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/lstrings.h"
16 #include "support/qstring_helpers.h"
17 #include "support/unicode.h"
25 #include "support/lassert.h"
32 docstring const from_ascii(char const * ascii)
35 for (char const * c = ascii; *c; ++c) {
36 LASSERT(static_cast<unsigned char>(*c) < 0x80, /**/);
43 docstring const from_ascii(string const & ascii)
45 int const len = ascii.length();
46 for (int i = 0; i < len; ++i)
47 LASSERT(static_cast<unsigned char>(ascii[i]) < 0x80, /**/);
48 return docstring(ascii.begin(), ascii.end());
52 string const to_ascii(docstring const & ucs4)
54 int const len = ucs4.length();
57 for (int i = 0; i < len; ++i) {
58 LASSERT(ucs4[i] < 0x80, /**/);
59 ascii[i] = static_cast<char>(ucs4[i]);
65 IconvProcessor & utf8ToUcs4()
67 static IconvProcessor iconv(ucs4_codeset, "UTF-8");
73 void utf8_to_ucs4(string const & utf8, docstring & ucs4)
75 size_t n = utf8.size();
76 // as utf8 is a multi-byte encoding, there would be at most
82 int maxoutsize = n * 4;
83 // basic_string::data() is not recognized by some old gcc version
84 // so we use &(ucs4[0]) instead.
85 char * outbuf = (char *)(&(ucs4[0]));
86 int bytes = utf8ToUcs4().convert(utf8.c_str(), n, outbuf, maxoutsize);
88 // adjust to the real converted size
93 docstring const from_utf8(string const & utf8)
96 utf8_to_ucs4(utf8, ucs4);
101 string const to_utf8(docstring const & ucs4)
103 vector<char> const utf8 =
104 ucs4_to_utf8(ucs4.data(), ucs4.size());
105 return string(utf8.begin(), utf8.end());
109 docstring const from_local8bit(string const & s)
111 return qstring_to_ucs4(QString::fromLocal8Bit(s.data(), s.length()));
115 /// Exception thrown by to_local8bit if the string could not be converted
116 class to_local8bit_failure : public bad_cast {
118 to_local8bit_failure() throw() : bad_cast() {}
119 virtual ~to_local8bit_failure() throw() {}
120 virtual const char* what() const throw()
122 return "A string could not be converted from unicode to the local 8 bit encoding.";
127 string const to_local8bit(docstring const & s)
129 // This conversion can fail, depending on input.
132 QByteArray const local = toqstr(s).toLocal8Bit();
133 if (local.size() == 0)
134 throw to_local8bit_failure();
135 return string(local.begin(), local.end());
139 docstring const from_filesystem8bit(string const & s)
141 QByteArray const encoded(s.c_str(), s.length());
142 return qstring_to_ucs4(QFile::decodeName(encoded));
146 string const to_filesystem8bit(docstring const & s)
148 QByteArray const encoded = QFile::encodeName(toqstr(s));
149 return string(encoded.begin(), encoded.end());
153 docstring const normalize_c(docstring const & s)
155 return qstring_to_ucs4(toqstr(s).normalized(QString::NormalizationForm_C));
159 bool operator==(lyx::docstring const & l, char const * r)
161 lyx::docstring::const_iterator it = l.begin();
162 lyx::docstring::const_iterator end = l.end();
163 for (; it != end; ++it, ++r) {
164 LASSERT(static_cast<unsigned char>(*r) < 0x80, /**/);
167 if (*it != static_cast<lyx::docstring::value_type>(*r))
174 lyx::docstring operator+(lyx::docstring const & l, char const * r)
177 for (char const * c = r; *c; ++c) {
178 LASSERT(static_cast<unsigned char>(*c) < 0x80, /**/);
185 lyx::docstring operator+(char const * l, lyx::docstring const & r)
188 for (char const * c = l; *c; ++c) {
189 LASSERT(static_cast<unsigned char>(*c) < 0x80, /**/);
197 lyx::docstring operator+(lyx::docstring const & l, char r)
199 LASSERT(static_cast<unsigned char>(r) < 0x80, /**/);
201 s += docstring::value_type(r);
206 lyx::docstring operator+(char l, lyx::docstring const & r)
208 LASSERT(static_cast<unsigned char>(l) < 0x80, /**/);
209 return lyx::docstring::value_type(l) + r;
213 lyx::docstring & operator+=(lyx::docstring & l, char const * r)
215 for (char const * c = r; *c; ++c) {
216 LASSERT(static_cast<unsigned char>(*c) < 0x80, /**/);
223 lyx::docstring & operator+=(lyx::docstring & l, char r)
225 LASSERT(static_cast<unsigned char>(r) < 0x80, /**/);
232 #if ! defined(USE_WCHAR_T) && defined(__GNUC__)
234 // gcc does not have proper locale facets for lyx::char_type if
235 // sizeof(wchar_t) == 2, so we have to implement them on our own.
