1 #include <u.h>
2 #include <libc.h>
3 #include <bio.h>
4 #include "dict.h"
5 #include "kuten.h"
6 
7 /*
8  * Routines for handling dictionaries in the "Languages of the World"
9  * format.  worldnextoff *must* be called with <address of valid entry>+1.
10  */
11 
12 #define	GSHORT(p)	(((p)[0]<<8)|(p)[1])
13 
14 static void	putchar(int, int*);
15 
16 #define	NONE	0xffff
17 
18 /* adapted from jhelling@cs.ruu.nl (Jeroen Hellingman) */
19 
20 static Rune chartab[] = {
21 
22 /*00*/	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,
23 	NONE,	NONE,	L'\n',	0xe6,	0xf8,	0xe5,	0xe4,	0xf6,
24 /*10*/	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,	NONE,
25 	NONE,	NONE,	NONE,	0xc6,	0xd8,	0xc5,	0xc4,	0xd6,
26 
27 /*20*/	0x20,	0x21,	0x22,	0x23,	0x24,	0x25,	0x26,	'\'',
28 	0x28,	0x29,	0x2a,	0x2b,	0x2c,	0x2d,	0x2e,	0x2f,
29 /*30*/  0x30,	0x31,	0x32,	0x33,	0x34,	0x35,	0x36,	0x37,
30 	0x38,	0x39,	0x3a,	0x3b,	0x3c,	0x3d,	0x3e,	0x3f,
31 /*40*/  0x40,	0x41,	0x42,	0x43,	0x44,	0x45,	0x46,	0x47,
32 	0x48,	0x49,	0x4a,	0x4b,	L'L',	0x4d,	0x4e,	0x4f,
33 /*50*/	0x50,	0x51,	0x52,	0x53,	0x54,	0x55,	0x56,	0x57,
34 	0x58,	0x59,	0x5a,	0x5b,	L'\\',	0x5d,	0x5e,	0x5f,
35 /*60*/	0x60,	0x61,	0x62,	0x63,	0x64,	0x65,	0x66,	0x67,
36 	0x68,	0x69,	0x6a,	0x6b,	0x6c,	0x6d,	0x6e,	0x6f,
37 /*70*/	0x70,	0x71,	0x72,	0x73,	0x74,	0x75,	0x76,	0x77,
38 	0x78,	0x79,	0x7a,	0x7b,	0x7c,	0x7d,	0x7e,	NONE,
39 
40 /*80*/	0xc7,	0xfc,	0xe9,	0xe2,	0xe4,	0xe0,	0xe5,	0xe7,
41 	0xea,	0xeb,	0xe8,	0xef,	0xee,	0xec,	0xc4,	0xc5,
42 /*90*/	0xc9,	0xe6,	0xc6,	0xf4,	0xf6,	0xf2,	0xfb,	0xf9,
43 	0xff,	0xd6,	0xdc,	0xa2,	0xa3,	0xa5,	0x20a7,	0x283,
44 /*a0*/	0xe1,	0xed,	0xf3,	0xfa,	0xf1,	0xd1,	0xaa,	0xba,
45 	0xbf,	0x2310,	0xac,	0xbd,	0xbc,	0xa1,	0xab,	0xbb,
46 
47 /*b0*/	0x254,	0x259,	0xf0,	0x283,	0x292,	0x14b,	0x251,	0x7a,
48 	0x26a,	0xf0,	0x292,	0xe3,	0x153,	0x169,	0x28c,	0x265,
49 /*c0*/	0x280,	0xeb,	0x6c,	0x28c,	0xf5,	0xf1,	0x152,	NONE,
50 	NONE,	0x53,	0x73,	0x5a,	0x7a,	NONE,	NONE,	NONE,
51 /*d0*/	0xdf,	NONE,	NONE,	0x101,	0x12b,	0x16b,	0x113,	0x14d,	
52 	NONE,	NONE,	NONE,	0x20,	NONE,	NONE,	NONE,	NONE,
53 
54 /*e0*/	0x3b1,	0x3b2,	0x3b3,	0x3c0,	0x3a3,	0x3c3,	0xb5,	0x3c4,
55 	0x3a6,	0x398,	0x3a9,	0x3b4,	0x221e,	0xd8,	0x3b5,	0x2229,
56 /*f0*/	0x2261,	0xb1,	0x2265,	0x2264,	0x2320,	0x2321,	0xf7,	0x2248,
57 	0xb0,	0x2219,	0xb7,	NONE,	NONE,	NONE,	NONE,	NONE,
58 };
59 
60 enum{ Utf, Kanahi, Kanalo=Kanahi+1, GBhi, GBlo=GBhi+1, };
61 
62 void
