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 static IconvProcessor iconv(ucs4_codeset, "UTF-8");
62 size_t n = utf8.size();
63 // as utf8 is a multi-byte encoding, there would be at most
69 int maxoutsize = n * 4;
70 // basic_string::data() is not recognized by some old gcc version
71 // so we use &(ucs4[0]) instead.
72 char * outbuf = (char *)(&(ucs4[0]));
73 int bytes = iconv.convert(utf8.c_str(), n, outbuf, maxoutsize);
75 // adjust to the real converted size
80 docstring const from_utf8(std::string const & utf8)
83 utf8_to_ucs4(utf8, ucs4);
88 std::string const to_utf8(docstring const & ucs4)
90 std::vector<char> const utf8 =
91 ucs4_to_utf8(ucs4.data(), ucs4.size());
92 return std::string(utf8.begin(), utf8.end());
96 bool operator==(lyx::docstring const & l, char const * r)
98 int const len = l.length();
99 for (int i = 0; i < len; ++i) {
100 BOOST_ASSERT(static_cast<unsigned char>(r[i]) < 0x80);
103 if (l[i] != lyx::docstring::value_type(r[i]))
106 return r[len] == '\0';
110 lyx::docstring operator+(lyx::docstring const & l, char const * r)
113 for (char const * c = r; *c; ++c) {
114 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
121 lyx::docstring operator+(char const * l, lyx::docstring const & r)
124 for (char const * c = l; *c; ++c) {
125 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
133 lyx::docstring operator+(lyx::docstring const & l, char r)
135 BOOST_ASSERT(static_cast<unsigned char>(r) < 0x80);
136 return l + lyx::docstring::value_type(r);
140 lyx::docstring operator+(char l, lyx::docstring const & r)
142 BOOST_ASSERT(static_cast<unsigned char>(l) < 0x80);
143 return lyx::docstring::value_type(l) + r;
147 lyx::docstring & operator+=(lyx::docstring & l, char const * r)
149 for (char const * c = r; *c; ++c) {
150 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
157 lyx::docstring & operator+=(lyx::docstring & l, char r)
159 BOOST_ASSERT(static_cast<unsigned char>(r) < 0x80);
166 #if (!defined(HAVE_WCHAR_T) || SIZEOF_WCHAR_T != 4) && defined(__GNUC__)
168 // gcc does not have proper locale facets for lyx::char_type if
169 // sizeof(wchar_t) == 2, so we have to implement them on our own.
172 // We get undefined references to these virtual methods. This looks like
173 // a bug in gcc. The implementation here does not do anything useful, since
174 // it is overriden in ascii_ctype_facet.
176 template<> ctype<lyx::char_type>::~ctype() {}
178 ctype<lyx::char_type>::do_is(ctype<lyx::char_type>::mask, lyx::char_type) const { return false; }
179 template<> lyx::char_type const *
180 ctype<lyx::char_type>::do_is(const lyx::char_type *, const lyx::char_type *, ctype<lyx::char_type>::mask *) const { return 0; }
181 template<> const lyx::char_type *
182 ctype<lyx::char_type>::do_scan_is(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
183 template<> const lyx::char_type *
184 ctype<lyx::char_type>::do_scan_not(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
185 template<> lyx::char_type ctype<lyx::char_type>::do_toupper(lyx::char_type) const { return 0; }
186 template<> const lyx::char_type * ctype<lyx::char_type>::do_toupper(lyx::char_type *, lyx::char_type const *) const { return 0; }
187 template<> lyx::char_type ctype<lyx::char_type>::do_tolower(lyx::char_type) const { return 0; }
188 template<> const lyx::char_type * ctype<lyx::char_type>::do_tolower(lyx::char_type *, lyx::char_type const *) const { return 0; }
189 template<> lyx::char_type ctype<lyx::char_type>::do_widen(char) const { return 0; }
190 template<> const char *
191 ctype<lyx::char_type>::do_widen(const char *, const char *, lyx::char_type *) const { return 0; }
193 ctype<lyx::char_type>::do_narrow(const lyx::char_type, char) const { return 0; }
194 template<> const lyx::char_type *
195 ctype<lyx::char_type>::do_narrow(const lyx::char_type *, const lyx::char_type *, char, char *) const { return 0; }
201 class ctype_failure : public std::bad_cast {
203 ctype_failure() throw() : std::bad_cast() {}
204 virtual ~ctype_failure() throw() {}
205 virtual const char* what() const throw()
207 return "The ctype<lyx::char_type> locale facet does only support ASCII characters on this platform.";
212 class num_put_failure : public std::bad_cast {
214 num_put_failure() throw() : std::bad_cast() {}
215 virtual ~num_put_failure() throw() {}
216 virtual const char* what() const throw()
218 return "The num_put locale facet does only support ASCII characters on this platform.";
223 /// ctype facet for UCS4 characters. The implementation does only support pure
224 /// ASCII, since we do not need anything else for now.
225 /// The code is partly stolen from std::ctype<wchar_t> from gcc.
