#include "support/unicode.h"
#include "support/debug.h"
+#include <QThreadStorage>
+
#include <iconv.h>
#include <boost/cstdint.hpp>
#include <cerrno>
-#include <iomanip>
-#include <ostream>
#include <map>
+#include <ostream>
+//Needed in MSVC
#include <string>
-using std::endl;
-using std::map;
-using std::make_pair;
-using std::string;
-using std::vector;
+
+using namespace std;
namespace {
static const iconv_t invalid_cd = (iconv_t)(-1);
-struct IconvProcessor::Impl
+class IconvProcessor::Impl
{
- Impl(std::string const & to, std::string const & from)
+public:
+ // noncopyable because iconv_close() is called in destructor
+ Impl(Impl const &) = delete;
+ Impl & operator=(Impl const &) = delete;
+
+ Impl(string const & to, string const & from)
: cd(invalid_cd), tocode_(to), fromcode_(from)
{}
~Impl()
{
if (cd != invalid_cd && iconv_close(cd) == -1)
- LYXERR0("Error returned from iconv_close(" << errno << ")");
+ LYXERR0("Error returned from iconv_close(" << errno << ')');
}
iconv_t cd;
- std::string tocode_;
- std::string fromcode_;
+ string tocode_;
+ string fromcode_;
};
}
-void IconvProcessor::operator=(IconvProcessor const & other)
+IconvProcessor & IconvProcessor::operator=(IconvProcessor const & other)
{
- if (&other != this)
+ if (&other != this) {
+ delete pimpl_;
pimpl_ = new Impl(other.pimpl_->tocode_, other.pimpl_->fromcode_);
+ }
+ return *this;
}
int res = iconv(pimpl_->cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
- //lyxerr << std::dec;
+ // flush out remaining data. This is needed because iconv sometimes
+ // holds back chars in the stream, waiting for a combination character
+ // (see e.g. http://sources.redhat.com/bugzilla/show_bug.cgi?id=1124)
+ iconv(pimpl_->cd, NULL, NULL, &outbuf, &outbytesleft);
+
+ //lyxerr << dec;
//lyxerr << "Inbytesleft: " << inbytesleft << endl;
//lyxerr << "Outbytesleft: " << outbytesleft << endl;
<< " has been encountered in the input.\n"
<< "When converting from " << pimpl_->fromcode_
<< " to " << pimpl_->tocode_ << ".\n";
- lyxerr << "Input:" << std::hex;
+ lyxerr << "Input:" << hex;
for (size_t i = 0; i < buflen; ++i) {
// char may be signed, avoid output of
// something like 0xffffffc2
*reinterpret_cast<unsigned char const *>(buf + i);
lyxerr << " 0x" << (unsigned int)b;
}
- lyxerr << std::dec << endl;
+ lyxerr << dec << endl;
break;
case EINVAL:
lyxerr << "EINVAL An incomplete multibyte sequence"
<< " has been encountered in the input.\n"
<< "When converting from " << pimpl_->fromcode_
<< " to " << pimpl_->tocode_ << ".\n";
- lyxerr << "Input:" << std::hex;
+ lyxerr << "Input:" << hex;
for (size_t i = 0; i < buflen; ++i) {
// char may be signed, avoid output of
// something like 0xffffffc2
*reinterpret_cast<unsigned char const *>(buf + i);
lyxerr << " 0x" << (unsigned int)b;
}
- lyxerr << std::dec << endl;
+ lyxerr << dec << endl;
break;
default:
lyxerr << "\tSome other error: " << errno << endl;
}
+std::string IconvProcessor::from() const
+{
+ return pimpl_->fromcode_;
+}
+
+
+std::string IconvProcessor::to() const
+{
+ return pimpl_->tocode_;
+}
+
+
namespace {
char const * inbuf = reinterpret_cast<char const *>(buf);
size_t inbytesleft = buflen * sizeof(InType);
- size_t const outsize = 32768;
- static char out[outsize];
- char * outbuf = out;
-
- int bytes = processor.convert(inbuf, inbytesleft, outbuf, outsize);
+ static QThreadStorage<std::vector<char> *> static_outbuf;
+ if (!static_outbuf.hasLocalData())
+ static_outbuf.setLocalData(new std::vector<char>(32768));
+ std::vector<char> & outbuf = *static_outbuf.localData();
+ // The number of UCS4 code points in buf is at most inbytesleft.
+ // The output encoding will use at most
+ // max_encoded_bytes(pimpl_->tocode_) per UCS4 code point.
+ size_t maxoutbufsize = max_encoded_bytes(processor.to()) * inbytesleft;
+ if (outbuf.size() < maxoutbufsize)
+ outbuf.resize(maxoutbufsize);
+
+ int bytes = processor.convert(inbuf, inbytesleft, &outbuf[0], outbuf.size());
if (bytes <= 0)
// Conversion failed
// FIXME Maybe throw an exception and handle that in the caller?
