fdo#48017 WIN32 long path support in Hyphen and MyThes

Change-Id: Ifb068efb553ed24a7caf65dbab28726bdeced0e6
This commit is contained in:
László Németh
2014-05-23 01:06:50 +02:00
parent 7818974103
commit 6d06aa8ba8
8 changed files with 130 additions and 51 deletions
+1
View File
@@ -17,6 +17,7 @@ $(eval $(call gb_UnpackedTarball_add_patches,hyphen,\
external/hyphen/hyphen-android.patch \
external/hyphen/hyphen-rhmin.patch \
external/hyphen/hyphen-build.patch \
external/hyphen/hyphen-fdo48017-wfopen.patch \
))
# vim: set noet sw=4 ts=4:
+46
View File
@@ -0,0 +1,46 @@
diff -u hyphen/hyphen.c build/hyphen/hyphen.c
--- hyphen/hyphen.c 2014-05-22 00:28:13.164587974 +0200
+++ build/hyphen/hyphen.c 2014-05-22 11:08:55.000000000 +0200
@@ -44,6 +44,11 @@
#include <unistd.h> /* for exit */
#endif
+#ifdef _WIN32
+#include <windows.h>
+#include <wchar.h>
+#endif
+
#define noVERBOSE
/* calculate hyphenmin values with long ligature length (2 or 3 characters
@@ -371,6 +376,21 @@
}
}
+FILE * hnj_fopen(const char * path, const char * mode) {
+#ifdef _WIN32
+#define WIN32_LONG_PATH_PREFIX "\\\\?\\"
+ if (strncmp(path, WIN32_LONG_PATH_PREFIX, 4) == 0) {
+ int len = MultiByteToWideChar(CP_UTF8, 0, path, -1, NULL, 0);
+ wchar_t *buff = (wchar_t *) malloc(len * sizeof(wchar_t));
+ MultiByteToWideChar(CP_UTF8, 0, path, -1, buff, len);
+ FILE * f = _wfopen(buff, (strcmp(mode, "r") == 0) ? L"r" : L"rb");
+ free(buff);
+ return f;
+ }
+#endif
+ return fopen(path, mode);
+}
+
HyphenDict *
hnj_hyphen_load (const char *fn)
{
@@ -383,7 +403,7 @@
HashEntry *e;
int state_num = 0;
- f = fopen (fn, "r");
+ f = hnj_fopen (fn, "r");
if (f == NULL)
return NULL;
+1
View File
@@ -15,6 +15,7 @@ $(eval $(call gb_UnpackedTarball_add_patches,mythes,\
external/mythes/mythes-1.2.0-vanilla-th-gen-idx.patch \
external/mythes/mythes-1.2.0-android.patch \
external/mythes/mythes-ssizet.patch \
external/mythes/mythes-fdo48017-wfopen.patch \
))
# vim: set noet sw=4 ts=4:
+72
View File
@@ -0,0 +1,72 @@
diff -u mythes/mythes.cxx build/mythes/mythes.cxx
--- mythes/mythes.cxx 2014-05-22 00:27:38.508588487 +0200
+++ build/mythes/mythes.cxx 2014-05-22 10:07:06.107547417 +0200
@@ -8,6 +8,11 @@
#include "mythes.hxx"
+#ifdef _WIN32
+#include <windows.h>
+#include <wchar.h>
+#endif
+
MyThes::MyThes(const char* idxpath, const char * datpath)
{
nw = 0;
@@ -35,7 +40,7 @@
{
// open the index file
- FILE * pifile = fopen(idxpath,"r");
+ FILE * pifile = myfopen(idxpath,"r");
if (!pifile) {
return 0;
}
@@ -90,7 +95,7 @@
fclose(pifile);
/* next open the data file */
- pdfile = fopen(datpath,"r");
+ pdfile = myfopen(datpath,"r");
if (!pdfile) {
return 0;
}
@@ -370,3 +375,17 @@
return -1;
}
+FILE * MyThes::myfopen(const char * path, const char * mode) {
+#ifdef _WIN32
+#define WIN32_LONG_PATH_PREFIX "\\\\?\\"
+ if (strncmp(path, WIN32_LONG_PATH_PREFIX, 4) == 0) {
+ int len = MultiByteToWideChar(CP_UTF8, 0, path, -1, NULL, 0);
+ wchar_t *buff = (wchar_t *) malloc(len * sizeof(wchar_t));
+ MultiByteToWideChar(CP_UTF8, 0, path, -1, buff, len);
+ FILE * f = _wfopen(buff, (strcmp(mode, "r") == 0) ? L"r" : L"rb");
+ free(buff);
+ return f;
+ }
+#endif
+ return fopen(path, mode);
+}
diff -u mythes/mythes.hxx build/mythes/mythes.hxx
--- mythes/mythes.hxx 2010-03-04 12:56:23.000000000 +0100
+++ build/mythes/mythes.hxx 2014-05-22 10:11:14.363543731 +0200
@@ -30,6 +30,7 @@
MyThes & operator = (const MyThes &);
public:
+ // use UTF-8 encoded paths in WIN32 environment
MyThes(const char* idxpath, const char* datpath);
~MyThes();
@@ -66,6 +67,9 @@
// return index of char in string
int mystr_indexOfChar(const char * d, int c);
+ // fopen or _wfopen
+ FILE * myfopen(const char * path, const char * mode);
+
};
#endif
@@ -285,14 +285,12 @@ Reference< XHyphenatedWord > SAL_CALL Hyphenator::hyphenate( const OUString& aWo
OUString dictpath;
osl::FileBase::getSystemPathFromFileURL( DictFN, dictpath );
OString sTmp( OU2ENC( dictpath, osl_getThreadTextEncoding() ) );
#if defined(WNT)
// workaround for Windows specific problem that the
// path length in calls to 'fopen' is limted to somewhat
// about 120+ characters which will usually be exceed when
// using dictionaries as extensions.