226 class ascii_ctype_facet : public std::ctype<lyx::char_type>
229 typedef lyx::char_type char_type;
230 typedef wctype_t wmask_type;
231 explicit ascii_ctype_facet(size_t refs = 0) : std::ctype<char_type>(refs)
233 M_initialize_ctype();
238 wint_t M_widen[1 + static_cast<unsigned char>(-1)];
240 wmask_type M_wmask[16];
241 wmask_type M_convert_to_wmask(const mask m) const
245 case space: ret = wctype("space"); break;
246 case print: ret = wctype("print"); break;
247 case cntrl: ret = wctype("cntrl"); break;
248 case upper: ret = wctype("upper"); break;
249 case lower: ret = wctype("lower"); break;
250 case alpha: ret = wctype("alpha"); break;
251 case digit: ret = wctype("digit"); break;
252 case punct: ret = wctype("punct"); break;
253 case xdigit: ret = wctype("xdigit"); break;
254 case alnum: ret = wctype("alnum"); break;
255 case graph: ret = wctype("graph"); break;
256 default: ret = wmask_type();
260 void M_initialize_ctype()
263 for (i = 0; i < 128; ++i) {
264 const int c = wctob(i);
268 M_narrow[i] = static_cast<char>(c);
274 for (size_t i = 0; i < sizeof(M_widen) / sizeof(wint_t); ++i)
275 M_widen[i] = btowc(i);
277 for (size_t i = 0; i <= 15; ++i) {
278 M_bit[i] = static_cast<mask>(1 << i);
279 M_wmask[i] = M_convert_to_wmask(M_bit[i]);
282 virtual ~ascii_ctype_facet() {}
283 char_type do_toupper(char_type c) const
286 throw ctype_failure();
287 return toupper(static_cast<int>(c));
289 char_type const * do_toupper(char_type * lo, char_type const * hi) const
293 throw ctype_failure();
294 *lo = toupper(static_cast<int>(*lo));
299 char_type do_tolower(char_type c) const
302 throw ctype_failure();
305 char_type const * do_tolower(char_type * lo, char_type const * hi) const
309 throw ctype_failure();
315 bool do_is(mask m, char_type c) const
318 throw ctype_failure();
319 // The code below works because c is in the ASCII range.
320 // We could not use iswctype() which is designed for a 2byte
321 // whar_t without encoding conversion otherwise.
323 // Generically, 15 (instead of 10) since we don't know the numerical
324 // encoding of the various categories in /usr/include/ctype.h.
325 const size_t bitmasksize = 15;
326 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
327 if (m & M_bit[bitcur] &&
328 iswctype(static_cast<int>(c), M_wmask[bitcur])) {
334 char_type const * do_is(char_type const * lo, char_type const * hi, mask * vec) const
336 for (;lo < hi; ++vec, ++lo) {
338 throw ctype_failure();
339 // The code below works because c is in the ASCII range.
340 // We could not use iswctype() which is designed for a 2byte
341 // whar_t without encoding conversion otherwise.
342 // Generically, 15 (instead of 10) since we don't know the numerical
343 // encoding of the various categories in /usr/include/ctype.h.
344 const size_t bitmasksize = 15;
346 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
347 if (iswctype(static_cast<int>(*lo), M_wmask[bitcur]))
353 char_type const * do_scan_is(mask m, char_type const * lo, char_type const * hi) const
355 while (lo < hi && !this->do_is(m, *lo))
359 char_type const * do_scan_not(mask m, char_type const * lo, char_type const * hi) const
361 while (lo < hi && this->do_is(m, *lo) != 0)
365 char_type do_widen(char c) const
367 if (static_cast<unsigned char>(c) < 0x80)
369 throw ctype_failure();
371 const char* do_widen(const char* lo, const char* hi, char_type* dest) const
374 if (static_cast<unsigned char>(*lo) >= 0x80)
375 throw ctype_failure();
382 char do_narrow(char_type wc, char) const
385 return static_cast<char>(wc);
386 throw ctype_failure();
388 const char_type * do_narrow(const char_type * lo, const char_type * hi, char, char * dest) const
392 *dest = static_cast<char>(*lo);
394 throw ctype_failure();
403 /// Facet for outputting numbers to odocstreams as ascii.
404 /// Here we simply need defining the virtual do_put functions.
405 class ascii_num_put_facet : public std::num_put<lyx::char_type, std::ostreambuf_iterator<lyx::char_type, std::char_traits<lyx::char_type> > >
407 typedef std::ostreambuf_iterator<lyx::char_type, std::char_traits<lyx::char_type> > iter_type;
409 ascii_num_put_facet(size_t refs = 0) : std::num_put<lyx::char_type, iter_type>(refs) {}
411 /// Facet for converting numbers to ascii strings.
412 class string_num_put_facet : public std::num_put<char, std::basic_string<char>::iterator>
415 string_num_put_facet() : std::num_put<char, std::basic_string<char>::iterator>(1) {}
420 do_put(iter_type oit, std::ios_base & b, char_type fill, long v) const
423 throw num_put_failure();
426 // 64 is large enough
428 string_num_put_facet f;
429 std::string::const_iterator cit = s.begin();
430 std::string::const_iterator end =
431 f.put(s.begin(), b, fill, v);
432 for (; cit != end; ++cit, ++oit)
440 /// class to add our facets to the global locale
441 class locale_initializer {
446 std::locale const loc1(global, new ascii_ctype_facet);
447 std::locale const loc2(loc1, new ascii_num_put_facet);
448 std::locale::global(loc2);
455 /// make sure that our facets get used
456 static locale_initializer initializer;