ regexec (Regex regex_fptr) str = do - withCString str $ \cstr -> do + withString str $ \cstr -> do withForeignPtr regex_fptr $ \regex_ptr -> do nsub <- (#peek regex_t, re_nsub) regex_ptr let nsub_int = fromIntegral (nsub :: CSize) @@ -125,7 +137,7 @@ matched_parts :: String -> Ptr CRegMatch matched_parts string p_match = do start <- (#peek regmatch_t, rm_so) p_match :: IO (#type regoff_t) end <- (#peek regmatch_t, rm_eo) p_match :: IO (#type regoff_t) - let s = fromIntegral start; e = fromIntegral end + let s = fromIntegral start `div` charSize; e = fromIntegral end `div` charSize return ( take s string, take (e-s) (drop s string), drop e string ) @@ -137,7 +149,7 @@ unpack string p_match = do -- the subexpression may not have matched at all, perhaps because it -- was optional. In this case, the offsets are set to -1. if (start == -1) then return "" else do - return (take (fromIntegral (end-start)) (drop (fromIntegral start) string)) + return (take (fromIntegral (end-start) `div` charSize) (drop (fromIntegral start `div` charSize) string)) -- ----------------------------------------------------------------------------- -- The POSIX regex C interface @@ -176,6 +188,35 @@ unpack string p_match = do type CRegMatch = () +#ifdef HAVE_ONIGURUMA + +charSize :: Int +charSize = sizeOf (0 :: Word32) + +withString :: String -> (Ptr Word32 -> IO a) -> IO a +withString = withArray0 0 . fmap (fromIntegral . fromEnum) + +foreign import ccall unsafe "onigposix.h regcomp" + c_regcomp :: Ptr CRegex -> Ptr a -> CInt -> IO CInt + +foreign import ccall unsafe "onigposix.h ®free" + ptr_regfree :: FunPtr (Ptr CRegex -> IO ()) + +foreign import ccall unsafe "onigposix.h regexec" + c_regexec :: Ptr CRegex -> Ptr a -> CSize + -> Ptr CRegMatch -> CInt -> IO CInt + +foreign import ccall unsafe "onigUtils.h" + onigSetDefaultEncodingUTF32 :: IO CInt + +#else + +charSize :: Int +charSize = 1 + +withString :: String -> (CString -> IO a) -> IO a +withString = withCString + -- GHC and Hugs get the appropriate include file from the OPTIONS -- pragma generated by hsc2hs from the above #include. -- Implementations following the FFI spec have to specify it in the @@ -214,3 +255,5 @@ foreign import ccall unsafe "regex/regex c_regexec :: Ptr CRegex -> CString -> CSize -> Ptr CRegMatch -> CInt -> IO CInt #endif + +#endif diff -uprN ghc-6.4.orig/libraries/base/cbits/Makefile base/cbits/Makefile --- ghc-6.4.orig/libraries/base/cbits/Makefile 2005-02-01 09:52:22.000000000 +0900 +++ ghc-6.4/libraries/base/cbits/Makefile 2005-05-04 23:52:31.000000000 +0900 @@ -8,12 +8,15 @@ HC = $(GHC_INPLACE) UseGhcForCc = YES +ifneq "$(HaveOniguruma)" "YES" +EXCLUDED_SRCS += onigUtils.c # If there is no system-supplied POSIX regex library, use our own ifneq "$(HavePosixRegex)" "YES" ALL_DIRS += regex SRC_CC_OPTS += -Iregex EXCLUDED_SRCS += regex/engine.c endif +endif ifneq "$(TARGETPLATFORM)" "i386-unknown-mingw32" EXCLUDED_SRCS += consUtils.c diff -uprN ghc-6.4.orig/libraries/base/cbits/onigUtils.c base/cbits/onigUtils.c --- ghc-6.4.orig/libraries/base/cbits/onigUtils.c 1970-01-01 09:00:00.000000000 +0900 +++ ghc-6.4/libraries/base/cbits/onigUtils.c 2005-05-05 00:07:45.000000000 +0900 @@ -0,0 +1,12 @@ +#include +#include "onigUtils.h" +#include "HsBase.h" + +int onigSetDefaultEncodingUTF32(void) +{ +#ifdef WORDS_BIGENDIAN + return onigenc_set_default_encoding(ONIG_ENCODING_UTF32_BE); +#else + return