1 %token CHAR CCL NCCL STR DELIM SCON ITER NEWE NULLS
2 %left SCON '/' NEWE
3 %left '|'
4 %left '$' '^'
5 %left CHAR CCL NCCL '(' '.' STR NULLS
6 %left ITER
7 %left CAT
8 %left '*' '+' '?'
9 
10 %{
11 # include "ldefs.h"
12 #define YYSTYPE union _yystype_
13 union _yystype_
14 {
15 	int	i;
16 	uchar	*cp;
17 };
18 %}
19 %%
20 %{
21 int i;
22 int j,k;
23 int g;
24 uchar *p;
25 %}
26 acc	:	lexinput
27 	={	
28 # ifdef DEBUG
29 		if(debug) sect2dump();
30 # endif
31 	}
32 	;
33 lexinput:	defns delim prods end
34 	|	defns delim end
35 	={
36 		if(!funcflag)phead2();
37 		funcflag = TRUE;
38 	}
39 	| error
40 	={
41 # ifdef DEBUG
42 		if(debug) {
43 			sect1dump();
44 			sect2dump();
45 			}
46 # endif
47 		}
48 	;
49 end:		delim | ;
50 defns:	defns STR STR
51 	={	strcpy((char*)dp,(char*)$2.cp);
52 		def[dptr] = dp;
53 		dp += strlen((char*)$2.cp) + 1;
54 		strcpy((char*)dp,(char*)$3.cp);
55 		subs[dptr++] = dp;
56 		if(dptr >= DEFSIZE)
57 			error("Too many definitions");
58 		dp += strlen((char*)$3.cp) + 1;
59 		if(dp >= dchar+DEFCHAR)
60 			error("Definitions too long");
61 		subs[dptr]=def[dptr]=0;	/* for lookup - require ending null */
62 	}
63 	|
64 	;
65 delim:	DELIM
66 	={
67 # ifdef DEBUG
68 		if(sect == DEFSECTION && debug) sect1dump();
69 # endif
70 		sect++;
71 		}
72 	;
73 prods:	prods pr
74 	={	$$.i = mn2(RNEWE,$1.i,$2.i);
75 		}
76 	|	pr
77 	={	$$.i = $1.i;}
78 	;
79 pr:	r NEWE
80 	={
81 		if(divflg == TRUE)
82 			i = mn1(S1FINAL,casecount);
83 		else i = mn1(FINAL,casecount);
84 		$$.i = mn2(RCAT,$1.i,i);
85 		divflg = FALSE;
86 		casecount++;
87 		}
88 	| error NEWE
89 	={
90 # ifdef DEBUG
91 		if(debug) sect2dump();
92 # endif
93 		}
94 r:	CHAR
95 	={	$$.i = mn0($1.i); }
96 	| STR
97 	={
98 		p = $1.cp;
99 		i = mn0(*p++);
100 		while(*p)
101 			i = mn2(RSTR,i,*p++);
102 		$$.i = i;
103 		}
104 	| '.'
105 	={	symbol['\n'] = 0;
106 		if(psave == FALSE){
107 			p = ccptr;
108 			psave = ccptr;
109 			for(i=1;i<'\n';i++){
110 				symbol[i] = 1;
111 				*ccptr++ = i;
112 				}
113 			for(i='\n'+1;i<NCH;i++){
114 				symbol[i] = 1;
115 				*ccptr++ = i;
116 				}
117 			*ccptr++ = 0;
118 			if(ccptr > ccl+CCLSIZE)
119 				error("Too many large character classes");
120 			}
121 		else
122 			p = psave;
123 		$$.i = mn1(RCCL,(int)p);
124 		cclinter(1);
125 		}
126 	| CCL
127 	={	$$.i = mn1(RCCL,$1.i); }
128 	| NCCL
129 	={	$$.i = mn1(RNCCL,$1.i); }
130 	| r '*'
131 	={	$$.i = mn1(STAR,$1.i); }
132 	| r '+'
133 	={	$$.i = mn1(PLUS,$1.i); }
134 	| r '?'
