xref: /illumos-gate/usr/src/tools/cscope-fast/find.c (revision a28480febf31f0e61debac062a55216a98a05a92)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License, Version 1.0 only
6  * (the "License").  You may not use this file except in compliance
7  * with the License.
8  *
9  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
10  * or http://www.opensolaris.org/os/licensing.
11  * See the License for the specific language governing permissions
12  * and limitations under the License.
13  *
14  * When distributing Covered Code, include this CDDL HEADER in each
15  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
16  * If applicable, add the following below this CDDL HEADER, with the
17  * fields enclosed by brackets "[]" replaced with your own identifying
18  * information: Portions Copyright [yyyy] [name of copyright owner]
19  *
20  * CDDL HEADER END
21  */
22 /*	Copyright (c) 1988 AT&T	*/
23 /*	  All Rights Reserved  	*/
24 
25 
26 /*
27  * Copyright 2004 Sun Microsystems, Inc.  All rights reserved.
28  * Use is subject to license terms.
29  */
30 
31 #pragma ident	"%Z%%M%	%I%	%E% SMI"
32 
33 /*
34  * 	cscope - interactive C symbol or text cross-reference
35  *
36  *	searching functions
37  */
38 
39 #include <unistd.h>
40 #include <stdio.h>
41 #include <libgen.h>
42 #include "global.h"
43 #include "vp.h"
44 
45 /*
46  * most of these functions have been optimized so their innermost loops have
47  * only one test for the desired character by putting the char and
48  * an end-of-block marker (\0) at the end of the disk block buffer.
49  * When the inner loop exits on the char, an outer loop will see if
50  * the char is followed by a \0.  If so, it will read the next block
51  * and restart the inner loop.
52  */
53 
54 char	block[BUFSIZ + 2];		/* leave room for end-of-block mark */
55 int	blocklen;			/* length of disk block read */
56 char	blockmark;			/* mark character to be searched for */
57 long	blocknumber;			/* block number */
58 char	*blockp;			/* pointer to current char in block */
59 char	lastfilepath[PATHLEN + 1];	/* last file that full path was */
60 					/* computed for */
61 
62 static	char	cpattern[PATLEN + 1];	/* compressed pattern */
63 static	long	lastfcnoffset;		/* last function name offset */
64 static	long	postingsfound;		/* retrieved number of postings */
65 static	char	*regexp;		/* regular expression */
66 static	POSTING	*postingp;		/* retrieved posting set pointer */
67 static	long	searchcount;		/* count of files searched */
68 static	long	starttime;		/* start time for progress messages */
69 
70 static POSTING *getposting(void);
71 static void putsource(FILE *output);
72 static void putref(char *file, char *function);
73 static void findcalledbysub(char *file);
74 static void findterm(void);
75 static void fileprogress(void);
76 static void putpostingref(POSTING *p);
77 static void putline(FILE *output);
78 static char *strtolower(char *s);
79 static char *filepath(char *file);
80 
81 /* find the symbol in the cross-reference */
82 
83 void
84 findsymbol(void)
85 {
86 	char	file[PATHLEN + 1];	/* source file name */
87 	char	function[PATLEN + 1];	/* function name */
88 	char	macro[PATLEN + 1];	/* macro name */
89 	char	symbol[PATLEN + 1];	/* symbol name */
90 	char	*cp;
91 	char	c;
92 	char	*s;
93 
94 	if (invertedindex == YES) {
95 		long	lastline = 0;
96 		POSTING	*p;
97 
98 		findterm();
99 		while ((p = getposting()) != NULL) {
100 			if (p->type != INCLUDE && p->lineoffset != lastline) {
101 				putpostingref(p);
102 				lastline = p->lineoffset;
103 			}
104 		}
105 		return;
106 	}
107 	(void) scanpast('\t');		/* find the end of the header */
108 	skiprefchar();			/* skip the file marker */
109 	getstring(file);		/* save the file name */
110 	*function = '\0';
111 	/* a macro can be inside a function, but not vice versa */
112 	*macro = '\0';
113 
114 	/* find the next symbol */
115 	/* note: this code was expanded in-line for speed */
116 	/* while (scanpast('\n') != NULL) { */
117 	/* other macros were replaced by code using cp instead of blockp */
118 	cp = blockp;
119 	for (;;) {
120 		setmark('\n');
121 		do {	/* innermost loop optimized to only one test */
122 			while (*cp != '\n') {
