3 * This file is part of LyX, the document processor.
4 * Licence details can be found in the file COPYING.
8 * Full author contact details are available in file CREDITS.
13 #include "docstring.h"
14 #include "qstring_helpers.h"
23 #include <boost/assert.hpp>
29 docstring const from_ascii(char const * ascii)
32 for (char const * c = ascii; *c; ++c) {
33 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
40 docstring const from_ascii(std::string const & ascii)
42 int const len = ascii.length();
43 for (int i = 0; i < len; ++i)
44 BOOST_ASSERT(static_cast<unsigned char>(ascii[i]) < 0x80);
45 return docstring(ascii.begin(), ascii.end());
49 std::string const to_ascii(docstring const & ucs4)
51 int const len = ucs4.length();
54 for (int i = 0; i < len; ++i) {
55 BOOST_ASSERT(ucs4[i] < 0x80);
56 ascii[i] = static_cast<char>(ucs4[i]);
62 IconvProcessor & utf8ToUcs4()
64 static IconvProcessor iconv(ucs4_codeset, "UTF-8");
70 void utf8_to_ucs4(std::string const & utf8, docstring & ucs4)
72 size_t n = utf8.size();
73 // as utf8 is a multi-byte encoding, there would be at most
79 int maxoutsize = n * 4;
80 // basic_string::data() is not recognized by some old gcc version
81 // so we use &(ucs4[0]) instead.
82 char * outbuf = (char *)(&(ucs4[0]));
83 int bytes = utf8ToUcs4().convert(utf8.c_str(), n, outbuf, maxoutsize);
85 // adjust to the real converted size
90 docstring const from_utf8(std::string const & utf8)
93 utf8_to_ucs4(utf8, ucs4);
98 std::string const to_utf8(docstring const & ucs4)
100 std::vector<char> const utf8 =
101 ucs4_to_utf8(ucs4.data(), ucs4.size());
102 return std::string(utf8.begin(), utf8.end());
106 docstring const from_local8bit(std::string const & s)
108 return qstring_to_ucs4(QString::fromLocal8Bit(s.data(), s.length()));
112 const char* to_local8bit_failure::what() const throw()
114 return "A string could not be converted from unicode to the local 8 bit encoding.";
118 std::string const to_local8bit(docstring const & s)
120 // This conversion can fail, depending on input.
122 return std::string();
123 QByteArray const local = toqstr(s).toLocal8Bit();
124 if (local.size() == 0)
125 throw to_local8bit_failure();
126 return std::string(local.begin(), local.end());
130 docstring const from_filesystem8bit(std::string const & s)
132 QByteArray const encoded(s.c_str(), s.length());
133 return qstring_to_ucs4(QFile::decodeName(encoded));
137 std::string const to_filesystem8bit(docstring const & s)
139 QByteArray const encoded = QFile::encodeName(toqstr(s));
140 return std::string(encoded.begin(), encoded.end());
144 docstring const normalize_c(docstring const & s)
146 return qstring_to_ucs4(toqstr(s).normalized(QString::NormalizationForm_C));
150 bool operator==(lyx::docstring const & l, char const * r)
152 lyx::docstring::const_iterator it = l.begin();
153 lyx::docstring::const_iterator end = l.end();
154 for (; it != end; ++it, ++r) {
155 BOOST_ASSERT(static_cast<unsigned char>(*r) < 0x80);
158 if (*it != static_cast<lyx::docstring::value_type>(*r))
165 lyx::docstring operator+(lyx::docstring const & l, char const * r)
168 for (char const * c = r; *c; ++c) {
169 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
176 lyx::docstring operator+(char const * l, lyx::docstring const & r)
179 for (char const * c = l; *c; ++c) {
180 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
188 lyx::docstring operator+(lyx::docstring const & l, char r)
190 BOOST_ASSERT(static_cast<unsigned char>(r) < 0x80);
191 return l + lyx::docstring::value_type(r);
195 lyx::docstring operator+(char l, lyx::docstring const & r)
197 BOOST_ASSERT(static_cast<unsigned char>(l) < 0x80);
198 return lyx::docstring::value_type(l) + r;
202 lyx::docstring & operator+=(lyx::docstring & l, char const * r)
204 for (char const * c = r; *c; ++c) {
205 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
212 lyx::docstring & operator+=(lyx::docstring & l, char r)
214 BOOST_ASSERT(static_cast<unsigned char>(r) < 0x80);
221 #if ! defined(USE_WCHAR_T) && defined(__GNUC__)
223 // gcc does not have proper locale facets for lyx::char_type if
224 // sizeof(wchar_t) == 2, so we have to implement them on our own.
