1 bb7ff349 2009-07-08 mt /*
  2 bb7ff349 2009-07-08 mt  * Parse /lib/keyboard to create latin1.h table for kernel.
  3 bb7ff349 2009-07-08 mt  * mklatinkbd -r prints an array of integers rather than a Rune string literal.
  4 bb7ff349 2009-07-08 mt  */
  5 bb7ff349 2009-07-08 mt 
  6 bb7ff349 2009-07-08 mt #include <u.h>
  7 bb7ff349 2009-07-08 mt #include <libc.h>
  8 bb7ff349 2009-07-08 mt #include <bio.h>
  9 bb7ff349 2009-07-08 mt #include <ctype.h>
 10 bb7ff349 2009-07-08 mt 
 11 bb7ff349 2009-07-08 mt int rflag;
 12 bb7ff349 2009-07-08 mt 
 13 bb7ff349 2009-07-08 mt enum {
 14 bb7ff349 2009-07-08 mt 	MAXLD = 2,	/* latin1.c assumes this is 2 */
 15 bb7ff349 2009-07-08 mt };
 16 bb7ff349 2009-07-08 mt 
 17 bb7ff349 2009-07-08 mt char *head = ""
 18 bb7ff349 2009-07-08 mt "/*\n"
 19 bb7ff349 2009-07-08 mt " * This is automatically generated by %s from /lib/keyboard\n"
 20 bb7ff349 2009-07-08 mt " * Edit /lib/keyboard instead.\n"
 21 bb7ff349 2009-07-08 mt " */\n";
 22 bb7ff349 2009-07-08 mt 
 23 bb7ff349 2009-07-08 mt /*
 24 bb7ff349 2009-07-08 mt  * latin1.c assumes that strlen(ld) is at most 2.
 25 bb7ff349 2009-07-08 mt  * It also assumes that latintab[i].ld can be a prefix of latintab[j].ld
 26 bb7ff349 2009-07-08 mt  * only when j < i.  We ensure this by sorting the output by prefix length.
 27 bb7ff349 2009-07-08 mt  * The so array is indexed by the character value.
 28 bb7ff349 2009-07-08 mt  */
 29 bb7ff349 2009-07-08 mt 
 30 bb7ff349 2009-07-08 mt typedef struct Trie	Trie;
 31 bb7ff349 2009-07-08 mt struct Trie {
 32 bb7ff349 2009-07-08 mt 	int n; /* of characters r */
 33 bb7ff349 2009-07-08 mt 	char seq[MAXLD+1];
 34 bb7ff349 2009-07-08 mt 	Rune r[256];
 35 bb7ff349 2009-07-08 mt 	Trie *link[256];
 36 bb7ff349 2009-07-08 mt };
 37 bb7ff349 2009-07-08 mt 
 38 bb7ff349 2009-07-08 mt Trie *root;
 39 bb7ff349 2009-07-08 mt 
 40 bb7ff349 2009-07-08 mt Trie*
 41 bb7ff349 2009-07-08 mt mktrie(char *seq)
 42 bb7ff349 2009-07-08 mt {
 43 bb7ff349 2009-07-08 mt 	uchar *q;
 44 bb7ff349 2009-07-08 mt 	Trie **tp;
 45 bb7ff349 2009-07-08 mt 
 46 bb7ff349 2009-07-08 mt 	if(root == nil) {
 47 bb7ff349 2009-07-08 mt 		root = malloc(sizeof *root);
 48 bb7ff349 2009-07-08 mt 		memset(root, 0, sizeof *root);
 49 bb7ff349 2009-07-08 mt 	}
 50 bb7ff349 2009-07-08 mt 
 51 bb7ff349 2009-07-08 mt 	assert(seq[0] != '\0');
 52 bb7ff349 2009-07-08 mt 
 53 bb7ff349 2009-07-08 mt 	tp = &root;
 54 bb7ff349 2009-07-08 mt 	for(q=(uchar*)seq; *(q+1) != '\0'; q++) {
 55 bb7ff349 2009-07-08 mt 		tp = &(*tp)->link[*q];