123 				++cp;
124 			}
125 		} while (*(cp + 1) == '\0' && (cp = readblock()) != NULL);
126 
127 		/* skip the found character */
128 		if (cp != NULL && *(++cp + 1) == '\0') {
129 			cp = readblock();
130 		}
131 		if (cp == NULL) {
132 			break;
133 		}
134 		/* look for a source file or function name */
135 		if (*cp == '\t') {
136 			blockp = cp;
137 			switch (getrefchar()) {
138 
139 			case NEWFILE:		/* file name */
140 
141 				/* save the name */
142 				skiprefchar();
143 				getstring(file);
144 
145 				/* check for the end of the symbols */
146 				if (*file == '\0') {
147 					return;
148 				}
149 				fileprogress();
150 				/* FALLTHROUGH */
151 
152 			case FCNEND:		/* function end */
153 				*function = '\0';
154 				goto notmatched;	/* don't match name */
155 
156 			case FCNDEF:		/* function name */
157 				s = function;
158 				break;
159 
160 			case DEFINE:		/* could be a macro */
161 				if (fileversion >= 10) {
162 					s = macro;
163 				} else {
164 					s = symbol;
165 				}
166 				break;
167 
168 			case DEFINEEND:
169 				*macro = '\0';
170 				goto notmatched;	/* don't match name */
171 
172 			case INCLUDE:		/* #include file */
173 				goto notmatched;	/* don't match name */
174 			default:		/* other symbol */
175 				s = symbol;
176 			}
177 			/* save the name */
178 			skiprefchar();
179 			getstring(s);
180 
181 			/* see if this is a regular expression pattern */
182 			if (regexp != NULL) {
183 				if (caseless == YES) {
184 					s = strtolower(s);
185 				}
186 				if (*s != '\0' && regex(regexp, s) != NULL) {
187 					goto matched;
188 				}
189 			}
190 			/* match the symbol to the text pattern */
191 			else if (strequal(pattern, s)) {
192 				goto matched;
193 			}
194 			goto notmatched;
195 		}
196 		/* if this is a regular expression pattern */
197 		if (regexp != NULL) {
198 			c = *cp;
199 			if (c & 0200) {	/* digraph char? */
200 				c = dichar1[(c & 0177) / 8];
201 			}
202 			/* if this is a symbol */
203 			if (isalpha(c) || c == '_') {
204 				blockp = cp;
205 				getstring(symbol);
206 				s = symbol;
207 				if (caseless == YES) {
208 					s = strtolower(s);
209 				}
210 				/* match the symbol to the regular expression */
211 				if (regex(regexp, s) != NULL) {
212 					goto matched;
213 				}
214 				goto notmatched;
215 			}
216 		}
217 		/* match the character to the text pattern */
218 		else if (*cp == cpattern[0]) {
219 			blockp = cp;
220 
221 			/* match the rest of the symbol to the text pattern */
222 			if (matchrest()) {
223 				s = NULL;
224 matched:
225 				/*
226 				 * output the file, calling function or macro,
227 				 * and source line
228 				 */
229 				if (*macro != '\0' && s != macro) {
230 					putref(file, macro);
231 				} else if (s != function) {
232 					putref(file, function);
233 				} else {
234 					putref(file, "");
235 				}
236 				if (blockp == NULL) {
237 					return;
238 				}
239 			}
240 notmatched:
241 			cp = blockp;
242 		}
243 	}
244 	blockp = cp;
245 }
246 
247 /* find the function definition or #define */
248 
249 void
250 finddef(void)
251 {
252 	char	file[PATHLEN + 1];	/* source file name */
253 	char	function[PATLEN + 1];	/* function name */
254 	char	macro[PATLEN + 1];	/* macro name */
255 	char	symbol[PATLEN + 1];	/* symbol name */
256 	char	*s;
257 
258 	if (invertedindex == YES) {
259 		POSTING	*p;
260 
261 		findterm();
262 		while ((p = getposting()) != NULL) {
263 			switch (p->type) {
264 			case DEFINE:		/* could be a macro */
265 			case FCNDEF:
266 			case CLASSDEF:
267 			case ENUMDEF:
268 			case MEMBERDEF:
269 			case STRUCTDEF:
270 			case TYPEDEF:
271 			case UNIONDEF:
272 			case GLOBALDEF:		/* other global definition */
273 			case LOCALDEF:		/* other local definition */
274 			case PARAMETER:
275 				putpostingref(p);
276 			}
277 		}
278 		return;
279 	}
280 	/* find the next file name or definition */
281 	*function = '\0';
282 	/* a macro can be inside a function, but not vice versa */
283 	*macro = '\0';
284 
285 	while (scanpast('\t') != NULL) {
286 		switch (*blockp) {
287 
288 		case NEWFILE:
289 			skiprefchar();	/* save file name */
290 			getstring(file);
291 			if (*file == '\0') {	/* if end of symbols */
