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 docstring const from_utf8(std::string const & utf8)
45 std::vector<lyx::char_type> const ucs4 =
46 utf8_to_ucs4(utf8.data(), utf8.size());
47 return docstring(ucs4.begin(), ucs4.end());
51 std::string const to_utf8(docstring const & ucs4)
53 std::vector<char> const utf8 =
54 ucs4_to_utf8(ucs4.data(), ucs4.size());
55 return std::string(utf8.begin(), utf8.end());
61 bool operator==(lyx::docstring const & l, char const * r)
63 int const len = l.length();
64 for (int i = 0; i < len; ++i) {
65 BOOST_ASSERT(static_cast<unsigned char>(r[i]) < 0x80);
68 if (l[i] != lyx::docstring::value_type(r[i]))
71 return r[len] == '\0';
75 lyx::docstring operator+(lyx::docstring const & l, char const * r)
78 for (char const * c = r; *c; ++c) {
79 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
86 lyx::docstring operator+(char const * l, lyx::docstring const & r)
89 for (char const * c = l; *c; ++c) {
90 BOOST_ASSERT(static_cast<unsigned char>(*c) < 0x80);
98 lyx::docstring operator+(lyx::docstring const & l, char r)
100 BOOST_ASSERT(static_cast<unsigned char>(r) < 0x80);
101 return l + lyx::docstring::value_type(r);
105 lyx::docstring operator+(char l, lyx::docstring const & r)
107 BOOST_ASSERT(static_cast<unsigned char>(l) < 0x80);
108 return lyx::docstring::value_type(l) + r;
112 #if (!defined(HAVE_WCHAR_T) || SIZEOF_WCHAR_T != 4) && defined(__GNUC__)
114 // gcc does not have proper locale facets for lyx::char_type if
115 // sizeof(wchar_t) == 2, so we have to implement them on our own.
118 // We get undefined references to these virtual methods. This looks like
119 // a bug in gcc. The implementation here does not do anything useful, since
120 // it is overriden in ascii_ctype_facet.
122 template<> ctype<lyx::char_type>::~ctype() {}
124 ctype<lyx::char_type>::do_is(ctype<lyx::char_type>::mask, lyx::char_type) const { return false; }
125 template<> lyx::char_type const *
126 ctype<lyx::char_type>::do_is(const lyx::char_type *, const lyx::char_type *, ctype<lyx::char_type>::mask *) const { return 0; }
127 template<> const lyx::char_type *
128 ctype<lyx::char_type>::do_scan_is(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
129 template<> const lyx::char_type *
130 ctype<lyx::char_type>::do_scan_not(ctype<lyx::char_type>::mask, const lyx::char_type *, const lyx::char_type *) const { return 0; }
131 template<> lyx::char_type ctype<lyx::char_type>::do_toupper(lyx::char_type) const { return 0; }
132 template<> const lyx::char_type * ctype<lyx::char_type>::do_toupper(lyx::char_type *, lyx::char_type const *) const { return 0; }
133 template<> lyx::char_type ctype<lyx::char_type>::do_tolower(lyx::char_type) const { return 0; }
134 template<> const lyx::char_type * ctype<lyx::char_type>::do_tolower(lyx::char_type *, lyx::char_type const *) const { return 0; }
135 template<> lyx::char_type ctype<lyx::char_type>::do_widen(char) const { return 0; }
136 template<> const char *
137 ctype<lyx::char_type>::do_widen(const char *, const char *, lyx::char_type *) const { return 0; }
139 ctype<lyx::char_type>::do_narrow(const lyx::char_type, char) const { return 0; }
140 template<> const lyx::char_type *
141 ctype<lyx::char_type>::do_narrow(const lyx::char_type *, const lyx::char_type *, char, char *) const { return 0; }
147 class ctype_failure : public std::bad_cast {
149 ctype_failure() throw() : std::bad_cast() {}
150 virtual ~ctype_failure() throw() {}
151 virtual const char* what() const throw()
153 return "The ctype<lyx::char_type> locale facet does only support ASCII characters on this platform.";
158 /// ctype facet for UCS4 characters. The implementation does only support pure
159 /// ASCII, since we do not need anything else for now.
160 /// The code is partly stolen from std::ctype<wchar_t> from gcc.
