anne/lyx_mirror
1
0
Fork 0
mirror of https://git.lyx.org/repos/lyx.git synced 2024-11-11 05:33:33 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Some further work and simplification. Now the conversion engine is only restarted on error.

Browse Source 
git-svn-id: svn://svn.lyx.org/lyx/lyx-devel/trunk@14967 a592a061-630c-0410-9148-cb99ea01b6c8
This commit is contained in:
Lars Gullik Bjønnes 2006-09-10 16:59:36 +00:00
parent 8fd2299744
commit 2a06b31354
1 changed files with 54 additions and 77 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

131
src/support/unicode.C
View File

@ -20,10 +20,8 @@
#include <cerrno> #include <cerrno>
#include <iomanip> #include <iomanip>
#include <string>
using std::endl; using std::endl;
using std::string;
namespace { namespace {
@ -35,36 +33,42 @@ namespace {
char const * ucs2_codeset = "UCS-2LE"; char const * ucs2_codeset = "UCS-2LE";
#endif #endif
std::vector<char> template<typename RetType, typename InType>
iconv_convert(std::string const & tocode, std::string const & fromcode, std::vector<RetType>
std::vector<char> const & buf) iconv_convert(iconv_t * cd,
char const * tocode,
char const * fromcode,
InType const * buf,
size_t buflen)
{ {
if (buf.empty()) if (buflen == 0)
return std::vector<char>(); return std::vector<RetType>();
iconv_t cd = iconv_open(tocode.c_str(), fromcode.c_str()); if (*cd == (iconv_t)(-1)) {
if (cd == (iconv_t)(-1)) { *cd = iconv_open(tocode, fromcode);
lyxerr << "Error returned from iconv_open" << endl; if (*cd == (iconv_t)(-1)) {
switch (errno) { lyxerr << "Error returned from iconv_open" << endl;
case EINVAL: switch (errno) {
lyxerr << "EINVAL The conversion from " << fromcode case EINVAL:
<< " to " << tocode lyxerr << "EINVAL The conversion from " << fromcode
<< " is not supported by the implementation." << " to " << tocode
<< endl; << " is not supported by the implementation."
break; << endl;
default: break;
lyxerr << "\tSome other error: " << errno << endl; default:
break; lyxerr << "\tSome other error: " << errno << endl;
break;
}
} }
} }
char ICONV_CONST * inbuf = const_cast<char ICONV_CONST *>(&buf[0]); char ICONV_CONST * inbuf = const_cast<char ICONV_CONST *>(reinterpret_cast<char const *>(buf));
size_t inbytesleft = buf.size(); size_t inbytesleft = buflen * sizeof(InType);
static char out[1000]; static char out[1000];
char * outbuf = out; char * outbuf = out;
size_t outbytesleft = 1000; size_t outbytesleft = 1000;
size_t res = iconv(cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft); size_t res = iconv(*cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
if (res == (size_t)(-1)) { if (res == (size_t)(-1)) {
lyxerr << "Error returned from iconv" << endl; lyxerr << "Error returned from iconv" << endl;
@ -78,7 +82,7 @@ iconv_convert(std::string const & tocode, std::string const & fromcode,
<< "When converting from " << fromcode << "When converting from " << fromcode
<< " to " << tocode << ".\n"; << " to " << tocode << ".\n";
lyxerr << "Input: " << std::hex; lyxerr << "Input: " << std::hex;
for (size_t i = 0; i < buf.size(); ++i) { for (size_t i = 0; i < buflen; ++i) {
unsigned char const b = buf[i]; unsigned char const b = buf[i];
lyxerr << "0x" << int(b) << " "; lyxerr << "0x" << int(b) << " ";
} }
@ -90,7 +94,7 @@ iconv_convert(std::string const & tocode, std::string const & fromcode,
<< "When converting from " << fromcode << "When converting from " << fromcode
<< " to " << tocode << ".\n"; << " to " << tocode << ".\n";
lyxerr << "Input: " << std::hex; lyxerr << "Input: " << std::hex;
for (size_t i = 0; i < buf.size(); ++i) { for (size_t i = 0; i < buflen; ++i) {
unsigned char const b = buf[i]; unsigned char const b = buf[i];
lyxerr << "0x" << int(b) << " "; lyxerr << "0x" << int(b) << " ";
} }
@ -100,11 +104,12 @@ iconv_convert(std::string const & tocode, std::string const & fromcode,
lyxerr << "\tSome other error: " << errno << endl; lyxerr << "\tSome other error: " << errno << endl;
break; break;
} }
} // We got an error so we close down the conversion engine
if (iconv_close(*cd) == -1) {
if (iconv_close(cd) == -1) { lyxerr << "Error returned from iconv_close("
lyxerr << "Error returned from iconv_close(" << errno << ")" << endl;
<< errno << ")" << endl; }
*cd = (iconv_t)(-1);
} }
//lyxerr << std::dec; //lyxerr << std::dec;
