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"
19 #include <boost/assert.hpp>
25 docstring const from_ascii(char const * ascii)
28 for (char const * c = ascii; *c; ++c) {
29 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
36 docstring const from_ascii(std::string const & ascii)
38 int const len = ascii.length();
39 for (int i = 0; i < len; ++i)
40 BOOST_ASSERT(static_cast<unsigned char>(ascii[i]) < 0x80);
41 return docstring(ascii.begin(), ascii.end());
45 std::string const to_ascii(docstring const & ucs4)
47 int const len = ucs4.length();
50 for (int i = 0; i < len; ++i) {
51 BOOST_ASSERT(ucs4[i] < 0x80);
52 ascii[i] = static_cast<char>(ucs4[i]);
58 void utf8_to_ucs4(std::string const & utf8, docstring & ucs4)
60 // FIXME (Abdel 17/11/06): static data are evil!
61 // This function cannot be used in the final exit process on Mac because
62 // static data are already destroyed at this stage.
63 // One solution would be to instantiate the utf8 to ucs4 IconvProcessor as a
64 // singleton inside the LyX main class to ensure that it does not get
65 // destroyed too early.
66 static IconvProcessor iconv(ucs4_codeset, "UTF-8");
68 size_t n = utf8.size();
69 // as utf8 is a multi-byte encoding, there would be at most
75 int maxoutsize = n * 4;
76 // basic_string::data() is not recognized by some old gcc version
77 // so we use &(ucs4[0]) instead.
78 char * outbuf = (char *)(&(ucs4[0]));
79 int bytes = iconv.convert(utf8.c_str(), n, outbuf, maxoutsize);
81 // adjust to the real converted size
86 docstring const from_utf8(std::string const & utf8)
89 utf8_to_ucs4(utf8, ucs4);
94 std::string const to_utf8(docstring const & ucs4)
96 std::vector<char> const utf8 =
97 ucs4_to_utf8(ucs4.data(), ucs4.size());
98 return std::string(utf8.begin(), utf8.end());
102 bool operator==(lyx::docstring const & l, char const * r)
104 int const len = l.length();
105 for (int i = 0; i < len; ++i) {
106 BOOST_ASSERT(static_cast<unsigned char>(r[i]) < 0x80);
109 if (l[i] != lyx::docstring::value_type(r[i]))
112 return r[len] == '\0';
116 lyx::docstring operator+(lyx::docstring const & l, char const * r)
119 for (char const * c = r; *c; ++c) {
120 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
127 lyx::docstring operator+(char const * l, lyx::docstring const & r)
130 for (char const * c = l; *c; ++c) {
131 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
139 lyx::docstring operator+(lyx::docstring const & l, char r)
141 BOOST_ASSERT(static_cast<unsigned char>(r) < 0x80);
142 return l + lyx::docstring::value_type(r);
146 lyx::docstring operator+(char l, lyx::docstring const & r)
148 BOOST_ASSERT(static_cast<unsigned char>(l) < 0x80);
149 return lyx::docstring::value_type(l) + r;
153 lyx::docstring & operator+=(lyx::docstring & l, char const * r)
155 for (char const * c = r; *c; ++c) {
156 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
163 lyx::docstring & operator+=(lyx::docstring & l, char r)
165 BOOST_ASSERT(static_cast<unsigned char>(r) < 0x80);
172 #if (!defined(HAVE_WCHAR_T) || SIZEOF_WCHAR_T != 4) && defined(__GNUC__)
174 // gcc does not have proper locale facets for lyx::char_type if
175 // sizeof(wchar_t) == 2, so we have to implement them on our own.
178 // We get undefined references to these virtual methods. This looks like
179 // a bug in gcc. The implementation here does not do anything useful, since
180 // it is overriden in ascii_ctype_facet.
