entity.c

Revision 428, 37.8 kB (checked in by walter, 2 years ago)
remove tabs, any trailing spaces
Property svn:eol-style set to `native`

Line
1	// Copyright (c) 1999-2009 by Digital Mars
2	// All Rights Reserved
3	// written by Walter Bright
4	// http://www.digitalmars.com
5	// License for redistribution is by either the Artistic License
6	// in artistic.txt, or the GNU General Public License in gnu.txt.
7	// See the included readme.txt for details.
8
9
10	#include <string.h>
11
12	/*********************************************
13	* Convert from named entity to its encoding.
14	* For reference:
15	* http://www.htmlhelp.com/reference/html40/entities/
16	* http://www.w3.org/TR/1999/REC-html401-19991224/sgml/entities.html
17	*/
18
19	struct NameId
20	{
21	const char *name;
22	unsigned short value;
23	};
24
25	#if IN_GCC
26	static NameId namesA[]={
27	"Aacgr", 0x0386,
28	"aacgr", 0x03AC,
29	"Aacute", 0x00C1,
30	"aacute", 0x00E1,
31	"Abreve", 0x0102,
32	"abreve", 0x0103,
33	"Acirc", 0x00C2,
34	"acirc", 0x00E2,
35	"acute", 0x00B4,
36	"Acy", 0x0410,
37	"acy", 0x0430,
38	"AElig", 0x00C6,
39	"aelig", 0x00E6,
40	"Agr", 0x0391,
41	"agr", 0x03B1,
42	"Agrave", 0x00C0,
43	"agrave", 0x00E0,
44	"aleph", 0x2135,
45	"alpha", 0x03B1,
46	"Amacr", 0x0100,
47	"amacr", 0x0101,
48	"amalg", 0x2210,
49	"amp", 0x0026,
50	"and", 0x2227,
51	"ang", 0x2220,
52	"ang90", 0x221F,
53	"angmsd", 0x2221,
54	"angsph", 0x2222,
55	"angst", 0x212B,
56	"Aogon", 0x0104,
57	"aogon", 0x0105,
58	"ap", 0x2248,
59	"ape", 0x224A,
60	"apos", 0x0027,
61	"Aring", 0x00C5,
62	"aring", 0x00E5,
63	"ast", 0x002A,
64	"asymp", 0x224D,
65	"Atilde", 0x00C3,
66	"atilde", 0x00E3,
67	"Auml", 0x00C4,
68	"auml", 0x00E4,
69	NULL, 0
70	};
71
72	static NameId namesB[]={
73	"barwed", 0x22BC,
74	"Barwed", 0x2306,
75	"bcong", 0x224C,
76	"Bcy", 0x0411,
77	"bcy", 0x0431,
78	"becaus", 0x2235,
79	"bepsi", 0x220D,
80	"bernou", 0x212C,
81	"beta", 0x03B2,
82	"beth", 0x2136,
83	"Bgr", 0x0392,
84	"bgr", 0x03B2,
85	"blank", 0x2423,
86	"blk12", 0x2592,
87	"blk14", 0x2591,
88	"blk34", 0x2593,
89	"block", 0x2588,
90	"bottom", 0x22A5,
91	"bowtie", 0x22C8,
92	"boxdl", 0x2510,
93	"boxDL", 0x2555,
94	"boxdL", 0x2556,
95	"boxDl", 0x2557,
96	"boxdr", 0x250C,
97	"boxDR", 0x2552,
98	"boxDr", 0x2553,
99	"boxdR", 0x2554,
100	"boxh", 0x2500,
101	"boxH", 0x2550,
102	"boxhd", 0x252C,
103	"boxhD", 0x2564,
104	"boxHD", 0x2565,
105	"boxHd", 0x2566,
106	"boxhu", 0x2534,
107	"boxhU", 0x2567,
108	"boxHU", 0x2568,
109	"boxHu", 0x2569,
110	"boxul", 0x2518,
111	"boxUL", 0x255B,
112	"boxUl", 0x255C,
113	"boxuL", 0x255D,
114	"boxur", 0x2514,
115	"boxUR", 0x2558,
116	"boxuR", 0x2559,
117	"boxUr", 0x255A,
118	"boxv", 0x2502,
119	"boxV", 0x2551,
120	"boxvh", 0x253C,
121	"boxvH", 0x256A,
122	"boxVH", 0x256B,
123	"boxVh", 0x256C,
124	"boxvl", 0x2524,
125	"boxvL", 0x2561,
126	"boxVL", 0x2562,
127	"boxVl", 0x2563,
128	"boxvr", 0x251C,
129	"boxvR", 0x255E,
130	"boxVR", 0x255F,
131	"boxVr", 0x2560,
132	"bprime", 0x2035,
133	"breve", 0x02D8,
134	"brvbar", 0x00A6,
135	"bsim", 0x223D,
136	"bsime", 0x22CD,
137	"bsol", 0x005C,
138	"bull", 0x2022,
139	"bump", 0x224E,
140	"bumpe", 0x224F,
141	NULL, 0
142	};
143
144	static NameId namesC[]={
145	"Cacute", 0x0106,
146	"cacute", 0x0107,
147	"cap", 0x2229,
148	"Cap", 0x22D2,
149	"caret", 0x2041,
150	"caron", 0x02C7,
151	"Ccaron", 0x010C,
152	"ccaron", 0x010D,
153	"Ccedil", 0x00C7,
154	"ccedil", 0x00E7,
155	"Ccirc", 0x0108,
156	"ccirc", 0x0109,
157	"Cdot", 0x010A,
158	"cdot", 0x010B,
