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 if (int n = strlen(ascii)) {
39 LASSERT(static_cast<unsigned char>(ascii[n]) < 0x80, /**/);
46 docstring const from_ascii(string const & ascii)
48 int const len = ascii.length();
49 for (int i = 0; i < len; ++i)
50 LASSERT(static_cast<unsigned char>(ascii[i]) < 0x80, /**/);
51 return docstring(ascii.begin(), ascii.end());
55 string const to_ascii(docstring const & ucs4)
57 int const len = ucs4.length();
60 for (int i = 0; i < len; ++i) {
61 LASSERT(ucs4[i] < 0x80, /**/);
62 ascii[i] = static_cast<char>(ucs4[i]);
68 IconvProcessor & utf8ToUcs4()
70 static IconvProcessor iconv(ucs4_codeset, "UTF-8");
76 void utf8_to_ucs4(string const & utf8, docstring & ucs4)
78 size_t n = utf8.size();
79 // as utf8 is a multi-byte encoding, there would be at most
85 int maxoutsize = n * 4;
86 // basic_string::data() is not recognized by some old gcc version
87 // so we use &(ucs4[0]) instead.
88 char * outbuf = (char *)(&(ucs4[0]));
89 int bytes = utf8ToUcs4().convert(utf8.c_str(), n, outbuf, maxoutsize);
91 // adjust to the real converted size
96 docstring const from_utf8(string const & utf8)
99 utf8_to_ucs4(utf8, ucs4);
104 string const to_utf8(docstring const & ucs4)
106 vector<char> const utf8 = ucs4_to_utf8(ucs4.data(), ucs4.size());
107 return string(utf8.begin(), utf8.end());
111 docstring const from_local8bit(string const & s)
113 return qstring_to_ucs4(QString::fromLocal8Bit(s.data(), s.length()));
117 /// Exception thrown by to_local8bit if the string could not be converted
118 class to_local8bit_failure : public bad_cast {
120 to_local8bit_failure() throw() : bad_cast() {}
121 virtual ~to_local8bit_failure() throw() {}
122 virtual const char* what() const throw()
124 return "A string could not be converted from unicode to the local 8 bit encoding.";
129 string const to_local8bit(docstring const & s)
131 // This conversion can fail, depending on input.
134 QByteArray const local = toqstr(s).toLocal8Bit();
135 if (local.size() == 0)
136 throw to_local8bit_failure();
137 return string(local.begin(), local.end());
141 docstring const from_filesystem8bit(string const & s)
143 QByteArray const encoded(s.c_str(), s.length());
144 return qstring_to_ucs4(QFile::decodeName(encoded));
148 string const to_filesystem8bit(docstring const & s)
150 QByteArray const encoded = QFile::encodeName(toqstr(s));
151 return string(encoded.begin(), encoded.end());
155 string const to_iconv_encoding(docstring const & s, string const & encoding)
157 std::vector<char> const encoded =
158 ucs4_to_eightbit(s.data(), s.length(), encoding);
159 return string(encoded.begin(), encoded.end());
163 docstring const from_iconv_encoding(string const & s, string const & encoding)
165 std::vector<char_type> const ucs4 =
166 eightbit_to_ucs4(s.data(), s.length(), encoding);
167 return docstring(ucs4.begin(), ucs4.end());
171 docstring const normalize_c(docstring const & s)
173 return qstring_to_ucs4(toqstr(s).normalized(QString::NormalizationForm_C));
177 bool operator==(lyx::docstring const & l, char const * r)
179 lyx::docstring::const_iterator it = l.begin();
180 lyx::docstring::const_iterator end = l.end();
181 for (; it != end; ++it, ++r) {
182 LASSERT(static_cast<unsigned char>(*r) < 0x80, /**/);
185 if (*it != static_cast<lyx::docstring::value_type>(*r))
192 lyx::docstring operator+(lyx::docstring const & l, char const * r)
195 for (char const * c = r; *c; ++c) {
196 LASSERT(static_cast<unsigned char>(*c) < 0x80, /**/);
203 lyx::docstring operator+(char const * l, lyx::docstring const & r)
206 for (char const * c = l; *c; ++c) {
207 LASSERT(static_cast<unsigned char>(*c) < 0x80, /**/);
215 lyx::docstring operator+(lyx::docstring const & l, char r)
217 LASSERT(static_cast<unsigned char>(r) < 0x80, /**/);
219 s += docstring::value_type(r);
224 lyx::docstring operator+(char l, lyx::docstring const & r)
226 LASSERT(static_cast<unsigned char>(l) < 0x80, /**/);
227 return lyx::docstring::value_type(l) + r;
231 lyx::docstring & operator+=(lyx::docstring & l, char const * r)
233 for (char const * c = r; *c; ++c) {
234 LASSERT(static_cast<unsigned char>(*c) < 0x80, /**/);
241 lyx::docstring & operator+=(lyx::docstring & l, char r)
243 LASSERT(static_cast<unsigned char>(r) < 0x80, /**/);
250 #if ! defined(USE_WCHAR_T) && defined(__GNUC__)
252 // gcc does not have proper locale facets for lyx::char_type if
253 // sizeof(wchar_t) == 2, so we have to implement them on our own.