182 template<> ctype<lyx::char_type>::~ctype() {}
184 ctype<lyx::char_type>::do_is(ctype<lyx::char_type>::mask, lyx::char_type) const { return false; }
185 template<> lyx::char_type const *
186 ctype<lyx::char_type>::do_is(const lyx::char_type *, const lyx::char_type *, ctype<lyx::char_type>::mask *) const { return 0; }
187 template<> const lyx::char_type *
188 ctype<lyx::char_type>::do_scan_is(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
189 template<> const lyx::char_type *
190 ctype<lyx::char_type>::do_scan_not(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
191 template<> lyx::char_type ctype<lyx::char_type>::do_toupper(lyx::char_type) const { return 0; }
192 template<> const lyx::char_type * ctype<lyx::char_type>::do_toupper(lyx::char_type *, lyx::char_type const *) const { return 0; }
193 template<> lyx::char_type ctype<lyx::char_type>::do_tolower(lyx::char_type) const { return 0; }
194 template<> const lyx::char_type * ctype<lyx::char_type>::do_tolower(lyx::char_type *, lyx::char_type const *) const { return 0; }
195 template<> lyx::char_type ctype<lyx::char_type>::do_widen(char) const { return 0; }
196 template<> const char *
197 ctype<lyx::char_type>::do_widen(const char *, const char *, lyx::char_type *) const { return 0; }
199 ctype<lyx::char_type>::do_narrow(const lyx::char_type, char) const { return 0; }
200 template<> const lyx::char_type *
201 ctype<lyx::char_type>::do_narrow(const lyx::char_type *, const lyx::char_type *, char, char *) const { return 0; }
207 class ctype_failure : public std::bad_cast {
209 ctype_failure() throw() : std::bad_cast() {}
210 virtual ~ctype_failure() throw() {}
211 virtual const char* what() const throw()
213 return "The ctype<lyx::char_type> locale facet does only support ASCII characters on this platform.";
218 class num_put_failure : public std::bad_cast {
220 num_put_failure() throw() : std::bad_cast() {}
221 virtual ~num_put_failure() throw() {}
222 virtual const char* what() const throw()
224 return "The num_put locale facet does only support ASCII characters on this platform.";
229 /// ctype facet for UCS4 characters. The implementation does only support pure
230 /// ASCII, since we do not need anything else for now.
231 /// The code is partly stolen from std::ctype<wchar_t> from gcc.
232 class ascii_ctype_facet : public std::ctype<lyx::char_type>
235 typedef lyx::char_type char_type;
236 typedef wctype_t wmask_type;
237 explicit ascii_ctype_facet(size_t refs = 0) : std::ctype<char_type>(refs)
239 M_initialize_ctype();
244 wint_t M_widen[1 + static_cast<unsigned char>(-1)];
246 wmask_type M_wmask[16];
247 wmask_type M_convert_to_wmask(const mask m) const
251 case space: ret = wctype("space"); break;
252 case print: ret = wctype("print"); break;
253 case cntrl: ret = wctype("cntrl"); break;
254 case upper: ret = wctype("upper"); break;
255 case lower: ret = wctype("lower"); break;
256 case alpha: ret = wctype("alpha"); break;
257 case digit: ret = wctype("digit"); break;
258 case punct: ret = wctype("punct"); break;
259 case xdigit: ret = wctype("xdigit"); break;
260 case alnum: ret = wctype("alnum"); break;
261 case graph: ret = wctype("graph"); break;
262 default: ret = wmask_type();
266 void M_initialize_ctype()
269 for (i = 0; i < 128; ++i) {
270 const int c = wctob(i);
274 M_narrow[i] = static_cast<char>(c);
280 for (size_t i = 0; i < sizeof(M_widen) / sizeof(wint_t); ++i)
281 M_widen[i] = btowc(i);
283 for (size_t i = 0; i <= 15; ++i) {
284 M_bit[i] = static_cast<mask>(1 << i);
285 M_wmask[i] = M_convert_to_wmask(M_bit[i]);
288 virtual ~ascii_ctype_facet() {}
289 char_type do_toupper(char_type c) const
292 throw ctype_failure();
293 return toupper(static_cast<int>(c));
295 char_type const * do_toupper(char_type * lo, char_type const * hi) const
299 throw ctype_failure();
300 *lo = toupper(static_cast<int>(*lo));
305 char_type do_tolower(char_type c) const
308 throw ctype_failure();
311 char_type const * do_tolower(char_type * lo, char_type const * hi) const
315 throw ctype_failure();
321 bool do_is(mask m, char_type c) const
324 throw ctype_failure();
325 // The code below works because c is in the ASCII range.
326 // We could not use iswctype() which is designed for a 2byte
327 // whar_t without encoding conversion otherwise.
329 // Generically, 15 (instead of 10) since we don't know the numerical
330 // encoding of the various categories in /usr/include/ctype.h.
331 const size_t bitmasksize = 15;
332 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
333 if (m & M_bit[bitcur] &&
334 iswctype(static_cast<int>(c), M_wmask[bitcur])) {
340 char_type const * do_is(char_type const * lo, char_type const * hi, mask * vec) const
342 for (;lo < hi; ++vec, ++lo) {
344 throw ctype_failure();
345 // The code below works because c is in the ASCII range.
