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 "support/docstring.h"
15 #include "support/lassert.h"
16 #include "support/lstrings.h"
17 #include "support/qstring_helpers.h"
18 #include "support/unicode.h"
31 docstring const from_ascii(char const * ascii)
34 int n = strlen(ascii);
39 LASSERT(static_cast<unsigned char>(ascii[n]) < 0x80, /**/);
45 docstring const from_ascii(string const & ascii)
47 int const len = ascii.length();
48 for (int i = 0; i < len; ++i)
49 LASSERT(static_cast<unsigned char>(ascii[i]) < 0x80, /**/);
50 return docstring(ascii.begin(), ascii.end());
54 string const to_ascii(docstring const & ucs4)
56 int const len = ucs4.length();
59 for (int i = 0; i < len; ++i) {
60 LASSERT(ucs4[i] < 0x80, /**/);
61 ascii[i] = static_cast<char>(ucs4[i]);
67 IconvProcessor & utf8ToUcs4()
69 static IconvProcessor iconv(ucs4_codeset, "UTF-8");
75 void utf8_to_ucs4(string const & utf8, docstring & ucs4)
77 size_t n = utf8.size();
78 // as utf8 is a multi-byte encoding, there would be at most
84 int maxoutsize = n * 4;
85 // basic_string::data() is not recognized by some old gcc version
86 // so we use &(ucs4[0]) instead.
87 char * outbuf = (char *)(&(ucs4[0]));
88 int bytes = utf8ToUcs4().convert(utf8.c_str(), n, outbuf, maxoutsize);
90 // adjust to the real converted size
95 docstring const from_utf8(string const & utf8)
98 utf8_to_ucs4(utf8, ucs4);
103 string const to_utf8(docstring const & ucs4)
105 vector<char> const utf8 = ucs4_to_utf8(ucs4.data(), ucs4.size());
106 return string(utf8.begin(), utf8.end());
110 docstring const from_local8bit(string const & s)
112 return qstring_to_ucs4(QString::fromLocal8Bit(s.data(), s.length()));
116 /// Exception thrown by to_local8bit if the string could not be converted
117 class to_local8bit_failure : public bad_cast {
119 to_local8bit_failure() throw() : bad_cast() {}
120 virtual ~to_local8bit_failure() throw() {}
121 virtual const char* what() const throw()
123 return "A string could not be converted from unicode to the local 8 bit encoding.";
128 string const to_local8bit(docstring const & s)
130 // This conversion can fail, depending on input.
133 QByteArray const local = toqstr(s).toLocal8Bit();
134 if (local.size() == 0)
135 throw to_local8bit_failure();
136 return string(local.begin(), local.end());
140 docstring const from_filesystem8bit(string const & s)
142 QByteArray const encoded(s.c_str(), s.length());
143 return qstring_to_ucs4(QFile::decodeName(encoded));
147 string const to_filesystem8bit(docstring const & s)
149 QByteArray const encoded = QFile::encodeName(toqstr(s));
150 return string(encoded.begin(), encoded.end());
154 docstring const normalize_c(docstring const & s)
156 return qstring_to_ucs4(toqstr(s).normalized(QString::NormalizationForm_C));
160 bool operator==(lyx::docstring const & l, char const * r)
162 lyx::docstring::const_iterator it = l.begin();
163 lyx::docstring::const_iterator end = l.end();
164 for (; it != end; ++it, ++r) {
165 LASSERT(static_cast<unsigned char>(*r) < 0x80, /**/);
168 if (*it != static_cast<lyx::docstring::value_type>(*r))
175 lyx::docstring operator+(lyx::docstring const & l, char const * r)
178 for (char const * c = r; *c; ++c) {
179 LASSERT(static_cast<unsigned char>(*c) < 0x80, /**/);
186 lyx::docstring operator+(char const * l, lyx::docstring const & r)
189 for (char const * c = l; *c; ++c) {
190 LASSERT(static_cast<unsigned char>(*c) < 0x80, /**/);
198 lyx::docstring operator+(lyx::docstring const & l, char r)
200 LASSERT(static_cast<unsigned char>(r) < 0x80, /**/);
202 s += docstring::value_type(r);
207 lyx::docstring operator+(char l, lyx::docstring const & r)
209 LASSERT(static_cast<unsigned char>(l) < 0x80, /**/);
210 return lyx::docstring::value_type(l) + r;
214 lyx::docstring & operator+=(lyx::docstring & l, char const * r)
216 for (char const * c = r; *c; ++c) {
217 LASSERT(static_cast<unsigned char>(*c) < 0x80, /**/);
224 lyx::docstring & operator+=(lyx::docstring & l, char r)
226 LASSERT(static_cast<unsigned char>(r) < 0x80, /**/);
233 #if ! defined(USE_WCHAR_T) && defined(__GNUC__)
235 // gcc does not have proper locale facets for lyx::char_type if
236 // sizeof(wchar_t) == 2, so we have to implement them on our own.