292 				return;
293 			}
294 			fileprogress();
295 			/* FALLTHROUGH */
296 
297 		case FCNEND:		/* function end */
298 			*function = '\0';
299 			break;
300 
301 		case FCNDEF:		/* function name */
302 			s = function;
303 			goto def;
304 
305 		case DEFINE:		/* could be a macro */
306 			if (fileversion >= 10) {
307 				s = macro;
308 			} else {
309 				s = symbol;
310 			}
311 			goto def;
312 
313 		case DEFINEEND:
314 			*macro = '\0';
315 			break;
316 
317 		case CLASSDEF:
318 		case ENUMDEF:
319 		case MEMBERDEF:
320 		case STRUCTDEF:
321 		case TYPEDEF:
322 		case UNIONDEF:
323 		case GLOBALDEF:		/* other global definition */
324 		case LOCALDEF:		/* other local definition */
325 		case PARAMETER:
326 			s = symbol;
327 		def:
328 			/* save the name */
329 			skiprefchar();
330 			getstring(s);
331 
332 			/* see if this is a regular expression pattern */
333 			if (regexp != NULL) {
334 				if (caseless == YES) {
335 					s = strtolower(s);
336 				}
337 				if (*s != '\0' && regex(regexp, s) != NULL) {
338 					goto matched;
339 				}
340 			} else if (strequal(pattern, s)) {
341 				/* match the symbol to the text pattern */
342 matched:
343 				/*
344 				 * output the file, calling function or macro,
345 				 * and source line
346 				 */
347 				if (*macro != '\0' && s != macro) {
348 					putref(file, macro);
349 				} else if (s != function) {
350 					putref(file, function);
351 				} else {
352 					putref(file, "");
353 				}
354 			}
355 		}
356 	}
357 }
358 
359 /* find all function definitions (used by samuel only) */
360 
361 void
362 findallfcns(void)
363 {
364 	char	file[PATHLEN + 1];	/* source file name */
365 	char	function[PATLEN + 1];	/* function name */
366 
367 	/* find the next file name or definition */
368 	while (scanpast('\t') != NULL) {
369 		switch (*blockp) {
370 		case NEWFILE:
371 			skiprefchar();	/* save file name */
372 			getstring(file);
373 			if (*file == '\0') {	/* if end of symbols */
374 				return;
375 			}
376 			fileprogress();
377 			break;
378 
379 		case FCNDEF:
380 		case CLASSDEF:
381 			skiprefchar();	/* save function name */
382 			getstring(function);
383 
384 			/* output the file, function and source line */
385 			putref(file, function);
386 			break;
387 		}
388 	}
389 }
390 
391 /* find the functions called by this function */
392 
393 void
394 findcalledby(void)
395 {
396 	char	file[PATHLEN + 1];	/* source file name */
397 
398 	if (invertedindex == YES) {
399 		POSTING	*p;
400 
401 		findterm();
402 		while ((p = getposting()) != NULL) {
403 			switch (p->type) {
404 			case DEFINE:		/* could be a macro */
405 			case FCNDEF:
406 				if (dbseek(p->lineoffset) != -1 &&
407 				    scanpast('\t') != NULL) {	/* skip def */
408 					findcalledbysub(srcfiles[p->fileindex]);
409 				}
410 			}
411 		}
412 		return;
413 	}
414 	/* find the function definition(s) */
415 	while (scanpast('\t') != NULL) {
416 		switch (*blockp) {
417 		case NEWFILE:
418 			skiprefchar();	/* save file name */
419 			getstring(file);
420 			if (*file == '\0') {	/* if end of symbols */
421 				return;
422 			}
423 			fileprogress();
424 			break;
425 
426 		case DEFINE:		/* could be a macro */
427 			if (fileversion < 10) {
428 				break;
429 			}
430 			/* FALLTHROUGH */
431 
432 		case FCNDEF:
433 			skiprefchar();	/* match name to pattern */
434 			if (match()) {
435 				findcalledbysub(file);
436 			}
437 			break;
438 		}
439 	}
440 }
441 
442 static void
443 findcalledbysub(char *file)
444 {
445 	/* find the next function call or the end of this function */
446 	while (scanpast('\t') != NULL) {
447 		switch (*blockp) {
448 		case DEFINE:		/* #define inside a function */
449 			if (fileversion >= 10) {	/* skip it */
450 				while (scanpast('\t') != NULL &&
451 				    *blockp != DEFINEEND)
452 					;
453 			}
454 			break;
455 		case FCNCALL:		/* function call */
456 
457 			/* output the file name */
458 			(void) fprintf(refsfound, "%s ", filepath(file));
459 
460 			/* output the function name */
461 			skiprefchar();
462 			putline(refsfound);
463 			(void) putc(' ', refsfound);
464 
465 			/* output the source line */
466 			putsource(refsfound);
467 			break;
468 