161 class ascii_ctype_facet : public std::ctype<lyx::char_type>
164 typedef lyx::char_type char_type;
165 typedef wctype_t wmask_type;
166 explicit ascii_ctype_facet(size_t refs = 0) : std::ctype<char_type>(refs)
168 M_initialize_ctype();
173 wint_t M_widen[1 + static_cast<unsigned char>(-1)];
175 wmask_type M_wmask[16];
176 wmask_type M_convert_to_wmask(const mask m) const
180 case space: ret = wctype("space"); break;
181 case print: ret = wctype("print"); break;
182 case cntrl: ret = wctype("cntrl"); break;
183 case upper: ret = wctype("upper"); break;
184 case lower: ret = wctype("lower"); break;
185 case alpha: ret = wctype("alpha"); break;
186 case digit: ret = wctype("digit"); break;
187 case punct: ret = wctype("punct"); break;
188 case xdigit: ret = wctype("xdigit"); break;
189 case alnum: ret = wctype("alnum"); break;
190 case graph: ret = wctype("graph"); break;
191 default: ret = wmask_type();
195 void M_initialize_ctype()
198 for (i = 0; i < 128; ++i) {
199 const int c = wctob(i);
203 M_narrow[i] = static_cast<char>(c);
209 for (size_t i = 0; i < sizeof(M_widen) / sizeof(wint_t); ++i)
210 M_widen[i] = btowc(i);
212 for (size_t i = 0; i <= 15; ++i) {
213 M_bit[i] = static_cast<mask>(1 << i);
214 M_wmask[i] = M_convert_to_wmask(M_bit[i]);
217 virtual ~ascii_ctype_facet() {}
218 char_type do_toupper(char_type c) const
221 throw ctype_failure();
222 return toupper(static_cast<int>(c));
224 char_type const * do_toupper(char_type * lo, char_type const * hi) const
228 throw ctype_failure();
229 *lo = toupper(static_cast<int>(*lo));
234 char_type do_tolower(char_type c) const
237 throw ctype_failure();
240 char_type const * do_tolower(char_type * lo, char_type const * hi) const
244 throw ctype_failure();
250 bool do_is(mask m, char_type c) const
253 throw ctype_failure();
254 // The code below works because c is in the ASCII range.
255 // We could not use iswctype() which is designed for a 2byte
256 // whar_t without encoding conversion otherwise.
258 // Generically, 15 (instead of 10) since we don't know the numerical
259 // encoding of the various categories in /usr/include/ctype.h.
260 const size_t bitmasksize = 15;
261 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
262 if (m & M_bit[bitcur] &&
263 iswctype(static_cast<int>(c), M_wmask[bitcur])) {
269 char_type const * do_is(char_type const * lo, char_type const * hi, mask * vec) const
271 for (;lo < hi; ++vec, ++lo) {
273 throw ctype_failure();
274 // The code below works because c is in the ASCII range.
275 // We could not use iswctype() which is designed for a 2byte
276 // whar_t without encoding conversion otherwise.
277 // Generically, 15 (instead of 10) since we don't know the numerical
278 // encoding of the various categories in /usr/include/ctype.h.
279 const size_t bitmasksize = 15;
281 for (size_t bitcur = 0; bitcur <= bitmasksize; ++bitcur)
282 if (iswctype(static_cast<int>(*lo), M_wmask[bitcur]))
288 char_type const * do_scan_is(mask m, char_type const * lo, char_type const * hi) const
290 while (lo < hi && !this->do_is(m, *lo))
294 char_type const * do_scan_not(mask m, char_type const * lo, char_type const * hi) const
296 while (lo < hi && this->do_is(m, *lo) != 0)
300 char_type do_widen(char c) const
302 if (static_cast<unsigned char>(c) < 0x80)
304 throw ctype_failure();
306 const char* do_widen(const char* lo, const char* hi, char_type* dest) const
309 if (static_cast<unsigned char>(*lo) >= 0x80)
310 throw ctype_failure();
317 char do_narrow(char_type wc, char) const
320 return static_cast<char>(wc);
321 throw ctype_failure();
323 const char_type * do_narrow(const char_type * lo, const char_type * hi, char, char * dest) const
327 *dest = static_cast<char>(*lo);
329 throw ctype_failure();
338 /// class to add our ascii_ctype_facet to the global locale
339 class locale_initializer {
344 std::locale const loc(global, new ascii_ctype_facet);
345 std::locale::global(loc);
352 /// make sure that our ascii_ctype_facet gets used
353 static locale_initializer initializer;