159	"cedil", 0x00B8,
160	"cent", 0x00A2,
161	"CHcy", 0x0427,
162	"chcy", 0x0447,
163	"check", 0x2713,
164	"chi", 0x03C7,
165	"cir", 0x25CB,
166	"circ", 0x005E,
167	"cire", 0x2257,
168	"clubs", 0x2663,
169	"colon", 0x003A,
170	"colone", 0x2254,
171	"comma", 0x002C,
172	"commat", 0x0040,
173	"comp", 0x2201,
174	"compfn", 0x2218,
175	"cong", 0x2245,
176	"conint", 0x222E,
177	"coprod", 0x2210,
178	"copy", 0x00A9,
179	"copysr", 0x2117,
180	"cross", 0x2717,
181	"cuepr", 0x22DE,
182	"cuesc", 0x22DF,
183	"cularr", 0x21B6,
184	"cup", 0x222A,
185	"Cup", 0x22D3,
186	"cupre", 0x227C,
187	"curarr", 0x21B7,
188	"curren", 0x00A4,
189	"cuvee", 0x22CE,
190	"cuwed", 0x22CF,
191	NULL, 0
192	};
193
194	static NameId namesD[]={
195	"dagger", 0x2020,
196	"Dagger", 0x2021,
197	"daleth", 0x2138,
198	"darr", 0x2193,
199	"dArr", 0x21D3,
200	"darr2", 0x21CA,
201	"dash", 0x2010,
202	"dashv", 0x22A3,
203	"dblac", 0x02DD,
204	"Dcaron", 0x010E,
205	"dcaron", 0x010F,
206	"Dcy", 0x0414,
207	"dcy", 0x0434,
208	"deg", 0x00B0,
209	"Delta", 0x0394,
210	"delta", 0x03B4,
211	"Dgr", 0x0394,
212	"dgr", 0x03B4,
213	"dharl", 0x21C3,
214	"dharr", 0x21C2,
215	"diam", 0x22C4,
216	"diams", 0x2666,
217	"die", 0x00A8,
218	"divide", 0x00F7,
219	"divonx", 0x22C7,
220	"DJcy", 0x0402,
221	"djcy", 0x0452,
222	"dlarr", 0x2199,
223	"dlcorn", 0x231E,
224	"dlcrop", 0x230D,
225	"dollar", 0x0024,
226	"Dot", 0x00A8,
227	"dot", 0x02D9,
228	"DotDot", 0x20DC,
229	"drarr", 0x2198,
230	"drcorn", 0x231F,
231	"drcrop", 0x230C,
232	"DScy", 0x0405,
233	"dscy", 0x0455,
234	"Dstrok", 0x0110,
235	"dstrok", 0x0111,
236	"dtri", 0x25BF,
237	"dtrif", 0x25BE,
238	"DZcy", 0x040F,
239	"dzcy", 0x045F,
240	NULL, 0
241	};
242
243	static NameId namesE[]={
244	"Eacgr", 0x0388,
245	"eacgr", 0x03AD,
246	"Eacute", 0x00C9,
247	"eacute", 0x00E9,
248	"Ecaron", 0x011A,
249	"ecaron", 0x011B,
250	"ecir", 0x2256,
251	"Ecirc", 0x00CA,
252	"ecirc", 0x00EA,
253	"ecolon", 0x2255,
254	"Ecy", 0x042D,
255	"ecy", 0x044D,
256	"Edot", 0x0116,
257	"edot", 0x0117,
258	"eDot", 0x2251,
259	"EEacgr", 0x0389,
260	"eeacgr", 0x03AE,
261	"EEgr", 0x0397,
262	"eegr", 0x03B7,
263	"efDot", 0x2252,
264	"Egr", 0x0395,
265	"egr", 0x03B5,
266	"Egrave", 0x00C8,
267	"egrave", 0x00E8,
268	"egs", 0x22DD,
269	"ell", 0x2113,
270	"els", 0x22DC,
271	"Emacr", 0x0112,
272	"emacr", 0x0113,
273	"empty", 0x2205,
274	"emsp", 0x2003,
275	"emsp13", 0x2004,
276	"emsp14", 0x2005,
277	"ENG", 0x014A,
278	"eng", 0x014B,
279	"ensp", 0x2002,
280	"Eogon", 0x0118,
281	"eogon", 0x0119,
282	"epsi", 0x220A,
283	"epsis", 0x220A,
284	"epsiv", 0x03B5,
285	"equals", 0x003D,
286	"equiv", 0x2261,
287	"erDot", 0x2253,
288	"esdot", 0x2250,
289	"eta", 0x03B7,
290	"ETH", 0x00D0,
291	"eth", 0x00F0,
292	"Euml", 0x00CB,
293	"euml", 0x00EB,
294	"excl", 0x0021,
295	"exist", 0x2203,
296	NULL, 0
297	};
298
299	static NameId namesF[]={
300	"Fcy", 0x0424,
301	"fcy", 0x0444,
302	"female", 0x2640,
303	"ffilig", 0xFB03,
304	"fflig", 0xFB00,
305	"ffllig", 0xFB04,
306	"filig", 0xFB01,
307	"flat", 0x266D,
308	"fllig", 0xFB02,
309	"fnof", 0x0192,
310	"forall", 0x2200,
311	"fork", 0x22D4,
312	"frac12", 0x00BD,
313	"frac13", 0x2153,
314	"frac14", 0x00BC,
315	"frac15", 0x2155,
316	"frac16", 0x2159,
317	"frac18", 0x215B,
318	"frac23", 0x2154,
319	"frac25", 0x2156,
320	"frac34", 0x00BE,
321	"frac35", 0x2157,
322	"frac38", 0x215C,
323	"frac45", 0x2158,
324	"frac56", 0x215A,
325	"frac58", 0x215D,
326	"frac78", 0x215E,
327	"frown", 0x2322,
328	NULL, 0
329	};
330
331	static NameId namesG[]={
