Amebis/wxWidgets
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fixed crash in iconv_open() and cleaned up the code a bit further

Browse Source 
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@13652 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
This commit is contained in:
Vadim Zeitlin
2002-01-19 13:15:56 +00:00
parent 855d6be778
commit 36acb88072
1 changed files with 266 additions and 206 deletions
Download Patch File Download Diff File Expand all files Collapse all files

188
src/common/strconv.cpp
View File

@@ -497,6 +497,14 @@ extern long wxCharsetToCodepage(const wxChar *name)
#endif // Win32 #endif // Win32
// ============================================================================
// wxCharacterSet and derived classes
// ============================================================================
// ----------------------------------------------------------------------------
// wxCharacterSet is the ABC for the classes below
// ----------------------------------------------------------------------------
class wxCharacterSet class wxCharacterSet
{ {
public: public:
@@ -514,6 +522,10 @@ public:
const wxChar*cname; const wxChar*cname;
}; };
// ----------------------------------------------------------------------------
// ID_CharSet: implementation of wxCharacterSet using an existing wxMBConv
// ----------------------------------------------------------------------------
class ID_CharSet : public wxCharacterSet class ID_CharSet : public wxCharacterSet
{ {
public: public:
@@ -533,10 +545,11 @@ public:
}; };
#ifdef HAVE_ICONV_H // ============================================================================
// The classes doing conversion using the iconv_xxx() functions
// ============================================================================
bool g_wcNeedsSwap = FALSE; #ifdef HAVE_ICONV_H
static const char *g_wcCharset = NULL;
// VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG // VS: glibc 2.1.3 is broken in that iconv() conversion to/from UCS4 fails with E2BIG
// if output buffer is _exactly_ as big as needed. Such case is (unless there's // if output buffer is _exactly_ as big as needed. Such case is (unless there's
@@ -552,33 +565,70 @@ static const char *g_wcCharset = NULL;
#define ICONV_FAILED(cres, bufLeft) (cres == (size_t)-1) #define ICONV_FAILED(cres, bufLeft) (cres == (size_t)-1)
#endif #endif
#ifdef WX_ICONV_TAKES_CHAR
#define ICONV_CHAR_CAST(x) (char **)x
#else
#define ICONV_CHAR_CAST(x) (const char **)x
#endif
// ----------------------------------------------------------------------------
// IC_CharSet: encapsulates an iconv character set
// ----------------------------------------------------------------------------
class IC_CharSet : public wxCharacterSet class IC_CharSet : public wxCharacterSet
{ {
public: public:
IC_CharSet(const wxChar *name) IC_CharSet(const wxChar *name);
virtual ~IC_CharSet();
virtual size_t MB2WC(wchar_t *buf, const char *psz, size_t n);
virtual size_t WC2MB(char *buf, const wchar_t *psz, size_t n);
bool usable() const
{ return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); }
protected:
// the iconv handlers used to translate from multibyte to wide char and in
// the other direction