@ -112,22 +117,8 @@ iconv_convert(std::string const & tocode, std::string const & fromcode,
//lyxerr << "Outbytesleft: " << outbytesleft << endl; //lyxerr << "Outbytesleft: " << outbytesleft << endl;
int bytes = 1000 - outbytesleft; int bytes = 1000 - outbytesleft;
std::vector<char> outvec(out, out + bytes); RetType const * tmp = reinterpret_cast<RetType const *>(out);
return outvec; return std::vector<RetType>(tmp, tmp + bytes / sizeof(RetType));
}
std::vector<boost::uint32_t> bytes_to_ucs4(std::vector<char> const & bytes)
{
boost::uint32_t const * tmp = reinterpret_cast<uint32_t const *>(&bytes[0]);
return std::vector<boost::uint32_t>(tmp, tmp + bytes.size() / 4);
}
std::vector<unsigned short> bytes_to_ucs2(std::vector<char> const & bytes)
{
unsigned short const * tmp = reinterpret_cast<unsigned short const *>(&bytes[0]);
return std::vector<unsigned short>(tmp, tmp + bytes.size() / 2);
} }
} // anon namespace } // anon namespace
@ -135,69 +126,55 @@ std::vector<unsigned short> bytes_to_ucs2(std::vector<char> const & bytes)
std::vector<boost::uint32_t> utf8_to_ucs4(std::vector<char> const & utf8str) std::vector<boost::uint32_t> utf8_to_ucs4(std::vector<char> const & utf8str)
{ {
//lyxerr << "Buff = " << string(utf8str.begin(), utf8str.end()) static iconv_t cd = (iconv_t)(-1);
// << " (" << utf8str.size() << ")" << endl; return iconv_convert<boost::uint32_t>(&cd, ucs4_codeset, "UTF-8",
//lyxerr << "Res = " << string(res.begin(), res.end()) &utf8str[0], utf8str.size());
// << " (" << res.size() << ")" << endl;
std::vector<char> res = iconv_convert(ucs4_codeset, "UTF-8", utf8str);
return bytes_to_ucs4(res);
} }
std::vector<boost::uint32_t> std::vector<boost::uint32_t>
ucs2_to_ucs4(std::vector<unsigned short> const & ucs2str) ucs2_to_ucs4(std::vector<unsigned short> const & ucs2str)
{ {
char const * tin = reinterpret_cast<char const *>(&ucs2str[0]); static iconv_t cd = (iconv_t)(-1);
std::vector<char> in(tin, tin + ucs2str.size() * 2); return iconv_convert<boost::uint32_t>(&cd, ucs4_codeset, ucs2_codeset,
std::vector<char> res = iconv_convert(ucs4_codeset, ucs2_codeset, in); &ucs2str[0], ucs2str.size());
return bytes_to_ucs4(res);
} }
std::vector<unsigned short> std::vector<unsigned short>
ucs4_to_ucs2(std::vector<boost::uint32_t> const & ucs4str) ucs4_to_ucs2(std::vector<boost::uint32_t> const & ucs4str)
{ {
char const * tin = reinterpret_cast<char const *>(&ucs4str[0]); return ucs4_to_ucs2(&ucs4str[0], ucs4str.size());
std::vector<char> in(tin, tin + ucs4str.size() * 4);
std::vector<char> res = iconv_convert(ucs2_codeset, ucs4_codeset, in);
return bytes_to_ucs2(res);
} }
std::vector<unsigned short> std::vector<unsigned short>
ucs4_to_ucs2(boost::uint32_t const * s, size_t ls) ucs4_to_ucs2(boost::uint32_t const * s, size_t ls)
{ {
char const * tin = reinterpret_cast<char const *>(s); static iconv_t cd = (iconv_t)(-1);
std::vector<char> in(tin, tin + ls * 4); return iconv_convert<unsigned short>(&cd, ucs2_codeset, ucs4_codeset,
std::vector<char> res = iconv_convert(ucs2_codeset, ucs4_codeset, in); s, ls);
return bytes_to_ucs2(res);
} }
unsigned short unsigned short
ucs4_to_ucs2(boost::uint32_t c) ucs4_to_ucs2(boost::uint32_t c)
{ {
char const * tin = reinterpret_cast<char const *>(&c); boost::uint32_t tmp[] = { c, 0 };
std::vector<char> in(tin, tin + 4); return ucs4_to_ucs2(tmp, 1)[0];
std::vector<char> res = iconv_convert(ucs2_codeset, ucs4_codeset, in);
return bytes_to_ucs2(res)[0];
} }
std::vector<char> ucs4_to_utf8(std::vector<boost::uint32_t> const & ucs4str) std::vector<char> ucs4_to_utf8(std::vector<boost::uint32_t> const & ucs4str)
{ {
char const * tin = reinterpret_cast<char const *>(&ucs4str[0]); static iconv_t cd = (iconv_t)(-1);
std::vector<char> in(tin, tin + ucs4str.size() * 4); return iconv_convert<char>(&cd, "UTF-8", ucs4_codeset,
std::vector<char> res = iconv_convert("UTF-8", ucs4_codeset, in); &ucs4str[0], ucs4str.size());
return res;
} }
std::vector<char> ucs4_to_utf8(boost::uint32_t c) std::vector<char> ucs4_to_utf8(boost::uint32_t c)
{ {
char const * tin = reinterpret_cast<char const *>(&c); static iconv_t cd = (iconv_t)(-1);
std::vector<char> in(tin, tin + 4); return iconv_convert<char>(&cd, "UTF-8", ucs4_codeset, &c, 1);
std::vector<char> res = iconv_convert("UTF-8", ucs4_codeset, in);
return res;
} }