332	"gacute", 0x01F5,
333	"Gamma", 0x0393,
334	"gamma", 0x03B3,
335	"gammad", 0x03DC,
336	"gap", 0x2273,
337	"Gbreve", 0x011E,
338	"gbreve", 0x011F,
339	"Gcedil", 0x0122,
340	"Gcirc", 0x011C,
341	"gcirc", 0x011D,
342	"Gcy", 0x0413,
343	"gcy", 0x0433,
344	"Gdot", 0x0120,
345	"gdot", 0x0121,
346	"ge", 0x2265,
347	"gE", 0x2267,
348	"gel", 0x22DB,
349	"gEl", 0x22DB,
350	"ges", 0x2265,
351	"Gg", 0x22D9,
352	"Ggr", 0x0393,
353	"ggr", 0x03B3,
354	"gimel", 0x2137,
355	"GJcy", 0x0403,
356	"gjcy", 0x0453,
357	"gl", 0x2277,
358	"gnap", 0xE411,
359	"gne", 0x2269,
360	"gnE", 0x2269,
361	"gnsim", 0x22E7,
362	"grave", 0x0060,
363	"gsdot", 0x22D7,
364	"gsim", 0x2273,
365	"gt", 0x003E,
366	"Gt", 0x226B,
367	"gvnE", 0x2269,
368	NULL, 0
369	};
370
371	static NameId namesH[]={
372	"hairsp", 0x200A,
373	"half", 0x00BD,
374	"hamilt", 0x210B,
375	"HARDcy", 0x042A,
376	"hardcy", 0x044A,
377	"harr", 0x2194,
378	"hArr", 0x21D4,
379	"harrw", 0x21AD,
380	"Hcirc", 0x0124,
381	"hcirc", 0x0125,
382	"hearts", 0x2665,
383	"hellip", 0x2026,
384	"horbar", 0x2015,
385	"Hstrok", 0x0126,
386	"hstrok", 0x0127,
387	"hybull", 0x2043,
388	"hyphen", 0x002D,
389	NULL, 0
390	};
391
392	static NameId namesI[]={
393	"Iacgr", 0x038A,
394	"iacgr", 0x03AF,
395	"Iacute", 0x00CD,
396	"iacute", 0x00ED,
397	"Icirc", 0x00CE,
398	"icirc", 0x00EE,
399	"Icy", 0x0418,
400	"icy", 0x0438,
401	"idiagr", 0x0390,
402	"Idigr", 0x03AA,
403	"idigr", 0x03CA,
404	"Idot", 0x0130,
405	"IEcy", 0x0415,
406	"iecy", 0x0435,
407	"iexcl", 0x00A1,
408	"iff", 0x21D4,
409	"Igr", 0x0399,
410	"igr", 0x03B9,
411	"Igrave", 0x00CC,
412	"igrave", 0x00EC,
413	"IJlig", 0x0132,
414	"ijlig", 0x0133,
415	"Imacr", 0x012A,
416	"imacr", 0x012B,
417	"image", 0x2111,
418	"incare", 0x2105,
419	"infin", 0x221E,
420	"inodot", 0x0131,
421	"int", 0x222B,
422	"intcal", 0x22BA,
423	"IOcy", 0x0401,
424	"iocy", 0x0451,
425	"Iogon", 0x012E,
426	"iogon", 0x012F,
427	"iota", 0x03B9,
428	"iquest", 0x00BF,
429	"isin", 0x220A,
430	"Itilde", 0x0128,
431	"itilde", 0x0129,
432	"Iukcy", 0x0406,
433	"iukcy", 0x0456,
434	"Iuml", 0x00CF,
435	"iuml", 0x00EF,
436	NULL, 0
437	};
438
439	static NameId namesJ[]={
440	"Jcirc", 0x0134,
441	"jcirc", 0x0135,
442	"Jcy", 0x0419,
443	"jcy", 0x0439,
444	"Jsercy", 0x0408,
445	"jsercy", 0x0458,
446	"Jukcy", 0x0404,
447	"jukcy", 0x0454,
448	NULL, 0
449	};
450
451	static NameId namesK[]={
452	"kappa", 0x03BA,
453	"kappav", 0x03F0,
454	"Kcedil", 0x0136,
455	"kcedil", 0x0137,
456	"Kcy", 0x041A,
457	"kcy", 0x043A,
458	"Kgr", 0x039A,
459	"kgr", 0x03BA,
460	"kgreen", 0x0138,
461	"KHcy", 0x0425,
462	"khcy", 0x0445,
463	"KHgr", 0x03A7,
464	"khgr", 0x03C7,
465	"KJcy", 0x040C,
466	"kjcy", 0x045C,
467	NULL, 0
468	};
469
470	static NameId namesL[]={
471	"lAarr", 0x21DA,
472	"Lacute", 0x0139,
473	"lacute", 0x013A,
474	"lagran", 0x2112,
475	"Lambda", 0x039B,
476	"lambda", 0x03BB,
477	"lang", 0x3008,
478	"lap", 0x2272,
479	"laquo", 0x00AB,
480	"larr", 0x2190,
481	"Larr", 0x219E,
482	"lArr", 0x21D0,
483	"larr2", 0x21C7,
484	"larrhk", 0x21A9,
485	"larrlp", 0x21AB,
486	"larrtl", 0x21A2,
487	"Lcaron", 0x013D,
488	"lcaron", 0x013E,
489	"Lcedil", 0x013B,
490	"lcedil", 0x013C,
491	"lceil", 0x2308,
492	"lcub", 0x007B,
493	"Lcy", 0x041B,
494	"lcy", 0x043B,
495	"ldot", 0x22D6,
496	"ldquo", 0x201C,
497	"ldquor", 0x201E,
498	"le", 0x2264,
499	"lE", 0x2266,
500	"leg", 0x22DA,
501	"lEg", 0x22DA,
502	"les", 0x2264,
503	"lfloor", 0x230A,
504	"lg", 0x2276,
505	"Lgr", 0x039B,
506	"lgr", 0x03BB,
507	"lhard", 0x21BD,
