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"
18 #include <boost/assert.hpp>
23 docstring const from_ascii(char const * ascii)
26 for (char const * c = ascii; *c; ++c) {
27 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
34 docstring const from_ascii(std::string const & ascii)
36 int const len = ascii.length();
37 for (int i = 0; i < len; ++i)
38 BOOST_ASSERT(static_cast<unsigned char>(ascii[i]) < 0x80);
39 return docstring(ascii.begin(), ascii.end());
43 std::string const to_ascii(docstring const & ucs4)
45 int const len = ucs4.length();
48 for (int i = 0; i < len; ++i) {
49 BOOST_ASSERT(ucs4[i] < 0x80);
50 ascii[i] = static_cast<char>(ucs4[i]);
56 docstring const from_utf8(std::string const & utf8)
58 std::vector<lyx::char_type> const ucs4 =
59 utf8_to_ucs4(utf8.data(), utf8.size());
60 return docstring(ucs4.begin(), ucs4.end());
64 std::string const to_utf8(docstring const & ucs4)
66 std::vector<char> const utf8 =
67 ucs4_to_utf8(ucs4.data(), ucs4.size());
68 return std::string(utf8.begin(), utf8.end());
72 bool operator==(lyx::docstring const & l, char const * r)
74 int const len = l.length();
75 for (int i = 0; i < len; ++i) {
76 BOOST_ASSERT(static_cast<unsigned char>(r[i]) < 0x80);
79 if (l[i] != lyx::docstring::value_type(r[i]))
82 return r[len] == '\0';
86 lyx::docstring operator+(lyx::docstring const & l, char const * r)
89 for (char const * c = r; *c; ++c) {
90 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
97 lyx::docstring operator+(char const * l, lyx::docstring const & r)
100 for (char const * c = l; *c; ++c) {
101 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
109 lyx::docstring operator+(lyx::docstring const & l, char r)
111 BOOST_ASSERT(static_cast<unsigned char>(r) < 0x80);
112 return l + lyx::docstring::value_type(r);
116 lyx::docstring operator+(char l, lyx::docstring const & r)
118 BOOST_ASSERT(static_cast<unsigned char>(l) < 0x80);
119 return lyx::docstring::value_type(l) + r;
123 lyx::docstring & operator+=(lyx::docstring & l, char const * r)
125 for (char const * c = r; *c; ++c) {
126 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
133 lyx::docstring & operator+=(lyx::docstring & l, char r)
135 BOOST_ASSERT(static_cast<unsigned char>(r) < 0x80);
142 #if (!defined(HAVE_WCHAR_T) || SIZEOF_WCHAR_T != 4) && defined(__GNUC__)
144 // gcc does not have proper locale facets for lyx::char_type if
145 // sizeof(wchar_t) == 2, so we have to implement them on our own.
148 // We get undefined references to these virtual methods. This looks like
149 // a bug in gcc. The implementation here does not do anything useful, since
150 // it is overriden in ascii_ctype_facet.
152 template<> ctype<lyx::char_type>::~ctype() {}
154 ctype<lyx::char_type>::do_is(ctype<lyx::char_type>::mask, lyx::char_type) const { return false; }
155 template<> lyx::char_type const *
156 ctype<lyx::char_type>::do_is(const lyx::char_type *, const lyx::char_type *, ctype<lyx::char_type>::mask *) const { return 0; }
157 template<> const lyx::char_type *
158 ctype<lyx::char_type>::do_scan_is(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
159 template<> const lyx::char_type *
160 ctype<lyx::char_type>::do_scan_not(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
161 template<> lyx::char_type ctype<lyx::char_type>::do_toupper(lyx::char_type) const { return 0; }
162 template<> const lyx::char_type * ctype<lyx::char_type>::do_toupper(lyx::char_type *, lyx::char_type const *) const { return 0; }
163 template<> lyx::char_type ctype<lyx::char_type>::do_tolower(lyx::char_type) const { return 0; }
164 template<> const lyx::char_type * ctype<lyx::char_type>::do_tolower(lyx::char_type *, lyx::char_type const *) const { return 0; }
165 template<> lyx::char_type ctype<lyx::char_type>::do_widen(char) const { return 0; }
166 template<> const char *
167 ctype<lyx::char_type>::do_widen(const char *, const char *, lyx::char_type *) const { return 0; }
169 ctype<lyx::char_type>::do_narrow(const lyx::char_type, char) const { return 0; }
170 template<> const lyx::char_type *
171 ctype<lyx::char_type>::do_narrow(const lyx::char_type *, const lyx::char_type *, char, char *) const { return 0; }
177 class ctype_failure : public std::bad_cast {
179 ctype_failure() throw() : std::bad_cast() {}
180 virtual ~ctype_failure() throw() {}
181 virtual const char* what() const throw()
183 return "The ctype<lyx::char_type> locale facet does only support ASCII characters on this platform.";
188 /// ctype facet for UCS4 characters. The implementation does only support pure
189 /// ASCII, since we do not need anything else for now.
