1 /*
2  * Copyright (c) 2019, 2020 Tracey Emery <tracey@openbsd.org>
3  * Copyright (c) 2004, 2005 Esben Norby <norby@openbsd.org>
4  * Copyright (c) 2004 Ryan McBride <mcbride@openbsd.org>
5  * Copyright (c) 2002, 2003, 2004 Henning Brauer <henning@openbsd.org>
6  * Copyright (c) 2001 Markus Friedl.  All rights reserved.
7  * Copyright (c) 2001 Daniel Hartmeier.  All rights reserved.
8  * Copyright (c) 2001 Theo de Raadt.  All rights reserved.
9  *
10  * Permission to use, copy, modify, and distribute this software for any
11  * purpose with or without fee is hereby granted, provided that the above
12  * copyright notice and this permission notice appear in all copies.
13  *
14  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
15  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
16  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
17  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
18  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
19  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
20  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
21  */
22 
23 %{
24 #include <sys/types.h>
25 #include <sys/queue.h>
26 
27 #include <ctype.h>
28 #include <err.h>
29 #include <limits.h>
30 #include <stdarg.h>
31 #include <stdio.h>
32 #include <stdlib.h>
33 #include <string.h>
34 
35 #include "got_error.h"
36 #include "gotweb.h"
37 
38 TAILQ_HEAD(files, file)		 files = TAILQ_HEAD_INITIALIZER(files);
39 static struct file {
40 	TAILQ_ENTRY(file)	 entry;
41 	FILE			*stream;
42 	char			*name;
43 	size_t	 		 ungetpos;
44 	size_t			 ungetsize;
45 	u_char			*ungetbuf;
46 	int			 eof_reached;
47 	int			 lineno;
48 } *file, *topfile;
49 static const struct got_error*	pushfile(struct file**, const char *);
50 int		 popfile(void);
51 int		 yyparse(void);
52 int		 yylex(void);
53 int		 yyerror(const char *, ...)
54     __attribute__((__format__ (printf, 1, 2)))
55     __attribute__((__nonnull__ (1)));
56 int		 kw_cmp(const void *, const void *);
57 int		 lookup(char *);
58 int		 igetc(void);
59 int		 lgetc(int);
60 void		 lungetc(int);
61 int		 findeol(void);
62 
63 TAILQ_HEAD(symhead, sym)	 symhead = TAILQ_HEAD_INITIALIZER(symhead);
64 struct sym {
65 	TAILQ_ENTRY(sym)	 entry;
66 	int			 used;
67 	int			 persist;
68 	char			*nam;
69 	char			*val;
70 };
71 
72 int	 symset(const char *, const char *, int);
73 char	*symget(const char *);
74 
75 const struct got_error* gerror = NULL;
76 struct gotweb_config		*gw_conf;
77 
78 typedef struct {
79 	union {
80 		int64_t		 number;
81 		char		*string;
82 	} v;
83 	int lineno;
84 } YYSTYPE;
85 
86 %}
87 
88 %token	GOT_WWW_PATH GOT_MAX_REPOS GOT_SITE_NAME GOT_SITE_OWNER GOT_SITE_LINK
89 %token	GOT_LOGO GOT_LOGO_URL GOT_SHOW_REPO_OWNER GOT_SHOW_REPO_AGE
90 %token	GOT_SHOW_REPO_DESCRIPTION GOT_MAX_REPOS_DISPLAY GOT_REPOS_PATH
91 %token	GOT_MAX_COMMITS_DISPLAY ERROR GOT_SHOW_SITE_OWNER
92 %token	GOT_SHOW_REPO_CLONEURL
93 %token	<v.string>	STRING
94 %token	<v.number>	NUMBER
