op public repos

Blob

Date:: Tue Nov 25 03:37:45 2003 UTC
Message:: add dict
Actions:: History | Blame | Raw File
1 #include <u.h>
2 #include <libc.h>
3 #include <bio.h>
4 #include "dict.h"
5 
6 /*
7  * Robert Électronique.
8  */
9 
10 enum
11 {
12 	CIT = MULTIE+1,	/* citation ptr followed by long int and ascii label */
13 	BROM,		/* bold roman */
14 	ITON,		/* start italic */
15 	ROM,		/* roman */
16 	SYM,		/* symbol font? */
17 	HEL,		/* helvetica */
18 	BHEL,		/* helvetica bold */
19 	SMALL,		/* smaller? */
20 	ITOFF,		/* end italic */
21 	SUP,		/* following character is superscript */
22 	SUB		/* following character is subscript */
23 };
24 
25 static Rune intab[256] = {
26 	/*0*/	/*1*/	/*2*/	/*3*/	/*4*/	/*5*/	/*6*/	/*7*/
27 /*00*/	NONE,	0x263a,	0x263b,	0x2665,	0x2666,	0x2663,	0x2660,	0x2022,
28 	0x25d8,	0x298,	L'\n',	0x2642,	0x2640,	0x266a,	0x266b,	0x203b,
29 /*10*/	0x21e8,	0x21e6,	0x2195,	0x203c,	0xb6,	0xa7,	0x2043,	0x21a8,
30 	0x2191,	0x2193,	0x2192,	0x2190,	0x2319,	0x2194,	0x25b4,	0x25be,
31 /*20*/	0x20,	0x21,	0x22,	0x23,	0x24,	0x25,	0x26,	L'\'',
32 	0x28,	0x29,	0x2a,	0x2b,	0x2c,	0x2d,	0x2e,	0x2f,
33 /*30*/	0x30,	0x31,	0x32,	0x33,	0x34,	0x35,	0x36,	0x37,
34 	0x38,	0x39,	0x3a,	0x3b,	0x3c,	0x3d,	0x3e,	0x3f,
35 /*40*/	0x40,	0x41,	0x42,	0x43,	0x44,	0x45,	0x46,	0x47,
36 	0x48,	0x49,	0x4a,	0x4b,	L'L',	0x4d,	0x4e,	0x4f,
37 /*50*/	0x50,	0x51,	0x52,	0x53,	0x54,	0x55,	0x56,	0x57,
38 	0x58,	0x59,	0x5a,	0x5b,	L'\\',	0x5d,	0x5e,	0x5f,
39 /*60*/	0x60,	0x61,	0x62,	0x63,	0x64,	0x65,	0x66,	0x67,
40 	0x68,	0x69,	0x6a,	0x6b,	0x6c,	0x6d,	0x6e,	0x6f,
41 /*70*/	0x70,	0x71,	0x72,	0x73,	0x74,	0x75,	0x76,	0x77,
42 	0x78,	0x79,	0x7a,	0x7b,	0x7c,	0x7d,	0x7e,	0x7f,
43 /*80*/	0xc7,	0xfc,	0xe9,	0xe2,	0xe4,	0xe0,	0xe5,	0xe7,
44 	0xea,	0xeb,	0xe8,	0xef,	0xee,	0xec,	0xc4,	0xc5,
45 /*90*/	0xc9,	0xe6,	0xc6,	0xf4,	0xf6,	0xf2,	0xfb,	0xf9,
46 	0xff,	0xd6,	0xdc,	0xa2,	0xa3,	0xa5,	0x20a7,	0x283,
47 /*a0*/	0xe1,	0xed,	0xf3,	0xfa,	0xf1,	0xd1,	0xaa,	0xba,
48 	0xbf,	0x2310,	0xac,	0xbd,	0xbc,	0xa1,	0xab,	0xbb,
49 /*b0*/	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,
50 	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,
51 /*c0*/	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,
52 	CIT,	BROM,	NONE,	ITON,	ROM,	SYM,	HEL,	BHEL,
53 /*d0*/	NONE,	SMALL,	ITOFF,	SUP,	SUB,	NONE,	NONE,	NONE,
54 	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,
55 /*e0*/	0x3b1,	0xdf,	0x3b3,	0x3c0,	0x3a3,	0x3c3,	0xb5,	0x3c4,
56 	0x3a6,	0x398,	0x3a9,	0x3b4,	0x221e,	0xd8,	0x3b5,	0x2229,
57 /*f0*/	0x2261,	0xb1,	0x2265,	0x2264,	0x2320,	0x2321,	0xf7,	0x2248,
58 	0xb0,	0x2219,	0xb7,	0x221a,	0x207f,	0xb2,	0x220e,	0xa0,
59 };
60 
61 static Rune suptab[] = {