239 // We get undefined references to these virtual methods. This looks like
240 // a bug in gcc. The implementation here does not do anything useful, since
241 // it is overriden in ascii_ctype_facet.
243 template<> ctype<lyx::char_type>::~ctype() {}
245 ctype<lyx::char_type>::do_is(ctype<lyx::char_type>::mask, lyx::char_type) const { return false; }
246 template<> lyx::char_type const *
247 ctype<lyx::char_type>::do_is(const lyx::char_type *, const lyx::char_type *, ctype<lyx::char_type>::mask *) const { return 0; }
248 template<> const lyx::char_type *
249 ctype<lyx::char_type>::do_scan_is(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
250 template<> const lyx::char_type *
251 ctype<lyx::char_type>::do_scan_not(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
252 template<> lyx::char_type ctype<lyx::char_type>::do_toupper(lyx::char_type) const { return 0; }
253 template<> const lyx::char_type * ctype<lyx::char_type>::do_toupper(lyx::char_type *, lyx::char_type const *) const { return 0; }
254 template<> lyx::char_type ctype<lyx::char_type>::do_tolower(lyx::char_type) const { return 0; }
255 template<> const lyx::char_type * ctype<lyx::char_type>::do_tolower(lyx::char_type *, lyx::char_type const *) const { return 0; }
256 template<> lyx::char_type ctype<lyx::char_type>::do_widen(char) const { return 0; }
257 template<> const char *
258 ctype<lyx::char_type>::do_widen(const char *, const char *, lyx::char_type *) const { return 0; }
260 ctype<lyx::char_type>::do_narrow(const lyx::char_type, char) const { return 0; }
261 template<> const lyx::char_type *
262 ctype<lyx::char_type>::do_narrow(const lyx::char_type *, const lyx::char_type *, char, char *) const { return 0; }
268 class ctype_failure : public bad_cast {
270 ctype_failure() throw() : bad_cast() {}
271 virtual ~ctype_failure() throw() {}
272 virtual const char* what() const throw()
274 return "The ctype<lyx::char_type> locale facet does only support ASCII characters on this platform.";
279 class num_put_failure : public bad_cast {
281 num_put_failure() throw() : bad_cast() {}
282 virtual ~num_put_failure() throw() {}
283 virtual const char* what() const throw()
285 return "The num_put locale facet does only support ASCII characters on this platform.";
290 /// ctype facet for UCS4 characters. The implementation does only support pure
291 /// ASCII, since we do not need anything else for now.
292 /// The code is partly stolen from ctype<wchar_t> from gcc.
