3 * This file is part of LyX, the document processor.
4 * Licence details can be found in the file COPYING.
6 * \author Lars Gullik Bjønnes
8 * Full author contact details are available in file CREDITS.
10 * A collection of unicode conversion functions, using iconv.
15 #include "support/unicode.h"
16 #include "support/debug.h"
20 #include <boost/cstdint.hpp>
32 #ifdef WORDS_BIGENDIAN
33 char const * utf16_codeset = "UTF16-BE";
35 char const * utf16_codeset = "UTF16-LE";
43 #ifdef WORDS_BIGENDIAN
44 char const * ucs4_codeset = "UCS-4BE";
46 char const * ucs4_codeset = "UCS-4LE";
49 static const iconv_t invalid_cd = (iconv_t)(-1);
52 struct IconvProcessor::Impl
54 Impl(std::string const & to, std::string const & from)
55 : cd(invalid_cd), tocode_(to), fromcode_(from)
60 if (cd != invalid_cd && iconv_close(cd) == -1)
61 LYXERR0("Error returned from iconv_close(" << errno << ")");
66 std::string fromcode_;
70 IconvProcessor::IconvProcessor(char const * tocode, char const * fromcode)
71 : pimpl_(new IconvProcessor::Impl(tocode, fromcode))
76 IconvProcessor::IconvProcessor(IconvProcessor const & other)
77 : pimpl_(new IconvProcessor::Impl(other.pimpl_->tocode_, other.pimpl_->fromcode_))
82 IconvProcessor::~IconvProcessor()
88 void IconvProcessor::operator=(IconvProcessor const & other)
91 pimpl_ = new Impl(other.pimpl_->tocode_, other.pimpl_->fromcode_);
95 bool IconvProcessor::init()
97 if (pimpl_->cd != invalid_cd)
100 pimpl_->cd = iconv_open(pimpl_->tocode_.c_str(), pimpl_->fromcode_.c_str());
101 if (pimpl_->cd != invalid_cd)
104 lyxerr << "Error returned from iconv_open" << endl;
107 lyxerr << "EINVAL The conversion from " << pimpl_->fromcode_
108 << " to " << pimpl_->tocode_
109 << " is not supported by the implementation."
113 lyxerr << "\tSome other error: " << errno << endl;
120 int IconvProcessor::convert(char const * buf, size_t buflen,
121 char * outbuf, size_t maxoutsize)
126 if (pimpl_->cd == invalid_cd) {
131 char ICONV_CONST * inbuf = const_cast<char ICONV_CONST *>(buf);
132 size_t inbytesleft = buflen;
133 size_t outbytesleft = maxoutsize;
135 int res = iconv(pimpl_->cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
137 //lyxerr << std::dec;
138 //lyxerr << "Inbytesleft: " << inbytesleft << endl;
139 //lyxerr << "Outbytesleft: " << outbytesleft << endl;
142 // Everything went well.
143 return maxoutsize - outbytesleft;
145 // There are some errors in the conversion
146 lyxerr << "Error returned from iconv" << endl;
149 lyxerr << "E2BIG There is not sufficient room at *outbuf." << endl;
152 lyxerr << "EILSEQ An invalid multibyte sequence"
153 << " has been encountered in the input.\n"
154 << "When converting from " << pimpl_->fromcode_
155 << " to " << pimpl_->tocode_ << ".\n";
156 lyxerr << "Input:" << std::hex;
157 for (size_t i = 0; i < buflen; ++i) {
158 // char may be signed, avoid output of
159 // something like 0xffffffc2
160 boost::uint32_t const b =
161 *reinterpret_cast<unsigned char const *>(buf + i);
162 lyxerr << " 0x" << (unsigned int)b;
164 lyxerr << std::dec << endl;
167 lyxerr << "EINVAL An incomplete multibyte sequence"
168 << " has been encountered in the input.\n"
169 << "When converting from " << pimpl_->fromcode_
170 << " to " << pimpl_->tocode_ << ".\n";
171 lyxerr << "Input:" << std::hex;
172 for (size_t i = 0; i < buflen; ++i) {
173 // char may be signed, avoid output of
174 // something like 0xffffffc2
175 boost::uint32_t const b =
176 *reinterpret_cast<unsigned char const *>(buf + i);
177 lyxerr << " 0x" << (unsigned int)b;
179 lyxerr << std::dec << endl;
182 lyxerr << "\tSome other error: " << errno << endl;
185 // We got an error so we close down the conversion engine
186 if (iconv_close(pimpl_->cd) == -1) {
187 lyxerr << "Error returned from iconv_close("
188 << errno << ")" << endl;
190 pimpl_->cd = invalid_cd;
198 template<typename RetType, typename InType>
200 iconv_convert(IconvProcessor & processor, InType const * buf, size_t buflen)
203 return vector<RetType>();
205 char const * inbuf = reinterpret_cast<char const *>(buf);
206 size_t inbytesleft = buflen * sizeof(InType);
208 size_t const outsize = 32768;
209 static char out[outsize];
212 int bytes = processor.convert(inbuf, inbytesleft, outbuf, outsize);
215 // FIXME Maybe throw an exception and handle that in the caller?