95 %type	<v.number>	boolean
96 %%
97 
98 grammar		: /* empty */
99 		| grammar '\n'
100 		| grammar main '\n'
101 		;
102 
103 boolean		: STRING {
104 			if (strcasecmp($1, "true") == 0 ||
105 			    strcasecmp($1, "on") == 0 ||
106 			    strcasecmp($1, "yes") == 0)
107 				$$ = 1;
108 			else if (strcasecmp($1, "false") == 0 ||
109 			    strcasecmp($1, "off") == 0 ||
110 			    strcasecmp($1, "no") == 0)
111 				$$ = 0;
112 			else {
113 				yyerror("invalid boolean value '%s'", $1);
114 				free($1);
115 				YYERROR;
116 			}
117 			free($1);
118 		}
119 		;
120 main		: GOT_REPOS_PATH STRING {
121 			gw_conf->got_repos_path = strdup($2);
122 			if (gw_conf->got_repos_path == NULL) {
123 				free($2);
124 				yyerror("strdup");
125 				YYERROR;
126 			}
127 			free($2);
128 		}
129 		| GOT_WWW_PATH STRING {
130 			gw_conf->got_www_path = strdup($2);
131 			if (gw_conf->got_www_path == NULL) {
132 				free($2);
133 				yyerror("strdup");
134 				YYERROR;
135 			}
136 			free($2);
137 		}
138 		| GOT_MAX_REPOS NUMBER {
139 			if ($2 > 0)
140 				gw_conf->got_max_repos = $2;
141 		}
142 		| GOT_SITE_NAME STRING {
143 			gw_conf->got_site_name = strdup($2);
144 			if (gw_conf->got_site_name == NULL) {
145 				free($2);
146 				yyerror("strdup");
147 				YYERROR;
148 			}
149 			free($2);
150 		}
151 		| GOT_SITE_OWNER STRING {
152 			gw_conf->got_site_owner = strdup($2);
153 			if (gw_conf->got_site_owner == NULL) {
154 				free($2);
155 				yyerror("strdup");
156 				YYERROR;
157 			}
158 			free($2);
159 		}
160 		| GOT_SITE_LINK STRING {
161 			gw_conf->got_site_link = strdup($2);
162 			if (gw_conf->got_site_link == NULL) {
163 				free($2);
164 				yyerror("strdup");
165 				YYERROR;
166 			}
167 			free($2);
168 		}
169 		| GOT_LOGO STRING {
170 			gw_conf->got_logo = strdup($2);
171 			if (gw_conf->got_logo== NULL) {
172 				free($2);
173 				yyerror("strdup");
174 				YYERROR;
175 			}
176 			free($2);
177 		}
178 		| GOT_LOGO_URL STRING {
179 			gw_conf->got_logo_url = strdup($2);
180 			if (gw_conf->got_logo_url== NULL) {
181 				free($2);
182 				yyerror("strdup");
183 				YYERROR;
184 			}
185 			free($2);
186 		}
187 		| GOT_SHOW_SITE_OWNER boolean {
188 			gw_conf->got_show_site_owner = $2;
189 		}
190 		| GOT_SHOW_REPO_OWNER boolean {
191 			gw_conf->got_show_repo_owner = $2;
192 		}
193 		| GOT_SHOW_REPO_AGE boolean {
194 			gw_conf->got_show_repo_age = $2;
195 		}
196 		| GOT_SHOW_REPO_DESCRIPTION boolean {
197 			gw_conf->got_show_repo_description = $2;
198 		}
199 		| GOT_SHOW_REPO_CLONEURL boolean {
200 			gw_conf->got_show_repo_cloneurl = $2;
201 		}
202 		| GOT_MAX_REPOS_DISPLAY NUMBER {
203 			if ($2 > 0)
204 				gw_conf->got_max_repos_display = $2;
205 		}
206 		| GOT_MAX_COMMITS_DISPLAY NUMBER {
207 			if ($2 > 0)
208 				gw_conf->got_max_commits_display = $2;
209 		}
210 		;
211 %%
212 
213 struct keywords {
214 	const char	*k_name;
215 	int		 k_val;
216 };
217 
218 int
219 yyerror(const char *fmt, ...)