238 // We get undefined references to these virtual methods. This looks like
239 // a bug in gcc. The implementation here does not do anything useful, since
240 // it is overriden in ascii_ctype_facet.
242 template<> ctype<lyx::char_type>::~ctype() {}
244 ctype<lyx::char_type>::do_is(ctype<lyx::char_type>::mask, lyx::char_type) const { return false; }
245 template<> lyx::char_type const *
246 ctype<lyx::char_type>::do_is(const lyx::char_type *, const lyx::char_type *, ctype<lyx::char_type>::mask *) const { return 0; }
247 template<> const lyx::char_type *
248 ctype<lyx::char_type>::do_scan_is(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
249 template<> const lyx::char_type *
250 ctype<lyx::char_type>::do_scan_not(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
251 template<> lyx::char_type ctype<lyx::char_type>::do_toupper(lyx::char_type) const { return 0; }
252 template<> const lyx::char_type * ctype<lyx::char_type>::do_toupper(lyx::char_type *, lyx::char_type const *) const { return 0; }
253 template<> lyx::char_type ctype<lyx::char_type>::do_tolower(lyx::char_type) const { return 0; }
254 template<> const lyx::char_type * ctype<lyx::char_type>::do_tolower(lyx::char_type *, lyx::char_type const *) const { return 0; }
255 template<> lyx::char_type ctype<lyx::char_type>::do_widen(char) const { return 0; }
256 template<> const char *
257 ctype<lyx::char_type>::do_widen(const char *, const char *, lyx::char_type *) const { return 0; }
259 ctype<lyx::char_type>::do_narrow(const lyx::char_type, char) const { return 0; }
260 template<> const lyx::char_type *
261 ctype<lyx::char_type>::do_narrow(const lyx::char_type *, const lyx::char_type *, char, char *) const { return 0; }
267 class ctype_failure : public bad_cast {
269 ctype_failure() throw() : bad_cast() {}
270 virtual ~ctype_failure() throw() {}
271 virtual const char* what() const throw()
273 return "The ctype<lyx::char_type> locale facet does only support ASCII characters on this platform.";
278 class num_put_failure : public bad_cast {
280 num_put_failure() throw() : bad_cast() {}
281 virtual ~num_put_failure() throw() {}
282 virtual const char* what() const throw()
284 return "The num_put locale facet does only support ASCII characters on this platform.";
289 /// ctype facet for UCS4 characters. The implementation does only support pure
290 /// ASCII, since we do not need anything else for now.
291 /// The code is partly stolen from ctype<wchar_t> from gcc.