227 // We get undefined references to these virtual methods. This looks like
228 // a bug in gcc. The implementation here does not do anything useful, since
229 // it is overriden in ascii_ctype_facet.
231 template<> ctype<lyx::char_type>::~ctype() {}
233 ctype<lyx::char_type>::do_is(ctype<lyx::char_type>::mask, lyx::char_type) const { return false; }
234 template<> lyx::char_type const *
235 ctype<lyx::char_type>::do_is(const lyx::char_type *, const lyx::char_type *, ctype<lyx::char_type>::mask *) const { return 0; }
236 template<> const lyx::char_type *
237 ctype<lyx::char_type>::do_scan_is(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
238 template<> const lyx::char_type *
239 ctype<lyx::char_type>::do_scan_not(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
240 template<> lyx::char_type ctype<lyx::char_type>::do_toupper(lyx::char_type) const { return 0; }
241 template<> const lyx::char_type * ctype<lyx::char_type>::do_toupper(lyx::char_type *, lyx::char_type const *) const { return 0; }
242 template<> lyx::char_type ctype<lyx::char_type>::do_tolower(lyx::char_type) const { return 0; }
243 template<> const lyx::char_type * ctype<lyx::char_type>::do_tolower(lyx::char_type *, lyx::char_type const *) const { return 0; }
244 template<> lyx::char_type ctype<lyx::char_type>::do_widen(char) const { return 0; }
245 template<> const char *
246 ctype<lyx::char_type>::do_widen(const char *, const char *, lyx::char_type *) const { return 0; }
248 ctype<lyx::char_type>::do_narrow(const lyx::char_type, char) const { return 0; }
249 template<> const lyx::char_type *
250 ctype<lyx::char_type>::do_narrow(const lyx::char_type *, const lyx::char_type *, char, char *) const { return 0; }
256 class ctype_failure : public std::bad_cast {
258 ctype_failure() throw() : std::bad_cast() {}
259 virtual ~ctype_failure() throw() {}
260 virtual const char* what() const throw()
262 return "The ctype<lyx::char_type> locale facet does only support ASCII characters on this platform.";
267 class num_put_failure : public std::bad_cast {
269 num_put_failure() throw() : std::bad_cast() {}
270 virtual ~num_put_failure() throw() {}
271 virtual const char* what() const throw()
273 return "The num_put locale facet does only support ASCII characters on this platform.";
278 /// ctype facet for UCS4 characters. The implementation does only support pure
279 /// ASCII, since we do not need anything else for now.
280 /// The code is partly stolen from std::ctype<wchar_t> from gcc.