135 	={	$$.i = mn1(QUEST,$1.i); }
136 	| r '|' r
137 	={	$$.i = mn2(BAR,$1.i,$3.i); }
138 	| r r %prec CAT
139 	={	$$.i = mn2(RCAT,$1.i,$2.i); }
140 	| r '/' r
141 	={	if(!divflg){
142 			j = mn1(S2FINAL,-casecount);
143 			i = mn2(RCAT,$1.i,j);
144 			$$.i = mn2(DIV,i,$3.i);
145 			}
146 		else {
147 			$$.i = mn2(RCAT,$1.i,$3.i);
148 			warning("Extra slash removed");
149 			}
150 		divflg = TRUE;
151 		}
152 	| r ITER ',' ITER '}'
153 	={	if($2.i > $4.i){
154 			i = $2.i;
155 			$2.i = $4.i;
156 			$4.i = i;
157 			}
158 		if($4.i <= 0)
159 			warning("Iteration range must be positive");
160 		else {
161 			j = $1.i;
162 			for(k = 2; k<=$2.i;k++)
163 				j = mn2(RCAT,j,dupl($1.i));
164 			for(i = $2.i+1; i<=$4.i; i++){
165 				g = dupl($1.i);
166 				for(k=2;k<=i;k++)
167 					g = mn2(RCAT,g,dupl($1.i));
168 				j = mn2(BAR,j,g);
169 				}
170 			$$.i = j;
171 			}
172 	}
173 	| r ITER '}'
174 	={
175 		if($2.i < 0)warning("Can't have negative iteration");
176 		else if($2.i == 0) $$.i = mn0(RNULLS);
177 		else {
178 			j = $1.i;
179 			for(k=2;k<=$2.i;k++)
180 				j = mn2(RCAT,j,dupl($1.i));
181 			$$.i = j;
182 			}
183 		}
184 	| r ITER ',' '}'
185 	={
186 				/* from n to infinity */
187 		if($2.i < 0)warning("Can't have negative iteration");
188 		else if($2.i == 0) $$.i = mn1(STAR,$1.i);
189 		else if($2.i == 1)$$.i = mn1(PLUS,$1.i);
190 		else {		/* >= 2 iterations minimum */
191 			j = $1.i;
192 			for(k=2;k<$2.i;k++)
193 				j = mn2(RCAT,j,dupl($1.i));
194 			k = mn1(PLUS,dupl($1.i));
195 			$$.i = mn2(RCAT,j,k);
196 			}
197 		}
198 	| SCON r
199 	={	$$.i = mn2(RSCON,$2.i,$1.i); }
200 	| '^' r
201 	={	$$.i = mn1(CARAT,$2.i); }
202 	| r '$'
203 	={	i = mn0('\n');
204 		if(!divflg){
205 			j = mn1(S2FINAL,-casecount);
206 			k = mn2(RCAT,$1.i,j);
207 			$$.i = mn2(DIV,k,i);
208 			}
209 		else $$.i = mn2(RCAT,$1.i,i);
210 		divflg = TRUE;
211 		}
212 	| '(' r ')'
213 	={	$$.i = $2.i; }
214 	|	NULLS
215 	={	$$.i = mn0(RNULLS); }
216 	;
217 %%
218 int
219 yylex(void)
220 {
221 	uchar *p;
222 	int c, i;
223 	uchar  *t, *xp;
224 	int n, j, k, x;
225 	static int sectbegin;
226 	static uchar token[TOKENSIZE];
227 	static int iter;
228 
229 # ifdef DEBUG
230 	yylval.i = 0;
231 # endif
232 
233 	if(sect == DEFSECTION) {		/* definitions section */
234 		while(!eof) {
235 			if(prev == '\n'){		/* next char is at beginning of line */
236 				getl(p=buf);
237 				switch(*p){
238 				case '%':
239 					switch(*(p+1)){
240 					case '%':
241 						lgate();
242 						Bprint(&fout,"#define YYNEWLINE %d\n",'\n');
243 						Bprint(&fout,"int\nyylex(void){\nint nstr; extern int yyprevious;\nif(yyprevious){}\n");
244 						sectbegin = TRUE;
245 						i = treesize*(sizeof(*name)+sizeof(*left)+
246 							sizeof(*right)+sizeof(*nullstr)+sizeof(*parent))+ALITTLEEXTRA;
247 						p = myalloc(i,1);
248 						if(p == 0)
249 							error("Too little core for parse tree");
250 						free(p);
251 						name = myalloc(treesize,sizeof(*name));
252 						left = myalloc(treesize,sizeof(*left));
253 						right = myalloc(treesize,sizeof(*right));
254 						nullstr = myalloc(treesize,sizeof(*nullstr));
255 						parent = myalloc(treesize,sizeof(*parent));
256 						if(name == 0 || left == 0 || right == 0 || parent == 0 || nullstr == 0)
257 							error("Too little core for parse tree");
258 						return(freturn(DELIM));
259 					case 'p': case 'P':	/* has overridden number of positions */
260 						while(*p && !isdigit(*p))p++;
261 						maxpos = atol((char*)p);
262 # ifdef DEBUG
263 						if (debug) print("positions (%%p) now %d\n",maxpos);