256 // We get undefined references to these virtual methods. This looks like
257 // a bug in gcc. The implementation here does not do anything useful, since
258 // it is overriden in ascii_ctype_facet.
260 template<> ctype<lyx::char_type>::~ctype() {}
262 ctype<lyx::char_type>::do_is(ctype<lyx::char_type>::mask, lyx::char_type) const { return false; }
263 template<> lyx::char_type const *
264 ctype<lyx::char_type>::do_is(const lyx::char_type *, const lyx::char_type *, ctype<lyx::char_type>::mask *) const { return 0; }
265 template<> const lyx::char_type *
266 ctype<lyx::char_type>::do_scan_is(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
267 template<> const lyx::char_type *
268 ctype<lyx::char_type>::do_scan_not(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
269 template<> lyx::char_type ctype<lyx::char_type>::do_toupper(lyx::char_type) const { return 0; }
270 template<> const lyx::char_type * ctype<lyx::char_type>::do_toupper(lyx::char_type *, lyx::char_type const *) const { return 0; }
271 template<> lyx::char_type ctype<lyx::char_type>::do_tolower(lyx::char_type) const { return 0; }
272 template<> const lyx::char_type * ctype<lyx::char_type>::do_tolower(lyx::char_type *, lyx::char_type const *) const { return 0; }
273 template<> lyx::char_type ctype<lyx::char_type>::do_widen(char) const { return 0; }
274 template<> const char *
275 ctype<lyx::char_type>::do_widen(const char *, const char *, lyx::char_type *) const { return 0; }
277 ctype<lyx::char_type>::do_narrow(const lyx::char_type, char) const { return 0; }
278 template<> const lyx::char_type *
279 ctype<lyx::char_type>::do_narrow(const lyx::char_type *, const lyx::char_type *, char, char *) const { return 0; }
285 class ctype_failure : public bad_cast {
287 ctype_failure() throw() : bad_cast() {}
288 virtual ~ctype_failure() throw() {}
289 virtual const char* what() const throw()
291 return "The ctype<lyx::char_type> locale facet does only support ASCII characters on this platform.";
296 class num_put_failure : public bad_cast {
298 num_put_failure() throw() : bad_cast() {}
299 virtual ~num_put_failure() throw() {}
300 virtual const char* what() const throw()
302 return "The num_put locale facet does only support ASCII characters on this platform.";
307 /// ctype facet for UCS4 characters. The implementation does only support pure
308 /// ASCII, since we do not need anything else for now.
309 /// The code is partly stolen from ctype<wchar_t> from gcc.