346 // We could not use iswctype() which is designed for a 2byte
347 // whar_t without encoding conversion otherwise.
348 // Generically, 15 (instead of 10) since we don't know the numerical
349 // encoding of the various categories in /usr/include/ctype.h.
350 const size_t bitmasksize = 15;
352 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
353 if (iswctype(static_cast<int>(*lo), M_wmask[bitcur]))
359 char_type const * do_scan_is(mask m, char_type const * lo, char_type const * hi) const
361 while (lo < hi && !this->do_is(m, *lo))
365 char_type const * do_scan_not(mask m, char_type const * lo, char_type const * hi) const
367 while (lo < hi && this->do_is(m, *lo) != 0)
371 char_type do_widen(char c) const
373 if (static_cast<unsigned char>(c) < 0x80)
375 throw ctype_failure();
377 const char* do_widen(const char* lo, const char* hi, char_type* dest) const
380 if (static_cast<unsigned char>(*lo) >= 0x80)
381 throw ctype_failure();
388 char do_narrow(char_type wc, char) const
391 return static_cast<char>(wc);
392 throw ctype_failure();
394 const char_type * do_narrow(const char_type * lo, const char_type * hi, char, char * dest) const
398 *dest = static_cast<char>(*lo);
400 throw ctype_failure();
409 /// Facet for outputting numbers to odocstreams as ascii.
410 /// Here we simply need defining the virtual do_put functions.
411 class ascii_num_put_facet : public std::num_put<lyx::char_type, std::ostreambuf_iterator<lyx::char_type, std::char_traits<lyx::char_type> > >
413 typedef std::ostreambuf_iterator<lyx::char_type, std::char_traits<lyx::char_type> > iter_type;
415 ascii_num_put_facet(size_t refs = 0) : std::num_put<lyx::char_type, iter_type>(refs) {}
417 /// Facet for converting numbers to ascii strings.
418 class string_num_put_facet : public std::num_put<char, std::basic_string<char>::iterator>
421 string_num_put_facet() : std::num_put<char, std::basic_string<char>::iterator>(1) {}
426 do_put(iter_type oit, std::ios_base & b, char_type fill, long v) const
429 throw num_put_failure();
432 // 64 is large enough
434 string_num_put_facet f;
435 std::string::const_iterator cit = s.begin();
436 std::string::const_iterator end =
437 f.put(s.begin(), b, fill, v);
438 for (; cit != end; ++cit, ++oit)
446 /// Facet for inputting ascii representations of numbers from idocstreams.
447 /// Here we simply need defining the virtual do_get functions.
448 class ascii_num_get_facet : public std::num_get<lyx::char_type, std::istreambuf_iterator<lyx::char_type, std::char_traits<lyx::char_type> > >
450 typedef std::istreambuf_iterator<lyx::char_type, std::char_traits<lyx::char_type> > iter_type;
452 ascii_num_get_facet(size_t refs = 0) : std::num_get<lyx::char_type, iter_type>(refs) {}
454 /// Facet for converting ascii representation of numbers to a value.
455 class string_num_get_facet : public std::num_get<char, std::basic_string<char>::iterator>
458 string_num_get_facet() : std::num_get<char, std::basic_string<char>::iterator>(1) {}
462 bool isNumpunct(lyx::char_type const c) const
464 /// Only account for the standard numpunct "C" locale facet.
465 return c < 0x80 && (c == '-' || c == '+' || isdigit(c)
466 || ('a' <= c && c <= 'f') || ('A' <= c && c <= 'F')
467 || c == 'x' || c == 'X');
472 do_get(iter_type iit, iter_type eit, std::ios_base & b,
473 std::ios_base::iostate & err, long & v) const
477 for (int i = 0; iit != eit && isNumpunct(*iit); ++i, ++iit)
478 s[i] = static_cast<char>(*iit);
479 string_num_get_facet f;
480 f.get(s.begin(), s.end(), b, err, v);
487 /// class to add our facets to the global locale
488 class locale_initializer {
493 std::locale const loc1(global, new ascii_ctype_facet);
494 std::locale const loc2(loc1, new ascii_num_put_facet);
495 std::locale const loc3(loc2, new ascii_num_get_facet);
496 std::locale::global(loc3);
503 /// make sure that our facets get used
504 static locale_initializer initializer;