281 class ascii_ctype_facet : public std::ctype<lyx::char_type>
284 typedef lyx::char_type char_type;
285 typedef wctype_t wmask_type;
286 explicit ascii_ctype_facet(size_t refs = 0) : std::ctype<char_type>(refs)
288 M_initialize_ctype();
293 wint_t M_widen[1 + static_cast<unsigned char>(-1)];
295 wmask_type M_wmask[16];
296 wmask_type M_convert_to_wmask(const mask m) const
300 case space: ret = wctype("space"); break;
301 case print: ret = wctype("print"); break;
302 case cntrl: ret = wctype("cntrl"); break;
303 case upper: ret = wctype("upper"); break;
304 case lower: ret = wctype("lower"); break;
305 case alpha: ret = wctype("alpha"); break;
306 case digit: ret = wctype("digit"); break;
307 case punct: ret = wctype("punct"); break;
308 case xdigit: ret = wctype("xdigit"); break;
309 case alnum: ret = wctype("alnum"); break;
310 case graph: ret = wctype("graph"); break;
311 default: ret = wmask_type();
315 void M_initialize_ctype()
318 for (i = 0; i < 128; ++i) {
319 const int c = wctob(i);
323 M_narrow[i] = static_cast<char>(c);
329 for (size_t i = 0; i < sizeof(M_widen) / sizeof(wint_t); ++i)
330 M_widen[i] = btowc(i);
332 for (size_t i = 0; i <= 15; ++i) {
333 M_bit[i] = static_cast<mask>(1 << i);
334 M_wmask[i] = M_convert_to_wmask(M_bit[i]);
337 virtual ~ascii_ctype_facet() {}
338 char_type do_toupper(char_type c) const
341 throw ctype_failure();
342 return toupper(static_cast<int>(c));
344 char_type const * do_toupper(char_type * lo, char_type const * hi) const
348 throw ctype_failure();
349 *lo = toupper(static_cast<int>(*lo));
354 char_type do_tolower(char_type c) const
357 throw ctype_failure();
360 char_type const * do_tolower(char_type * lo, char_type const * hi) const
364 throw ctype_failure();
370 bool do_is(mask m, char_type c) const
373 throw ctype_failure();
374 // The code below works because c is in the ASCII range.
375 // We could not use iswctype() which is designed for a 2byte
376 // whar_t without encoding conversion otherwise.
378 // Generically, 15 (instead of 10) since we don't know the numerical
379 // encoding of the various categories in /usr/include/ctype.h.
380 const size_t bitmasksize = 15;
381 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
382 if (m & M_bit[bitcur] &&
383 iswctype(static_cast<int>(c), M_wmask[bitcur])) {
389 char_type const * do_is(char_type const * lo, char_type const * hi, mask * vec) const
391 for (;lo < hi; ++vec, ++lo) {
393 throw ctype_failure();
394 // The code below works because c is in the ASCII range.
395 // We could not use iswctype() which is designed for a 2byte
396 // whar_t without encoding conversion otherwise.
397 // Generically, 15 (instead of 10) since we don't know the numerical
398 // encoding of the various categories in /usr/include/ctype.h.
399 const size_t bitmasksize = 15;
401 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
402 if (iswctype(static_cast<int>(*lo), M_wmask[bitcur]))
408 char_type const * do_scan_is(mask m, char_type const * lo, char_type const * hi) const
410 while (lo < hi && !this->do_is(m, *lo))
414 char_type const * do_scan_not(mask m, char_type const * lo, char_type const * hi) const
416 while (lo < hi && this->do_is(m, *lo) != 0)
420 char_type do_widen(char c) const
422 if (static_cast<unsigned char>(c) < 0x80)
424 throw ctype_failure();
426 const char* do_widen(const char* lo, const char* hi, char_type* dest) const
429 if (static_cast<unsigned char>(*lo) >= 0x80)
430 throw ctype_failure();
437 char do_narrow(char_type wc, char) const
440 return static_cast<char>(wc);
441 throw ctype_failure();
443 const char_type * do_narrow(const char_type * lo, const char_type * hi, char, char * dest) const
447 *dest = static_cast<char>(*lo);
449 throw ctype_failure();
458 /// Facet for outputting numbers to odocstreams as ascii.
459 /// Here we simply need defining the virtual do_put functions.
460 class ascii_num_put_facet : public std::num_put<lyx::char_type, std::ostreambuf_iterator<lyx::char_type, std::char_traits<lyx::char_type> > >
462 typedef std::ostreambuf_iterator<lyx::char_type, std::char_traits<lyx::char_type> > iter_type;
464 ascii_num_put_facet(size_t refs = 0) : std::num_put<lyx::char_type, iter_type>(refs) {}
466 /// Facet for converting numbers to ascii strings.
