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());
74 bool operator==(lyx::docstring const & l, char const * r)
76 int const len = l.length();
77 for (int i = 0; i < len; ++i) {
78 BOOST_ASSERT(static_cast<unsigned char>(r[i]) < 0x80);
81 if (l[i] != lyx::docstring::value_type(r[i]))
84 return r[len] == '\0';
88 lyx::docstring operator+(lyx::docstring const & l, char const * r)
91 for (char const * c = r; *c; ++c) {
92 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
99 lyx::docstring operator+(char const * l, lyx::docstring const & r)
102 for (char const * c = l; *c; ++c) {
103 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
111 lyx::docstring operator+(lyx::docstring const & l, char r)
113 BOOST_ASSERT(static_cast<unsigned char>(r) < 0x80);
114 return l + lyx::docstring::value_type(r);
118 lyx::docstring operator+(char l, lyx::docstring const & r)
120 BOOST_ASSERT(static_cast<unsigned char>(l) < 0x80);
121 return lyx::docstring::value_type(l) + r;
125 lyx::docstring & operator+=(lyx::docstring & l, char const * r)
127 for (char const * c = r; *c; ++c) {
128 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
135 lyx::docstring & operator+=(lyx::docstring & l, char r)
137 BOOST_ASSERT(static_cast<unsigned char>(r) < 0x80);
143 #if (!defined(HAVE_WCHAR_T) || SIZEOF_WCHAR_T != 4) && defined(__GNUC__)
145 // gcc does not have proper locale facets for lyx::char_type if
146 // sizeof(wchar_t) == 2, so we have to implement them on our own.
149 // We get undefined references to these virtual methods. This looks like
150 // a bug in gcc. The implementation here does not do anything useful, since
151 // it is overriden in ascii_ctype_facet.
153 template<> ctype<lyx::char_type>::~ctype() {}
155 ctype<lyx::char_type>::do_is(ctype<lyx::char_type>::mask, lyx::char_type) const { return false; }
156 template<> lyx::char_type const *
157 ctype<lyx::char_type>::do_is(const lyx::char_type *, const lyx::char_type *, ctype<lyx::char_type>::mask *) const { return 0; }
158 template<> const lyx::char_type *
159 ctype<lyx::char_type>::do_scan_is(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
160 template<> const lyx::char_type *
161 ctype<lyx::char_type>::do_scan_not(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
162 template<> lyx::char_type ctype<lyx::char_type>::do_toupper(lyx::char_type) const { return 0; }
163 template<> const lyx::char_type * ctype<lyx::char_type>::do_toupper(lyx::char_type *, lyx::char_type const *) const { return 0; }
164 template<> lyx::char_type ctype<lyx::char_type>::do_tolower(lyx::char_type) const { return 0; }
165 template<> const lyx::char_type * ctype<lyx::char_type>::do_tolower(lyx::char_type *, lyx::char_type const *) const { return 0; }
166 template<> lyx::char_type ctype<lyx::char_type>::do_widen(char) const { return 0; }
167 template<> const char *
168 ctype<lyx::char_type>::do_widen(const char *, const char *, lyx::char_type *) const { return 0; }
170 ctype<lyx::char_type>::do_narrow(const lyx::char_type, char) const { return 0; }
171 template<> const lyx::char_type *
172 ctype<lyx::char_type>::do_narrow(const lyx::char_type *, const lyx::char_type *, char, char *) const { return 0; }
178 class ctype_failure : public std::bad_cast {
180 ctype_failure() throw() : std::bad_cast() {}
181 virtual ~ctype_failure() throw() {}
182 virtual const char* what() const throw()
184 return "The ctype<lyx::char_type> locale facet does only support ASCII characters on this platform.";
189 /// ctype facet for UCS4 characters. The implementation does only support pure
190 /// ASCII, since we do not need anything else for now.
191 /// The code is partly stolen from std::ctype<wchar_t> from gcc.