469 		case DEFINEEND:		/* #define end */
470 		case FCNEND:		/* function end */
471 		case FCNDEF:		/* function end (pre 9.5) */
472 		case NEWFILE:		/* file end */
473 			return;
474 		}
475 	}
476 }
477 
478 /* find the functions calling this function */
479 
480 void
481 findcalling(void)
482 {
483 	char	file[PATHLEN + 1];	/* source file name */
484 	char	function[PATLEN + 1];	/* function name */
485 	char	macro[PATLEN + 1];	/* macro name */
486 
487 	if (invertedindex == YES) {
488 		POSTING	*p;
489 
490 		findterm();
491 		while ((p = getposting()) != NULL) {
492 			if (p->type == FCNCALL) {
493 				putpostingref(p);
494 			}
495 		}
496 		return;
497 	}
498 	/* find the next file name or function definition */
499 	/* a macro can be inside a function, but not vice versa */
500 	*macro = '\0';
501 
502 	while (scanpast('\t') != NULL) {
503 		switch (*blockp) {
504 		case NEWFILE:		/* save file name */
505 			skiprefchar();
506 			getstring(file);
507 			if (*file == '\0') {	/* if end of symbols */
508 				return;
509 			}
510 			fileprogress();
511 			/* FALLTHROUGH */
512 		case FCNEND:		/* function end */
513 			*function = '\0';
514 			break;
515 		case DEFINE:		/* could be a macro */
516 			if (fileversion >= 10) {
517 				skiprefchar();
518 				getstring(macro);
519 			}
520 			break;
521 
522 		case DEFINEEND:
523 			*macro = '\0';
524 			break;
525 
526 		case FCNDEF:		/* save calling function name */
527 			skiprefchar();
528 			getstring(function);
529 			break;
530 		case FCNCALL:		/* match function called to pattern */
531 			skiprefchar();
532 			if (match()) {
533 				/* output the file, calling function or */
534 				/* macro, and source */
535 				if (*macro != '\0') {
536 					putref(file, macro);
537 				} else {
538 					putref(file, function);
539 				}
540 			}
541 		}
542 	}
543 }
544 
545 /* find direct assignment to, and increment and decrement of, this variable */
546 
547 void
548 findassignments(void)
549 {
550 	char	file[PATHLEN + 1];	/* source file name */
551 	char	function[PATLEN + 1];	/* function name */
552 	char	macro[PATLEN + 1];	/* macro name */
553 
554 	if (fileversion < 13) {
555 		putmsg("Database built with cscope version < 13 does not "
556 		    "have assignment information");
557 		(void) sleep(3);
558 		return;
559 	}
560 #if CTRACE
561 	ctroff();
562 #endif
563 	if (invertedindex == YES) {
564 		POSTING	*p;
565 
566 		findterm();
567 		while ((p = getposting()) != NULL) {
568 			switch (p->type) {
569 			case ASSIGNMENT:
570 			case GLOBALDEF:		/* can have initializer */
571 			case LOCALDEF:		/* can have initializer */
572 			case PARAMETER:		/* initial value */
573 				putpostingref(p);
574 			}
575 		}
576 		return;
577 	}
578 	/* find the next file name or function definition */
579 	/* a macro can be inside a function, but not vice versa */
580 	*macro = '\0';
581 
582 	while (scanpast('\t') != NULL) {
583 		switch (*blockp) {
584 		case NEWFILE:		/* save file name */
585 			skiprefchar();
586 			getstring(file);
587 			if (*file == '\0') {	/* if end of symbols */
588 				return;
589 			}
590 			fileprogress();
591 			/* FALLTHROUGH */
592 		case FCNEND:		/* function end */
593 			*function = '\0';
594 			break;
595 		case DEFINE:		/* could be a macro */
596 			if (fileversion >= 10) {
597 				skiprefchar();
598 				getstring(macro);
599 			}
600 			break;
601 
602 		case DEFINEEND:
603 			*macro = '\0';
604 			break;
605 
606 		case FCNDEF:		/* save calling function name */
607 			skiprefchar();
608 			getstring(function);
609 			break;
610 		case ASSIGNMENT:	/* match assignment to pattern */
611 		case GLOBALDEF:		/* can have initializer */
612 		case LOCALDEF:		/* can have initializer */
613 		case PARAMETER:		/* initial value */
614 			skiprefchar();
615 			if (match()) {
616 				/* output the file, calling function or */
617 				/* macro, and source */
618 				if (*macro != '\0') {
619 					putref(file, macro);
620 				} else {
621 					putref(file, function);
622 				}
623 			}
624 		}
625 	}
626 }
627 
628 /* find the grep pattern in the source files */
629 
630 char *
631 findgreppat(void)
632 {
633 	char	egreppat[2 * PATLEN];
634 	char	*cp, *pp;
635 