220 {
221 	va_list		 ap;
222 	char		*msg;
223 	char		*err = NULL;
224 
225 	va_start(ap, fmt);
226 	if (vasprintf(&msg, fmt, ap) == -1) {
227 		gerror =  got_error_from_errno("vasprintf");
228 		return 0;
229 	}
230 	va_end(ap);
231 	if (asprintf(&err, "%s:%d: %s", file->name, yylval.lineno, msg) == -1) {
232 		gerror = got_error_from_errno("asprintf");
233 		return(0);
234 	}
235 	gerror = got_error_msg(GOT_ERR_PARSE_CONFIG, strdup(err));
236 	free(msg);
237 	free(err);
238 	return(0);
239 }
240 
241 int
242 kw_cmp(const void *k, const void *e)
243 {
244 	return (strcmp(k, ((const struct keywords *)e)->k_name));
245 }
246 
247 int
248 lookup(char *s)
249 {
250 	/* This has to be sorted always. */
251 	static const struct keywords keywords[] = {
252 		{ "got_logo",			GOT_LOGO },
253 		{ "got_logo_url",		GOT_LOGO_URL },
254 		{ "got_max_commits_display",	GOT_MAX_COMMITS_DISPLAY },
255 		{ "got_max_repos",		GOT_MAX_REPOS },
256 		{ "got_max_repos_display",	GOT_MAX_REPOS_DISPLAY },
257 		{ "got_repos_path",		GOT_REPOS_PATH },
258 		{ "got_show_repo_age",		GOT_SHOW_REPO_AGE },
259 		{ "got_show_repo_cloneurl",	GOT_SHOW_REPO_CLONEURL },
260 		{ "got_show_repo_description",	GOT_SHOW_REPO_DESCRIPTION },
261 		{ "got_show_repo_owner",	GOT_SHOW_REPO_OWNER },
262 		{ "got_show_site_owner",	GOT_SHOW_SITE_OWNER },
263 		{ "got_site_link",		GOT_SITE_LINK },
264 		{ "got_site_name",		GOT_SITE_NAME },
265 		{ "got_site_owner",		GOT_SITE_OWNER },
266 		{ "got_www_path",		GOT_WWW_PATH },
267 	};
268 	const struct keywords	*p;
269 
270 	p = bsearch(s, keywords, sizeof(keywords)/sizeof(keywords[0]),
271 	    sizeof(keywords[0]), kw_cmp);
272 
273 	if (p)
274 		return (p->k_val);
275 	else
276 		return (STRING);
277 }
278 
279 #define START_EXPAND	1
280 #define DONE_EXPAND	2
281 
282 static int	expanding;
283 
284 int
285 igetc(void)
286 {
287 	int	c;
288 
289 	while (1) {
290 		if (file->ungetpos > 0)
291 			c = file->ungetbuf[--file->ungetpos];
292 		else
293 			c = getc(file->stream);
294 
295 		if (c == START_EXPAND)
296 			expanding = 1;
297 		else if (c == DONE_EXPAND)
298 			expanding = 0;
299 		else
300 			break;
301 	}
302 	return (c);
303 }
304 
305 int
306 lgetc(int quotec)
307 {
308 	int		c, next;
309 
310 	if (quotec) {
311 		if ((c = igetc()) == EOF) {
312 			yyerror("reached end of file while parsing "
313 			    "quoted string");
314 			if (file == topfile || popfile() == EOF)
315 				return (EOF);
316 			return (quotec);
317 		}
318 		return (c);
319 	}
320 
321 	while ((c = igetc()) == '\\') {
322 		next = igetc();
323 		if (next != '\n') {
324 			c = next;
325 			break;
326 		}
327 		yylval.lineno = file->lineno;
328 		file->lineno++;
329 	}
330 
331 	if (c == EOF) {
332 		/*
333 		 * Fake EOL when hit EOF for the first time. This gets line
334 		 * count right if last line in included file is syntactically
335 		 * invalid and has no newline.