190 /// The code is partly stolen from std::ctype<wchar_t> from gcc.
191 class ascii_ctype_facet : public std::ctype<lyx::char_type>
194 typedef lyx::char_type char_type;
195 typedef wctype_t wmask_type;
196 explicit ascii_ctype_facet(size_t refs = 0) : std::ctype<char_type>(refs)
198 M_initialize_ctype();
203 wint_t M_widen[1 + static_cast<unsigned char>(-1)];
205 wmask_type M_wmask[16];
206 wmask_type M_convert_to_wmask(const mask m) const
210 case space: ret = wctype("space"); break;
211 case print: ret = wctype("print"); break;
212 case cntrl: ret = wctype("cntrl"); break;
213 case upper: ret = wctype("upper"); break;
214 case lower: ret = wctype("lower"); break;
215 case alpha: ret = wctype("alpha"); break;
216 case digit: ret = wctype("digit"); break;
217 case punct: ret = wctype("punct"); break;
218 case xdigit: ret = wctype("xdigit"); break;
219 case alnum: ret = wctype("alnum"); break;
220 case graph: ret = wctype("graph"); break;
221 default: ret = wmask_type();
225 void M_initialize_ctype()
228 for (i = 0; i < 128; ++i) {
229 const int c = wctob(i);
233 M_narrow[i] = static_cast<char>(c);
239 for (size_t i = 0; i < sizeof(M_widen) / sizeof(wint_t); ++i)
240 M_widen[i] = btowc(i);
242 for (size_t i = 0; i <= 15; ++i) {
243 M_bit[i] = static_cast<mask>(1 << i);
244 M_wmask[i] = M_convert_to_wmask(M_bit[i]);
247 virtual ~ascii_ctype_facet() {}
248 char_type do_toupper(char_type c) const
251 throw ctype_failure();
252 return toupper(static_cast<int>(c));
254 char_type const * do_toupper(char_type * lo, char_type const * hi) const
258 throw ctype_failure();
259 *lo = toupper(static_cast<int>(*lo));
264 char_type do_tolower(char_type c) const
267 throw ctype_failure();
270 char_type const * do_tolower(char_type * lo, char_type const * hi) const
274 throw ctype_failure();
280 bool do_is(mask m, char_type c) const
283 throw ctype_failure();
284 // The code below works because c is in the ASCII range.
285 // We could not use iswctype() which is designed for a 2byte
286 // whar_t without encoding conversion otherwise.
288 // Generically, 15 (instead of 10) since we don't know the numerical
289 // encoding of the various categories in /usr/include/ctype.h.
290 const size_t bitmasksize = 15;
291 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
292 if (m & M_bit[bitcur] &&
293 iswctype(static_cast<int>(c), M_wmask[bitcur])) {
299 char_type const * do_is(char_type const * lo, char_type const * hi, mask * vec) const
301 for (;lo < hi; ++vec, ++lo) {
303 throw ctype_failure();
304 // The code below works because c is in the ASCII range.
305 // We could not use iswctype() which is designed for a 2byte
306 // whar_t without encoding conversion otherwise.
307 // Generically, 15 (instead of 10) since we don't know the numerical
308 // encoding of the various categories in /usr/include/ctype.h.
309 const size_t bitmasksize = 15;
311 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
312 if (iswctype(static_cast<int>(*lo), M_wmask[bitcur]))
318 char_type const * do_scan_is(mask m, char_type const * lo, char_type const * hi) const
320 while (lo < hi && !this->do_is(m, *lo))
324 char_type const * do_scan_not(mask m, char_type const * lo, char_type const * hi) const
326 while (lo < hi && this->do_is(m, *lo) != 0)
330 char_type do_widen(char c) const
332 if (static_cast<unsigned char>(c) < 0x80)
334 throw ctype_failure();
336 const char* do_widen(const char* lo, const char* hi, char_type* dest) const
339 if (static_cast<unsigned char>(*lo) >= 0x80)
340 throw ctype_failure();
347 char do_narrow(char_type wc, char) const
350 return static_cast<char>(wc);
351 throw ctype_failure();
353 const char_type * do_narrow(const char_type * lo, const char_type * hi, char, char * dest) const
357 *dest = static_cast<char>(*lo);
359 throw ctype_failure();
368 /// class to add our ascii_ctype_facet to the global locale
369 class locale_initializer {
374 std::locale const loc(global, new ascii_ctype_facet);
375 std::locale::global(loc);
382 /// make sure that our ascii_ctype_facet gets used
383 static locale_initializer initializer;