636 	/* translate egrep special characters in the regular expression */
637 	cp = egreppat;
638 	for (pp = pattern; *pp != '\0'; ++pp) {
639 		if (strchr("+?|()", *pp) != NULL) {
640 			*cp++ = '\\';
641 		}
642 		*cp++ = *pp;
643 	}
644 	*cp = '\0';
645 
646 	/* search the source files */
647 	return (findegreppat(egreppat));
648 }
649 
650 /* find this regular expression in the source files */
651 
652 char *
653 findegreppat(char *egreppat)
654 {
655 	int	i;
656 	char	*egreperror;
657 	char	msg[MSGLEN + 1];
658 
659 	/* compile the pattern */
660 	if ((egreperror = egrepinit(egreppat)) == NULL) {
661 
662 		/* search the files */
663 		for (i = 0; i < nsrcfiles; ++i) {
664 			char *file = filepath(srcfiles[i]);
665 			fileprogress();
666 			if (egrep(file, refsfound, "%s <unknown> %ld ") < 0) {
667 				(void) sprintf(msg, "Cannot open file %s",
668 				    file);
669 				putmsg2(msg);
670 			}
671 		}
672 	}
673 	return (egreperror);
674 }
675 
676 /* find matching file names */
677 
678 void
679 findfile(void)
680 {
681 	int	i;
682 	char	*s;
683 
684 	for (i = 0; i < nsrcfiles; ++i) {
685 		s = srcfiles[i];
686 		if (caseless == YES) {
687 			s = strtolower(s);
688 		}
689 		if (regex(regexp, s) != NULL) {
690 			(void) fprintf(refsfound, "%s <unknown> 1 <unknown>\n",
691 				filepath(srcfiles[i]));
692 		}
693 	}
694 }
695 
696 /* find files #including this file */
697 
698 void
699 findinclude(void)
700 {
701 	char	file[PATHLEN + 1];	/* source file name */
702 
703 	if (invertedindex == YES) {
704 		POSTING	*p;
705 
706 		findterm();
707 		while ((p = getposting()) != NULL) {
708 			if (p->type == INCLUDE) {
709 				putpostingref(p);
710 			}
711 		}
712 		return;
713 	}
714 	/* find the next file name or function definition */
715 	while (scanpast('\t') != NULL) {
716 		switch (*blockp) {
717 
718 		case NEWFILE:		/* save file name */
719 			skiprefchar();
720 			getstring(file);
721 			if (*file == '\0') {	/* if end of symbols */
722 				return;
723 			}
724 			fileprogress();
725 			break;
726 
727 		case INCLUDE:		/* match function called to pattern */
728 			skiprefchar();
729 			/* skip global or local #include marker */
730 			skiprefchar();
731 			if (match()) {
732 				/* output the file and source line */
733 				putref(file, "");
734 			}
735 		}
736 	}
737 }
738 
739 /* initialize */
740 
741 FINDINIT
742 findinit(void)
743 {
744 	char	buf[PATLEN + 3];
745 	BOOL	isregexp = NO;
746 	int	i;
747 	char	*s;
748 	unsigned c;
749 
750 	/* remove trailing white space */
751 	for (s = pattern + strlen(pattern) - 1; isspace(*s); --s) {
752 		*s = '\0';
753 	}
754 	/* allow a partial match for a file name */
755 	if (field == FILENAME || field == INCLUDES) {
756 		/* allow types.h to match #include <sys/types.h> */
757 		if (invertedindex == YES && field == INCLUDES &&
758 		    strncmp(pattern, ".*", 2) != 0) {
759 			(void) sprintf(pattern, ".*%s", strcpy(buf, pattern));
760 		}
761 		if ((regexp = regcmp(pattern, (char *)NULL)) == NULL) {
762 			return (REGCMPERROR);
763 		}
764 		return (NOERROR);
765 	}
766 	/* see if the pattern is a regular expression */
767 	if (strpbrk(pattern, "^.[{*+$") != NULL) {
768 		isregexp = YES;
769 	} else {
770 		/* check for a valid C symbol */
771 		s = pattern;
772 		if (!isalpha(*s) && *s != '_') {
773 			return (NOTSYMBOL);
774 		}
775 		while (*++s != '\0') {
776 			if (!isalnum(*s) && *s != '_') {
777 				return (NOTSYMBOL);
778 			}
779 		}
780 		/*
781 		 * look for use of the -T option (truncate symbol to 8
782 		 * characters) on a database not built with -T
783 		 */
784 		if (truncatesyms == YES && isuptodate == YES &&
785 		    dbtruncated == NO && s - pattern >= 8) {
786 			(void) strcpy(pattern + 8, ".*");
787 			isregexp = YES;
788 		}
789 	}
790 	/* if this is a regular expression or letter case is to be ignored */
791 	/* or there is an inverted index */
792 	if (isregexp == YES || caseless == YES || invertedindex == YES) {
793 
794 		/* remove a leading ^ */
795 		s = pattern;
796 		if (*s == '^') {
797 			(void) strcpy(newpat, s + 1);
798 			(void) strcpy(s, newpat);
799 		}
800 		/* remove a trailing $ */
801 		i = strlen(s) - 1;
802 		if (s[i] == '$') {
803 			s[i] = '\0';
804 		}