336 		 */
337 		if (file->eof_reached == 0) {
338 			file->eof_reached = 1;
339 			return ('\n');
340 		}
341 		while (c == EOF) {
342 			if (file == topfile || popfile() == EOF)
343 				return (EOF);
344 			c = igetc();
345 		}
346 	}
347 	return (c);
348 }
349 
350 void
351 lungetc(int c)
352 {
353 	if (c == EOF)
354 		return;
355 
356 	if (file->ungetpos >= file->ungetsize) {
357 		void *p = reallocarray(file->ungetbuf, file->ungetsize, 2);
358 		if (p == NULL)
359 			err(1, "%s", __func__);
360 		file->ungetbuf = p;
361 		file->ungetsize *= 2;
362 	}
363 	file->ungetbuf[file->ungetpos++] = c;
364 }
365 
366 int
367 findeol(void)
368 {
369 	int	c;
370 
371 	/* Skip to either EOF or the first real EOL. */
372 	while (1) {
373 		c = lgetc(0);
374 		if (c == '\n') {
375 			file->lineno++;
376 			break;
377 		}
378 		if (c == EOF)
379 			break;
380 	}
381 	return (ERROR);
382 }
383 
384 int
385 yylex(void)
386 {
387 	unsigned char	 buf[8096];
388 	unsigned char	*p, *val;
389 	int		 quotec, next, c;
390 	int		 token;
391 
392 top:
393 	p = buf;
394 	while ((c = lgetc(0)) == ' ' || c == '\t')
395 		; /* nothing */
396 
397 	yylval.lineno = file->lineno;
398 	if (c == '#')
399 		while ((c = lgetc(0)) != '\n' && c != EOF)
400 			; /* nothing */
401 	if (c == '$' && !expanding) {
402 		while (1) {
403 			if ((c = lgetc(0)) == EOF)
404 				return (0);
405 
406 			if (p + 1 >= buf + sizeof(buf) - 1) {
407 				yyerror("string too long");
408 				return (findeol());
409 			}
410 			if (isalnum(c) || c == '_') {
411 				*p++ = c;
412 				continue;
413 			}
414 			*p = '\0';
415 			lungetc(c);
416 			break;
417 		}
418 		val = symget(buf);
419 		if (val == NULL) {
420 			yyerror("macro '%s' not defined", buf);
421 			return (findeol());
422 		}
423 		p = val + strlen(val) - 1;
424 		lungetc(DONE_EXPAND);
425 		while (p >= val) {
426 			lungetc(*p);
427 			p--;
428 		}
429 		lungetc(START_EXPAND);
430 		goto top;
431 	}
432 
433 	switch (c) {
434 	case '\'':
435 	case '"':
436 		quotec = c;
437 		while (1) {
438 			if ((c = lgetc(quotec)) == EOF)
439 				return (0);
440 			if (c == '\n') {
441 				file->lineno++;
442 				continue;
443 			} else if (c == '\\') {
444 				if ((next = lgetc(quotec)) == EOF)
445 					return (0);
446 				if (next == quotec || c == ' ' || c == '\t')
447 					c = next;
448 				else if (next == '\n') {
449 					file->lineno++;
450 					continue;
451 				} else
452 					lungetc(next);
453 			} else if (c == quotec) {
454 				*p = '\0';
455 				break;
456 			} else if (c == '\0') {
457 				yyerror("syntax error");
458 				return (findeol());
459 			}
460 			if (p + 1 >= buf + sizeof(buf) - 1) {
461 				yyerror("string too long");
462 				return (findeol());
463 			}
464 			*p++ = c;
465 		}
466 		yylval.v.string = strdup(buf);
467 		if (yylval.v.string == NULL)
468 			err(1, "%s", __func__);
469 		return (STRING);
470 	}
471 
472 #define allowed_to_end_number(x) \
473 	(isspace(x) || x == ')' || x ==',' || x == '/' || x == '}' || x == '=')
474 
475 	if (c == '-' || isdigit(c)) {
476 		do {
477 			*p++ = c;
478 			if ((unsigned)(p-buf) >= sizeof(buf)) {
479 				yyerror("string too long");
480 				return (findeol());
481 			}
482 		} while ((c = lgetc(0)) != EOF && isdigit(c));