 56 bb7ff349 2009-07-08 mt 		if(*tp == nil) {
 57 bb7ff349 2009-07-08 mt 			*tp = malloc(sizeof(**tp));
 58 bb7ff349 2009-07-08 mt 			assert(*tp != nil);
 59 bb7ff349 2009-07-08 mt 			memset(*tp, 0, sizeof(**tp));
 60 bb7ff349 2009-07-08 mt 			strcpy((*tp)->seq, seq);
 61 bb7ff349 2009-07-08 mt 			(*tp)->seq[q+1-(uchar*)seq] = '\0';
 62 bb7ff349 2009-07-08 mt 		}
 63 bb7ff349 2009-07-08 mt 	}
 64 bb7ff349 2009-07-08 mt 
 65 bb7ff349 2009-07-08 mt 	assert(*tp != nil);
 66 bb7ff349 2009-07-08 mt 	return *tp;
 67 bb7ff349 2009-07-08 mt }
 68 bb7ff349 2009-07-08 mt 
 69 bb7ff349 2009-07-08 mt /* add character sequence s meaning rune r */
 70 bb7ff349 2009-07-08 mt void
 71 bb7ff349 2009-07-08 mt insert(char *s, Rune r)
 72 bb7ff349 2009-07-08 mt {
 73 bb7ff349 2009-07-08 mt 	uchar lastc;
 74 bb7ff349 2009-07-08 mt 	int len;
 75 bb7ff349 2009-07-08 mt 	Trie *t;
 76 bb7ff349 2009-07-08 mt 
 77 bb7ff349 2009-07-08 mt 	len = strlen(s);
 78 bb7ff349 2009-07-08 mt 	lastc = (uchar)s[len-1];
 79 bb7ff349 2009-07-08 mt 
 80 bb7ff349 2009-07-08 mt 	t = mktrie(s);
 81 bb7ff349 2009-07-08 mt 	if(t->r[lastc]) {
 82 bb7ff349 2009-07-08 mt 		fprint(2, "warning: table duplicate: %s is %C and %C\n", s, t->r[lastc], r);
 83 bb7ff349 2009-07-08 mt 		return;
 84 bb7ff349 2009-07-08 mt 	}
 85 bb7ff349 2009-07-08 mt 	t->r[lastc] = r;
 86 bb7ff349 2009-07-08 mt 	t->n++;
 87 bb7ff349 2009-07-08 mt }
 88 bb7ff349 2009-07-08 mt 
 89 bb7ff349 2009-07-08 mt void
 90 bb7ff349 2009-07-08 mt cprintchar(Biobuf *b, int c)
 91 bb7ff349 2009-07-08 mt {
 92 bb7ff349 2009-07-08 mt 	/* print a byte c safe for a C string. */
 93 bb7ff349 2009-07-08 mt 	switch(c) {
 94 bb7ff349 2009-07-08 mt 	case '\'':
 95 bb7ff349 2009-07-08 mt 	case '\"':
 96 bb7ff349 2009-07-08 mt 	case '\\':
 97 bb7ff349 2009-07-08 mt 		Bprint(b, "\\%c", c);
 98 bb7ff349 2009-07-08 mt 		break;
 99 bb7ff349 2009-07-08 mt 	case '\t':
100 bb7ff349 2009-07-08 mt 		Bprint(b, "\\t");
101 bb7ff349 2009-07-08 mt 		break;
102 bb7ff349 2009-07-08 mt 	default:
103 bb7ff349 2009-07-08 mt 		if(isascii(c) && isprint(c))
104 bb7ff349 2009-07-08 mt 			Bprint(b, "%c", c);
105 bb7ff349 2009-07-08 mt 		else
106 bb7ff349 2009-07-08 mt 			Bprint(b, "\\x%.2x", c);
107 bb7ff349 2009-07-08 mt 		break;
108 bb7ff349 2009-07-08 mt 	}
109 bb7ff349 2009-07-08 mt }
110 bb7ff349 2009-07-08 mt 
111 bb7ff349 2009-07-08 mt void
112 bb7ff349 2009-07-08 mt cprints(Biobuf *b, char *p)
113 bb7ff349 2009-07-08 mt {
114 bb7ff349 2009-07-08 mt 	while(*p != '\0')