264 # endif
265 						if(report == 2)report = 1;
266 						continue;
267 					case 'n': case 'N':	/* has overridden number of states */
268 						while(*p && !isdigit(*p))p++;
269 						nstates = atol((char*)p);
270 # ifdef DEBUG
271 						if(debug)print( " no. states (%%n) now %d\n",nstates);
272 # endif
273 						if(report == 2)report = 1;
274 						continue;
275 					case 'e': case 'E':		/* has overridden number of tree nodes */
276 						while(*p && !isdigit(*p))p++;
277 						treesize = atol((char*)p);
278 # ifdef DEBUG
279 						if (debug) print("treesize (%%e) now %d\n",treesize);
280 # endif
281 						if(report == 2)report = 1;
282 						continue;
283 					case 'o': case 'O':
284 						while (*p && !isdigit(*p))p++;
285 						outsize = atol((char*)p);
286 						if (report ==2) report=1;
287 						continue;
288 					case 'a': case 'A':		/* has overridden number of transitions */
289 						while(*p && !isdigit(*p))p++;
290 						if(report == 2)report = 1;
291 						ntrans = atol((char*)p);
292 # ifdef DEBUG
293 						if (debug)print("N. trans (%%a) now %d\n",ntrans);
294 # endif
295 						continue;
296 					case 'k': case 'K': /* overriden packed char classes */
297 						while (*p && !isdigit(*p))p++;
298 						if (report==2) report=1;
299 						free(pchar);
300 						pchlen = atol((char*)p);
301 # ifdef DEBUG
302 						if (debug) print( "Size classes (%%k) now %d\n",pchlen);
303 # endif
304 						pchar=pcptr=myalloc(pchlen, sizeof(*pchar));
305 						continue;
306 					case '{':
307 						lgate();
308 						while(getl(p) && strcmp((char*)p,"%}") != 0)
309 							Bprint(&fout, "%s\n",(char*)p);
310 						if(p[0] == '%') continue;
311 						error("Premature eof");
312 					case 's': case 'S':		/* start conditions */
313 						lgate();
314 						while(*p && strchr(" \t,", *p) == 0) p++;
315 						n = TRUE;
316 						while(n){
317 							while(*p && strchr(" \t,", *p)) p++;
318 							t = p;
319 							while(*p && strchr(" \t,", *p) == 0)p++;
320 							if(!*p) n = FALSE;
321 							*p++ = 0;
322 							if (*t == 0) continue;
323 							i = sptr*2;
324 							Bprint(&fout,"#define %s %d\n",(char*)t,i);
325 							strcpy((char*)sp, (char*)t);
326 							sname[sptr++] = sp;
327 							sname[sptr] = 0;	/* required by lookup */
328 							if(sptr >= STARTSIZE)
329 								error("Too many start conditions");
330 							sp += strlen((char*)sp) + 1;
331 							if(sp >= stchar+STARTCHAR)
332 								error("Start conditions too long");
333 						}
334 						continue;
335 					default:
336 						warning("Invalid request %s",p);
337 						continue;
338 					}	/* end of switch after seeing '%' */
339 				case ' ': case '\t':		/* must be code */
340 					lgate();
341 					Bprint(&fout, "%s\n",(char*)p);
342 					continue;
343 				default:		/* definition */
344 					while(*p && !isspace(*p)) p++;
345 					if(*p == 0)
346 						continue;
347 					prev = *p;
348 					*p = 0;
349 					bptr = p+1;
350 					yylval.cp = buf;
351 					if(isdigit(buf[0]))
352 						warning("Substitution strings may not begin with digits");
353 					return(freturn(STR));
354 				}
355 			}
356 			/* still sect 1, but prev != '\n' */
357 			else {
358 				p = bptr;
359 				while(*p && isspace(*p)) p++;
360 				if(*p == 0)
361 					warning("No translation given - null string assumed");
362 				strcpy((char*)token, (char*)p);
363 				yylval.cp = token;
364 				prev = '\n';
365 				return(freturn(STR));
366 			}
367 		}
368 		/* end of section one processing */
369 	} else if(sect == RULESECTION){		/* rules and actions */
