1 /*
2  * Copyright © 2011,2012,2014  Google, Inc.
3  *
4  *  This is part of HarfBuzz, a text shaping library.
5  *
6  * Permission is hereby granted, without written agreement and without
7  * license or royalty fees, to use, copy, modify, and distribute this
8  * software and its documentation for any purpose, provided that the
9  * above copyright notice and the following two paragraphs appear in
10  * all copies of this software.
11  *
12  * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
13  * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
14  * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
15  * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
16  * DAMAGE.
17  *
18  * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
19  * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
20  * FITNESS FOR A PARTICULAR PURPOSE.  THE SOFTWARE PROVIDED HEREUNDER IS
21  * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
22  * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
23  *
24  * Google Author(s): Behdad Esfahbod
25  */
26 
27 #ifndef HB_UTF_HH
28 #define HB_UTF_HH
29 
30 #include "hb.hh"
31 
32 #include "hb-open-type.hh"
33 
34 
35 struct hb_utf8_t
36 {
37   typedef uint8_t codepoint_t;
38 
39   static const codepoint_t *
nexthb_utf8_t40   next (const codepoint_t *text,
41 	const codepoint_t *end,
42 	hb_codepoint_t *unicode,
43 	hb_codepoint_t replacement)
44   {
45     /* Written to only accept well-formed sequences.
46      * Based on ideas from ICU's U8_NEXT.
47      * Generates one "replacement" for each ill-formed byte. */
48 
49     hb_codepoint_t c = *text++;
50 
51     if (c > 0x7Fu)
52     {
53       if (hb_in_range<hb_codepoint_t> (c, 0xC2u, 0xDFu)) /* Two-byte */
54       {
55 	unsigned int t1;
56 	if (likely (text < end &&
57 		    (t1 = text[0] - 0x80u) <= 0x3Fu))
58 	{
59 	  c = ((c&0x1Fu)<<6) | t1;
60 	  text++;
61 	}
62 	else
63 	  goto error;
64       }
65       else if (hb_in_range<hb_codepoint_t> (c, 0xE0u, 0xEFu)) /* Three-byte */
66       {
67 	unsigned int t1, t2;
68 	if (likely (1 < end - text &&
69 		    (t1 = text[0] - 0x80u) <= 0x3Fu &&
70 		    (t2 = text[1] - 0x80u) <= 0x3Fu))
71 	{
72 	  c = ((c&0xFu)<<12) | (t1<<6) | t2;
73 	  if (unlikely (c < 0x0800u || hb_in_range<hb_codepoint_t> (c, 0xD800u, 0xDFFFu)))
74 	    goto error;
75 	  text += 2;
76 	}
77 	else
78 	  goto error;
79       }
80       else if (hb_in_range<hb_codepoint_t> (c, 0xF0u, 0xF4u)) /* Four-byte */
81       {
82 	unsigned int t1, t2, t3;
83 	if (likely (2 < end - text &&
84 		    (t1 = text[0] - 0x80u) <= 0x3Fu &&
85 		    (t2 = text[1] - 0x80u) <= 0x3Fu &&
86 		    (t3 = text[2] - 0x80u) <= 0x3Fu))
87 	{
88 	  c = ((c&0x7u)<<18) | (t1<<12) | (t2<<6) | t3;
89 	  if (unlikely (!hb_in_range<hb_codepoint_t> (c, 0x10000u, 0x10FFFFu)))
90 	    goto error;
91 	  text += 3;
92 	}
93 	else
94 	  goto error;
95       }
96       else
97 	goto error;
98     }
99 
100     *unicode = c;
101     return text;
102 
103   error:
104     *unicode = replacement;
105     return text;
106   }
107 
108   static const codepoint_t *
prevhb_utf8_t109   prev (const codepoint_t *text,
110 	const codepoint_t *start,
111 	hb_codepoint_t *unicode,
112 	hb_codepoint_t replacement)
113   {
114     const codepoint_t *end = text--;
115     while (start < text && (*text & 0xc0) == 0x80 && end - text < 4)
116       text--;
117 
118     if (likely (next (text, end, unicode, replacement) == end))
119       return text;
120 
121     *unicode = replacement;
122     return end - 1;
123   }
124 
125   static unsigned int
strlenhb_utf8_t126   strlen (const codepoint_t *text)
127   { return ::strlen ((const char *) text); }
128 
129   static unsigned int
encode_lenhb_utf8_t130   encode_len (hb_codepoint_t unicode)
131   {
132     if (unicode <   0x0080u) return 1;
133     if (unicode <   0x0800u) return 2;
134     if (unicode <  0x10000u) return 3;
135     if (unicode < 0x110000u) return 4;
136     return 3;
137   }
138 
139   static codepoint_t *
encodehb_utf8_t140   encode (codepoint_t *text,
141 	  const codepoint_t *end,
142 	  hb_codepoint_t unicode)
143   {