292 class ascii_ctype_facet : public ctype<lyx::char_type>
295 typedef lyx::char_type char_type;
296 typedef wctype_t wmask_type;
297 explicit ascii_ctype_facet(size_t refs = 0) : ctype<char_type>(refs)
299 M_initialize_ctype();
304 wint_t M_widen[1 + static_cast<unsigned char>(-1)];
306 wmask_type M_wmask[16];
307 wmask_type M_convert_to_wmask(const mask m) const
311 case space: ret = wctype("space"); break;
312 case print: ret = wctype("print"); break;
313 case cntrl: ret = wctype("cntrl"); break;
314 case upper: ret = wctype("upper"); break;
315 case lower: ret = wctype("lower"); break;
316 case alpha: ret = wctype("alpha"); break;
317 case digit: ret = wctype("digit"); break;
318 case punct: ret = wctype("punct"); break;
319 case xdigit: ret = wctype("xdigit"); break;
320 case alnum: ret = wctype("alnum"); break;
321 case graph: ret = wctype("graph"); break;
322 default: ret = wmask_type();
326 void M_initialize_ctype()
329 for (i = 0; i < 128; ++i) {
330 const int c = wctob(i);
334 M_narrow[i] = static_cast<char>(c);
340 for (size_t i = 0; i < sizeof(M_widen) / sizeof(wint_t); ++i)
341 M_widen[i] = btowc(i);
343 for (size_t i = 0; i <= 15; ++i) {
344 M_bit[i] = static_cast<mask>(1 << i);
345 M_wmask[i] = M_convert_to_wmask(M_bit[i]);
348 virtual ~ascii_ctype_facet() {}
349 char_type do_toupper(char_type c) const
352 throw ctype_failure();
353 return toupper(static_cast<int>(c));
355 char_type const * do_toupper(char_type * lo, char_type const * hi) const
359 throw ctype_failure();
360 *lo = toupper(static_cast<int>(*lo));
365 char_type do_tolower(char_type c) const
368 throw ctype_failure();
371 char_type const * do_tolower(char_type * lo, char_type const * hi) const
375 throw ctype_failure();
381 bool do_is(mask m, char_type c) const
384 throw ctype_failure();
385 // The code below works because c is in the ASCII range.
386 // We could not use iswctype() which is designed for a 2byte
387 // whar_t without encoding conversion otherwise.
389 // Generically, 15 (instead of 10) since we don't know the numerical
390 // encoding of the various categories in /usr/include/ctype.h.
391 const size_t bitmasksize = 15;
392 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
393 if (m & M_bit[bitcur] &&
394 iswctype(static_cast<int>(c), M_wmask[bitcur])) {
400 char_type const * do_is(char_type const * lo, char_type const * hi, mask * vec) const
402 for (;lo < hi; ++vec, ++lo) {
404 throw ctype_failure();
405 // The code below works because c is in the ASCII range.
406 // We could not use iswctype() which is designed for a 2byte
407 // whar_t without encoding conversion otherwise.
408 // Generically, 15 (instead of 10) since we don't know the numerical
409 // encoding of the various categories in /usr/include/ctype.h.
410 const size_t bitmasksize = 15;
412 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
413 if (iswctype(static_cast<int>(*lo), M_wmask[bitcur]))
419 char_type const * do_scan_is(mask m, char_type const * lo, char_type const * hi) const
421 while (lo < hi && !this->do_is(m, *lo))
425 char_type const * do_scan_not(mask m, char_type const * lo, char_type const * hi) const
427 while (lo < hi && this->do_is(m, *lo) != 0)
431 char_type do_widen(char c) const
433 if (static_cast<unsigned char>(c) < 0x80)
435 throw ctype_failure();
437 const char* do_widen(const char* lo, const char* hi, char_type* dest) const
440 if (static_cast<unsigned char>(*lo) >= 0x80)
441 throw ctype_failure();
448 char do_narrow(char_type wc, char) const
451 return static_cast<char>(wc);
452 throw ctype_failure();
454 const char_type * do_narrow(const char_type * lo, const char_type * hi, char, char * dest) const
458 *dest = static_cast<char>(*lo);
460 throw ctype_failure();
469 /// Facet for outputting numbers to odocstreams as ascii.
470 /// Here we simply need defining the virtual do_put functions.
471 class ascii_num_put_facet : public num_put<lyx::char_type, ostreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > >
473 typedef ostreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > iter_type;
475 ascii_num_put_facet(size_t refs = 0) : num_put<lyx::char_type, iter_type>(refs) {}
477 /// Facet for converting numbers to ascii strings.