805 		/* if requested, try to truncate a C symbol pattern */
806 		if (truncatesyms == YES && strpbrk(s, "[{*+") == NULL) {
807 			s[8] = '\0';
808 		}
809 		/* must be an exact match */
810 		/*
811 		 * note: regcmp doesn't recognize ^*keypad$ as an syntax error
812 		 * unless it is given as a single arg
813 		 */
814 		(void) sprintf(buf, "^%s$", s);
815 		if ((regexp = regcmp(buf, (char *)NULL)) == NULL) {
816 			return (REGCMPERROR);
817 		}
818 	} else {
819 		/* if requested, truncate a C symbol pattern */
820 		if (truncatesyms == YES && field <= CALLING) {
821 			pattern[8] = '\0';
822 		}
823 		/* compress the string pattern for matching */
824 		s = cpattern;
825 		for (i = 0; (c = pattern[i]) != '\0'; ++i) {
826 			if (dicode1[c] && dicode2[(unsigned)pattern[i + 1]]) {
827 				c = (0200 - 2) + dicode1[c] +
828 				    dicode2[(unsigned)pattern[i + 1]];
829 				++i;
830 			}
831 			*s++ = (char)c;
832 		}
833 		*s = '\0';
834 	}
835 	return (NOERROR);
836 }
837 
838 void
839 findcleanup(void)
840 {
841 	/* discard any regular expression */
842 	if (regexp != NULL) {
843 		free(regexp);
844 		regexp = NULL;
845 	}
846 }
847 
848 /* find this term, which can be a regular expression */
849 
850 static void
851 findterm(void)
852 {
853 	char	*s;
854 	int	len;
855 	char	prefix[PATLEN + 1];
856 	char	term[PATLEN + 1];
857 
858 	npostings = 0;		/* will be non-zero after database built */
859 	lastfcnoffset = 0;	/* clear the last function name found */
860 	boolclear();		/* clear the posting set */
861 
862 	/* get the string prefix (if any) of the regular expression */
863 	(void) strcpy(prefix, pattern);
864 	if ((s = strpbrk(prefix, ".[{*+")) != NULL) {
865 		*s = '\0';
866 	}
867 	/* if letter case is to be ignored */
868 	if (caseless == YES) {
869 
870 		/*
871 		 * convert the prefix to upper case because it is lexically
872 		 * less than lower case
873 		 */
874 		s = prefix;
875 		while (*s != '\0') {
876 			*s = toupper(*s);
877 			++s;
878 		}
879 	}
880 	/* find the term lexically >= the prefix */
881 	(void) invfind(&invcontrol, prefix);
882 	if (caseless == YES) {	/* restore lower case */
883 		(void) strcpy(prefix, strtolower(prefix));
884 	}
885 	/*
886 	 * a null prefix matches the null term in the inverted index,
887 	 * so move to the first real term
888 	 */
889 	if (*prefix == '\0') {
890 		(void) invforward(&invcontrol);
891 	}
892 	len = strlen(prefix);
893 	do {
894 		(void) invterm(&invcontrol, term);	/* get the term */
895 		s = term;
896 		if (caseless == YES) {
897 			s = strtolower(s);	/* make it lower case */
898 		}
899 		/* if it matches */
900 		if (regex(regexp, s) != NULL) {
901 			/* add it's postings to the set */
902 			if ((postingp = boolfile(&invcontrol,
903 			    &npostings, OR)) == NULL) {
904 				break;
905 			}
906 		} else if (len > 0) {
907 			/* if there is a prefix */
908 
909 			/*
910 			 * if ignoring letter case and the term is out of the
911 			 * range of possible matches
912 			 */
913 			if (caseless == YES) {
914 				if (strncmp(term, prefix, len) > 0) {
915 					break;	/* stop searching */
916 				}
917 			}
918 			/* if using letter case and the prefix doesn't match */
919 			else if (strncmp(term, prefix, len) != 0) {
920 				break;	/* stop searching */
921 			}
922 		}
923 		/* display progress about every three seconds */
924 		if (++searchcount % 50 == 0) {
925 			progress("%ld of %ld symbols matched",
926 			    searchcount, totalterms);
927 		}
928 	} while (invforward(&invcontrol));	/* while didn't wrap around */
929 
930 	/* initialize the progress message for retrieving the references */
931 	initprogress();
932 	postingsfound = npostings;
933 }
934 
935 /* display the file search progress about every three seconds */
936 
937 static void
938 fileprogress(void)
939 {
940 	if (++searchcount % 10 == 0) {
941 		progress("%ld of %ld files searched", searchcount,
942 		    (long)nsrcfiles);
943 	}
944 }
945 
946 /* initialize the progress message */
947 
948 void
949 initprogress(void)
950 {
951 	searchcount = 0;
952 	starttime = time((long *)NULL);
953 }
954 
955 /* display the progress every three seconds */
956 
957 void
958 progress(char *format, long n1, long n2)
