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