iconv_t m2w,
w2m;
private:
// the name (for iconv_open()) of a wide char charset - if none is
// available on this machine, it will remain NULL
static const char *ms_wcCharsetName;
// true if the wide char encoding we use (i.e. ms_wcCharsetName) has
// different endian-ness than the native one
static bool ms_wcNeedsSwap = FALSE;
};
const char *IC_CharSet::ms_wcCharsetName = NULL;
IC_CharSet::IC_CharSet(const wxChar *name)
: wxCharacterSet(name) : wxCharacterSet(name)
{ {
// check for charset that represents wchar_t: // check for charset that represents wchar_t:
if (g_wcCharset == NULL) if (ms_wcCharsetName == NULL)
{ {
g_wcNeedsSwap = FALSE; ms_wcNeedsSwap = FALSE;
// try charset with explicit bytesex info (e.g. "UCS-4LE"): // try charset with explicit bytesex info (e.g. "UCS-4LE"):
g_wcCharset = WC_NAME_BEST; ms_wcCharsetName = WC_NAME_BEST;
m2w = iconv_open(g_wcCharset, wxConvLibc.cWX2MB(name)); m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name));
if (m2w == (iconv_t)-1) if (m2w == (iconv_t)-1)
{ {
// try charset w/o bytesex info (e.g. "UCS4") // try charset w/o bytesex info (e.g. "UCS4")
// and check for bytesex ourselves: // and check for bytesex ourselves:
g_wcCharset = WC_NAME; ms_wcCharsetName = WC_NAME;
m2w = iconv_open(g_wcCharset, wxConvLibc.cWX2MB(name)); m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name));
// last bet, try if it knows WCHAR_T pseudo-charset // last bet, try if it knows WCHAR_T pseudo-charset
if (m2w == (iconv_t)-1) if (m2w == (iconv_t)-1)
{ {
g_wcCharset = "WCHAR_T"; ms_wcCharsetName = "WCHAR_T";
m2w = iconv_open(g_wcCharset, wxConvLibc.cWX2MB(name)); m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name));
} }
if (m2w != (iconv_t)-1) if (m2w != (iconv_t)-1)
@@ -596,37 +646,41 @@ public:
wbufPtr = wbuf; wbufPtr = wbuf;
bufPtr = buf; bufPtr = buf;
#ifdef WX_ICONV_TAKES_CHAR res = iconv(m2w, ICONV_CHAR_CAST(&bufPtr), &insz,
res = iconv(m2w, (char**)&bufPtr, &insz, (char**)&wbufPtr, &outsz); (char**)&wbufPtr, &outsz);
#else
res = iconv(m2w, (const char**)&bufPtr, &insz, (char**)&wbufPtr, &outsz);
#endif
if (ICONV_FAILED(res, insz)) if (ICONV_FAILED(res, insz))
{ {
g_wcCharset = NULL; ms_wcCharsetName = NULL;
wxLogLastError(wxT("iconv")); wxLogLastError(wxT("iconv"));
wxLogError(_("Convertion to charset '%s' doesn't work."), name); wxLogError(_("Convertion to charset '%s' doesn't work."), name);
} }
else else
{ {
g_wcNeedsSwap = (wbuf[0] != (wchar_t)buf[0]); ms_wcNeedsSwap = wbuf[0] != (wchar_t)buf[0];
} }
} }
else else
{ {
g_wcCharset = NULL; ms_wcCharsetName = NULL;
wxLogError(_("Don't know how to convert to/from charset '%s'."), name); wxLogError(_("Impossible to convert to/from charset '%s'."), name);
} }
} }
wxLogTrace(wxT("strconv"), wxT("wchar_t charset is '%s', needs swap: %i"), g_wcCharset, g_wcNeedsSwap); wxLogTrace(wxT("strconv"), wxT("wchar_t charset is '%s', needs swap: %i"), ms_wcCharsetName, ms_wcNeedsSwap);
} }
else else // we already have ms_wcCharsetName
m2w = iconv_open(g_wcCharset, wxConvLibc.cWX2MB(name)); {
m2w = iconv_open(ms_wcCharsetName, wxConvLibc.cWX2MB(name));
w2m = iconv_open(wxConvLibc.cWX2MB(name), g_wcCharset);
} }
~IC_CharSet() // NB: don't ever pass NULL to iconv_open(), it may crash!