293 class ascii_ctype_facet : public ctype<lyx::char_type>
296 typedef lyx::char_type char_type;
297 typedef wctype_t wmask_type;
298 explicit ascii_ctype_facet(size_t refs = 0) : ctype<char_type>(refs)
300 M_initialize_ctype();
305 wint_t M_widen[1 + static_cast<unsigned char>(-1)];
307 wmask_type M_wmask[16];
308 wmask_type M_convert_to_wmask(const mask m) const
312 case space: ret = wctype("space"); break;
313 case print: ret = wctype("print"); break;
314 case cntrl: ret = wctype("cntrl"); break;
315 case upper: ret = wctype("upper"); break;
316 case lower: ret = wctype("lower"); break;
317 case alpha: ret = wctype("alpha"); break;
318 case digit: ret = wctype("digit"); break;
319 case punct: ret = wctype("punct"); break;
320 case xdigit: ret = wctype("xdigit"); break;
321 case alnum: ret = wctype("alnum"); break;
322 case graph: ret = wctype("graph"); break;
323 default: ret = wmask_type();
327 void M_initialize_ctype()
330 for (i = 0; i < 128; ++i) {
331 const int c = wctob(i);
335 M_narrow[i] = static_cast<char>(c);
341 for (size_t i = 0; i < sizeof(M_widen) / sizeof(wint_t); ++i)
342 M_widen[i] = btowc(i);
344 for (size_t i = 0; i <= 15; ++i) {
345 M_bit[i] = static_cast<mask>(1 << i);
346 M_wmask[i] = M_convert_to_wmask(M_bit[i]);
349 virtual ~ascii_ctype_facet() {}
350 char_type do_toupper(char_type c) const
353 throw ctype_failure();
354 return toupper(static_cast<int>(c));
356 char_type const * do_toupper(char_type * lo, char_type const * hi) const
360 throw ctype_failure();
361 *lo = toupper(static_cast<int>(*lo));
366 char_type do_tolower(char_type c) const
369 throw ctype_failure();
372 char_type const * do_tolower(char_type * lo, char_type const * hi) const
376 throw ctype_failure();
382 bool do_is(mask m, char_type c) const
385 throw ctype_failure();
386 // The code below works because c is in the ASCII range.
387 // We could not use iswctype() which is designed for a 2byte
388 // whar_t without encoding conversion otherwise.
390 // Generically, 15 (instead of 10) since we don't know the numerical
391 // encoding of the various categories in /usr/include/ctype.h.
392 const size_t bitmasksize = 15;
393 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
394 if (m & M_bit[bitcur] &&
395 iswctype(static_cast<int>(c), M_wmask[bitcur])) {
401 char_type const * do_is(char_type const * lo, char_type const * hi, mask * vec) const
403 for (;lo < hi; ++vec, ++lo) {
405 throw ctype_failure();
406 // The code below works because c is in the ASCII range.
407 // We could not use iswctype() which is designed for a 2byte
408 // whar_t without encoding conversion otherwise.
409 // Generically, 15 (instead of 10) since we don't know the numerical
410 // encoding of the various categories in /usr/include/ctype.h.
411 const size_t bitmasksize = 15;
413 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
414 if (iswctype(static_cast<int>(*lo), M_wmask[bitcur]))
420 char_type const * do_scan_is(mask m, char_type const * lo, char_type const * hi) const
422 while (lo < hi && !this->do_is(m, *lo))
426 char_type const * do_scan_not(mask m, char_type const * lo, char_type const * hi) const
428 while (lo < hi && this->do_is(m, *lo) != 0)
432 char_type do_widen(char c) const
434 if (static_cast<unsigned char>(c) < 0x80)
436 throw ctype_failure();
438 const char* do_widen(const char* lo, const char* hi, char_type* dest) const
441 if (static_cast<unsigned char>(*lo) >= 0x80)
442 throw ctype_failure();
449 char do_narrow(char_type wc, char) const
452 return static_cast<char>(wc);
453 throw ctype_failure();
455 const char_type * do_narrow(const char_type * lo, const char_type * hi, char, char * dest) const
459 *dest = static_cast<char>(*lo);
461 throw ctype_failure();
470 /// Facet for outputting numbers to odocstreams as ascii.
471 /// Here we simply need defining the virtual do_put functions.
472 class ascii_num_put_facet : public num_put<lyx::char_type, ostreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > >
474 typedef ostreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > iter_type;
476 ascii_num_put_facet(size_t refs = 0) : num_put<lyx::char_type, iter_type>(refs) {}
478 /// Facet for converting numbers to ascii strings.