508	"lharu", 0x21BC,
509	"lhblk", 0x2584,
510	"LJcy", 0x0409,
511	"ljcy", 0x0459,
512	"Ll", 0x22D8,
513	"Lmidot", 0x013F,
514	"lmidot", 0x0140,
515	"lnap", 0xE2A2,
516	"lne", 0x2268,
517	"lnE", 0x2268,
518	"lnsim", 0x22E6,
519	"lowast", 0x2217,
520	"lowbar", 0x005F,
521	"loz", 0x25CA,
522	"lozf", 0x2726,
523	"lpar", 0x0028,
524	"lrarr2", 0x21C6,
525	"lrhar2", 0x21CB,
526	"lsh", 0x21B0,
527	"lsim", 0x2272,
528	"lsqb", 0x005B,
529	"lsquo", 0x2018,
530	"lsquor", 0x201A,
531	"Lstrok", 0x0141,
532	"lstrok", 0x0142,
533	"lt", 0x003C,
534	"Lt", 0x226A,
535	"lthree", 0x22CB,
536	"ltimes", 0x22C9,
537	"ltri", 0x25C3,
538	"ltrie", 0x22B4,
539	"ltrif", 0x25C2,
540	"lvnE", 0x2268,
541	NULL, 0
542	};
543
544	static NameId namesM[]={
545	"macr", 0x00AF,
546	"male", 0x2642,
547	"malt", 0x2720,
548	"map", 0x21A6,
549	"marker", 0x25AE,
550	"Mcy", 0x041C,
551	"mcy", 0x043C,
552	"mdash", 0x2014,
553	"Mgr", 0x039C,
554	"mgr", 0x03BC,
555	"micro", 0x00B5,
556	"mid", 0x2223,
557	"middot", 0x00B7,
558	"minus", 0x2212,
559	"minusb", 0x229F,
560	"mldr", 0x2026,
561	"mnplus", 0x2213,
562	"models", 0x22A7,
563	"mu", 0x03BC,
564	"mumap", 0x22B8,
565	NULL, 0
566	};
567
568	static NameId namesN[]={
569	"nabla", 0x2207,
570	"Nacute", 0x0143,
571	"nacute", 0x0144,
572	"nap", 0x2249,
573	"napos", 0x0149,
574	"natur", 0x266E,
575	// "nbsp", 0x00A0,
576	"nbsp", 32, // make non-breaking space appear as space
577	"Ncaron", 0x0147,
578	"ncaron", 0x0148,
579	"Ncedil", 0x0145,
580	"ncedil", 0x0146,
581	"ncong", 0x2247,
582	"Ncy", 0x041D,
583	"ncy", 0x043D,
584	"ndash", 0x2013,
585	"ne", 0x2260,
586	"nearr", 0x2197,
587	"nequiv", 0x2262,
588	"nexist", 0x2204,
589	"nge", 0x2271,
590	"ngE", 0x2271,
591	"nges", 0x2271,
592	"Ngr", 0x039D,
593	"ngr", 0x03BD,
594	"ngt", 0x226F,
595	"nharr", 0x21AE,
596	"nhArr", 0x21CE,
597	"ni", 0x220D,
598	"NJcy", 0x040A,
599	"njcy", 0x045A,
600	"nlarr", 0x219A,
601	"nlArr", 0x21CD,
602	"nldr", 0x2025,
603	"nle", 0x2270,
604	"nlE", 0x2270,
605	"nles", 0x2270,
606	"nlt", 0x226E,
607	"nltri", 0x22EA,
608	"nltrie", 0x22EC,
609	"nmid", 0x2224,
610	"not", 0x00AC,
611	"notin", 0x2209,
612	"npar", 0x2226,
613	"npr", 0x2280,
614	"npre", 0x22E0,
615	"nrarr", 0x219B,
616	"nrArr", 0x21CF,
617	"nrtri", 0x22EB,
618	"nrtrie", 0x22ED,
619	"nsc", 0x2281,
620	"nsce", 0x22E1,
621	"nsim", 0x2241,
622	"nsime", 0x2244,
623	"nsmid", 0xE2AA,
624	"nspar", 0x2226,
625	"nsub", 0x2284,
626	"nsube", 0x2288,
627	"nsubE", 0x2288,
628	"nsup", 0x2285,
629	"nsupe", 0x2289,
630	"nsupE", 0x2289,
631	"Ntilde", 0x00D1,
632	"ntilde", 0x00F1,
633	"nu", 0x03BD,
634	"num", 0x0023,
635	"numero", 0x2116,
636	"numsp", 0x2007,
637	"nvdash", 0x22AC,
638	"nvDash", 0x22AD,
639	"nVdash", 0x22AE,
640	"nVDash", 0x22AF,
641	"nwarr", 0x2196,
642	NULL, 0
643	};
644
645	static NameId namesO[]={
646	"Oacgr", 0x038C,
647	"oacgr", 0x03CC,
648	"Oacute", 0x00D3,
649	"oacute", 0x00F3,
650	"oast", 0x229B,
651	"ocir", 0x229A,
652	"Ocirc", 0x00D4,
653	"ocirc", 0x00F4,
654	"Ocy", 0x041E,
655	"ocy", 0x043E,
656	"odash", 0x229D,
657	"Odblac", 0x0150,
658	"odblac", 0x0151,
659	"odot", 0x2299,
660	"OElig", 0x0152,
661	"oelig", 0x0153,
662	"ogon", 0x02DB,
663	"Ogr", 0x039F,
664	"ogr", 0x03BF,
665	"Ograve", 0x00D2,
666	"ograve", 0x00F2,
667	"OHacgr", 0x038F,
668	"ohacgr", 0x03CE,
669	"OHgr", 0x03A9,
670	"ohgr", 0x03C9,
671	"ohm", 0x2126,
672	"olarr", 0x21BA,
673	"Omacr", 0x014C,
674	"omacr", 0x014D,
675	"Omega", 0x03A9,
676	"omega", 0x03C9,
677	"ominus", 0x2296,
678	"oplus", 0x2295,