370 		while(!eof){
371 			switch(c=gch()){
372 			case '\0':
373 				return(freturn(0));
374 			case '\n':
375 				if(prev == '\n') continue;
376 				x = NEWE;
377 				break;
378 			case ' ':
379 			case '\t':
380 				if(sectbegin == TRUE){
381 					cpyact();
382 					while((c=gch()) && c != '\n');
383 					continue;
384 				}
385 				if(!funcflag)phead2();
386 				funcflag = TRUE;
387 				Bprint(&fout,"case %d:\n",casecount);
388 				if(cpyact())
389 					Bprint(&fout,"break;\n");
390 				while((c=gch()) && c != '\n');
391 				if(peek == ' ' || peek == '\t' || sectbegin == TRUE){
392 					warning("Executable statements should occur right after %%");
393 					continue;
394 				}
395 				x = NEWE;
396 				break;
397 			case '%':
398 				if(prev != '\n') goto character;
399 				if(peek == '{'){	/* included code */
400 					getl(buf);
401 					while(!eof && getl(buf) && strcmp("%}",(char*)buf) != 0)
402 						Bprint(&fout,"%s\n",(char*)buf);
403 					continue;
404 				}
405 				if(peek == '%'){
406 					gch();
407 					gch();
408 					x = DELIM;
409 					break;
410 				}
411 				goto character;
412 			case '|':
413 				if(peek == ' ' || peek == '\t' || peek == '\n'){
414 					Bprint(&fout,"%d\n",30000+casecount++);
415 					continue;
416 				}
417 				x = '|';
418 				break;
419 			case '$':
420 				if(peek == '\n' || peek == ' ' || peek == '\t' || peek == '|' || peek == '/'){
421 					x = c;
422 					break;
423 				}
424 				goto character;
425 			case '^':
426 				if(prev != '\n' && scon != TRUE) goto character;	/* valid only at line begin */
427 				x = c;
428 				break;
429 			case '?':
430 			case '+':
431 			case '.':
432 			case '*':
433 			case '(':
434 			case ')':
435 			case ',':
436 			case '/':
437 				x = c;
438 				break;
439 			case '}':
440 				iter = FALSE;
441 				x = c;
442 				break;
443 			case '{':	/* either iteration or definition */
444 				if(isdigit(c=gch())){	/* iteration */
445 					iter = TRUE;
446 				ieval:
447 					i = 0;
448 					while(isdigit(c)){
449 						token[i++] = c;
450 						c = gch();
451 					}
452 					token[i] = 0;
453 					yylval.i = atol((char*)token);
454 					munputc(c);
455 					x = ITER;
456 					break;
457 				} else {		/* definition */
458 					i = 0;
459 					while(c && c!='}'){
460 						token[i++] = c;
461 						c = gch();
462 					}
463 					token[i] = 0;
464 					i = lookup(token,def);
465 					if(i < 0)
466 						warning("Definition %s not found",token);
467 					else
468 						munputs(subs[i]);
469 					continue;
470 				}
471 			case '<':		/* start condition ? */
472 				if(prev != '\n')		/* not at line begin, not start */
473 					goto character;
474 				t = slptr;
475 				do {
476 					i = 0;
477 					c = gch();
478 					while(c != ',' && c && c != '>'){
479 						token[i++] = c;
480 						c = gch();
481 					}
482 					token[i] = 0;
483 					if(i == 0)
484 						goto character;
485 					i = lookup(token,sname);
486 					if(i < 0) {
487 						warning("Undefined start condition %s",token);
488 						continue;
489 					}
490 					*slptr++ = i+1;
491 				} while(c && c != '>');
492 				*slptr++ = 0;
493 				/* check if previous value re-usable */
494 				for (xp=slist; xp<t; ){
495 					if (strcmp((char*)xp, (char*)t)==0)
496 						break;
497 					while (*xp++);
498 				}
499 				if (xp<t){
500 					/* re-use previous pointer to string */
501 					slptr=t;
502 					t=xp;
503 				}
504 				if(slptr > slist+STARTSIZE)		/* note not packed ! */
505 					error("Too many start conditions used");
506 				yylval.cp = t;