if ( ms_wcCharsetName )
{
w2m = iconv_open(wxConvLibc.cWX2MB(name), ms_wcCharsetName);
}
}
IC_CharSet::~IC_CharSet()
{ {
if ( m2w != (iconv_t)-1 ) if ( m2w != (iconv_t)-1 )
iconv_close(m2w); iconv_close(m2w);
@@ -634,7 +688,7 @@ public:
iconv_close(w2m); iconv_close(w2m);
} }
size_t MB2WC(wchar_t *buf, const char *psz, size_t n) size_t IC_CharSet::MB2WC(wchar_t *buf, const char *psz, size_t n)
{ {
size_t inbuf = strlen(psz); size_t inbuf = strlen(psz);
size_t outbuf = n * SIZEOF_WCHAR_T; size_t outbuf = n * SIZEOF_WCHAR_T;
@@ -646,14 +700,12 @@ public:
if (buf) if (buf)
{ {
// have destination buffer, convert there // have destination buffer, convert there
#ifdef WX_ICONV_TAKES_CHAR cres = iconv(m2w,
cres = iconv(m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf); ICONV_CHAR_CAST(&pszPtr), &inbuf,
#else (char**)&bufPtr, &outbuf);
cres = iconv(m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf);
#endif
res = n - (outbuf / SIZEOF_WCHAR_T); res = n - (outbuf / SIZEOF_WCHAR_T);
if (g_wcNeedsSwap) if (ms_wcNeedsSwap)
{ {
// convert to native endianness // convert to native endianness
WC_BSWAP(buf /* _not_ bufPtr */, res) WC_BSWAP(buf /* _not_ bufPtr */, res)
@@ -666,12 +718,13 @@ public:
wchar_t tbuf[8]; wchar_t tbuf[8];
res = 0; res = 0;
do { do {
bufPtr = tbuf; outbuf = 8*SIZEOF_WCHAR_T; bufPtr = tbuf;
#ifdef WX_ICONV_TAKES_CHAR outbuf = 8*SIZEOF_WCHAR_T;
cres = iconv( m2w, (char**)&pszPtr, &inbuf, (char**)&bufPtr, &outbuf );
#else cres = iconv(m2w,
cres = iconv( m2w, &pszPtr, &inbuf, (char**)&bufPtr, &outbuf ); ICONV_CHAR_CAST(&pszPtr), &inbuf,
#endif (char**)&bufPtr, &outbuf );
res += 8-(outbuf/SIZEOF_WCHAR_T); res += 8-(outbuf/SIZEOF_WCHAR_T);
} while ((cres==(size_t)-1) && (errno==E2BIG)); } while ((cres==(size_t)-1) && (errno==E2BIG));
} }
@@ -686,7 +739,7 @@ public:
return res; return res;
} }
size_t WC2MB(char *buf, const wchar_t *psz, size_t n) size_t IC_CharSet::WC2MB(char *buf, const wchar_t *psz, size_t n)
{ {
#if defined(__BORLANDC__) && (__BORLANDC__ > 0x530) #if defined(__BORLANDC__) && (__BORLANDC__ > 0x530)
size_t inbuf = std::wcslen(psz) * SIZEOF_WCHAR_T; size_t inbuf = std::wcslen(psz) * SIZEOF_WCHAR_T;
@@ -698,7 +751,7 @@ public:
wchar_t *tmpbuf = 0; wchar_t *tmpbuf = 0;
if (g_wcNeedsSwap) if (ms_wcNeedsSwap)
{ {
// need to copy to temp buffer to switch endianness // need to copy to temp buffer to switch endianness
// this absolutely doesn't rock! // this absolutely doesn't rock!