679	"or", 0x2228,
680	"orarr", 0x21BB,
681	"order", 0x2134,
682	"ordf", 0x00AA,
683	"ordm", 0x00BA,
684	"oS", 0x24C8,
685	"Oslash", 0x00D8,
686	"oslash", 0x00F8,
687	"osol", 0x2298,
688	"Otilde", 0x00D5,
689	"otilde", 0x00F5,
690	"otimes", 0x2297,
691	"Ouml", 0x00D6,
692	"ouml", 0x00F6,
693	NULL, 0
694	};
695
696	static NameId namesP[]={
697	"par", 0x2225,
698	"para", 0x00B6,
699	"part", 0x2202,
700	"Pcy", 0x041F,
701	"pcy", 0x043F,
702	"percnt", 0x0025,
703	"period", 0x002E,
704	"permil", 0x2030,
705	"perp", 0x22A5,
706	"Pgr", 0x03A0,
707	"pgr", 0x03C0,
708	"PHgr", 0x03A6,
709	"phgr", 0x03C6,
710	"Phi", 0x03A6,
711	"phis", 0x03C6,
712	"phiv", 0x03D5,
713	"phmmat", 0x2133,
714	"phone", 0x260E,
715	"Pi", 0x03A0,
716	"pi", 0x03C0,
717	"piv", 0x03D6,
718	"planck", 0x210F,
719	"plus", 0x002B,
720	"plusb", 0x229E,
721	"plusdo", 0x2214,
722	"plusmn", 0x00B1,
723	"pound", 0x00A3,
724	"pr", 0x227A,
725	"prap", 0x227E,
726	"pre", 0x227C,
727	"prime", 0x2032,
728	"Prime", 0x2033,
729	"prnap", 0x22E8,
730	"prnE", 0xE2B3,
731	"prnsim", 0x22E8,
732	"prod", 0x220F,
733	"prop", 0x221D,
734	"prsim", 0x227E,
735	"PSgr", 0x03A8,
736	"psgr", 0x03C8,
737	"Psi", 0x03A8,
738	"psi", 0x03C8,
739	"puncsp", 0x2008,
740	NULL, 0
741	};
742
743	static NameId namesQ[]={
744	"quest", 0x003F,
745	"quot", 0x0022,
746	NULL, 0
747	};
748
749	static NameId namesR[]={
750	"rAarr", 0x21DB,
751	"Racute", 0x0154,
752	"racute", 0x0155,
753	"radic", 0x221A,
754	"rang", 0x3009,
755	"raquo", 0x00BB,
756	"rarr", 0x2192,
757	"Rarr", 0x21A0,
758	"rArr", 0x21D2,
759	"rarr2", 0x21C9,
760	"rarrhk", 0x21AA,
761	"rarrlp", 0x21AC,
762	"rarrtl", 0x21A3,
763	"rarrw", 0x219D,
764	"Rcaron", 0x0158,
765	"rcaron", 0x0159,
766	"Rcedil", 0x0156,
767	"rcedil", 0x0157,
768	"rceil", 0x2309,
769	"rcub", 0x007D,
770	"Rcy", 0x0420,
771	"rcy", 0x0440,
772	"rdquo", 0x201D,
773	"rdquor", 0x201C,
774	"real", 0x211C,
775	"rect", 0x25AD,
776	"reg", 0x00AE,
777	"rfloor", 0x230B,
778	"Rgr", 0x03A1,
779	"rgr", 0x03C1,
780	"rhard", 0x21C1,
781	"rharu", 0x21C0,
782	"rho", 0x03C1,
783	"rhov", 0x03F1,
784	"ring", 0x02DA,
785	"rlarr2", 0x21C4,
786	"rlhar2", 0x21CC,
787	"rpar", 0x0029,
788	"rpargt", 0xE291,
789	"rsh", 0x21B1,
790	"rsqb", 0x005D,
791	"rsquo", 0x2019,
792	"rsquor", 0x2018,
793	"rthree", 0x22CC,
794	"rtimes", 0x22CA,
795	"rtri", 0x25B9,
796	"rtrie", 0x22B5,
797	"rtrif", 0x25B8,
798	"rx", 0x211E,
799	NULL, 0
800	};
801
802	static NameId namesS[]={
803	"Sacute", 0x015A,
804	"sacute", 0x015B,
805	"samalg", 0x2210,
806	"sbsol", 0xFE68,
807	"sc", 0x227B,
808	"scap", 0x227F,
809	"Scaron", 0x0160,
810	"scaron", 0x0161,
811	"sccue", 0x227D,
812	"sce", 0x227D,
813	"Scedil", 0x015E,
814	"scedil", 0x015F,
815	"Scirc", 0x015C,
816	"scirc", 0x015D,
817	"scnap", 0x22E9,
818	"scnE", 0xE2B5,
819	"scnsim", 0x22E9,
820	"scsim", 0x227F,
821	"Scy", 0x0421,
822	"scy", 0x0441,
823	"sdot", 0x22C5,
824	"sdotb", 0x22A1,
825	"sect", 0x00A7,
826	"semi", 0x003B,
827	"setmn", 0x2216,
828	"sext", 0x2736,
829	"sfgr", 0x03C2,
830	"sfrown", 0x2322,
831	"Sgr", 0x03A3,
832	"sgr", 0x03C3,
833	"sharp", 0x266F,
834	"SHCHcy", 0x0429,
835	"shchcy", 0x0449,
836	"SHcy", 0x0428,
837	"shcy", 0x0448,
838	"shy", 0x00AD,
839	"Sigma", 0x03A3,
840	"sigma", 0x03C3,
841	"sigmav", 0x03C2,
842	"sim", 0x223C,
843	"sime", 0x2243,
844	"smid", 0xE301,
845	"smile", 0x2323,
846	"SOFTcy", 0x042C,
847	"softcy", 0x044C,
848	"sol", 0x002F,
849	"spades", 0x2660,
850	"spar", 0x2225,
851	"sqcap", 0x2293,
852	"sqcup", 0x2294,
853	"sqsub", 0x228F,