115 bb7ff349 2009-07-08 mt 		cprintchar(b, *p++);
116 bb7ff349 2009-07-08 mt }
117 bb7ff349 2009-07-08 mt 
118 bb7ff349 2009-07-08 mt 
119 bb7ff349 2009-07-08 mt void
120 bb7ff349 2009-07-08 mt printtrie(Biobuf *b, Trie *t)
121 bb7ff349 2009-07-08 mt {
122 bb7ff349 2009-07-08 mt 	int i;
123 bb7ff349 2009-07-08 mt 
124 bb7ff349 2009-07-08 mt 	for(i=0; i<256; i++)
125 bb7ff349 2009-07-08 mt 		if(t->link[i])
126 bb7ff349 2009-07-08 mt 			printtrie(b, t->link[i]);
127 bb7ff349 2009-07-08 mt 
128 bb7ff349 2009-07-08 mt 	if(t->n > 0) {
129 bb7ff349 2009-07-08 mt 		Bprint(b, "\t\"");
130 bb7ff349 2009-07-08 mt 		cprints(b, t->seq);
131 bb7ff349 2009-07-08 mt 		Bprint(b, "\", \"");
132 bb7ff349 2009-07-08 mt 		for(i=0; i<256; i++)
133 bb7ff349 2009-07-08 mt 			if(t->r[i])
134 bb7ff349 2009-07-08 mt 				cprintchar(b, i);
135 bb7ff349 2009-07-08 mt 		Bprint(b, "\",\t");
136 bb7ff349 2009-07-08 mt 		if(rflag) {
137 bb7ff349 2009-07-08 mt 			Bprint(b, "{");
138 bb7ff349 2009-07-08 mt 			for(i=0; i<256; i++)
139 bb7ff349 2009-07-08 mt 				if(t->r[i])
140 bb7ff349 2009-07-08 mt 					Bprint(b, " 0x%.4ux,", t->r[i]);
141 bb7ff349 2009-07-08 mt 			Bprint(b, " }");
142 bb7ff349 2009-07-08 mt 		} else {
143 bb7ff349 2009-07-08 mt 			Bprint(b, "L\"");
144 bb7ff349 2009-07-08 mt 			for(i=0; i<256; i++)
145 bb7ff349 2009-07-08 mt 				if(t->r[i])
146 bb7ff349 2009-07-08 mt 					Bprint(b, "%C", t->r[i]);
147 bb7ff349 2009-07-08 mt 			Bprint(b, "\"");
148 bb7ff349 2009-07-08 mt 		}
149 bb7ff349 2009-07-08 mt 		Bprint(b, ",\n");
150 bb7ff349 2009-07-08 mt 	}	
151 bb7ff349 2009-07-08 mt }
152 bb7ff349 2009-07-08 mt 
153 bb7ff349 2009-07-08 mt void
154 bb7ff349 2009-07-08 mt readfile(char *fname)
155 bb7ff349 2009-07-08 mt {
156 bb7ff349 2009-07-08 mt 	Biobuf *b;
157 bb7ff349 2009-07-08 mt 	char *line, *p;
158 bb7ff349 2009-07-08 mt 	char *seq;
159 bb7ff349 2009-07-08 mt 	int inseq;
160 bb7ff349 2009-07-08 mt 	int lineno;
161 bb7ff349 2009-07-08 mt 	Rune r;
162 bb7ff349 2009-07-08 mt 
163 bb7ff349 2009-07-08 mt 	if((b = Bopen(fname, OREAD)) == 0) {
164 bb7ff349 2009-07-08 mt 		fprint(2, "cannot open \"%s\": %r\n", fname);
165 bb7ff349 2009-07-08 mt 		exits("open");
166 bb7ff349 2009-07-08 mt 	}
167 bb7ff349 2009-07-08 mt 
168 bb7ff349 2009-07-08 mt 	lineno = 0;
169 bb7ff349 2009-07-08 mt 	while((line = Brdline(b, '\n')) != 0) {
170 bb7ff349 2009-07-08 mt 		lineno++;
171 bb7ff349 2009-07-08 mt 		if(line[0] == '#')
172 bb7ff349 2009-07-08 mt 			continue;
173 bb7ff349 2009-07-08 mt 