483 		lungetc(c);
484 		if (p == buf + 1 && buf[0] == '-')
485 			goto nodigits;
486 		if (c == EOF || allowed_to_end_number(c)) {
487 			const char *errstr = NULL;
488 
489 			*p = '\0';
490 			yylval.v.number = strtonum(buf, LLONG_MIN,
491 			    LLONG_MAX, &errstr);
492 			if (errstr) {
493 				yyerror("\"%s\" invalid number: %s",
494 				    buf, errstr);
495 				return (findeol());
496 			}
497 			return (NUMBER);
498 		} else {
499 nodigits:
500 			while (p > buf + 1)
501 				lungetc(*--p);
502 			c = *--p;
503 			if (c == '-')
504 				return (c);
505 		}
506 	}
507 
508 #define allowed_in_string(x) \
509 	(isalnum(x) || (ispunct(x) && x != '(' && x != ')' && \
510 	x != '{' && x != '}' && \
511 	x != '!' && x != '=' && x != '#' && \
512 	x != ','))
513 
514 	if (isalnum(c) || c == ':' || c == '_') {
515 		do {
516 			*p++ = c;
517 			if ((unsigned)(p-buf) >= sizeof(buf)) {
518 				yyerror("string too long");
519 				return (findeol());
520 			}
521 		} while ((c = lgetc(0)) != EOF && (allowed_in_string(c)));
522 		lungetc(c);
523 		*p = '\0';
524 		if ((token = lookup(buf)) == STRING)
525 			if ((yylval.v.string = strdup(buf)) == NULL)
526 				err(1, "%s", __func__);
527 		return (token);
528 	}
529 	if (c == '\n') {
530 		yylval.lineno = file->lineno;
531 		file->lineno++;
532 	}
533 	if (c == EOF)
534 		return (0);
535 	return (c);
536 }
537 
538 static const struct got_error*
539 pushfile(struct file **nfile, const char *name)
540 {
541 	const struct got_error* error = NULL;
542 
543 	if (((*nfile) = calloc(1, sizeof(struct file))) == NULL)
544 		return got_error_from_errno2(__func__, "calloc");
545 	if (((*nfile)->name = strdup(name)) == NULL) {
546 		free(nfile);
547 		return got_error_from_errno2(__func__, "strdup");
548 	}
549 	if (((*nfile)->stream = fopen((*nfile)->name, "r")) == NULL) {
550 		char *msg = NULL;
551 		if (asprintf(&msg, "%s", (*nfile)->name) == -1)
552 			return got_error_from_errno("asprintf");
553 		error = got_error_msg(GOT_ERR_NO_CONFIG_FILE, msg);
554 		free((*nfile)->name);
555 		free((*nfile));
556 		free(msg);
557 		return error;
558 	}
559 	(*nfile)->lineno = TAILQ_EMPTY(&files) ? 1 : 0;
560 	(*nfile)->ungetsize = 16;
561 	(*nfile)->ungetbuf = malloc((*nfile)->ungetsize);
562 	if ((*nfile)->ungetbuf == NULL) {
563 		fclose((*nfile)->stream);
564 		free((*nfile)->name);
565 		free((*nfile));
566 		return got_error_from_errno2(__func__, "malloc");
567 	}
568 	TAILQ_INSERT_TAIL(&files, (*nfile), entry);
569 	return error;
570 }
571 
572 int
573 popfile(void)
574 {
575 	struct file	*prev = NULL;
576 
577 	TAILQ_REMOVE(&files, file, entry);
578 	fclose(file->stream);
579 	free(file->name);
580 	free(file->ungetbuf);
581 	free(file);
582 	file = prev;
583 	return (file ? 0 : EOF);
584 }
585 
586 const struct got_error*
587 parse_gotweb_config(struct gotweb_config **gconf, const char *filename)
588 {
589 	gw_conf = malloc(sizeof(struct gotweb_config));
590 	if (gw_conf == NULL) {
591 		gerror = got_error_from_errno("malloc");
592 		goto done;
593 	}
594 	gw_conf->got_repos_path = strdup(D_GOTPATH);
595 	if (gw_conf->got_repos_path == NULL) {
596 		gerror = got_error_from_errno("strdup");
597 		goto done;
598 	}
599 	gw_conf->got_www_path = strdup(D_GOTWWW);