62 	['0'] 0x2070,	['1'] 0x2071,	['2'] 0x2072,	['3'] 0x2073,
63 	['4'] 0x2074,	['5'] 0x2075,	['6'] 0x2076,	['7'] 0x2077,
64 	['8'] 0x2078,	['9'] 0x2079,	['+'] 0x207a,	['-'] 0x207b,
65 	['='] 0x207c,	['('] 0x207d,	[')'] 0x207e,	['a'] 0xaa,
66 	['n'] 0x207f,	['o'] 0xba
67 };
68 
69 static Rune subtab[] = {
70 	['0'] 0x2080,	['1'] 0x2081,	['2'] 0x2082,	['3'] 0x2083,
71 	['4'] 0x2084,	['5'] 0x2085,	['6'] 0x2086,	['7'] 0x2087,
72 	['8'] 0x2088,	['9'] 0x2089,	['+'] 0x208a,	['-'] 0x208b,
73 	['='] 0x208c,	['('] 0x208d,	[')'] 0x208e
74 };
75 
76 #define	GSHORT(p)	(((p)[0]<<8) | (p)[1])
77 #define	GLONG(p)	(((p)[0]<<24) | ((p)[1]<<16) | ((p)[2]<<8) | (p)[3])
78 
79 static char	cfile[] = "/lib/dict/robert/cits.rob";
80 static char	dfile[] = "/lib/dict/robert/defs.rob";
81 static char	efile[] = "/lib/dict/robert/etym.rob";
82 static char	kfile[] = "/lib/dict/robert/_phon";
83 
84 static Biobuf *	cb;
85 static Biobuf *	db;
86 static Biobuf *	eb;
87 
88 static Biobuf *	Bouvrir(char*);
89 static void	citation(int, int);
90 static void	robertprintentry(Entry*, Entry*, int);
91 
92 void
93 robertindexentry(Entry e, int cmd)
94 {
95 	uchar *p = (uchar *)e.start;
96 	long ea, el, da, dl, fa;
97 	Entry def, etym;
98 
99 	ea = GLONG(&p[0]);
100 	el = GSHORT(&p[4]);
101 	da = GLONG(&p[6]);
102 	dl = GSHORT(&p[10]);
103 	fa = GLONG(&p[12]);
104 	USED(fa);
105 
106 	if(db == 0)
107 		db = Bouvrir(dfile);
108 	def.start = malloc(dl+1);
109 	def.end = def.start + dl;
110 	def.doff = da;
111 	Bseek(db, da, 0);
112 	Bread(db, def.start, dl);
113 	*def.end = 0;
114 	if(cmd == 'h'){
115 		robertprintentry(&def, 0, cmd);
116 	}else{
117 		if(eb == 0)
118 			eb = Bouvrir(efile);
119 		etym.start = malloc(el+1);
120 		etym.end = etym.start + el;
121 		etym.doff = ea;
122 		Bseek(eb, ea, 0);
123 		Bread(eb, etym.start, el);
124 		*etym.end = 0;
125 		robertprintentry(&def, &etym, cmd);
126 		free(etym.start);
127 	}
128 	free(def.start);
129 }
130 
131 static void
132 robertprintentry(Entry *def, Entry *etym, int cmd)
133 {
134 	uchar *p, *pe;
135 	Rune r; int c, n;
136 	int baseline = 0;
137 	int lineno = 0;
138 	int cit = 0;
139 
140 	p = (uchar *)def->start;
141 	pe = (uchar *)def->end;
142 	while(p < pe){
143 		if(cmd == 'r'){
144 			outchar(*p++);
145 			continue;
146 		}
147 		c = *p++;
148 		switch(r = intab[c]){	/* assign = */
149 		case BROM:
150 		case ITON:
151 		case ROM:
152 		case SYM:
153 		case HEL:
154 		case BHEL:
155 		case SMALL:
156 		case ITOFF:
157 		case NONE:
158 			if(debug)
159 				outprint("\\%.2ux", c);
160 			baseline = 0;
161 			break;
162 
163 		case SUP:
164 			baseline = 1;
165 			break;
166 
167 		case SUB:
168 			baseline = -1;
169 			break;
170 
171 		case CIT:
172 			n = p[0] | (p[1]<<8) | (p[2]<<16) | (p[3]<<24);
173 			p += 4;
174 			if(debug)
175 				outprint("[%d]", n);
176 			while(*p == ' ' || ('0'<=*p && *p<='9') || *p == '.'){
177 				if(debug)
178 					outchar(*p);
179 				++p;
180 			}
181 			++cit;
182 			outnl(2);
183 			citation(n, cmd);
184 			baseline = 0;
185 			break;
186 
187 		case '\n':
188 			outnl(0);
189 			baseline = 0;
190 			++lineno;
191 			break;
192 
193 		default:
194 			if(baseline > 0 && r < nelem(suptab))
195 				r = suptab[r];
196 			else if(baseline < 0 && r < nelem(subtab))
197 				r = subtab[r];
198 			if(cit){
199 				outchar('\n');
200 				cit = 0;
201 			}
202 			outrune(r);
203 			baseline = 0;
204 			break;
205 		}
206 		if(r == '\n'){
207 			if(cmd == 'h')
208 				break;
209 			if(lineno == 1 && etym)
210 				robertprintentry(etym, 0, cmd);
211 		}
212 	}
213 	outnl(0);
214 }
215 
216 static void
217 citation(int addr, int cmd)
218 {
219 	Entry cit;
220 
221 	if(cb == 0)
222 		cb = Bouvrir(cfile);
223 	Bseek(cb, addr, 0);
224 	cit.start = Brdline(cb, 0xc8);
225 	cit.end = cit.start + Blinelen(cb) - 1;
226 	cit.doff = addr;
227 	*cit.end = 0;
228 	robertprintentry(&cit, 0, cmd);
229 }
230 
231 long
232 robertnextoff(long fromoff)
233 {
234 	return (fromoff & ~15) + 16;
235 }
236 
237 void
238 robertprintkey(void)
239 {
240 	Biobuf *db;
241 	char *l;
242 
243 	db = Bouvrir(kfile);
244 	while(l = Brdline(db, '\n'))	/* assign = */
245 		Bwrite(bout, l, Blinelen(db));
246 	Bterm(db);
247 }
248 
249 void
250 robertflexentry(Entry e, int cmd)
251 {
252 	uchar *p, *pe;
253 	Rune r; int c;
254 	int lineno = 1;
255 
256 	p = (uchar *)e.start;
257 	pe = (uchar *)e.end;
258 	while(p < pe){
259 		if(cmd == 'r'){
260 			Bputc(bout, *p++);
261 			continue;
262 		}
263 		c = *p++;
264 		r = intab[c];
265 		if(r == '$')
266 			r = '\n';
267 		if(r == '\n'){
268 			++lineno;
269 			if(cmd == 'h' && lineno > 2)
270 				break;
271 		}
272 		if(cmd == 'h' && lineno < 2)
273 			continue;
274 		if(r > MULTIE){
275 			if(debug)
276 				Bprint(bout, "\\%.2ux", c);
277 			continue;
278 		}
279 		if(r < Runeself)
280 			Bputc(bout, r);
281 		else
282 			Bputrune(bout, r);
283 	}
284 	outnl(0);
285 }
286 
287 long
288 robertnextflex(long fromoff)
289 {
290 	int c;
291 
292 	if(Bseek(bdict, fromoff, 0) < 0)
293 		return -1;
294 	while((c = Bgetc(bdict)) >= 0){
295 		if(c == '$')
296 			return Boffset(bdict);
297 	}
298 	return -1;
299 }
300 
301 static Biobuf *
302 Bouvrir(char *fichier)
303 {
304 	Biobuf *db;
305 
306 	db = Bopen(fichier, OREAD);
307 	if(db == 0){
308 		fprint(2, "%s: impossible d'ouvrir %s: %r\n", argv0, fichier);
309 		exits("ouvrir");
310 	}
311 	return db;
312 }