467 class string_num_put_facet : public std::num_put<char, std::basic_string<char>::iterator>
470 string_num_put_facet() : std::num_put<char, std::basic_string<char>::iterator>(1) {}
475 do_put(iter_type oit, std::ios_base & b, char_type fill, bool v) const
477 return do_put_helper(oit, b, fill, v);
481 do_put(iter_type oit, std::ios_base & b, char_type fill, long v) const
483 return do_put_helper(oit, b, fill, v);
487 do_put(iter_type oit, std::ios_base & b, char_type fill, unsigned long v) const
489 return do_put_helper(oit, b, fill, v);
492 #ifdef _GLIBCXX_USE_LONG_LONG
494 do_put(iter_type oit, std::ios_base & b, char_type fill, long long v) const
496 return do_put_helper(oit, b, fill, v);
500 do_put(iter_type oit, std::ios_base & b, char_type fill, unsigned long long v) const
502 return do_put_helper(oit, b, fill, v);
507 do_put(iter_type oit, std::ios_base & b, char_type fill, double v) const
509 return do_put_helper(oit, b, fill, v);
513 do_put(iter_type oit, std::ios_base & b, char_type fill, long double v) const
515 return do_put_helper(oit, b, fill, v);
519 do_put(iter_type oit, std::ios_base & b, char_type fill, void const * v) const
521 return do_put_helper(oit, b, fill, v);
525 template <typename ValueType>
527 do_put_helper(iter_type oit, std::ios_base & b, char_type fill, ValueType v) const
530 throw num_put_failure();
532 std::streamsize const sz = b.width() > b.precision() ?
533 b.width() : b.precision();
534 // 64 is large enough, unless width or precision are bigger
535 std::streamsize const wd = (sz > 56 ? sz : 56) + 8;
536 std::string s(wd, '\0');
537 string_num_put_facet f;
538 std::string::const_iterator cit = s.begin();
539 std::string::const_iterator end =
540 f.put(s.begin(), b, fill, v);
541 for (; cit != end; ++cit, ++oit)
549 /// Facet for inputting ascii representations of numbers from idocstreams.
550 /// Here we simply need defining the virtual do_get functions.
551 class ascii_num_get_facet : public std::num_get<lyx::char_type, std::istreambuf_iterator<lyx::char_type, std::char_traits<lyx::char_type> > >
553 typedef std::istreambuf_iterator<lyx::char_type, std::char_traits<lyx::char_type> > iter_type;
555 ascii_num_get_facet(size_t refs = 0) : std::num_get<lyx::char_type, iter_type>(refs) {}
557 /// Facet for converting ascii representation of numbers to a value.
558 class string_num_get_facet : public std::num_get<char, std::basic_string<char>::iterator>
561 string_num_get_facet() : std::num_get<char, std::basic_string<char>::iterator>(1) {}
564 /// Numpunct facet defining the I/O format.
565 class numpunct_facet : public std::numpunct<char>
568 numpunct_facet() : std::numpunct<char>(1) {}
573 do_get(iter_type iit, iter_type eit, std::ios_base & b,
574 std::ios_base::iostate & err, bool & v) const
576 if (b.flags() & std::ios_base::boolalpha) {
578 lyx::docstring const truename = from_local8bit(p.truename());
579 lyx::docstring const falsename = from_local8bit(p.falsename());
584 size_t const tsize = truename.size();
585 size_t const fsize = falsename.size();
586 for (; iit != eit; ++iit) {
589 bool true_ok = lyx::support::prefixIs(truename, s);
590 bool false_ok = lyx::support::prefixIs(falsename, s);
591 if (!true_ok && !false_ok) {
596 if ((true_ok && n == tsize) ||
597 (false_ok && n == fsize)) {
603 err = std::ios_base::goodbit;
604 v = truename == s ? true : false;
606 err = std::ios_base::failbit;
608 err |= std::ios_base::eofbit;
612 iter_type end = this->do_get(iit, eit, b, err, l);
613 if (!(err & std::ios_base::failbit)) {
619 err |= std::ios_base::failbit;
626 do_get(iter_type iit, iter_type eit, std::ios_base & b,
627 std::ios_base::iostate & err, long & v) const
629 return do_get_integer(iit, eit, b, err, v);
633 do_get(iter_type iit, iter_type eit, std::ios_base & b,