854	"sqsube", 0x2291,
855	"sqsup", 0x2290,
856	"sqsupe", 0x2292,
857	"squ", 0x25A1,
858	"square", 0x25A1,
859	"squf", 0x25AA,
860	"ssetmn", 0x2216,
861	"ssmile", 0x2323,
862	"sstarf", 0x22C6,
863	"star", 0x22C6,
864	"starf", 0x2605,
865	"sub", 0x2282,
866	"Sub", 0x22D0,
867	"sube", 0x2286,
868	"subE", 0x2286,
869	"subne", 0x228A,
870	"subnE", 0x228A,
871	"sum", 0x2211,
872	"sung", 0x2669,
873	"sup", 0x2283,
874	"Sup", 0x22D1,
875	"sup1", 0x00B9,
876	"sup2", 0x00B2,
877	"sup3", 0x00B3,
878	"supe", 0x2287,
879	"supE", 0x2287,
880	"supne", 0x228B,
881	"supnE", 0x228B,
882	"szlig", 0x00DF,
883	NULL, 0
884	};
885
886	static NameId namesT[]={
887	"target", 0x2316,
888	"tau", 0x03C4,
889	"Tcaron", 0x0164,
890	"tcaron", 0x0165,
891	"Tcedil", 0x0162,
892	"tcedil", 0x0163,
893	"Tcy", 0x0422,
894	"tcy", 0x0442,
895	"tdot", 0x20DB,
896	"telrec", 0x2315,
897	"Tgr", 0x03A4,
898	"tgr", 0x03C4,
899	"there4", 0x2234,
900	"Theta", 0x0398,
901	"thetas", 0x03B8,
902	"thetav", 0x03D1,
903	"THgr", 0x0398,
904	"thgr", 0x03B8,
905	"thinsp", 0x2009,
906	"thkap", 0x2248,
907	"thksim", 0x223C,
908	"THORN", 0x00DE,
909	"thorn", 0x00FE,
910	"tilde", 0x02DC,
911	"times", 0x00D7,
912	"timesb", 0x22A0,
913	"top", 0x22A4,
914	"tprime", 0x2034,
915	"trade", 0x2122,
916	"trie", 0x225C,
917	"TScy", 0x0426,
918	"tscy", 0x0446,
919	"TSHcy", 0x040B,
920	"tshcy", 0x045B,
921	"Tstrok", 0x0166,
922	"tstrok", 0x0167,
923	"twixt", 0x226C,
924	NULL, 0
925	};
926
927	static NameId namesU[]={
928	"Uacgr", 0x038E,
929	"uacgr", 0x03CD,
930	"Uacute", 0x00DA,
931	"uacute", 0x00FA,
932	"uarr", 0x2191,
933	"uArr", 0x21D1,
934	"uarr2", 0x21C8,
935	"Ubrcy", 0x040E,
936	"ubrcy", 0x045E,
937	"Ubreve", 0x016C,
938	"ubreve", 0x016D,
939	"Ucirc", 0x00DB,
940	"ucirc", 0x00FB,
941	"Ucy", 0x0423,
942	"ucy", 0x0443,
943	"Udblac", 0x0170,
944	"udblac", 0x0171,
945	"udiagr", 0x03B0,
946	"Udigr", 0x03AB,
947	"udigr", 0x03CB,
948	"Ugr", 0x03A5,
949	"ugr", 0x03C5,
950	"Ugrave", 0x00D9,
951	"ugrave", 0x00F9,
952	"uharl", 0x21BF,
953	"uharr", 0x21BE,
954	"uhblk", 0x2580,
955	"ulcorn", 0x231C,
956	"ulcrop", 0x230F,
957	"Umacr", 0x016A,
958	"umacr", 0x016B,
959	"uml", 0x00A8,
960	"Uogon", 0x0172,
961	"uogon", 0x0173,
962	"uplus", 0x228E,
963	"upsi", 0x03C5,
964	"Upsi", 0x03D2,
965	"urcorn", 0x231D,
966	"urcrop", 0x230E,
967	"Uring", 0x016E,
968	"uring", 0x016F,
969	"Utilde", 0x0168,
970	"utilde", 0x0169,
971	"utri", 0x25B5,
972	"utrif", 0x25B4,
973	"Uuml", 0x00DC,
974	"uuml", 0x00FC,
975	NULL, 0
976	};
977
978	static NameId namesV[]={
979	"varr", 0x2195,
980	"vArr", 0x21D5,
981	"Vcy", 0x0412,
982	"vcy", 0x0432,
983	"vdash", 0x22A2,
984	"vDash", 0x22A8,
985	"Vdash", 0x22A9,
986	"veebar", 0x22BB,
987	"vellip", 0x22EE,
988	"verbar", 0x007C,
989	"Verbar", 0x2016,
990	"vltri", 0x22B2,
991	"vprime", 0x2032,
992	"vprop", 0x221D,
993	"vrtri", 0x22B3,
994	"vsubne", 0x228A,
995	"vsubnE", 0xE2B8,
996	"vsupne", 0x228B,
997	"vsupnE", 0x228B,
998	"Vvdash", 0x22AA,
999	NULL, 0
1000	};
1001
1002	static NameId namesW[]={
1003	"Wcirc", 0x0174,
1004	"wcirc", 0x0175,
1005	"wedgeq", 0x2259,
1006	"weierp", 0x2118,
1007	"wreath", 0x2240,
1008	NULL, 0
1009	};
1010
1011	static NameId namesX[]={
1012	"xcirc", 0x25CB,
1013	"xdtri", 0x25BD,
1014	"Xgr", 0x039E,
1015	"xgr", 0x03BE,
1016	"xharr", 0x2194,
1017	"xhArr", 0x2194,
1018	"Xi", 0x039E,
1019	"xi", 0x03BE,
1020	"xlArr", 0x21D0,
1021	"xrArr", 0x21D2,
1022	"xutri", 0x25B3,
1023	NULL, 0
1024	};
1025
1026	static NameId namesY[]={
1027	"Yacute", 0x00DD,