310 class ascii_ctype_facet : public ctype<lyx::char_type>
313 typedef lyx::char_type char_type;
314 typedef wctype_t wmask_type;
315 explicit ascii_ctype_facet(size_t refs = 0) : ctype<char_type>(refs)
317 M_initialize_ctype();
322 wint_t M_widen[1 + static_cast<unsigned char>(-1)];
324 wmask_type M_wmask[16];
325 wmask_type M_convert_to_wmask(const mask m) const
329 case space: ret = wctype("space"); break;
330 case print: ret = wctype("print"); break;
331 case cntrl: ret = wctype("cntrl"); break;
332 case upper: ret = wctype("upper"); break;
333 case lower: ret = wctype("lower"); break;
334 case alpha: ret = wctype("alpha"); break;
335 case digit: ret = wctype("digit"); break;
336 case punct: ret = wctype("punct"); break;
337 case xdigit: ret = wctype("xdigit"); break;
338 case alnum: ret = wctype("alnum"); break;
339 case graph: ret = wctype("graph"); break;
340 default: ret = wmask_type();
344 void M_initialize_ctype()
347 for (i = 0; i < 128; ++i) {
348 const int c = wctob(i);
352 M_narrow[i] = static_cast<char>(c);
358 for (size_t i = 0; i < sizeof(M_widen) / sizeof(wint_t); ++i)
359 M_widen[i] = btowc(i);
361 for (size_t i = 0; i <= 15; ++i) {
362 M_bit[i] = static_cast<mask>(1 << i);
363 M_wmask[i] = M_convert_to_wmask(M_bit[i]);
366 virtual ~ascii_ctype_facet() {}
367 char_type do_toupper(char_type c) const
370 throw ctype_failure();
371 return toupper(static_cast<int>(c));
373 char_type const * do_toupper(char_type * lo, char_type const * hi) const
377 throw ctype_failure();
378 *lo = toupper(static_cast<int>(*lo));
383 char_type do_tolower(char_type c) const
386 throw ctype_failure();
389 char_type const * do_tolower(char_type * lo, char_type const * hi) const
393 throw ctype_failure();
399 bool do_is(mask m, char_type c) const
402 throw ctype_failure();
403 // The code below works because c is in the ASCII range.
404 // We could not use iswctype() which is designed for a 2byte
405 // whar_t without encoding conversion otherwise.
407 // Generically, 15 (instead of 10) since we don't know the numerical
408 // encoding of the various categories in /usr/include/ctype.h.
409 const size_t bitmasksize = 15;
410 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
411 if (m & M_bit[bitcur] &&
412 iswctype(static_cast<int>(c), M_wmask[bitcur])) {
418 char_type const * do_is(char_type const * lo, char_type const * hi, mask * vec) const
420 for (;lo < hi; ++vec, ++lo) {
422 throw ctype_failure();
423 // The code below works because c is in the ASCII range.
424 // We could not use iswctype() which is designed for a 2byte
425 // whar_t without encoding conversion otherwise.
426 // Generically, 15 (instead of 10) since we don't know the numerical
427 // encoding of the various categories in /usr/include/ctype.h.
428 const size_t bitmasksize = 15;
430 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
431 if (iswctype(static_cast<int>(*lo), M_wmask[bitcur]))
437 char_type const * do_scan_is(mask m, char_type const * lo, char_type const * hi) const
439 while (lo < hi && !this->do_is(m, *lo))
443 char_type const * do_scan_not(mask m, char_type const * lo, char_type const * hi) const
445 while (lo < hi && this->do_is(m, *lo) != 0)
449 char_type do_widen(char c) const
451 if (static_cast<unsigned char>(c) < 0x80)
453 throw ctype_failure();
455 const char* do_widen(const char* lo, const char* hi, char_type* dest) const
458 if (static_cast<unsigned char>(*lo) >= 0x80)
459 throw ctype_failure();
466 char do_narrow(char_type wc, char) const
469 return static_cast<char>(wc);
470 throw ctype_failure();
472 const char_type * do_narrow(const char_type * lo, const char_type * hi, char, char * dest) const
476 *dest = static_cast<char>(*lo);
478 throw ctype_failure();
487 /// Facet for outputting numbers to odocstreams as ascii.
488 /// Here we simply need defining the virtual do_put functions.
489 class ascii_num_put_facet : public num_put<lyx::char_type, ostreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > >
491 typedef ostreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > iter_type;
493 ascii_num_put_facet(size_t refs = 0) : num_put<lyx::char_type, iter_type>(refs) {}
495 /// Facet for converting numbers to ascii strings.