634 std::ios_base::iostate & err, unsigned short & v) const
636 return do_get_integer(iit, eit, b, err, v);
640 do_get(iter_type iit, iter_type eit, std::ios_base & b,
641 std::ios_base::iostate & err, unsigned int & v) const
643 return do_get_integer(iit, eit, b, err, v);
647 do_get(iter_type iit, iter_type eit, std::ios_base & b,
648 std::ios_base::iostate & err, unsigned long & v) const
650 return do_get_integer(iit, eit, b, err, v);
653 #ifdef _GLIBCXX_USE_LONG_LONG
655 do_get(iter_type iit, iter_type eit, std::ios_base & b,
656 std::ios_base::iostate & err, long long & v) const
658 return do_get_integer(iit, eit, b, err, v);
662 do_get(iter_type iit, iter_type eit, std::ios_base & b,
663 std::ios_base::iostate & err, unsigned long long & v) const
665 return do_get_integer(iit, eit, b, err, v);
670 do_get(iter_type iit, iter_type eit, std::ios_base & b,
671 std::ios_base::iostate & err, float & v) const
673 return do_get_float(iit, eit, b, err, v);
677 do_get(iter_type iit, iter_type eit, std::ios_base & b,
678 std::ios_base::iostate & err, double & v) const
680 return do_get_float(iit, eit, b, err, v);
684 do_get(iter_type iit, iter_type eit, std::ios_base & b,
685 std::ios_base::iostate & err, long double & v) const
687 return do_get_float(iit, eit, b, err, v);
691 do_get(iter_type iit, iter_type eit, std::ios_base & b,
692 std::ios_base::iostate & err, void * & v) const
695 iter_type end = do_get_integer(iit, eit, b, err, val);
696 if (!(err & std::ios_base::failbit))
697 v = reinterpret_cast<void *>(val);
702 template <typename ValueType>
704 do_get_integer(iter_type iit, iter_type eit, std::ios_base & b,
705 std::ios_base::iostate & err, ValueType & v) const
709 for (; iit != eit && isNumpunct(*iit); ++iit)
710 s += static_cast<char>(*iit);
711 // We add another character, not part of the numpunct facet,
712 // in order to avoid setting the eofbit in the stream state,
713 // which would prevent any further read. The space seems a
716 string_num_get_facet f;
717 f.get(s.begin(), s.end(), b, err, v);
719 err |= std::ios_base::eofbit;
724 bool isNumpunct(lyx::char_type const c) const
726 /// Only account for the standard numpunct "C" locale facet.
727 return c < 0x80 && (c == '-' || c == '+' || isdigit(c)
728 || ('a' <= c && c <= 'f') || ('A' <= c && c <= 'F')
729 || c == 'x' || c == 'X');
732 template <typename ValueType>
734 do_get_float(iter_type iit, iter_type eit, std::ios_base & b,
735 std::ios_base::iostate & err, ValueType & v) const
737 // Gather a string of the form
738 // [+-]? [0-9]* .? [0-9]* ([eE] [+-]? [0-9]+)?
743 char const dot = p.decimal_point();
744 char const sep = p.thousands_sep();
745 // Get an optional sign
746 if (iit != eit && (*iit == '-' || *iit == '+')) {
747 s += static_cast<char>(*iit);
750 for (; iit != eit && isDigitOrSep(*iit, sep); ++iit)
751 s += static_cast<char>(*iit);
752 if (iit != eit && *iit == dot) {
755 for (; iit != eit && isDigitOrSep(*iit, 0); ++iit)
756 s += static_cast<char>(*iit);
757 if (iit != eit && (*iit == 'e' || *iit == 'E')) {
758 s += static_cast<char>(*iit);
760 for (; iit != eit && isDigitOrSep(*iit, 0); ++iit)
761 s += static_cast<char>(*iit);
765 string_num_get_facet f;
766 f.get(s.begin(), s.end(), b, err, v);
768 err |= std::ios_base::eofbit;
773 bool isDigitOrSep(lyx::char_type const c, char const sep) const
775 return (c >= '0' && c <= '9') || (c != 0 && c == sep);
780 /// class to add our facets to the global locale
781 class locale_initializer {
786 std::locale const loc1(global, new ascii_ctype_facet);
787 std::locale const loc2(loc1, new ascii_num_put_facet);
788 std::locale const loc3(loc2, new ascii_num_get_facet);
789 std::locale::global(loc3);
796 /// make sure that our facets get used
797 static locale_initializer initializer;