63 worldprintentry(Entry e, int cmd)
64 {
65 	int nh, state[3];
66 	uchar *p, *pe;
67 
68 	p = (uchar *)e.start;
69 	pe = (uchar *)e.end;
70 	nh = GSHORT(p);
71 	p += 6;
72 	if(cmd == 'h')
73 		pe = p+nh;
74 	state[0] = Utf;
75 	state[1] = 0;
76 	state[2] = 0;
77 	while(p < pe){
78 		if(cmd == 'r')
79 			outchar(*p++);
80 		else
81 			putchar(*p++, state);
82 	}
83 	outnl(0);
84 }
85 
86 long
87 worldnextoff(long fromoff)
88 {
89 	int nh, np, nd;
90 	uchar buf[6];
91 
92 	if(Bseek(bdict, fromoff-1, 0) < 0)
93 		return -1;
94 	if(Bread(bdict, buf, 6) != 6)
95 		return -1;
96 	nh = GSHORT(buf);
97 	np = GSHORT(buf+2);
98 	nd = GSHORT(buf+4);
99 	return fromoff-1 + 6 + nh + np + nd;
100 }
101 
102 static void
103 putchar(int c, int *state)
104 {
105 	int xflag = 0;
106 	Rune r;
107 	int hi, lo;
108 
109 	switch(state[0]){
110 	case Kanahi:
111 	case GBhi:
112 		if(CANS2JH(c) || c == 0xff){
113 			state[0]++;
114 			state[1] = c;
115 			break;
116 		}
117 		/* fall through */
118 	case Utf:
119 		if(c == 0xfe){
120 			state[0] = Kanahi;
121 			break;
122 		}else if(c == 0xff){
123 			state[0] = GBhi;
124 			break;
125 		}
126 		r = chartab[c];
127 		if(r < 0x80 && state[2] == 0)
128 			outchar(r);
129 		else if(r == NONE){
130 			switch(c){
131 			case 0xfb:
132 				if(!xflag){
133 					state[2] = 1;
134 					break;
135 				}
136 			case 0xfc:
137 				if(!xflag){
138 					state[2] = 0;
139 					break;
140 				}
141 			case 0x10:
142 			case 0xc7: case 0xc8:
143 			case 0xd8: case 0xd9: case 0xda:
144 			case 0xdc: case 0xdd: case 0xde: case 0xdf:
145 			case 0xfd:
146 				if(!xflag)
147 					break;
148 				/* fall through */
149 			default:
150 				outprint("\\%.2ux", c);
151 			}
152 		}else if(state[2] == 0)
153 			outrune(r);
154 		break;
155 	case Kanalo:
156 	case GBlo:
157 		if(state[1] == 0xff && c == 0xff){
158 			state[0] = Utf;
159 			break;
160 		}
161 		state[0]--;
162 		hi = state[1];
163 		lo = c;
164 		S2J(hi, lo);		/* convert to JIS */
165 		r = hi*100 + lo - 3232;	/* convert to jis208 */
166 		if(state[0] == Kanahi && r < JIS208MAX)
167 			r = tabjis208[r];
168 		else if(state[0] == GBhi && r < GB2312MAX)
169 			r = tabgb2312[r];
170 		else
171 			r = NONE;
172 		if(r == NONE)
173 			outprint("\\%.2ux\\%.2ux", state[1], c);
174 		else
175 			outrune(r);
176 		break;
177 	}
178 }
179 
180 void
181 worldprintkey(void)
182 {
183 	Bprint(bout, "No pronunciation key.\n");
184 }