1028	"yacute", 0x00FD,
1029	"YAcy", 0x042F,
1030	"yacy", 0x044F,
1031	"Ycirc", 0x0176,
1032	"ycirc", 0x0177,
1033	"Ycy", 0x042B,
1034	"ycy", 0x044B,
1035	"yen", 0x00A5,
1036	"YIcy", 0x0407,
1037	"yicy", 0x0457,
1038	"YUcy", 0x042E,
1039	"yucy", 0x044E,
1040	"yuml", 0x00FF,
1041	"Yuml", 0x0178,
1042	NULL, 0
1043	};
1044
1045	static NameId namesZ[]={
1046	"Zacute", 0x0179,
1047	"zacute", 0x017A,
1048	"Zcaron", 0x017D,
1049	"zcaron", 0x017E,
1050	"Zcy", 0x0417,
1051	"zcy", 0x0437,
1052	"Zdot", 0x017B,
1053	"zdot", 0x017C,
1054	"zeta", 0x03B6,
1055	"Zgr", 0x0396,
1056	"zgr", 0x03B6,
1057	"ZHcy", 0x0416,
1058	"zhcy", 0x0436,
1059	NULL, 0
1060	};
1061
1062	// @todo@ order namesTable and names? by frequency
1063	static NameId* namesTable[] = {
1064	namesA, namesB, namesC, namesD, namesE, namesF, namesG, namesH, namesI,
1065	namesJ, namesK, namesL, namesM, namesN, namesO, namesP, namesQ, namesR,
1066	namesS, namesT, namesU, namesV, namesW, namesX, namesY, namesZ, NULL
1067	};
1068
1069	int HtmlNamedEntity(unsigned char *p, int length)
1070	{
1071	int tableIndex = tolower(*p) - 'a';
1072	if (tableIndex >= 0 && tableIndex < 26) {
1073	NameId* names = namesTable[tableIndex];
1074	int i;
1075
1076	for (i = 0; names[i].name; i++){
1077	if (strncmp(names[i].name, (char *)p, length) == 0){
1078	return names[i].value;
1079	}
1080	}
1081	}
1082	error("unrecognized character entity \"%.*s\"", length, p);
1083	return -1;
1084	}
1085
1086	#else //TODO: Merge Walter's list with Thomas'
1087
1088	static NameId names[] =
1089	{
1090	// Entities
1091	"quot", 34,
1092	"amp", 38,
1093	"lt", 60,
1094	"gt", 62,
1095
1096	"OElig", 338,
1097	"oelig", 339,
1098	"Scaron", 352,
1099	"scaron", 353,
1100	"Yuml", 376,
1101	"circ", 710,
1102	"tilde", 732,
1103	"ensp", 8194,
1104	"emsp", 8195,
1105	"thinsp", 8201,
1106	"zwnj", 8204,
1107	"zwj", 8205,
1108	"lrm", 8206,
1109	"rlm", 8207,
1110	"ndash", 8211,
1111	"mdash", 8212,
1112	"lsquo", 8216,
1113	"rsquo", 8217,
1114	"sbquo", 8218,
1115	"ldquo", 8220,
1116	"rdquo", 8221,
1117	"bdquo", 8222,
1118	"dagger", 8224,
1119	"Dagger", 8225,
1120	"permil", 8240,
1121	"lsaquo", 8249,
1122	"rsaquo", 8250,
1123	"euro", 8364,
1124
1125	// Latin-1 (ISO-8859-1) Entities
1126	"nbsp", 160,
1127	"iexcl", 161,
1128	"cent", 162,
1129	"pound", 163,
1130	"curren", 164,
1131	"yen", 165,
1132	"brvbar", 166,
1133	"sect", 167,
1134	"uml", 168,
1135	"copy", 169,
1136	"ordf", 170,
1137	"laquo", 171,
1138	"not", 172,
1139	"shy", 173,
1140	"reg", 174,
1141	"macr", 175,
1142	"deg", 176,
1143	"plusmn", 177,
1144	"sup2", 178,
1145	"sup3", 179,
1146	"acute", 180,
1147	"micro", 181,
1148	"para", 182,
1149	"middot", 183,
1150	"cedil", 184,
1151	"sup1", 185,
1152	"ordm", 186,
1153	"raquo", 187,
1154	"frac14", 188,
1155	"frac12", 189,
1156	"frac34", 190,
1157	"iquest", 191,
1158	"Agrave", 192,
1159	"Aacute", 193,
1160	"Acirc", 194,
1161	"Atilde", 195,
1162	"Auml", 196,
1163	"Aring", 197,
1164	"AElig", 198,
1165	"Ccedil", 199,
1166	"Egrave", 200,
1167	"Eacute", 201,
1168	"Ecirc", 202,
1169	"Euml", 203,
1170	"Igrave", 204,
1171	"Iacute", 205,
1172	"Icirc", 206,
1173	"Iuml", 207,
1174	"ETH", 208,
1175	"Ntilde", 209,
1176	"Ograve", 210,
1177	"Oacute", 211,
1178	"Ocirc", 212,
1179	"Otilde", 213,
1180	"Ouml", 214,
1181	"times", 215,
1182	"Oslash", 216,
1183	"Ugrave", 217,
1184	"Uacute", 218,
1185	"Ucirc", 219,
1186	"Uuml", 220,
1187	"Yacute", 221,
1188	"THORN", 222,
1189	"szlig", 223,
1190	"agrave", 224,
1191	"aacute", 225,
1192	"acirc", 226,