144     if (unlikely (unicode >= 0xD800u && (unicode <= 0xDFFFu || unicode > 0x10FFFFu)))
145       unicode = 0xFFFDu;
146     if (unicode < 0x0080u)
147      *text++ = unicode;
148     else if (unicode < 0x0800u)
149     {
150       if (end - text >= 2)
151       {
152 	*text++ =  0xC0u + (0x1Fu & (unicode >>  6));
153 	*text++ =  0x80u + (0x3Fu & (unicode      ));
154       }
155     }
156     else if (unicode < 0x10000u)
157     {
158       if (end - text >= 3)
159       {
160 	*text++ =  0xE0u + (0x0Fu & (unicode >> 12));
161 	*text++ =  0x80u + (0x3Fu & (unicode >>  6));
162 	*text++ =  0x80u + (0x3Fu & (unicode      ));
163       }
164     }
165     else
166     {
167       if (end - text >= 4)
168       {
169 	*text++ =  0xF0u + (0x07u & (unicode >> 18));
170 	*text++ =  0x80u + (0x3Fu & (unicode >> 12));
171 	*text++ =  0x80u + (0x3Fu & (unicode >>  6));
172 	*text++ =  0x80u + (0x3Fu & (unicode      ));
173       }
174     }
175     return text;
176   }
177 };
178 
179 
180 template <typename TCodepoint>
181 struct hb_utf16_xe_t
182 {
183   static_assert (sizeof (TCodepoint) == 2, "");
184   typedef TCodepoint codepoint_t;
185 
186   static const codepoint_t *
nexthb_utf16_xe_t187   next (const codepoint_t *text,
188 	const codepoint_t *end,
189 	hb_codepoint_t *unicode,
190 	hb_codepoint_t replacement)
191   {
192     hb_codepoint_t c = *text++;
193 
194     if (likely (!hb_in_range<hb_codepoint_t> (c, 0xD800u, 0xDFFFu)))
195     {
196       *unicode = c;
197       return text;
198     }
199 
200     if (likely (c <= 0xDBFFu && text < end))
201     {
202       /* High-surrogate in c */
203       hb_codepoint_t l = *text;
204       if (likely (hb_in_range<hb_codepoint_t> (l, 0xDC00u, 0xDFFFu)))
205       {
206 	/* Low-surrogate in l */
207 	*unicode = (c << 10) + l - ((0xD800u << 10) - 0x10000u + 0xDC00u);
208 	 text++;
209 	 return text;
210       }
211     }
212 
213     /* Lonely / out-of-order surrogate. */
214     *unicode = replacement;
215     return text;
216   }
217 
218   static const codepoint_t *
prevhb_utf16_xe_t219   prev (const codepoint_t *text,
220 	const codepoint_t *start,
221 	hb_codepoint_t *unicode,
222 	hb_codepoint_t replacement)
223   {
224     hb_codepoint_t c = *--text;
225 
226     if (likely (!hb_in_range<hb_codepoint_t> (c, 0xD800u, 0xDFFFu)))
227     {
228       *unicode = c;
229       return text;
230     }
231 
232     if (likely (c >= 0xDC00u && start < text))
233     {
234       /* Low-surrogate in c */
235       hb_codepoint_t h = text[-1];
236       if (likely (hb_in_range<hb_codepoint_t> (h, 0xD800u, 0xDBFFu)))
237       {
238 	/* High-surrogate in h */
239 	*unicode = (h << 10) + c - ((0xD800u << 10) - 0x10000u + 0xDC00u);
240 	text--;
241 	return text;
242       }
243     }
244 
245     /* Lonely / out-of-order surrogate. */
246     *unicode = replacement;
247     return text;
248   }
249 
250 
251   static unsigned int
strlenhb_utf16_xe_t252   strlen (const codepoint_t *text)
253   {
254     unsigned int l = 0;
255     while (*text++) l++;
256     return l;
257   }
258 
259   static unsigned int
encode_lenhb_utf16_xe_t260   encode_len (hb_codepoint_t unicode)
261   {
262     return unicode < 0x10000 ? 1 : 2;
263   }
264 
265   static codepoint_t *
encodehb_utf16_xe_t266   encode (codepoint_t *text,
267 	  const codepoint_t *end,
268 	  hb_codepoint_t unicode)
269   {
270     if (unlikely (unicode >= 0xD800u && (unicode <= 0xDFFFu || unicode > 0x10FFFFu)))
271       unicode = 0xFFFDu;
272     if (unicode < 0x10000u)
273      *text++ = unicode;
274     else if (end - text >= 2)
275     {
276       unicode -= 0x10000u;
277       *text++ =  0xD800u + (unicode >> 10);
278       *text++ =  0xDC00u + (unicode & 0x03FFu);
279     }
280     return text;
281   }
282 };
283 
284 typedef hb_utf16_xe_t<uint16_t> hb_utf16_t;
285 typedef hb_utf16_xe_t<OT::HBUINT16> hb_utf16_be_t;
286 
287 
288 template <typename TCodepoint, bool validate=true>
289 struct hb_utf32_xe_t
290 {
291   static_assert (sizeof (TCodepoint) == 4, "");
292   typedef TCodepoint codepoint_t;
293 
294   static const TCodepoint *