@@ -713,11 +766,8 @@ public:
if (buf) if (buf)
{ {
// have destination buffer, convert there // have destination buffer, convert there
#ifdef WX_ICONV_TAKES_CHAR cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
cres = iconv( w2m, (char**)&psz, &inbuf, &buf, &outbuf );
#else
cres = iconv( w2m, (const char**)&psz, &inbuf, &buf, &outbuf );
#endif
res = n-outbuf; res = n-outbuf;
} }
else else
@@ -728,16 +778,14 @@ public:
res = 0; res = 0;
do { do {
buf = tbuf; outbuf = 16; buf = tbuf; outbuf = 16;
#ifdef WX_ICONV_TAKES_CHAR
cres = iconv( w2m, (char**)&psz, &inbuf, &buf, &outbuf ); cres = iconv( w2m, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
#else
cres = iconv( w2m, (const char**)&psz, &inbuf, &buf, &outbuf );
#endif
res += 16 - outbuf; res += 16 - outbuf;
} while ((cres==(size_t)-1) && (errno==E2BIG)); } while ((cres==(size_t)-1) && (errno==E2BIG));
} }
if (g_wcNeedsSwap) if (ms_wcNeedsSwap)
{ {
free(tmpbuf); free(tmpbuf);
} }
@@ -752,13 +800,11 @@ public:
return res; return res;
} }
bool usable() #endif // HAVE_ICONV_H
{ return (m2w != (iconv_t)-1) && (w2m != (iconv_t)-1); }
protected: // ============================================================================
iconv_t m2w, w2m; // Win32 conversion classes
}; // ============================================================================
#endif
#if defined(__WIN32__) && !defined(__WXMICROWIN__) #if defined(__WIN32__) && !defined(__WXMICROWIN__)
class CP_CharSet : public wxCharacterSet class CP_CharSet : public wxCharacterSet
@@ -796,6 +842,10 @@ public:
}; };
#endif // __WIN32__ #endif // __WIN32__
// ============================================================================
// wxEncodingConverter based conversion classes
// ============================================================================
#if wxUSE_FONTMAP #if wxUSE_FONTMAP
class EC_CharSet : public wxCharacterSet class EC_CharSet : public wxCharacterSet
@@ -844,6 +894,11 @@ public:
#endif // wxUSE_FONTMAP #endif // wxUSE_FONTMAP
// ----------------------------------------------------------------------------
// the function creating the wxCharacterSet for the specified charset on the
// current system, trying all possibilities
// ----------------------------------------------------------------------------
static wxCharacterSet *wxGetCharacterSet(const wxChar *name) static wxCharacterSet *wxGetCharacterSet(const wxChar *name)
{ {
wxCharacterSet *cset = NULL; wxCharacterSet *cset = NULL;
@@ -889,6 +944,10 @@ static wxCharacterSet *wxGetCharacterSet(const wxChar *name)
return NULL; return NULL;
} }
// ============================================================================
// wxCSConv implementation
// ============================================================================
wxCSConv::wxCSConv(const wxChar *charset) wxCSConv::wxCSConv(const wxChar *charset)
{ {
m_name = (wxChar *)NULL; m_name = (wxChar *)NULL;
@@ -972,6 +1031,9 @@ size_t wxCSConv::WC2MB(char *buf, const wchar_t *psz, size_t n) const
return len; return len;
} }
// VZ: are the classes below used at all??
#if 0
#ifdef HAVE_ICONV_H #ifdef HAVE_ICONV_H
class IC_CharSetConverter class IC_CharSetConverter
@@ -993,11 +1055,7 @@ public:
{ {
size_t inbuf = strlen(psz); size_t inbuf = strlen(psz);
size_t outbuf = n; size_t outbuf = n;
#ifdef WX_ICONV_TAKES_CHAR size_t res = iconv( cnv, ICONV_CHAR_CAST(&psz), &inbuf, &buf, &outbuf );
size_t res = iconv( cnv, (char**)&psz, &inbuf, &buf, &outbuf );
#else
size_t res = iconv( cnv, &psz, &inbuf, &buf, &outbuf );
#endif
if (res == (size_t)-1) if (res == (size_t)-1)
return (size_t)-1; return (size_t)-1;
return (n - outbuf); return (n - outbuf);
@@ -1026,6 +1084,8 @@ public:
wxEncodingConverter cnv; wxEncodingConverter cnv;
}; };
#endif // 0
#else // !wxUSE_WCHAR_T #else // !wxUSE_WCHAR_T
// ---------------------------------------------------------------------------- // ----------------------------------------------------------------------------