479 class string_num_put_facet : public num_put<char, basic_string<char>::iterator>
482 string_num_put_facet() : num_put<char, basic_string<char>::iterator>(1) {}
487 do_put(iter_type oit, ios_base & b, char_type fill, bool v) const
489 return do_put_helper(oit, b, fill, v);
493 do_put(iter_type oit, ios_base & b, char_type fill, long v) const
495 return do_put_helper(oit, b, fill, v);
499 do_put(iter_type oit, ios_base & b, char_type fill, unsigned long v) const
501 return do_put_helper(oit, b, fill, v);
504 #ifdef _GLIBCXX_USE_LONG_LONG
506 do_put(iter_type oit, ios_base & b, char_type fill, long long v) const
508 return do_put_helper(oit, b, fill, v);
512 do_put(iter_type oit, ios_base & b, char_type fill, unsigned long long v) const
514 return do_put_helper(oit, b, fill, v);
519 do_put(iter_type oit, ios_base & b, char_type fill, double v) const
521 return do_put_helper(oit, b, fill, v);
525 do_put(iter_type oit, ios_base & b, char_type fill, long double v) const
527 return do_put_helper(oit, b, fill, v);
531 do_put(iter_type oit, ios_base & b, char_type fill, void const * v) const
533 return do_put_helper(oit, b, fill, v);
537 template <typename ValueType>
539 do_put_helper(iter_type oit, ios_base & b, char_type fill, ValueType v) const
542 throw num_put_failure();
544 streamsize const sz = b.width() > b.precision() ?
545 b.width() : b.precision();
546 // 64 is large enough, unless width or precision are bigger
547 streamsize const wd = (sz > 56 ? sz : 56) + 8;
549 string_num_put_facet f;
550 string::const_iterator cit = s.begin();
551 string::const_iterator end =
552 f.put(s.begin(), b, fill, v);
553 for (; cit != end; ++cit, ++oit)
561 /// Facet for inputting ascii representations of numbers from idocstreams.
562 /// Here we simply need defining the virtual do_get functions.
563 class ascii_num_get_facet : public num_get<lyx::char_type, istreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > >
565 typedef istreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > iter_type;
567 ascii_num_get_facet(size_t refs = 0) : num_get<lyx::char_type, iter_type>(refs) {}
569 /// Facet for converting ascii representation of numbers to a value.
570 class string_num_get_facet : public num_get<char, basic_string<char>::iterator>
573 string_num_get_facet() : num_get<char, basic_string<char>::iterator>(1) {}
576 /// Numpunct facet defining the I/O format.
577 class numpunct_facet : public numpunct<char>
580 numpunct_facet() : numpunct<char>(1) {}
585 do_get(iter_type iit, iter_type eit, ios_base & b,
586 ios_base::iostate & err, bool & v) const
588 if (b.flags() & ios_base::boolalpha) {
590 lyx::docstring const truename = from_local8bit(p.truename());
591 lyx::docstring const falsename = from_local8bit(p.falsename());
596 size_t const tsize = truename.size();
597 size_t const fsize = falsename.size();
598 for (; iit != eit; ++iit) {
601 bool true_ok = support::prefixIs(truename, s);
602 bool false_ok = support::prefixIs(falsename, s);
603 if (!true_ok && !false_ok) {
608 if ((true_ok && n == tsize) ||
609 (false_ok && n == fsize)) {
615 err = ios_base::goodbit;
616 v = truename == s ? true : false;
618 err = ios_base::failbit;
620 err |= ios_base::eofbit;
624 iter_type end = this->do_get(iit, eit, b, err, l);
625 if (!(err & ios_base::failbit)) {
631 err |= ios_base::failbit;
638 do_get(iter_type iit, iter_type eit, ios_base & b,
639 ios_base::iostate & err, long & v) const
641 return do_get_integer(iit, eit, b, err, v);
645 do_get(iter_type iit, iter_type eit, ios_base & b,
646 ios_base::iostate & err, unsigned short & v) const