nexthb_utf32_xe_t295   next (const TCodepoint *text,
296 	const TCodepoint *end HB_UNUSED,
297 	hb_codepoint_t *unicode,
298 	hb_codepoint_t replacement)
299   {
300     hb_codepoint_t c = *unicode = *text++;
301     if (validate && unlikely (c >= 0xD800u && (c <= 0xDFFFu || c > 0x10FFFFu)))
302       *unicode = replacement;
303     return text;
304   }
305 
306   static const TCodepoint *
prevhb_utf32_xe_t307   prev (const TCodepoint *text,
308 	const TCodepoint *start HB_UNUSED,
309 	hb_codepoint_t *unicode,
310 	hb_codepoint_t replacement)
311   {
312     hb_codepoint_t c = *unicode = *--text;
313     if (validate && unlikely (c >= 0xD800u && (c <= 0xDFFFu || c > 0x10FFFFu)))
314       *unicode = replacement;
315     return text;
316   }
317 
318   static unsigned int
strlenhb_utf32_xe_t319   strlen (const TCodepoint *text)
320   {
321     unsigned int l = 0;
322     while (*text++) l++;
323     return l;
324   }
325 
326   static unsigned int
encode_lenhb_utf32_xe_t327   encode_len (hb_codepoint_t unicode HB_UNUSED)
328   {
329     return 1;
330   }
331 
332   static codepoint_t *
encodehb_utf32_xe_t333   encode (codepoint_t *text,
334 	  const codepoint_t *end HB_UNUSED,
335 	  hb_codepoint_t unicode)
336   {
337     if (validate && unlikely (unicode >= 0xD800u && (unicode <= 0xDFFFu || unicode > 0x10FFFFu)))
338       unicode = 0xFFFDu;
339     *text++ = unicode;
340     return text;
341   }
342 };
343 
344 typedef hb_utf32_xe_t<uint32_t> hb_utf32_t;
345 typedef hb_utf32_xe_t<uint32_t, false> hb_utf32_novalidate_t;
346 
347 
348 struct hb_latin1_t
349 {
350   typedef uint8_t codepoint_t;
351 
352   static const codepoint_t *
nexthb_latin1_t353   next (const codepoint_t *text,
354 	const codepoint_t *end HB_UNUSED,
355 	hb_codepoint_t *unicode,
356 	hb_codepoint_t replacement HB_UNUSED)
357   {
358     *unicode = *text++;
359     return text;
360   }
361 
362   static const codepoint_t *
prevhb_latin1_t363   prev (const codepoint_t *text,
364 	const codepoint_t *start HB_UNUSED,
365 	hb_codepoint_t *unicode,
366 	hb_codepoint_t replacement HB_UNUSED)
367   {
368     *unicode = *--text;
369     return text;
370   }
371 
372   static unsigned int
strlenhb_latin1_t373   strlen (const codepoint_t *text)
374   {
375     unsigned int l = 0;
376     while (*text++) l++;
377     return l;
378   }
379 
380   static unsigned int
encode_lenhb_latin1_t381   encode_len (hb_codepoint_t unicode HB_UNUSED)
382   {
383     return 1;
384   }
385 
386   static codepoint_t *
encodehb_latin1_t387   encode (codepoint_t *text,
388 	  const codepoint_t *end HB_UNUSED,
389 	  hb_codepoint_t unicode)
390   {
391     if (unlikely (unicode >= 0x0100u))
392       unicode = '?';
393     *text++ = unicode;
394     return text;
395   }
396 };
397 
398 
399 struct hb_ascii_t
400 {
401   typedef uint8_t codepoint_t;
402 
403   static const codepoint_t *
nexthb_ascii_t404   next (const codepoint_t *text,
405 	const codepoint_t *end HB_UNUSED,
406 	hb_codepoint_t *unicode,
407 	hb_codepoint_t replacement HB_UNUSED)
408   {
409     *unicode = *text++;
410     if (*unicode >= 0x0080u)
411       *unicode = replacement;
412     return text;
413   }
414 
415   static const codepoint_t *
prevhb_ascii_t416   prev (const codepoint_t *text,
417 	const codepoint_t *start HB_UNUSED,
418 	hb_codepoint_t *unicode,
419 	hb_codepoint_t replacement)
420   {
421     *unicode = *--text;
422     if (*unicode >= 0x0080u)
423       *unicode = replacement;
424     return text;
425   }
426 
427   static unsigned int
strlenhb_ascii_t428   strlen (const codepoint_t *text)
429   {
430     unsigned int l = 0;
431     while (*text++) l++;
432     return l;
433   }
434 
435   static unsigned int
encode_lenhb_ascii_t436   encode_len (hb_codepoint_t unicode HB_UNUSED)
437   {
438     return 1;
439   }
440 
441   static codepoint_t *
encodehb_ascii_t442   encode (codepoint_t *text,
443 	  const codepoint_t *end HB_UNUSED,
444 	  hb_codepoint_t unicode)
445   {
446     if (unlikely (unicode >= 0x0080u))
447       unicode = '?';
448     *text++ = unicode;
449     return text;
450   }
451 };
452 
453 #endif /* HB_UTF_HH */
454