600 	if (gw_conf->got_www_path == NULL) {
601 		gerror = got_error_from_errno("strdup");
602 		goto done;
603 	}
604 	gw_conf->got_site_name = strdup(D_SITENAME);
605 	if (gw_conf->got_site_name == NULL) {
606 		gerror = got_error_from_errno("strdup");
607 		goto done;
608 	}
609 	gw_conf->got_site_owner = strdup(D_SITEOWNER);
610 	if (gw_conf->got_site_owner == NULL) {
611 		gerror = got_error_from_errno("strdup");
612 		goto done;
613 	}
614 	gw_conf->got_site_link = strdup(D_SITELINK);
615 	if (gw_conf->got_site_link == NULL) {
616 		gerror = got_error_from_errno("strdup");
617 		goto done;
618 	}
619 	gw_conf->got_logo = strdup(D_GOTLOGO);
620 	if (gw_conf->got_logo == NULL) {
621 		gerror = got_error_from_errno("strdup");
622 		goto done;
623 	}
624 	gw_conf->got_logo_url = strdup(D_GOTURL);
625 	if (gw_conf->got_logo_url == NULL) {
626 		gerror = got_error_from_errno("strdup");
627 		goto done;
628 	}
629 	gw_conf->got_show_site_owner = D_SHOWSOWNER;
630 	gw_conf->got_show_repo_owner = D_SHOWROWNER;
631 	gw_conf->got_show_repo_age = D_SHOWAGE;
632 	gw_conf->got_show_repo_description = D_SHOWDESC;
633 	gw_conf->got_show_repo_cloneurl = D_SHOWURL;
634 	gw_conf->got_max_repos = D_MAXREPO;
635 	gw_conf->got_max_repos_display = D_MAXREPODISP;
636 	gw_conf->got_max_commits_display = D_MAXCOMMITDISP;
637 
638 	/*
639 	 * We don't require that the gotweb config file exists
640 	 * So reset gerror if it doesn't exist and goto done.
641 	 */
642 	gerror = pushfile(&file, filename);
643 	if (gerror && gerror->code == GOT_ERR_NO_CONFIG_FILE) {
644 		gerror = NULL;
645 		goto done;
646 	} else if (gerror)
647 		return gerror;
648 	topfile = file;
649 
650 	yyparse();
651 	popfile();
652 done:
653 	*gconf = gw_conf;
654 	return gerror;
655 }
656 
657 int
658 symset(const char *nam, const char *val, int persist)
659 {
660 	struct sym	*sym;
661 
662 	TAILQ_FOREACH(sym, &symhead, entry) {
663 		if (strcmp(nam, sym->nam) == 0)
664 			break;
665 	}
666 
667 	if (sym != NULL) {
668 		if (sym->persist == 1)
669 			return (0);
670 		else {
671 			free(sym->nam);
672 			free(sym->val);
673 			TAILQ_REMOVE(&symhead, sym, entry);
674 			free(sym);
675 		}
676 	}
677 	if ((sym = calloc(1, sizeof(*sym))) == NULL)
678 		return (-1);
679 
680 	sym->nam = strdup(nam);
681 	if (sym->nam == NULL) {
682 		free(sym);
683 		return (-1);
684 	}
685 	sym->val = strdup(val);
686 	if (sym->val == NULL) {
687 		free(sym->nam);
688 		free(sym);
689 		return (-1);
690 	}
691 	sym->used = 0;
692 	sym->persist = persist;
693 	TAILQ_INSERT_TAIL(&symhead, sym, entry);
694 	return (0);
695 }
696 
697 int
698 cmdline_symset(char *s)
699 {
700 	char	*sym, *val;
701 	int	ret;
702 	size_t	len;
703 
704 	if ((val = strrchr(s, '=')) == NULL)
705 		return (-1);
706 
707 	len = strlen(s) - strlen(val) + 1;
708 	if ((sym = malloc(len)) == NULL)
709 		errx(1, "cmdline_symset: malloc");
710 
711 	strlcpy(sym, s, len);
712 
713 	ret = symset(sym, val + 1, 1);
714 	free(sym);
715 
716 	return (ret);
717 }
718 
719 char *
720 symget(const char *nam)
721 {
722 	struct sym	*sym;
723 
724 	TAILQ_FOREACH(sym, &symhead, entry) {
725 		if (strcmp(nam, sym->nam) == 0) {
726 			sym->used = 1;
727 			return (sym->val);
728 		}
729 	}
730 	return (NULL);
731 }