return vector<RetType>();
- RetType const * tmp = reinterpret_cast<RetType const *>(out);
+ RetType const * tmp = reinterpret_cast<RetType const *>(&outbuf[0]);
return vector<RetType>(tmp, tmp + bytes / sizeof(RetType));
}
} // anon namespace
+IconvProcessor & utf8ToUcs4()
+{
+ static QThreadStorage<IconvProcessor *> processor;
+ if (!processor.hasLocalData())
+ processor.setLocalData(new IconvProcessor(ucs4_codeset, "UTF-8"));
+ return *processor.localData();
+}
+
+
vector<char_type> utf8_to_ucs4(vector<char> const & utf8str)
{
if (utf8str.empty())
vector<char_type>
utf8_to_ucs4(char const * utf8str, size_t ls)
{
- static IconvProcessor processor(ucs4_codeset, "UTF-8");
- return iconv_convert<char_type>(processor, utf8str, ls);
+ return iconv_convert<char_type>(utf8ToUcs4(), utf8str, ls);
}
vector<char_type>
utf16_to_ucs4(unsigned short const * s, size_t ls)
{
- static IconvProcessor processor(ucs4_codeset, utf16_codeset);
- return iconv_convert<char_type>(processor, s, ls);
+ static QThreadStorage<IconvProcessor *> processor;
+ if (!processor.hasLocalData())
+ processor.setLocalData(new IconvProcessor(ucs4_codeset, utf16_codeset));
+ return iconv_convert<char_type>(*processor.localData(), s, ls);
}
vector<unsigned short>
ucs4_to_utf16(char_type const * s, size_t ls)
{
- static IconvProcessor processor(utf16_codeset, ucs4_codeset);
- return iconv_convert<unsigned short>(processor, s, ls);
+ static QThreadStorage<IconvProcessor *> processor;
+ if (!processor.hasLocalData())
+ processor.setLocalData(new IconvProcessor(utf16_codeset, ucs4_codeset));
+ return iconv_convert<unsigned short>(*processor.localData(), s, ls);
+}
+
+
+IconvProcessor & ucs4ToUtf8()
+{
+ static QThreadStorage<IconvProcessor *> processor;
+ if (!processor.hasLocalData())
+ processor.setLocalData(new IconvProcessor("UTF-8", ucs4_codeset));
+ return *processor.localData();
}
vector<char>
ucs4_to_utf8(char_type c)
{
- static IconvProcessor processor("UTF-8", ucs4_codeset);
- return iconv_convert<char>(processor, &c, 1);
+ return iconv_convert<char>(ucs4ToUtf8(), &c, 1);
}
vector<char>
ucs4_to_utf8(char_type const * ucs4str, size_t ls)
{
- static IconvProcessor processor("UTF-8", ucs4_codeset);
- return iconv_convert<char>(processor, ucs4str, ls);
+ return iconv_convert<char>(ucs4ToUtf8(), ucs4str, ls);
}
vector<char_type>
eightbit_to_ucs4(char const * s, size_t ls, string const & encoding)
{
- static map<string, IconvProcessor> processors;
+ static QThreadStorage<map<string, IconvProcessor> *> static_processors;
+ if (!static_processors.hasLocalData())
+ static_processors.setLocalData(new map<string, IconvProcessor>);
+ map<string, IconvProcessor> & processors = *static_processors.localData();
if (processors.find(encoding) == processors.end()) {
IconvProcessor processor(ucs4_codeset, encoding.c_str());
processors.insert(make_pair(encoding, processor));
}
+namespace {
+
+map<string, IconvProcessor> & ucs4To8bitProcessors()
+{
+ static QThreadStorage<map<string, IconvProcessor> *> processors;
+ if (!processors.hasLocalData())
+ processors.setLocalData(new map<string, IconvProcessor>);
+ return *processors.localData();
+}
+
+}
+
+
vector<char>
ucs4_to_eightbit(char_type const * ucs4str, size_t ls, string const & encoding)
{
- static map<string, IconvProcessor> processors;
+ map<string, IconvProcessor> & processors(ucs4To8bitProcessors());
if (processors.find(encoding) == processors.end()) {
IconvProcessor processor(encoding.c_str(), ucs4_codeset);
processors.insert(make_pair(encoding, processor));
char ucs4_to_eightbit(char_type ucs4, string const & encoding)
{
- static map<string, IconvProcessor> processors;
+ map<string, IconvProcessor> & processors(ucs4To8bitProcessors());
map<string, IconvProcessor>::iterator it = processors.find(encoding);
if (it == processors.end()) {
IconvProcessor processor(encoding.c_str(), ucs4_codeset);
void ucs4_to_multibytes(char_type ucs4, vector<char> & out,
string const & encoding)
{
- static map<string, IconvProcessor> processors;
+ static QThreadStorage<map<string, IconvProcessor> *> static_processors;
+ if (!static_processors.hasLocalData())
+ static_processors.setLocalData(new map<string, IconvProcessor>);
+ map<string, IconvProcessor> & processors = *static_processors.localData();
map<string, IconvProcessor>::iterator it = processors.find(encoding);
if (it == processors.end()) {
IconvProcessor processor(encoding.c_str(), ucs4_codeset);
out.clear();
}
+int max_encoded_bytes(std::string const & encoding)
+{
+ // FIXME: this information should be transferred to lib/encodings
+ // UTF8 uses at most 4 bytes to represent one UCS4 code point
+ // (see RFC 3629). RFC 2279 specifies 6 bytes, but that
+ // information is outdated, and RFC 2279 has been superseded by
+ // RFC 3629.
+ // The CJK encodings use (different) multibyte representation as well.
+ // All other encodings encode one UCS4 code point in one byte
+ // (and can therefore only encode a subset of UCS4)
+ // Furthermore, all encodings that use shifting (like SJIS) do not work with
+ // iconv_codecvt_facet.
+ if (encoding == "UTF-8" ||
+ encoding == "GB" ||
+ encoding == "EUC-TW")
+ return 4;
+ else if (encoding == "EUC-JP")
+ return 3;
+ else if (encoding == "ISO-2022-JP")
+ return 8;
+ else if (encoding == "BIG5" ||
+ encoding == "EUC-KR" ||
+ encoding == "EUC-CN" ||
+ encoding == "SJIS" ||
+ encoding == "GBK")
+ return 2;
+ else
+ return 1;
+}
+
} // namespace lyx
projects / lyx.git / blobdiff