216 return vector<RetType>();
218 RetType const * tmp = reinterpret_cast<RetType const *>(out);
219 return vector<RetType>(tmp, tmp + bytes / sizeof(RetType));
225 vector<char_type> utf8_to_ucs4(vector<char> const & utf8str)
228 return vector<char_type>();
230 return utf8_to_ucs4(&utf8str[0], utf8str.size());
235 utf8_to_ucs4(char const * utf8str, size_t ls)
237 static IconvProcessor processor(ucs4_codeset, "UTF-8");
238 return iconv_convert<char_type>(processor, utf8str, ls);
243 utf16_to_ucs4(unsigned short const * s, size_t ls)
245 static IconvProcessor processor(ucs4_codeset, utf16_codeset);
246 return iconv_convert<char_type>(processor, s, ls);
250 vector<unsigned short>
251 ucs4_to_utf16(char_type const * s, size_t ls)
253 static IconvProcessor processor(utf16_codeset, ucs4_codeset);
254 return iconv_convert<unsigned short>(processor, s, ls);
259 ucs4_to_utf8(char_type c)
261 static IconvProcessor processor("UTF-8", ucs4_codeset);
262 return iconv_convert<char>(processor, &c, 1);
267 ucs4_to_utf8(vector<char_type> const & ucs4str)
270 return vector<char>();
272 return ucs4_to_utf8(&ucs4str[0], ucs4str.size());
277 ucs4_to_utf8(char_type const * ucs4str, size_t ls)
279 static IconvProcessor processor("UTF-8", ucs4_codeset);
280 return iconv_convert<char>(processor, ucs4str, ls);
285 eightbit_to_ucs4(char const * s, size_t ls, string const & encoding)
287 static map<string, IconvProcessor> processors;
288 if (processors.find(encoding) == processors.end()) {
289 IconvProcessor processor(ucs4_codeset, encoding.c_str());
290 processors.insert(make_pair(encoding, processor));
292 return iconv_convert<char_type>(processors[encoding], s, ls);
297 ucs4_to_eightbit(char_type const * ucs4str, size_t ls, string const & encoding)
299 static map<string, IconvProcessor> processors;
300 if (processors.find(encoding) == processors.end()) {
301 IconvProcessor processor(encoding.c_str(), ucs4_codeset);
302 processors.insert(make_pair(encoding, processor));
304 return iconv_convert<char>(processors[encoding], ucs4str, ls);
308 char ucs4_to_eightbit(char_type ucs4, string const & encoding)
310 static map<string, IconvProcessor> processors;
311 map<string, IconvProcessor>::iterator it = processors.find(encoding);
312 if (it == processors.end()) {
313 IconvProcessor processor(encoding.c_str(), ucs4_codeset);
314 it = processors.insert(make_pair(encoding, processor)).first;
318 int const bytes = it->second.convert((char *)(&ucs4), 4, &out, 1);
325 void ucs4_to_multibytes(char_type ucs4, vector<char> & out,
326 string const & encoding)
328 static map<string, IconvProcessor> processors;
329 map<string, IconvProcessor>::iterator it = processors.find(encoding);
330 if (it == processors.end()) {
331 IconvProcessor processor(encoding.c_str(), ucs4_codeset);
332 it = processors.insert(make_pair(encoding, processor)).first;
336 int bytes = it->second.convert((char *)(&ucs4), 4, &out[0], 4);