1193	"atilde", 227,
1194	"auml", 228,
1195	"aring", 229,
1196	"aelig", 230,
1197	"ccedil", 231,
1198	"egrave", 232,
1199	"eacute", 233,
1200	"ecirc", 234,
1201	"euml", 235,
1202	"igrave", 236,
1203	"iacute", 237,
1204	"icirc", 238,
1205	"iuml", 239,
1206	"eth", 240,
1207	"ntilde", 241,
1208	"ograve", 242,
1209	"oacute", 243,
1210	"ocirc", 244,
1211	"otilde", 245,
1212	"ouml", 246,
1213	"divide", 247,
1214	"oslash", 248,
1215	"ugrave", 249,
1216	"uacute", 250,
1217	"ucirc", 251,
1218	"uuml", 252,
1219	"yacute", 253,
1220	"thorn", 254,
1221	"yuml", 255,
1222
1223	// Symbols and Greek letter entities
1224	"fnof", 402,
1225	"Alpha", 913,
1226	"Beta", 914,
1227	"Gamma", 915,
1228	"Delta", 916,
1229	"Epsilon", 917,
1230	"Zeta", 918,
1231	"Eta", 919,
1232	"Theta", 920,
1233	"Iota", 921,
1234	"Kappa", 922,
1235	"Lambda", 923,
1236	"Mu", 924,
1237	"Nu", 925,
1238	"Xi", 926,
1239	"Omicron", 927,
1240	"Pi", 928,
1241	"Rho", 929,
1242	"Sigma", 931,
1243	"Tau", 932,
1244	"Upsilon", 933,
1245	"Phi", 934,
1246	"Chi", 935,
1247	"Psi", 936,
1248	"Omega", 937,
1249	"alpha", 945,
1250	"beta", 946,
1251	"gamma", 947,
1252	"delta", 948,
1253	"epsilon", 949,
1254	"zeta", 950,
1255	"eta", 951,
1256	"theta", 952,
1257	"iota", 953,
1258	"kappa", 954,
1259	"lambda", 955,
1260	"mu", 956,
1261	"nu", 957,
1262	"xi", 958,
1263	"omicron", 959,
1264	"pi", 960,
1265	"rho", 961,
1266	"sigmaf", 962,
1267	"sigma", 963,
1268	"tau", 964,
1269	"upsilon", 965,
1270	"phi", 966,
1271	"chi", 967,
1272	"psi", 968,
1273	"omega", 969,
1274	"thetasym", 977,
1275	"upsih", 978,
1276	"piv", 982,
1277	"bull", 8226,
1278	"hellip", 8230,
1279	"prime", 8242,
1280	"Prime", 8243,
1281	"oline", 8254,
1282	"frasl", 8260,
1283	"weierp", 8472,
1284	"image", 8465,
1285	"real", 8476,
1286	"trade", 8482,
1287	"alefsym", 8501,
1288	"larr", 8592,
1289	"uarr", 8593,
1290	"rarr", 8594,
1291	"darr", 8595,
1292	"harr", 8596,
1293	"crarr", 8629,
1294	"lArr", 8656,
1295	"uArr", 8657,
1296	"rArr", 8658,
1297	"dArr", 8659,
1298	"hArr", 8660,
1299	"forall", 8704,
1300	"part", 8706,
1301	"exist", 8707,
1302	"empty", 8709,
1303	"nabla", 8711,
1304	"isin", 8712,
1305	"notin", 8713,
1306	"ni", 8715,
1307	"prod", 8719,
1308	"sum", 8721,
1309	"minus", 8722,
1310	"lowast", 8727,
1311	"radic", 8730,
1312	"prop", 8733,
1313	"infin", 8734,
1314	"ang", 8736,
1315	"and", 8743,
1316	"or", 8744,
1317	"cap", 8745,
1318	"cup", 8746,
1319	"int", 8747,
1320	"there4", 8756,
1321	"sim", 8764,
1322	"cong", 8773,
1323	"asymp", 8776,
1324	"ne", 8800,
1325	"equiv", 8801,
1326	"le", 8804,
1327	"ge", 8805,
1328	"sub", 8834,
1329	"sup", 8835,
1330	"nsub", 8836,
1331	"sube", 8838,
1332	"supe", 8839,
1333	"oplus", 8853,
1334	"otimes", 8855,
1335	"perp", 8869,
1336	"sdot", 8901,
1337	"lceil", 8968,
1338	"rceil", 8969,
1339	"lfloor", 8970,
1340	"rfloor", 8971,
1341	"lang", 9001,
1342	"rang", 9002,
1343	"loz", 9674,
1344	"spades", 9824,
1345	"clubs", 9827,
1346	"hearts", 9829,
1347	"diams", 9830,
1348	};
1349
1350	int HtmlNamedEntity(unsigned char *p, int length)
1351	{
1352	int i;
1353
1354	// BUG: this is a dumb, slow linear search
1355	for (i = 0; i < sizeof(names) / sizeof(names[0]); i++)
1356	{
1357	// Entries are case sensitive
1358	if (memcmp(names[i].name, (char *)p, length) == 0 &&
1359	!names[i].name[length])
1360	return names[i].value;
1361	}
1362	return -1;
1363	}
1364
1365	#endif

Note: See TracBrowser for help on using the browser.

dmd

root/trunk/src/entity.c

Download in other formats: