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"
22 #if ! defined(USE_WCHAR_T) && defined(__GNUC__)
30 using lyx::support::isHexChar;
34 docstring const from_ascii(char const * ascii)
37 if (int n = strlen(ascii)) {
42 LASSERT(static_cast<unsigned char>(ascii[n]) < 0x80, /**/);
49 docstring const from_ascii(string const & ascii)
51 int const len = ascii.length();
52 for (int i = 0; i < len; ++i)
53 LASSERT(static_cast<unsigned char>(ascii[i]) < 0x80, /**/);
54 return docstring(ascii.begin(), ascii.end());
58 string const to_ascii(docstring const & ucs4)
60 int const len = ucs4.length();
63 for (int i = 0; i < len; ++i) {
64 LASSERT(ucs4[i] < 0x80, /**/);
65 ascii[i] = static_cast<char>(ucs4[i]);
71 IconvProcessor & utf8ToUcs4()
73 static IconvProcessor iconv(ucs4_codeset, "UTF-8");
79 void utf8_to_ucs4(string const & utf8, docstring & ucs4)
81 size_t n = utf8.size();
82 // as utf8 is a multi-byte encoding, there would be at most
88 int maxoutsize = n * 4;
89 // basic_string::data() is not recognized by some old gcc version
90 // so we use &(ucs4[0]) instead.
91 char * outbuf = (char *)(&(ucs4[0]));
92 int bytes = utf8ToUcs4().convert(utf8.c_str(), n, outbuf, maxoutsize);
94 // adjust to the real converted size
99 docstring const from_utf8(string const & utf8)
102 utf8_to_ucs4(utf8, ucs4);
107 string const to_utf8(docstring const & ucs4)
109 vector<char> const utf8 = ucs4_to_utf8(ucs4.data(), ucs4.size());
110 return string(utf8.begin(), utf8.end());
114 docstring const from_local8bit(string const & s)
116 return qstring_to_ucs4(QString::fromLocal8Bit(s.data(), s.length()));
120 /// Exception thrown by to_local8bit if the string could not be converted
121 class to_local8bit_failure : public bad_cast {
123 to_local8bit_failure() throw() : bad_cast() {}
124 virtual ~to_local8bit_failure() throw() {}
125 virtual const char* what() const throw()
127 return "A string could not be converted from unicode to the local 8 bit encoding.";
132 string const to_local8bit(docstring const & s)
134 // This conversion can fail, depending on input.
137 QByteArray const local = toqstr(s).toLocal8Bit();
139 throw to_local8bit_failure();
140 return string(local.begin(), local.end());
144 docstring const from_filesystem8bit(string const & s)
146 QByteArray const encoded(s.c_str(), s.length());
147 return qstring_to_ucs4(QFile::decodeName(encoded));
151 string const to_filesystem8bit(docstring const & s)
153 QByteArray const encoded = QFile::encodeName(toqstr(s));
154 return string(encoded.begin(), encoded.end());
158 string const to_iconv_encoding(docstring const & s, string const & encoding)
160 std::vector<char> const encoded =
161 ucs4_to_eightbit(s.data(), s.length(), encoding);
162 return string(encoded.begin(), encoded.end());
166 docstring const from_iconv_encoding(string const & s, string const & encoding)
168 std::vector<char_type> const ucs4 =
169 eightbit_to_ucs4(s.data(), s.length(), encoding);
170 return docstring(ucs4.begin(), ucs4.end());
174 docstring const normalize_c(docstring const & s)
176 return qstring_to_ucs4(toqstr(s).normalized(QString::NormalizationForm_C));
180 bool operator==(lyx::docstring const & l, char const * r)
182 lyx::docstring::const_iterator it = l.begin();
183 lyx::docstring::const_iterator end = l.end();
184 for (; it != end; ++it, ++r) {
185 LASSERT(static_cast<unsigned char>(*r) < 0x80, /**/);
188 if (*it != static_cast<lyx::docstring::value_type>(*r))
195 lyx::docstring operator+(lyx::docstring const & l, char const * r)
198 for (char const * c = r; *c; ++c) {
199 LASSERT(static_cast<unsigned char>(*c) < 0x80, /**/);
206 lyx::docstring operator+(char const * l, lyx::docstring const & r)
209 for (char const * c = l; *c; ++c) {
210 LASSERT(static_cast<unsigned char>(*c) < 0x80, /**/);
218 lyx::docstring operator+(lyx::docstring const & l, char r)
220 LASSERT(static_cast<unsigned char>(r) < 0x80, /**/);
222 s += docstring::value_type(r);
227 lyx::docstring operator+(char l, lyx::docstring const & r)
229 LASSERT(static_cast<unsigned char>(l) < 0x80, /**/);
230 return lyx::docstring::value_type(l) + r;
234 lyx::docstring & operator+=(lyx::docstring & l, char const * r)
236 for (char const * c = r; *c; ++c) {
237 LASSERT(static_cast<unsigned char>(*c) < 0x80, /**/);
244 lyx::docstring & operator+=(lyx::docstring & l, char r)
246 LASSERT(static_cast<unsigned char>(r) < 0x80, /**/);
253 #if ! defined(USE_WCHAR_T) && defined(__GNUC__)
255 // gcc does not have proper locale facets for lyx::char_type if
256 // sizeof(wchar_t) == 2, so we have to implement them on our own.
259 // We get undefined references to these virtual methods. This looks like
260 // a bug in gcc. The implementation here does not do anything useful, since
261 // it is overriden in ascii_ctype_facet.
263 template<> ctype<lyx::char_type>::~ctype() {}
265 ctype<lyx::char_type>::do_is(ctype<lyx::char_type>::mask, lyx::char_type) const { return false; }
266 template<> lyx::char_type const *
267 ctype<lyx::char_type>::do_is(const lyx::char_type *, const lyx::char_type *, ctype<lyx::char_type>::mask *) const { return 0; }
268 template<> const lyx::char_type *
269 ctype<lyx::char_type>::do_scan_is(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
270 template<> const lyx::char_type *
271 ctype<lyx::char_type>::do_scan_not(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
272 template<> lyx::char_type ctype<lyx::char_type>::do_toupper(lyx::char_type) const { return 0; }
273 template<> const lyx::char_type * ctype<lyx::char_type>::do_toupper(lyx::char_type *, lyx::char_type const *) const { return 0; }
274 template<> lyx::char_type ctype<lyx::char_type>::do_tolower(lyx::char_type) const { return 0; }
275 template<> const lyx::char_type * ctype<lyx::char_type>::do_tolower(lyx::char_type *, lyx::char_type const *) const { return 0; }
276 template<> lyx::char_type ctype<lyx::char_type>::do_widen(char) const { return 0; }
277 template<> const char *
278 ctype<lyx::char_type>::do_widen(const char *, const char *, lyx::char_type *) const { return 0; }
280 ctype<lyx::char_type>::do_narrow(const lyx::char_type, char) const { return 0; }
281 template<> const lyx::char_type *
282 ctype<lyx::char_type>::do_narrow(const lyx::char_type *, const lyx::char_type *, char, char *) const { return 0; }
288 class ctype_failure : public bad_cast {
290 ctype_failure() throw() : bad_cast() {}
291 virtual ~ctype_failure() throw() {}
292 virtual const char* what() const throw()
294 return "The ctype<lyx::char_type> locale facet does only support ASCII characters on this platform.";
299 class num_put_failure : public bad_cast {
301 num_put_failure() throw() : bad_cast() {}
302 virtual ~num_put_failure() throw() {}
303 virtual const char* what() const throw()
305 return "The num_put locale facet does only support ASCII characters on this platform.";
310 /// ctype facet for UCS4 characters. The implementation does only support pure
311 /// ASCII, since we do not need anything else for now.
312 /// The code is partly stolen from ctype<wchar_t> from gcc.
313 class ascii_ctype_facet : public ctype<lyx::char_type>
316 typedef lyx::char_type char_type;
317 typedef wctype_t wmask_type;
318 explicit ascii_ctype_facet(size_t refs = 0) : ctype<char_type>(refs)
320 M_initialize_ctype();
325 wint_t M_widen[1 + static_cast<unsigned char>(-1)];
327 wmask_type M_wmask[16];
328 wmask_type M_convert_to_wmask(const mask m) const
332 case space: ret = wctype("space"); break;
333 case print: ret = wctype("print"); break;
334 case cntrl: ret = wctype("cntrl"); break;
335 case upper: ret = wctype("upper"); break;
336 case lower: ret = wctype("lower"); break;
337 case alpha: ret = wctype("alpha"); break;
338 case digit: ret = wctype("digit"); break;
339 case punct: ret = wctype("punct"); break;
340 case xdigit: ret = wctype("xdigit"); break;
341 case alnum: ret = wctype("alnum"); break;
342 case graph: ret = wctype("graph"); break;
343 default: ret = wmask_type();
347 void M_initialize_ctype()
350 for (i = 0; i < 128; ++i) {
351 const int c = wctob(i);
355 M_narrow[i] = static_cast<char>(c);
361 for (size_t i = 0; i < sizeof(M_widen) / sizeof(wint_t); ++i)
362 M_widen[i] = btowc(i);
364 for (size_t i = 0; i <= 15; ++i) {
365 M_bit[i] = static_cast<mask>(1 << i);
366 M_wmask[i] = M_convert_to_wmask(M_bit[i]);
369 virtual ~ascii_ctype_facet() {}
370 char_type do_toupper(char_type c) const
373 throw ctype_failure();
374 return toupper(static_cast<int>(c));
376 char_type const * do_toupper(char_type * lo, char_type const * hi) const
380 throw ctype_failure();
381 *lo = toupper(static_cast<int>(*lo));
386 char_type do_tolower(char_type c) const
389 throw ctype_failure();
392 char_type const * do_tolower(char_type * lo, char_type const * hi) const
396 throw ctype_failure();
402 bool do_is(mask m, char_type c) const
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.
410 // Generically, 15 (instead of 10) since we don't know the numerical
411 // encoding of the various categories in /usr/include/ctype.h.
412 const size_t bitmasksize = 15;
413 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
414 if (m & M_bit[bitcur] &&
415 iswctype(static_cast<int>(c), M_wmask[bitcur])) {
421 char_type const * do_is(char_type const * lo, char_type const * hi, mask * vec) const
423 for (;lo < hi; ++vec, ++lo) {
425 throw ctype_failure();
426 // The code below works because c is in the ASCII range.
427 // We could not use iswctype() which is designed for a 2byte
428 // whar_t without encoding conversion otherwise.
429 // Generically, 15 (instead of 10) since we don't know the numerical
430 // encoding of the various categories in /usr/include/ctype.h.
431 const size_t bitmasksize = 15;
433 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
434 if (iswctype(static_cast<int>(*lo), M_wmask[bitcur]))
440 char_type const * do_scan_is(mask m, char_type const * lo, char_type const * hi) const
442 while (lo < hi && !this->do_is(m, *lo))
446 char_type const * do_scan_not(mask m, char_type const * lo, char_type const * hi) const
448 while (lo < hi && this->do_is(m, *lo) != 0)
452 char_type do_widen(char c) const
454 if (static_cast<unsigned char>(c) < 0x80)
456 throw ctype_failure();
458 const char* do_widen(const char* lo, const char* hi, char_type* dest) const
461 if (static_cast<unsigned char>(*lo) >= 0x80)
462 throw ctype_failure();
469 char do_narrow(char_type wc, char) const
472 return static_cast<char>(wc);
473 throw ctype_failure();
475 const char_type * do_narrow(const char_type * lo, const char_type * hi, char, char * dest) const
479 *dest = static_cast<char>(*lo);
481 throw ctype_failure();
490 /// Facet for outputting numbers to odocstreams as ascii.
491 /// Here we simply need defining the virtual do_put functions.
492 class ascii_num_put_facet : public num_put<lyx::char_type, ostreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > >
494 typedef ostreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > iter_type;
496 ascii_num_put_facet(size_t refs = 0) : num_put<lyx::char_type, iter_type>(refs) {}
498 /// Facet for converting numbers to ascii strings.
499 class string_num_put_facet : public num_put<char, basic_string<char>::iterator>
502 string_num_put_facet() : num_put<char, basic_string<char>::iterator>(1) {}
507 do_put(iter_type oit, ios_base & b, char_type fill, bool v) const
509 return do_put_helper(oit, b, fill, v);
513 do_put(iter_type oit, ios_base & b, char_type fill, long v) const
515 return do_put_helper(oit, b, fill, v);
519 do_put(iter_type oit, ios_base & b, char_type fill, unsigned long v) const
521 return do_put_helper(oit, b, fill, v);
524 #ifdef _GLIBCXX_USE_LONG_LONG
526 do_put(iter_type oit, ios_base & b, char_type fill, long long v) const
528 return do_put_helper(oit, b, fill, v);
532 do_put(iter_type oit, ios_base & b, char_type fill, unsigned long long v) const
534 return do_put_helper(oit, b, fill, v);
539 do_put(iter_type oit, ios_base & b, char_type fill, double v) const
541 return do_put_helper(oit, b, fill, v);
545 do_put(iter_type oit, ios_base & b, char_type fill, long double v) const
547 return do_put_helper(oit, b, fill, v);
551 do_put(iter_type oit, ios_base & b, char_type fill, void const * v) const
553 return do_put_helper(oit, b, fill, v);
557 template <typename ValueType>
559 do_put_helper(iter_type oit, ios_base & b, char_type fill, ValueType v) const
562 throw num_put_failure();
564 streamsize const sz = b.width() > b.precision() ?
565 b.width() : b.precision();
566 // 64 is large enough, unless width or precision are bigger
567 streamsize const wd = (sz > 56 ? sz : 56) + 8;
569 string_num_put_facet f;
570 string::const_iterator cit = s.begin();
571 string::const_iterator end =
572 f.put(s.begin(), b, fill, v);
573 for (; cit != end; ++cit, ++oit)
581 /// Facet for inputting ascii representations of numbers from idocstreams.
582 /// Here we simply need defining the virtual do_get functions.
583 class ascii_num_get_facet : public num_get<lyx::char_type, istreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > >
585 typedef istreambuf_iterator<lyx::char_type, char_traits<lyx::char_type> > iter_type;
587 ascii_num_get_facet(size_t refs = 0) : num_get<lyx::char_type, iter_type>(refs) {}
589 /// Facet for converting ascii representation of numbers to a value.
590 class string_num_get_facet : public num_get<char, basic_string<char>::iterator>
593 string_num_get_facet() : num_get<char, basic_string<char>::iterator>(1) {}
596 /// Numpunct facet defining the I/O format.
597 class numpunct_facet : public numpunct<char>
600 numpunct_facet() : numpunct<char>(1) {}
605 do_get(iter_type iit, iter_type eit, ios_base & b,
606 ios_base::iostate & err, bool & v) const
608 if (b.flags() & ios_base::boolalpha) {
610 lyx::docstring const truename = from_local8bit(p.truename());
611 lyx::docstring const falsename = from_local8bit(p.falsename());
616 size_t const tsize = truename.size();
617 size_t const fsize = falsename.size();
618 for (; iit != eit; ++iit) {
621 bool true_ok = support::prefixIs(truename, s);
622 bool false_ok = support::prefixIs(falsename, s);
623 if (!true_ok && !false_ok) {
628 if ((true_ok && n == tsize) ||
629 (false_ok && n == fsize)) {
635 err = ios_base::goodbit;
636 v = truename == s ? true : false;
638 err = ios_base::failbit;
640 err |= ios_base::eofbit;
644 iter_type end = this->do_get(iit, eit, b, err, l);
645 if (!(err & ios_base::failbit)) {
651 err |= ios_base::failbit;
658 do_get(iter_type iit, iter_type eit, ios_base & b,
659 ios_base::iostate & err, long & v) const
661 return do_get_integer(iit, eit, b, err, v);
665 do_get(iter_type iit, iter_type eit, ios_base & b,
666 ios_base::iostate & err, unsigned short & v) const
668 return do_get_integer(iit, eit, b, err, v);
672 do_get(iter_type iit, iter_type eit, ios_base & b,
673 ios_base::iostate & err, unsigned int & v) const
675 return do_get_integer(iit, eit, b, err, v);
679 do_get(iter_type iit, iter_type eit, ios_base & b,
680 ios_base::iostate & err, unsigned long & v) const
682 return do_get_integer(iit, eit, b, err, v);
685 #ifdef _GLIBCXX_USE_LONG_LONG
687 do_get(iter_type iit, iter_type eit, ios_base & b,
688 ios_base::iostate & err, long long & v) const
690 return do_get_integer(iit, eit, b, err, v);
694 do_get(iter_type iit, iter_type eit, ios_base & b,
695 ios_base::iostate & err, unsigned long long & v) const
697 return do_get_integer(iit, eit, b, err, v);
702 do_get(iter_type iit, iter_type eit, ios_base & b,
703 ios_base::iostate & err, float & v) const
705 return do_get_float(iit, eit, b, err, v);
709 do_get(iter_type iit, iter_type eit, ios_base & b,
710 ios_base::iostate & err, double & v) const
712 return do_get_float(iit, eit, b, err, v);
716 do_get(iter_type iit, iter_type eit, ios_base & b,
717 ios_base::iostate & err, long double & v) const
719 return do_get_float(iit, eit, b, err, v);
723 do_get(iter_type iit, iter_type eit, ios_base & b,
724 ios_base::iostate & err, void * & v) const
727 iter_type end = do_get_integer(iit, eit, b, err, val);
728 if (!(err & ios_base::failbit))
729 v = reinterpret_cast<void *>(val);
734 template <typename ValueType>
736 do_get_integer(iter_type iit, iter_type eit, ios_base & b,
737 ios_base::iostate & err, ValueType & v) const
741 for (; iit != eit && isNumpunct(*iit); ++iit)
742 s += static_cast<char>(*iit);
743 // We add another character, not part of the numpunct facet,
744 // in order to avoid setting the eofbit in the stream state,
745 // which would prevent any further read. The space seems a
748 string_num_get_facet f;
749 f.get(s.begin(), s.end(), b, err, v);
751 err |= ios_base::eofbit;
756 bool isNumpunct(lyx::char_type const c) const
758 /// Only account for the standard numpunct "C" locale facet.
759 return c == '-' || c == '+'
760 || c == 'x' || c == 'X'
764 template <typename ValueType>
766 do_get_float(iter_type iit, iter_type eit, ios_base & b,
767 ios_base::iostate & err, ValueType & v) const
769 // Gather a string of the form
770 // [+-]? [0-9]* .? [0-9]* ([eE] [+-]? [0-9]+)?
775 char const dot = p.decimal_point();
776 char const sep = p.thousands_sep();
777 // Get an optional sign
778 if (iit != eit && (*iit == '-' || *iit == '+')) {
779 s += static_cast<char>(*iit);
782 for (; iit != eit && isDigitOrSep(*iit, sep); ++iit)
783 s += static_cast<char>(*iit);
784 if (iit != eit && *iit == dot) {
787 for (; iit != eit && isDigitOrSep(*iit, 0); ++iit)
788 s += static_cast<char>(*iit);
789 if (iit != eit && (*iit == 'e' || *iit == 'E')) {
790 s += static_cast<char>(*iit);
792 for (; iit != eit && isDigitOrSep(*iit, 0); ++iit)
793 s += static_cast<char>(*iit);
797 string_num_get_facet f;
798 f.get(s.begin(), s.end(), b, err, v);
800 err |= ios_base::eofbit;
805 bool isDigitOrSep(lyx::char_type const c, char const sep) const
807 return (c >= '0' && c <= '9') || (c != 0 && c == sep);
812 /// class to add our facets to the global locale
813 class locale_initializer {
818 locale const loc1(global, new ascii_ctype_facet);
819 locale const loc2(loc1, new ascii_num_put_facet);
820 locale const loc3(loc2, new ascii_num_get_facet);
821 locale::global(loc3);
828 /// make sure that our facets get used
829 static locale_initializer initializer;