648 return do_get_integer(iit, eit, b, err, v);
652 do_get(iter_type iit, iter_type eit, ios_base & b,
653 ios_base::iostate & err, unsigned int & v) const
655 return do_get_integer(iit, eit, b, err, v);
659 do_get(iter_type iit, iter_type eit, ios_base & b,
660 ios_base::iostate & err, unsigned long & v) const
662 return do_get_integer(iit, eit, b, err, v);
665 #ifdef _GLIBCXX_USE_LONG_LONG
667 do_get(iter_type iit, iter_type eit, ios_base & b,
668 ios_base::iostate & err, long long & v) const
670 return do_get_integer(iit, eit, b, err, v);
674 do_get(iter_type iit, iter_type eit, ios_base & b,
675 ios_base::iostate & err, unsigned long long & v) const
677 return do_get_integer(iit, eit, b, err, v);
682 do_get(iter_type iit, iter_type eit, ios_base & b,
683 ios_base::iostate & err, float & v) const
685 return do_get_float(iit, eit, b, err, v);
689 do_get(iter_type iit, iter_type eit, ios_base & b,
690 ios_base::iostate & err, double & v) const
692 return do_get_float(iit, eit, b, err, v);
696 do_get(iter_type iit, iter_type eit, ios_base & b,
697 ios_base::iostate & err, long double & v) const
699 return do_get_float(iit, eit, b, err, v);
703 do_get(iter_type iit, iter_type eit, ios_base & b,
704 ios_base::iostate & err, void * & v) const
707 iter_type end = do_get_integer(iit, eit, b, err, val);
708 if (!(err & ios_base::failbit))
709 v = reinterpret_cast<void *>(val);
714 template <typename ValueType>
716 do_get_integer(iter_type iit, iter_type eit, ios_base & b,
717 ios_base::iostate & err, ValueType & v) const
721 for (; iit != eit && isNumpunct(*iit); ++iit)
722 s += static_cast<char>(*iit);
723 // We add another character, not part of the numpunct facet,
724 // in order to avoid setting the eofbit in the stream state,
725 // which would prevent any further read. The space seems a
728 string_num_get_facet f;
729 f.get(s.begin(), s.end(), b, err, v);
731 err |= ios_base::eofbit;
736 bool isNumpunct(lyx::char_type const c) const
738 /// Only account for the standard numpunct "C" locale facet.
739 return c < 0x80 && (c == '-' || c == '+' || isdigit(c)
740 || ('a' <= c && c <= 'f') || ('A' <= c && c <= 'F')
741 || c == 'x' || c == 'X');
744 template <typename ValueType>
746 do_get_float(iter_type iit, iter_type eit, ios_base & b,
747 ios_base::iostate & err, ValueType & v) const
749 // Gather a string of the form
750 // [+-]? [0-9]* .? [0-9]* ([eE] [+-]? [0-9]+)?
755 char const dot = p.decimal_point();
756 char const sep = p.thousands_sep();
757 // Get an optional sign
758 if (iit != eit && (*iit == '-' || *iit == '+')) {
759 s += static_cast<char>(*iit);
762 for (; iit != eit && isDigitOrSep(*iit, sep); ++iit)
763 s += static_cast<char>(*iit);
764 if (iit != eit && *iit == dot) {
767 for (; iit != eit && isDigitOrSep(*iit, 0); ++iit)
768 s += static_cast<char>(*iit);
769 if (iit != eit && (*iit == 'e' || *iit == 'E')) {
770 s += static_cast<char>(*iit);
772 for (; iit != eit && isDigitOrSep(*iit, 0); ++iit)
773 s += static_cast<char>(*iit);
777 string_num_get_facet f;
778 f.get(s.begin(), s.end(), b, err, v);
780 err |= ios_base::eofbit;
785 bool isDigitOrSep(lyx::char_type const c, char const sep) const
787 return (c >= '0' && c <= '9') || (c != 0 && c == sep);
792 /// class to add our facets to the global locale
793 class locale_initializer {
798 locale const loc1(global, new ascii_ctype_facet);
799 locale const loc2(loc1, new ascii_num_put_facet);
800 locale const loc3(loc2, new ascii_num_get_facet);
801 locale::global(loc3);
808 /// make sure that our facets get used
809 static locale_initializer initializer;