496 class string_num_put_facet : public num_put<char, basic_string<char>::iterator>
499 string_num_put_facet() : num_put<char, basic_string<char>::iterator>(1) {}
504 do_put(iter_type oit, ios_base & b, char_type fill, bool v) const
506 return do_put_helper(oit, b, fill, v);
510 do_put(iter_type oit, ios_base & b, char_type fill, long v) const
512 return do_put_helper(oit, b, fill, v);
516 do_put(iter_type oit, ios_base & b, char_type fill, unsigned long v) const
518 return do_put_helper(oit, b, fill, v);
521 #ifdef _GLIBCXX_USE_LONG_LONG
523 do_put(iter_type oit, ios_base & b, char_type fill, long long v) const
525 return do_put_helper(oit, b, fill, v);
529 do_put(iter_type oit, ios_base & b, char_type fill, unsigned long long v) const
531 return do_put_helper(oit, b, fill, v);
536 do_put(iter_type oit, ios_base & b, char_type fill, double v) const
538 return do_put_helper(oit, b, fill, v);
542 do_put(iter_type oit, ios_base & b, char_type fill, long double v) const
544 return do_put_helper(oit, b, fill, v);
548 do_put(iter_type oit, ios_base & b, char_type fill, void const * v) const
550 return do_put_helper(oit, b, fill, v);
554 template <typename ValueType>
556 do_put_helper(iter_type oit, ios_base & b, char_type fill, ValueType v) const
559 throw num_put_failure();
561 streamsize const sz = b.width() > b.precision() ?
562 b.width() : b.precision();
563 // 64 is large enough, unless width or precision are bigger
564 streamsize const wd = (sz > 56 ? sz : 56) + 8;
566 string_num_put_facet f;
567 string::const_iterator cit = s.begin();
568 string::const_iterator end =
569 f.put(s.begin(), b, fill, v);
570 for (; cit != end; ++cit, ++oit)
578 /// Facet for inputting ascii representations of numbers from idocstreams.
579 /// Here we simply need defining the virtual do_get functions.
580 class ascii_num_get_facet : public num_get<lyx::char_type, istreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > >
582 typedef istreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > iter_type;
584 ascii_num_get_facet(size_t refs = 0) : num_get<lyx::char_type, iter_type>(refs) {}
586 /// Facet for converting ascii representation of numbers to a value.
587 class string_num_get_facet : public num_get<char, basic_string<char>::iterator>
590 string_num_get_facet() : num_get<char, basic_string<char>::iterator>(1) {}
593 /// Numpunct facet defining the I/O format.
594 class numpunct_facet : public numpunct<char>
597 numpunct_facet() : numpunct<char>(1) {}
602 do_get(iter_type iit, iter_type eit, ios_base & b,
603 ios_base::iostate & err, bool & v) const
605 if (b.flags() & ios_base::boolalpha) {
607 lyx::docstring const truename = from_local8bit(p.truename());
608 lyx::docstring const falsename = from_local8bit(p.falsename());
613 size_t const tsize = truename.size();
614 size_t const fsize = falsename.size();
615 for (; iit != eit; ++iit) {
618 bool true_ok = support::prefixIs(truename, s);
619 bool false_ok = support::prefixIs(falsename, s);
620 if (!true_ok && !false_ok) {
625 if ((true_ok && n == tsize) ||
626 (false_ok && n == fsize)) {
632 err = ios_base::goodbit;
633 v = truename == s ? true : false;
635 err = ios_base::failbit;
637 err |= ios_base::eofbit;
641 iter_type end = this->do_get(iit, eit, b, err, l);
642 if (!(err & ios_base::failbit)) {
648 err |= ios_base::failbit;
655 do_get(iter_type iit, iter_type eit, ios_base & b,
656 ios_base::iostate & err, long & v) const
658 return do_get_integer(iit, eit, b, err, v);
662 do_get(iter_type iit, iter_type eit, ios_base & b,
663 ios_base::iostate & err, unsigned short & v) const