192 class ascii_ctype_facet : public std::ctype<lyx::char_type>
195 typedef lyx::char_type char_type;
196 typedef wctype_t wmask_type;
197 explicit ascii_ctype_facet(size_t refs = 0) : std::ctype<char_type>(refs)
199 M_initialize_ctype();
204 wint_t M_widen[1 + static_cast<unsigned char>(-1)];
206 wmask_type M_wmask[16];
207 wmask_type M_convert_to_wmask(const mask m) const
211 case space: ret = wctype("space"); break;
212 case print: ret = wctype("print"); break;
213 case cntrl: ret = wctype("cntrl"); break;
214 case upper: ret = wctype("upper"); break;
215 case lower: ret = wctype("lower"); break;
216 case alpha: ret = wctype("alpha"); break;
217 case digit: ret = wctype("digit"); break;
218 case punct: ret = wctype("punct"); break;
219 case xdigit: ret = wctype("xdigit"); break;
220 case alnum: ret = wctype("alnum"); break;
221 case graph: ret = wctype("graph"); break;
222 default: ret = wmask_type();
226 void M_initialize_ctype()
229 for (i = 0; i < 128; ++i) {
230 const int c = wctob(i);
234 M_narrow[i] = static_cast<char>(c);
240 for (size_t i = 0; i < sizeof(M_widen) / sizeof(wint_t); ++i)
241 M_widen[i] = btowc(i);
243 for (size_t i = 0; i <= 15; ++i) {
244 M_bit[i] = static_cast<mask>(1 << i);
245 M_wmask[i] = M_convert_to_wmask(M_bit[i]);
248 virtual ~ascii_ctype_facet() {}
249 char_type do_toupper(char_type c) const
252 throw ctype_failure();
253 return toupper(static_cast<int>(c));
255 char_type const * do_toupper(char_type * lo, char_type const * hi) const
259 throw ctype_failure();
260 *lo = toupper(static_cast<int>(*lo));
265 char_type do_tolower(char_type c) const
268 throw ctype_failure();
271 char_type const * do_tolower(char_type * lo, char_type const * hi) const
275 throw ctype_failure();
281 bool do_is(mask m, char_type c) const
284 throw ctype_failure();
285 // The code below works because c is in the ASCII range.
286 // We could not use iswctype() which is designed for a 2byte
287 // whar_t without encoding conversion otherwise.
289 // Generically, 15 (instead of 10) since we don't know the numerical
290 // encoding of the various categories in /usr/include/ctype.h.
291 const size_t bitmasksize = 15;
292 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
293 if (m & M_bit[bitcur] &&
294 iswctype(static_cast<int>(c), M_wmask[bitcur])) {
300 char_type const * do_is(char_type const * lo, char_type const * hi, mask * vec) const
302 for (;lo < hi; ++vec, ++lo) {
304 throw ctype_failure();
305 // The code below works because c is in the ASCII range.
306 // We could not use iswctype() which is designed for a 2byte
307 // whar_t without encoding conversion otherwise.
308 // Generically, 15 (instead of 10) since we don't know the numerical
309 // encoding of the various categories in /usr/include/ctype.h.
310 const size_t bitmasksize = 15;
312 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
313 if (iswctype(static_cast<int>(*lo), M_wmask[bitcur]))
319 char_type const * do_scan_is(mask m, char_type const * lo, char_type const * hi) const
321 while (lo < hi && !this->do_is(m, *lo))
325 char_type const * do_scan_not(mask m, char_type const * lo, char_type const * hi) const
327 while (lo < hi && this->do_is(m, *lo) != 0)
331 char_type do_widen(char c) const
333 if (static_cast<unsigned char>(c) < 0x80)
335 throw ctype_failure();
337 const char* do_widen(const char* lo, const char* hi, char_type* dest) const
340 if (static_cast<unsigned char>(*lo) >= 0x80)
341 throw ctype_failure();
348 char do_narrow(char_type wc, char) const
351 return static_cast<char>(wc);
352 throw ctype_failure();
354 const char_type * do_narrow(const char_type * lo, const char_type * hi, char, char * dest) const
358 *dest = static_cast<char>(*lo);
360 throw ctype_failure();
369 /// class to add our ascii_ctype_facet to the global locale
370 class locale_initializer {
375 std::locale const loc(global, new ascii_ctype_facet);
376 std::locale::global(loc);
383 /// make sure that our ascii_ctype_facet gets used
384 static locale_initializer initializer;