507 				x = SCON;
508 				break;
509 			case '"':
510 				i = 0;
511 				while((c=gch()) && c != '"' && c != '\n'){
512 					if(c == '\\') c = usescape(gch());
513 					token[i++] = c;
514 					if(i > TOKENSIZE){
515 						warning("String too long");
516 						i = TOKENSIZE-1;
517 						break;
518 					}
519 				}
520 				if(c == '\n') {
521 					yyline--;
522 					warning("Non-terminated string");
523 					yyline++;
524 				}
525 				token[i] = 0;
526 				if(i == 0)x = NULLS;
527 				else if(i == 1){
528 					yylval.i = token[0];
529 					x = CHAR;
530 				} else {
531 					yylval.cp = token;
532 					x = STR;
533 				}
534 				break;
535 			case '[':
536 				for(i=1;i<NCH;i++) symbol[i] = 0;
537 				x = CCL;
538 				if((c = gch()) == '^'){
539 					x = NCCL;
540 					c = gch();
541 				}
542 				while(c != ']' && c){
543 					if(c == '\\') c = usescape(gch());
544 					symbol[c] = 1;
545 					j = c;
546 					if((c=gch()) == '-' && peek != ']'){		/* range specified */
547 						c = gch();
548 						if(c == '\\') c = usescape(gch());
549 						k = c;
550 						if(j > k) {
551 							n = j;
552 							j = k;
553 							k = n;
554 						}
555 						if(!(('A' <= j && k <= 'Z') ||
556 						     ('a' <= j && k <= 'z') ||
557 						     ('0' <= j && k <= '9')))
558 							warning("Non-portable Character Class");
559 						for(n=j+1;n<=k;n++)
560 							symbol[n] = 1;		/* implementation dependent */
561 						c = gch();
562 					}
563 				}
564 				/* try to pack ccl's */
565 				i = 0;
566 				for(j=0;j<NCH;j++)
567 					if(symbol[j])token[i++] = j;
568 				token[i] = 0;
569 				p = ccl;
570 				while(p <ccptr && strcmp((char*)token,(char*)p) != 0)p++;
571 				if(p < ccptr)	/* found it */
572 					yylval.cp = p;
573 				else {
574 					yylval.cp = ccptr;
575 					strcpy((char*)ccptr,(char*)token);
576 					ccptr += strlen((char*)token) + 1;
577 					if(ccptr >= ccl+CCLSIZE)
578 						error("Too many large character classes");
579 				}
580 				cclinter(x==CCL);
581 				break;
582 			case '\\':
583 				c = usescape(gch());
584 			default:
585 			character:
586 				if(iter){	/* second part of an iteration */
587 					iter = FALSE;
588 					if('0' <= c && c <= '9')
589 						goto ieval;
590 				}
591 				if(isalpha(peek)){
592 					i = 0;
593 					yylval.cp = token;
594 					token[i++] = c;
595 					while(isalpha(peek))
596 						token[i++] = gch();
597 					if(peek == '?' || peek == '*' || peek == '+')
598 						munputc(token[--i]);
599 					token[i] = 0;
600 					if(i == 1){
601 						yylval.i = token[0];
602 						x = CHAR;
603 					}
604 					else x = STR;
605 				} else {
606 					yylval.i = c;
607 					x = CHAR;
608 				}
609 			}
610 			scon = FALSE;
611 			if(x == SCON)scon = TRUE;
612 			sectbegin = FALSE;
613 			return(freturn(x));
614 		}
615 	}
616 	/* section three */
617 	ptail();
618 # ifdef DEBUG
619 	if(debug)
620 		Bprint(&fout,"\n/*this comes from section three - debug */\n");
621 # endif
622 	while(getl(buf) && !eof)
623 		Bprint(&fout,"%s\n",(char*)buf);
624 	return(freturn(0));
625 }
626 /* end of yylex */
627 # ifdef DEBUG
628 int
629 freturn(int i)
630 {
631 	if(yydebug) {
632 		print("now return ");
633 		if(i < NCH) allprint(i);
634 		else print("%d",i);
635 		printf("   yylval = ");
636 		switch(i){
637 			case STR: case CCL: case NCCL:
638 				strpt(yylval.cp);
639 				break;
640 			case CHAR:
641 				allprint(yylval.i);
642 				break;
643 			default:
644 				print("%d",yylval.i);
645 				break;
646 		}
647 		print("\n");
648 	}
649 	return(i);
650 }
651 # endif