665 return do_get_integer(iit, eit, b, err, v);
669 do_get(iter_type iit, iter_type eit, ios_base & b,
670 ios_base::iostate & err, unsigned int & v) const
672 return do_get_integer(iit, eit, b, err, v);
676 do_get(iter_type iit, iter_type eit, ios_base & b,
677 ios_base::iostate & err, unsigned long & v) const
679 return do_get_integer(iit, eit, b, err, v);
682 #ifdef _GLIBCXX_USE_LONG_LONG
684 do_get(iter_type iit, iter_type eit, ios_base & b,
685 ios_base::iostate & err, long long & v) const
687 return do_get_integer(iit, eit, b, err, v);
691 do_get(iter_type iit, iter_type eit, ios_base & b,
692 ios_base::iostate & err, unsigned long long & v) const
694 return do_get_integer(iit, eit, b, err, v);
699 do_get(iter_type iit, iter_type eit, ios_base & b,
700 ios_base::iostate & err, float & v) const
702 return do_get_float(iit, eit, b, err, v);
706 do_get(iter_type iit, iter_type eit, ios_base & b,
707 ios_base::iostate & err, double & v) const
709 return do_get_float(iit, eit, b, err, v);
713 do_get(iter_type iit, iter_type eit, ios_base & b,
714 ios_base::iostate & err, long double & v) const
716 return do_get_float(iit, eit, b, err, v);
720 do_get(iter_type iit, iter_type eit, ios_base & b,
721 ios_base::iostate & err, void * & v) const
724 iter_type end = do_get_integer(iit, eit, b, err, val);
725 if (!(err & ios_base::failbit))
726 v = reinterpret_cast<void *>(val);
731 template <typename ValueType>
733 do_get_integer(iter_type iit, iter_type eit, ios_base & b,
734 ios_base::iostate & err, ValueType & v) const
738 for (; iit != eit && isNumpunct(*iit); ++iit)
739 s += static_cast<char>(*iit);
740 // We add another character, not part of the numpunct facet,
741 // in order to avoid setting the eofbit in the stream state,
742 // which would prevent any further read. The space seems a
745 string_num_get_facet f;
746 f.get(s.begin(), s.end(), b, err, v);
748 err |= ios_base::eofbit;
753 bool isNumpunct(lyx::char_type const c) const
755 /// Only account for the standard numpunct "C" locale facet.
756 return c < 0x80 && (c == '-' || c == '+' || isdigit(c)
757 || ('a' <= c && c <= 'f') || ('A' <= c && c <= 'F')
758 || c == 'x' || c == 'X');
761 template <typename ValueType>
763 do_get_float(iter_type iit, iter_type eit, ios_base & b,
764 ios_base::iostate & err, ValueType & v) const
766 // Gather a string of the form
767 // [+-]? [0-9]* .? [0-9]* ([eE] [+-]? [0-9]+)?
772 char const dot = p.decimal_point();
773 char const sep = p.thousands_sep();
774 // Get an optional sign
775 if (iit != eit && (*iit == '-' || *iit == '+')) {
776 s += static_cast<char>(*iit);
779 for (; iit != eit && isDigitOrSep(*iit, sep); ++iit)
780 s += static_cast<char>(*iit);
781 if (iit != eit && *iit == dot) {
784 for (; iit != eit && isDigitOrSep(*iit, 0); ++iit)
785 s += static_cast<char>(*iit);
786 if (iit != eit && (*iit == 'e' || *iit == 'E')) {
787 s += static_cast<char>(*iit);
789 for (; iit != eit && isDigitOrSep(*iit, 0); ++iit)
790 s += static_cast<char>(*iit);
794 string_num_get_facet f;
795 f.get(s.begin(), s.end(), b, err, v);
797 err |= ios_base::eofbit;
802 bool isDigitOrSep(lyx::char_type const c, char const sep) const
804 return (c >= '0' && c <= '9') || (c != 0 && c == sep);
809 /// class to add our facets to the global locale
810 class locale_initializer {
815 locale const loc1(global, new ascii_ctype_facet);
816 locale const loc2(loc1, new ascii_num_put_facet);
817 locale const loc3(loc2, new ascii_num_get_facet);
818 locale::global(loc3);
825 /// make sure that our facets get used
826 static locale_initializer initializer;