478 class string_num_put_facet : public num_put<char, basic_string<char>::iterator>
481 string_num_put_facet() : num_put<char, basic_string<char>::iterator>(1) {}
486 do_put(iter_type oit, ios_base & b, char_type fill, bool v) const
488 return do_put_helper(oit, b, fill, v);
492 do_put(iter_type oit, ios_base & b, char_type fill, long v) const
494 return do_put_helper(oit, b, fill, v);
498 do_put(iter_type oit, ios_base & b, char_type fill, unsigned long v) const
500 return do_put_helper(oit, b, fill, v);
503 #ifdef _GLIBCXX_USE_LONG_LONG
505 do_put(iter_type oit, ios_base & b, char_type fill, long long v) const
507 return do_put_helper(oit, b, fill, v);
511 do_put(iter_type oit, ios_base & b, char_type fill, unsigned long long v) const
513 return do_put_helper(oit, b, fill, v);
518 do_put(iter_type oit, ios_base & b, char_type fill, double v) const
520 return do_put_helper(oit, b, fill, v);
524 do_put(iter_type oit, ios_base & b, char_type fill, long double v) const
526 return do_put_helper(oit, b, fill, v);
530 do_put(iter_type oit, ios_base & b, char_type fill, void const * v) const
532 return do_put_helper(oit, b, fill, v);
536 template <typename ValueType>
538 do_put_helper(iter_type oit, ios_base & b, char_type fill, ValueType v) const
541 throw num_put_failure();
543 streamsize const sz = b.width() > b.precision() ?
544 b.width() : b.precision();
545 // 64 is large enough, unless width or precision are bigger
546 streamsize const wd = (sz > 56 ? sz : 56) + 8;
548 string_num_put_facet f;
549 string::const_iterator cit = s.begin();
550 string::const_iterator end =
551 f.put(s.begin(), b, fill, v);
552 for (; cit != end; ++cit, ++oit)
560 /// Facet for inputting ascii representations of numbers from idocstreams.
561 /// Here we simply need defining the virtual do_get functions.
562 class ascii_num_get_facet : public num_get<lyx::char_type, istreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > >
564 typedef istreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > iter_type;
566 ascii_num_get_facet(size_t refs = 0) : num_get<lyx::char_type, iter_type>(refs) {}
568 /// Facet for converting ascii representation of numbers to a value.
569 class string_num_get_facet : public num_get<char, basic_string<char>::iterator>
572 string_num_get_facet() : num_get<char, basic_string<char>::iterator>(1) {}
575 /// Numpunct facet defining the I/O format.
576 class numpunct_facet : public numpunct<char>
579 numpunct_facet() : numpunct<char>(1) {}
584 do_get(iter_type iit, iter_type eit, ios_base & b,
585 ios_base::iostate & err, bool & v) const
587 if (b.flags() & ios_base::boolalpha) {
589 lyx::docstring const truename = from_local8bit(p.truename());
590 lyx::docstring const falsename = from_local8bit(p.falsename());
595 size_t const tsize = truename.size();
596 size_t const fsize = falsename.size();
597 for (; iit != eit; ++iit) {
600 bool true_ok = support::prefixIs(truename, s);
601 bool false_ok = support::prefixIs(falsename, s);
602 if (!true_ok && !false_ok) {
607 if ((true_ok && n == tsize) ||
608 (false_ok && n == fsize)) {
614 err = ios_base::goodbit;
615 v = truename == s ? true : false;
617 err = ios_base::failbit;
619 err |= ios_base::eofbit;
623 iter_type end = this->do_get(iit, eit, b, err, l);
624 if (!(err & ios_base::failbit)) {
630 err |= ios_base::failbit;
637 do_get(iter_type iit, iter_type eit, ios_base & b,
638 ios_base::iostate & err, long & v) const
640 return do_get_integer(iit, eit, b, err, v);
644 do_get(iter_type iit, iter_type eit, ios_base & b,
645 ios_base::iostate & err, unsigned short & v) const