959 {
960 	char	msg[MSGLEN + 1];
961 	long	now;
962 
963 	/* print after 2 seconds so the average is nearer 3 seconds */
964 	if (linemode == NO && (now = time((long *)NULL)) - starttime >= 2) {
965 		starttime = now;
966 		(void) sprintf(msg, format, n1, n2);
967 		putmsg(msg);
968 	}
969 }
970 
971 /* match the pattern to the string */
972 
973 BOOL
974 match(void)
975 {
976 	char	string[PATLEN + 1];
977 	char	*s;
978 
979 	/* see if this is a regular expression pattern */
980 	if (regexp != NULL) {
981 		getstring(string);
982 		if (*string == '\0') {
983 			return (NO);
984 		}
985 		s = string;
986 		if (caseless == YES) {
987 			s = strtolower(s);
988 		}
989 		return (regex(regexp, s) ? YES : NO);
990 	}
991 	/* it is a string pattern */
992 	return ((BOOL)(*blockp == cpattern[0] && matchrest()));
993 }
994 
995 /* match the rest of the pattern to the name */
996 
997 BOOL
998 matchrest(void)
999 {
1000 	int	i = 1;
1001 
1002 	skiprefchar();
1003 	do {
1004 		while (*blockp == cpattern[i]) {
1005 			++blockp;
1006 			++i;
1007 		}
1008 	} while (*(blockp + 1) == '\0' && readblock() != NULL);
1009 
1010 	if (*blockp == '\n' && cpattern[i] == '\0') {
1011 		return (YES);
1012 	}
1013 	return (NO);
1014 }
1015 
1016 /* get the next posting for this term */
1017 
1018 static POSTING *
1019 getposting(void)
1020 {
1021 	if (npostings-- <= 0) {
1022 		return (NULL);
1023 	}
1024 	/* display progress about every three seconds */
1025 	if (++searchcount % 100 == 0) {
1026 		progress("%ld of %ld possible references retrieved",
1027 		    searchcount, postingsfound);
1028 	}
1029 	return (postingp++);
1030 }
1031 
1032 /* put the posting reference into the file */
1033 
1034 static void
1035 putpostingref(POSTING *p)
1036 {
1037 	static	char	function[PATLEN + 1];	/* function name */
1038 
1039 	if (p->fcnoffset == 0) {
1040 		*function = '\0';
1041 	} else if (p->fcnoffset != lastfcnoffset) {
1042 		if (dbseek(p->fcnoffset) != -1) {
1043 			getstring(function);
1044 			lastfcnoffset = p->fcnoffset;
1045 		}
1046 	}
1047 	if (dbseek(p->lineoffset) != -1) {
1048 		putref(srcfiles[p->fileindex], function);
1049 	}
1050 }
1051 
1052 /* put the reference into the file */
1053 
1054 static void
1055 putref(char *file, char *function)
1056 {
1057 	FILE	*output;
1058 
1059 	/* put global references first */
1060 	if (*function == '\0') {
1061 		function = "<global>";
1062 		output = refsfound;
1063 	} else {
1064 		output = nonglobalrefs;
1065 	}
1066 	if (fprintf(output, "%s %s ", filepath(file), function) == EOF) {
1067 		cannotwrite(temp1);
1068 		/* NOTREACHED */
1069 	}
1070 	putsource(output);
1071 }
1072 
1073 /* put the source line into the file */
1074 
1075 static void
1076 putsource(FILE *output)
1077 {
1078 	char	*cp, nextc = '\0';
1079 
1080 	if (fileversion <= 5) {
1081 		(void) scanpast(' ');
1082 		putline(output);
1083 		(void) putc('\n', output);
1084 		return;
1085 	}
1086 	/* scan back to the beginning of the source line */
1087 	cp = blockp;
1088 	while (*cp != '\n' || nextc != '\n') {
1089 		nextc = *cp;
1090 		if (--cp < block) {
1091 			/* read the previous block */
1092 			(void) dbseek((blocknumber - 1) * BUFSIZ);
1093 			cp = &block[BUFSIZ - 1];
1094 		}
1095 	}
1096 	/* there must be a double newline followed by a line number */
1097 	blockp = cp;
1098 	setmark(' ');	/* so getrefchar doesn't skip the last block char */
1099 	if (*blockp != '\n' || getrefchar() != '\n' ||
1100 	    !isdigit(getrefchar()) && fileversion >= 12) {
1101 		putmsg("Internal error: cannot get source line from database");
1102 		myexit(1);
1103 	}
1104 	/* until a double newline is found */
1105 	do {
1106 		/* skip a symbol type */
1107 		if (*blockp == '\t') {
1108 			skiprefchar();
1109 			skiprefchar();
1110 		}
1111 		/* output a piece of the source line */
1112 		putline(output);
1113 	} while (blockp != NULL && getrefchar() != '\n');
1114 	(void) putc('\n', output);
1115 }
1116 
1117 /* put the rest of the cross-reference line into the file */
1118 
1119 static void
1120 putline(FILE *output)
1121 {
1122 	char	*cp;
1123 	unsigned c;
1124 
1125 	setmark('\n');
1126 	cp = blockp;
1127 	do {
1128 		while ((c = *cp) != '\n') {
