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