647 return do_get_integer(iit, eit, b, err, v);
651 do_get(iter_type iit, iter_type eit, ios_base & b,
652 ios_base::iostate & err, unsigned int & v) const
654 return do_get_integer(iit, eit, b, err, v);
658 do_get(iter_type iit, iter_type eit, ios_base & b,
659 ios_base::iostate & err, unsigned long & v) const
661 return do_get_integer(iit, eit, b, err, v);
664 #ifdef _GLIBCXX_USE_LONG_LONG
666 do_get(iter_type iit, iter_type eit, ios_base & b,
667 ios_base::iostate & err, long long & v) const
669 return do_get_integer(iit, eit, b, err, v);
673 do_get(iter_type iit, iter_type eit, ios_base & b,
674 ios_base::iostate & err, unsigned long long & v) const
676 return do_get_integer(iit, eit, b, err, v);
681 do_get(iter_type iit, iter_type eit, ios_base & b,
682 ios_base::iostate & err, float & v) const
684 return do_get_float(iit, eit, b, err, v);
688 do_get(iter_type iit, iter_type eit, ios_base & b,
689 ios_base::iostate & err, double & v) const
691 return do_get_float(iit, eit, b, err, v);
695 do_get(iter_type iit, iter_type eit, ios_base & b,
696 ios_base::iostate & err, long double & v) const
698 return do_get_float(iit, eit, b, err, v);
702 do_get(iter_type iit, iter_type eit, ios_base & b,
703 ios_base::iostate & err, void * & v) const
706 iter_type end = do_get_integer(iit, eit, b, err, val);
707 if (!(err & ios_base::failbit))
708 v = reinterpret_cast<void *>(val);
713 template <typename ValueType>
715 do_get_integer(iter_type iit, iter_type eit, ios_base & b,
716 ios_base::iostate & err, ValueType & v) const
720 for (; iit != eit && isNumpunct(*iit); ++iit)
721 s += static_cast<char>(*iit);
722 // We add another character, not part of the numpunct facet,
723 // in order to avoid setting the eofbit in the stream state,
724 // which would prevent any further read. The space seems a
727 string_num_get_facet f;
728 f.get(s.begin(), s.end(), b, err, v);
730 err |= ios_base::eofbit;
735 bool isNumpunct(lyx::char_type const c) const
737 /// Only account for the standard numpunct "C" locale facet.
738 return c < 0x80 && (c == '-' || c == '+' || isdigit(c)
739 || ('a' <= c && c <= 'f') || ('A' <= c && c <= 'F')
740 || c == 'x' || c == 'X');
743 template <typename ValueType>
745 do_get_float(iter_type iit, iter_type eit, ios_base & b,
746 ios_base::iostate & err, ValueType & v) const
748 // Gather a string of the form
749 // [+-]? [0-9]* .? [0-9]* ([eE] [+-]? [0-9]+)?
754 char const dot = p.decimal_point();
755 char const sep = p.thousands_sep();
756 // Get an optional sign
757 if (iit != eit && (*iit == '-' || *iit == '+')) {
758 s += static_cast<char>(*iit);
761 for (; iit != eit && isDigitOrSep(*iit, sep); ++iit)
762 s += static_cast<char>(*iit);
763 if (iit != eit && *iit == dot) {
766 for (; iit != eit && isDigitOrSep(*iit, 0); ++iit)
767 s += static_cast<char>(*iit);
768 if (iit != eit && (*iit == 'e' || *iit == 'E')) {
769 s += static_cast<char>(*iit);
771 for (; iit != eit && isDigitOrSep(*iit, 0); ++iit)
772 s += static_cast<char>(*iit);
776 string_num_get_facet f;
777 f.get(s.begin(), s.end(), b, err, v);
779 err |= ios_base::eofbit;
784 bool isDigitOrSep(lyx::char_type const c, char const sep) const
786 return (c >= '0' && c <= '9') || (c != 0 && c == sep);
791 /// class to add our facets to the global locale
792 class locale_initializer {
797 locale const loc1(global, new ascii_ctype_facet);
798 locale const loc2(loc1, new ascii_num_put_facet);
799 locale const loc3(loc2, new ascii_num_get_facet);
800 locale::global(loc3);
807 /// make sure that our facets get used
808 static locale_initializer initializer;