174 bb7ff349 2009-07-08 mt 		r = strtol(line, nil, 16);
175 bb7ff349 2009-07-08 mt 		p = strchr(line, ' ');
176 bb7ff349 2009-07-08 mt 		if(r == 0 || p != line+4 || p[0] != ' ' || p[1] != ' ') {
177 bb7ff349 2009-07-08 mt 			fprint(2, "%s:%d: cannot parse line\n", fname, lineno);
178 bb7ff349 2009-07-08 mt 			continue;
179 bb7ff349 2009-07-08 mt 		}
180 bb7ff349 2009-07-08 mt 
181 bb7ff349 2009-07-08 mt 		p = line+6;
182 bb7ff349 2009-07-08 mt /*	00AE  Or rO       ®	registered trade mark sign	*/
183 bb7ff349 2009-07-08 mt 		for(inseq=1, seq=p; (uchar)*p < Runeself; p++) {
184 bb7ff349 2009-07-08 mt 			if(*p == '\0' || isspace(*p)) {
185 bb7ff349 2009-07-08 mt 				if(inseq && p-seq >= 2) {
186 bb7ff349 2009-07-08 mt 					*p = '\0';
187 bb7ff349 2009-07-08 mt 					inseq = 0;
188 bb7ff349 2009-07-08 mt 					insert(seq, r);
189 bb7ff349 2009-07-08 mt 					*p = ' ';
190 bb7ff349 2009-07-08 mt 				}
191 bb7ff349 2009-07-08 mt 				if(*p == '\0')
192 bb7ff349 2009-07-08 mt 					break;
193 bb7ff349 2009-07-08 mt 			} else {
194 bb7ff349 2009-07-08 mt 				if(!inseq) {
195 bb7ff349 2009-07-08 mt 					seq = p;
196 bb7ff349 2009-07-08 mt 					inseq = 1;
197 bb7ff349 2009-07-08 mt 				}
198 bb7ff349 2009-07-08 mt 			}
199 bb7ff349 2009-07-08 mt 		}
200 bb7ff349 2009-07-08 mt 	}
201 bb7ff349 2009-07-08 mt }
202 bb7ff349 2009-07-08 mt 
203 bb7ff349 2009-07-08 mt void
204 bb7ff349 2009-07-08 mt usage(void)
205 bb7ff349 2009-07-08 mt {
206 bb7ff349 2009-07-08 mt 	fprint(2, "usage: mklatinkbd [-r] [/lib/keyboard]\n");
207 bb7ff349 2009-07-08 mt 	exits("usage");
208 bb7ff349 2009-07-08 mt }
209 bb7ff349 2009-07-08 mt 
210 bb7ff349 2009-07-08 mt void
211 bb7ff349 2009-07-08 mt main(int argc, char **argv)
212 bb7ff349 2009-07-08 mt {
213 bb7ff349 2009-07-08 mt 	Biobuf bout;
214 bb7ff349 2009-07-08 mt 
215 bb7ff349 2009-07-08 mt 	ARGBEGIN{
216 bb7ff349 2009-07-08 mt 	case 'r':	/* print rune values */
217 bb7ff349 2009-07-08 mt 		rflag = 1;
218 bb7ff349 2009-07-08 mt 		break;
219 bb7ff349 2009-07-08 mt 	default:
220 bb7ff349 2009-07-08 mt 		usage();
221 bb7ff349 2009-07-08 mt 	}ARGEND
222 bb7ff349 2009-07-08 mt 
223 bb7ff349 2009-07-08 mt 	if(argc > 1)
224 bb7ff349 2009-07-08 mt 		usage();
225 bb7ff349 2009-07-08 mt 
226 bb7ff349 2009-07-08 mt 	readfile(argc == 1 ? argv[0] : "/fd/0");
227 bb7ff349 2009-07-08 mt 
228 bb7ff349 2009-07-08 mt 	Binit(&bout, 1, OWRITE);
229 bb7ff349 2009-07-08 mt 	if(root)
230 bb7ff349 2009-07-08 mt 		printtrie(&bout, root);
231 bb7ff349 2009-07-08 mt 	exits(0);
232 bb7ff349 2009-07-08 mt }