1129 			/* check for a compressed digraph */
1130 			if (c & 0200) {
1131 				c &= 0177;
1132 				(void) putc(dichar1[c / 8], output);
1133 				(void) putc(dichar2[c & 7], output);
1134 			} else if (c < ' ') {
1135 				/* a compressed keyword */
1136 				(void) fputs(keyword[c].text, output);
1137 				if (keyword[c].delim != '\0') {
1138 					(void) putc(' ', output);
1139 				}
1140 				if (keyword[c].delim == '(') {
1141 					(void) putc('(', output);
1142 				}
1143 			} else {
1144 				(void) putc((int)c, output);
1145 			}
1146 			++cp;
1147 		}
1148 	} while (*(cp + 1) == '\0' && (cp = readblock()) != NULL);
1149 	blockp = cp;
1150 }
1151 
1152 /* put the rest of the cross-reference line into the string */
1153 
1154 void
1155 getstring(char *s)
1156 {
1157 	char	*cp;
1158 	unsigned c;
1159 
1160 	setmark('\n');
1161 	cp = blockp;
1162 	do {
1163 		while ((c = *cp) != '\n') {
1164 			if (c & 0200) {
1165 				c &= 0177;
1166 				*s++ = dichar1[c / 8];
1167 				*s++ = dichar2[c & 7];
1168 			} else {
1169 				*s++ = (char)c;
1170 			}
1171 			++cp;
1172 		}
1173 	} while (*(cp + 1) == '\0' && (cp = readblock()) != NULL);
1174 	blockp = cp;
1175 	*s = '\0';
1176 }
1177 
1178 /* scan past the next occurence of this character in the cross-reference */
1179 
1180 char *
1181 scanpast(int c)
1182 {
1183 	char	*cp;
1184 
1185 	setmark(c);
1186 	cp = blockp;
1187 	do {	/* innermost loop optimized to only one test */
1188 		while (*cp != c) {
1189 			++cp;
1190 		}
1191 	} while (*(cp + 1) == '\0' && (cp = readblock()) != NULL);
1192 	blockp = cp;
1193 	if (cp != NULL) {
1194 		skiprefchar();	/* skip the found character */
1195 	}
1196 	return (blockp);
1197 }
1198 
1199 /* read a block of the cross-reference */
1200 
1201 char *
1202 readblock(void)
1203 {
1204 	/* read the next block */
1205 	blocklen = read(symrefs, block, BUFSIZ);
1206 	blockp = block;
1207 
1208 	/* add the search character and end-of-block mark */
1209 	block[blocklen] = blockmark;
1210 	block[blocklen + 1] = '\0';
1211 
1212 	/* return NULL on end-of-file */
1213 	if (blocklen == 0) {
1214 		blockp = NULL;
1215 	} else {
1216 		++blocknumber;
1217 	}
1218 	return (blockp);
1219 }
1220 
1221 /* seek to the database offset */
1222 
1223 long
1224 dbseek(long offset)
1225 {
1226 	long	n;
1227 	int	rc = 0;
1228 
1229 	if ((n = offset / BUFSIZ) != blocknumber) {
1230 		if ((rc = lseek(symrefs, n * BUFSIZ, 0)) == -1) {
1231 			myperror("Lseek failed");
1232 			(void) sleep(3);
1233 			return (rc);
1234 		}
1235 		(void) readblock();
1236 		blocknumber = n;
1237 	}
1238 	blockp = block + offset % BUFSIZ;
1239 	return (rc);
1240 }
1241 
1242 /* convert the string to lower case */
1243 
1244 static char *
1245 strtolower(char *s)
1246 {
1247 	static char buf[PATLEN + 1];
1248 	char *lp = buf;
1249 
1250 	while (*s != '\0') {
1251 		/*
1252 		 * note: s in not incremented in this line because the BSD
1253 		 * compatibility tolower macro evaluates its argument twice
1254 		 */
1255 		*lp++ = tolower(*s);
1256 		++s;
1257 	}
1258 	*lp = '\0';
1259 	return (buf);
1260 }
1261 
1262 /* if needed, convert a relative path to a full path */
1263 
1264 static char *
1265 filepath(char *file)
1266 {
1267 	static	char	path[PATHLEN + 1];
1268 	int	i;
1269 
1270 	if (*file != '/') {
1271 
1272 		/* if same file as last time, return the same path */
1273 		if (strequal(file, lastfilepath)) {
1274 			return (path);
1275 		}
1276 		(void) strcpy(lastfilepath, file);
1277 
1278 		/* if requested, prepend a path to a relative file path */
1279 		if (prependpath != NULL) {
1280 			(void) sprintf(path, "%s/%s", prependpath, file);
1281 			return (path);
1282 		}
1283 		/*
1284 		 * if the database was built with a view path, return a
1285 		 * full path so "cscope -d -f" does not have to be called
1286 		 * from the build directory with the same view path
1287 		 */
1288 		if (dbvpndirs > 1) {
1289 			for (i = 0; i < dbvpndirs; i++) {
1290 				(void) sprintf(path,
1291 				    "%s/%s", dbvpdirs[i], file);
1292 				if (access(path, READ) != -1) {
1293 					return (path);
1294 				}
1295 			}
1296 		}
1297 		(void) strcpy(path, file);	/* for lastfilepath check */
1298 	}
1299 	return (file);
1300 }
1301