sTmp = Win_GetShortPathName( dictpath );
// Hyphen waits UTF-8 encoded paths with \\?\ long path prefix.
OString sTmp = OUStringToOString(dicpath, RTL_TEXTENCODING_UTF8);
#else
OString sTmp( OU2ENC( dictpath, osl_getThreadTextEncoding() ) );
#endif
if ( ( dict = hnj_hyphen_load ( sTmp.getStr()) ) == NULL )
@@ -17,10 +17,6 @@
* the License at http://www.apache.org/licenses/LICENSE-2.0 .
*/
#if defined(WNT)
#include <windows.h>
#endif
#include <osl/thread.h>
#include <osl/file.hxx>
#include <tools/debug.hxx>
@@ -45,29 +41,6 @@
using ::com::sun::star::lang::Locale;
using namespace ::com::sun::star;
#if defined(WNT)
OString Win_GetShortPathName( const OUString &rLongPathName )
{
OString aRes;
sal_Unicode aShortBuffer[1024] = {0};
sal_Int32 nShortBufSize = SAL_N_ELEMENTS( aShortBuffer );
// use the version of 'GetShortPathName' that can deal with Unicode...
sal_Int32 nShortLen = GetShortPathNameW(
reinterpret_cast<LPCWSTR>( rLongPathName.getStr() ),
reinterpret_cast<LPWSTR>( aShortBuffer ),
nShortBufSize );
if (nShortLen < nShortBufSize) // conversion successful?
aRes = OString( OU2ENC( OUString( aShortBuffer, nShortLen ), osl_getThreadTextEncoding()) );
else
OSL_FAIL( "Win_GetShortPathName: buffer to short" );
return aRes;
}
#endif //defined(WNT)
// build list of old style diuctionaries (not as extensions) to use.
// User installed dictionaries (the ones residing in the user paths)
// will get precedence over system installed ones for the same language.
@@ -48,16 +48,6 @@ inline bool operator == ( const ::com::sun::star::lang::Locale &rL1, const ::com
rL1.Variant == rL2.Variant;
}
#if defined(WNT)
// to be use to get a short path name under Windows that still can be used with
// the 'fopen' call. This is necessary since under Windows there seems to be
// a restriction of only about 110-130 characters length to a path name in order
// for it to work with 'fopen'. And that length is usually easily exceeded
// when using extensions...
OString Win_GetShortPathName( const OUString &rLongPathName );
#endif
// temporary function, to be removed when new style dictionaries
// using configuration entries are fully implemented and provided
std::vector< SvtLinguConfigDictionaryEntry > GetOldStyleDics( const char * pDicType );
@@ -328,16 +328,14 @@ Sequence < Reference < ::com::sun::star::linguistic2::XMeaning > > SAL_CALL Thes
OUString nidx;
osl::FileBase::getSystemPathFromFileURL(datpath,ndat);
osl::FileBase::getSystemPathFromFileURL(idxpath,nidx);
OString aTmpidx(OU2ENC(nidx,osl_getThreadTextEncoding()));
OString aTmpdat(OU2ENC(ndat,osl_getThreadTextEncoding()));
#if defined(WNT)
// workaround for Windows specific problem that the
// path length in calls to 'fopen' is limted to somewhat
// about 120+ characters which will usually be exceed when
// using dictionaries as extensions.
aTmpidx = Win_GetShortPathName( nidx );
aTmpdat = Win_GetShortPathName( ndat );
// MyThes waits UTF-8 encoded paths with \\?\ long path prefix.
OString aTmpidx = OUStringToOString(nidx, RTL_TEXTENCODING_UTF8);
OString aTmpdat = OUStringToOString(ndat, RTL_TEXTENCODING_UTF8);
#else
OString aTmpidx(OU2ENC(nidx,osl_getThreadTextEncoding()));
OString aTmpdat(OU2ENC(ndat,osl_getThreadTextEncoding()));
#endif
aThes[i] = new MyThes(aTmpidx.getStr(),aTmpdat.getStr());