onigenc_set_default_encoding(ONIG_ENCODING_UTF32_LE); +#endif +} diff -uprN ghc-6.4.orig/libraries/base/config.mk.in base/config.mk.in --- ghc-6.4.orig/libraries/base/config.mk.in 2005-02-01 09:52:21.000000000 +0900 +++ ghc-6.4/libraries/base/config.mk.in 2005-05-05 00:49:11.000000000 +0900 @@ -1,3 +1,5 @@ +HaveOniguruma = @HaveOniguruma@ + # Regex library # (if present in libc use that one, otherwise use the one in the tree) # diff -uprN ghc-6.4.orig/libraries/base/configure.ac base/configure.ac --- ghc-6.4.orig/libraries/base/configure.ac 2005-03-07 19:34:38.000000000 +0900 +++ ghc-6.4/libraries/base/configure.ac 2005-05-05 00:15:39.000000000 +0900 @@ -15,6 +15,18 @@ AC_CHECK_HEADERS([ctype.h fcntl.h signal # rlim_t, because it will affect the result of that test. AC_SYS_LARGEFILE +AC_C_BIGENDIAN + +dnl ** check for Oniguruma +HaveOniguruma=NO +AC_CHECK_HEADERS([onigposix.h], + [AC_CHECK_LIB(onig, onig_init, + [HaveOniguruma=YES + AC_DEFINE(HAVE_ONIGURUMA, [], [Define to 1 if you have the oniguruma library.]) + ]) + ]) +AC_SUBST(HaveOniguruma) + dnl ** check for POSIX regex HavePosixRegex=NO AC_CHECK_HEADERS([regex.h], [AC_CHECK_FUNCS(regcomp, [HavePosixRegex=YES])]) @@ -88,6 +100,15 @@ case "$target" in *) EXTRA_LIBS= ;; esac + +if test "$HaveOniguruma" == "YES"; then + if test "x$EXTRA_LIBS" == "x" ; then + EXTRA_LIBS="onig" + else + EXTRA_LIBS="onig, $EXTRALIBS" + fi +fi + AC_SUBST([EXTRA_LIBS]) AC_CONFIG_FILES([config.mk base.buildinfo]) diff -uprN ghc-6.4.orig/libraries/base/include/HsBaseConfig.h.in base/include/HsBaseConfig.h.in --- ghc-6.4.orig/libraries/base/include/HsBaseConfig.h.in 2005-03-11 04:44:22.000000000 +0900 +++ ghc-6.4/libraries/base/include/HsBaseConfig.h.in 2005-05-04 23:58:44.000000000 +0900 @@ -435,6 +435,12 @@ /* Define to 1 if you have the header file. */ #undef HAVE_MEMORY_H +/* Define to 1 if you have the header file. */ +#undef HAVE_ONIGPOSIX_H + +/* Define to 1 if you have the oniguruma library. */ +#undef HAVE_ONIGURUMA + /* Define to 1 if you have the `readdir_r' function. */ #undef HAVE_READDIR_R @@ -603,6 +609,10 @@ /* Define to 1 if you have the ANSI C header files. */ #undef STDC_HEADERS +/* Define to 1 if your processor stores words with the most significant byte + first (like Motorola and SPARC, unlike Intel and VAX). */ +#undef WORDS_BIGENDIAN + /* Number of bits in a file offset, on hosts where this is settable. */ #undef _FILE_OFFSET_BITS diff -uprN ghc-6.4.orig/libraries/base/include/onigUtils.h base/include/onigUtils.h --- ghc-6.4.orig/libraries/base/include/onigUtils.h 1970-01-01 09:00:00.000000000 +0900 +++ ghc-6.4/libraries/base/include/onigUtils.h 2005-05-04 23:34:29.000000000 +0900 @@ -0,0 +1,6 @@ +#ifndef __ONIGUTILS_H__ +#define __ONIGUTILS_H__ + +extern int setOnigDefaultEncodingUTF32(void); + +#endif /* __ONIGUTILS_H__ */ diff -uprN ghc-6.4.orig/libraries/base/package.conf.in base/package.conf.in --- ghc-6.4.orig/libraries/base/package.conf.in 2005-02-21 19:24:09.000000000 +0900 +++ ghc-6.4/libraries/base/package.conf.in 2005-05-04 23:40:44.000000000 +0900 @@ -1,4 +1,5 @@ #include "ghcconfig.h" +#include "HsBaseConfig.h" name: PACKAGE version: VERSION @@ -184,6 +185,9 @@ extra-libraries: "HSbase_cbits" #if defined(mingw32_HOST_OS) || defined(__MINGW32__) || defined(_MSC_VER) , "wsock32", "msvcrt", "kernel32", "user32", "shell32" #endif +#ifdef HAVE_ONIGURUMA + , "onig" +#endif include-dirs: INCLUDE_DIR includes: HsBase.h