X-Git-Url: https://git.lyx.org/gitweb/?a=blobdiff_plain;f=src%2Fsupport%2Funicode.cpp;h=26f834a58bec98aa4c38f2155ebac98365b1b3c5;hb=329eae5605d7bc40a53ebafa817470d9739ff632;hp=09bf8c450eb4e44fad2593c5072d089ee5e0ff56;hpb=9160d824d87d27088db65afa0bcf0951c2db08cc;p=lyx.git diff --git a/src/support/unicode.cpp b/src/support/unicode.cpp index 09bf8c450e..26f834a58b 100644 --- a/src/support/unicode.cpp +++ b/src/support/unicode.cpp @@ -15,20 +15,20 @@ #include "support/unicode.h" #include "support/debug.h" +#include + #include #include #include -#include -#include #include +#include +//Needed in MSVC +#include -using std::endl; -using std::map; -using std::make_pair; -using std::string; -using std::vector; + +using namespace std; namespace { @@ -52,21 +52,25 @@ namespace lyx { static const iconv_t invalid_cd = (iconv_t)(-1); -struct IconvProcessor::Impl +class IconvProcessor::Impl { - Impl(std::string const & to, std::string const & from) + // noncopyable because iconv_close() is called in destructor + Impl(Impl const &); + Impl & operator=(Impl const &); +public: + 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_; }; @@ -88,10 +92,13 @@ IconvProcessor::~IconvProcessor() } -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; } @@ -137,7 +144,12 @@ int IconvProcessor::convert(char const * buf, size_t buflen, 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; @@ -156,7 +168,7 @@ int IconvProcessor::convert(char const * buf, size_t buflen, << " 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 @@ -164,14 +176,14 @@ int IconvProcessor::convert(char const * buf, size_t buflen, *reinterpret_cast(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 @@ -179,7 +191,7 @@ int IconvProcessor::convert(char const * buf, size_t buflen, *reinterpret_cast(buf + i); lyxerr << " 0x" << (unsigned int)b; } - lyxerr << std::dec << endl; + lyxerr << dec << endl; break; default: lyxerr << "\tSome other error: " << errno << endl; @@ -195,6 +207,18 @@ int IconvProcessor::convert(char const * buf, size_t buflen, } +std::string IconvProcessor::from() const +{ + return pimpl_->fromcode_; +} + + +std::string IconvProcessor::to() const +{ + return pimpl_->tocode_; +} + + namespace { @@ -208,23 +232,39 @@ iconv_convert(IconvProcessor & processor, InType const * buf, size_t buflen) char const * inbuf = reinterpret_cast(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 *> static_outbuf; + if (!static_outbuf.hasLocalData()) + static_outbuf.setLocalData(new std::vector(32768)); + std::vector & 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 const * tmp = reinterpret_cast(out); + RetType const * tmp = reinterpret_cast(&outbuf[0]); return vector(tmp, tmp + bytes / sizeof(RetType)); } } // anon namespace +IconvProcessor & utf8ToUcs4() +{ + static QThreadStorage processor; + if (!processor.hasLocalData()) + processor.setLocalData(new IconvProcessor(ucs4_codeset, "UTF-8")); + return *processor.localData(); +} + + vector utf8_to_ucs4(vector const & utf8str) { if (utf8str.empty()) @@ -237,32 +277,43 @@ vector utf8_to_ucs4(vector const & utf8str) vector utf8_to_ucs4(char const * utf8str, size_t ls) { - static IconvProcessor processor(ucs4_codeset, "UTF-8"); - return iconv_convert(processor, utf8str, ls); + return iconv_convert(utf8ToUcs4(), utf8str, ls); } vector utf16_to_ucs4(unsigned short const * s, size_t ls) { - static IconvProcessor processor(ucs4_codeset, utf16_codeset); - return iconv_convert(processor, s, ls); + static QThreadStorage processor; + if (!processor.hasLocalData()) + processor.setLocalData(new IconvProcessor(ucs4_codeset, utf16_codeset)); + return iconv_convert(*processor.localData(), s, ls); } vector ucs4_to_utf16(char_type const * s, size_t ls) { - static IconvProcessor processor(utf16_codeset, ucs4_codeset); - return iconv_convert(processor, s, ls); + static QThreadStorage processor; + if (!processor.hasLocalData()) + processor.setLocalData(new IconvProcessor(utf16_codeset, ucs4_codeset)); + return iconv_convert(*processor.localData(), s, ls); +} + + +IconvProcessor & ucs4ToUtf8() +{ + static QThreadStorage processor; + if (!processor.hasLocalData()) + processor.setLocalData(new IconvProcessor("UTF-8", ucs4_codeset)); + return *processor.localData(); } vector ucs4_to_utf8(char_type c) { - static IconvProcessor processor("UTF-8", ucs4_codeset); - return iconv_convert(processor, &c, 1); + return iconv_convert(ucs4ToUtf8(), &c, 1); } @@ -279,15 +330,17 @@ ucs4_to_utf8(vector const & ucs4str) vector ucs4_to_utf8(char_type const * ucs4str, size_t ls) { - static IconvProcessor processor("UTF-8", ucs4_codeset); - return iconv_convert(processor, ucs4str, ls); + return iconv_convert(ucs4ToUtf8(), ucs4str, ls); } vector eightbit_to_ucs4(char const * s, size_t ls, string const & encoding) { - static map processors; + static QThreadStorage *> static_processors; + if (!static_processors.hasLocalData()) + static_processors.setLocalData(new map); + map & processors = *static_processors.localData(); if (processors.find(encoding) == processors.end()) { IconvProcessor processor(ucs4_codeset, encoding.c_str()); processors.insert(make_pair(encoding, processor)); @@ -296,10 +349,23 @@ eightbit_to_ucs4(char const * s, size_t ls, string const & encoding) } +namespace { + +map & ucs4To8bitProcessors() +{ + static QThreadStorage *> processors; + if (!processors.hasLocalData()) + processors.setLocalData(new map); + return *processors.localData(); +} + +} + + vector ucs4_to_eightbit(char_type const * ucs4str, size_t ls, string const & encoding) { - static map processors; + map & processors(ucs4To8bitProcessors()); if (processors.find(encoding) == processors.end()) { IconvProcessor processor(encoding.c_str(), ucs4_codeset); processors.insert(make_pair(encoding, processor)); @@ -310,7 +376,7 @@ ucs4_to_eightbit(char_type const * ucs4str, size_t ls, string const & encoding) char ucs4_to_eightbit(char_type ucs4, string const & encoding) { - static map processors; + map & processors(ucs4To8bitProcessors()); map::iterator it = processors.find(encoding); if (it == processors.end()) { IconvProcessor processor(encoding.c_str(), ucs4_codeset); @@ -328,7 +394,10 @@ char ucs4_to_eightbit(char_type ucs4, string const & encoding) void ucs4_to_multibytes(char_type ucs4, vector & out, string const & encoding) { - static map processors; + static QThreadStorage *> static_processors; + if (!static_processors.hasLocalData()) + static_processors.setLocalData(new map); + map & processors = *static_processors.localData(); map::iterator it = processors.find(encoding); if (it == processors.end()) { IconvProcessor processor(encoding.c_str(), ucs4_codeset); @@ -343,4 +412,34 @@ void ucs4_to_multibytes(char_type ucs4, vector & out, 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