1 748b15fa 2021-03-12 op /*
 2 748b15fa 2021-03-12 op  * Copyright (c) 2021 Omar Polo <op@omarpolo.com>
 3 748b15fa 2021-03-12 op  *
 4 748b15fa 2021-03-12 op  * Permission to use, copy, modify, and distribute this software for any
 5 748b15fa 2021-03-12 op  * purpose with or without fee is hereby granted, provided that the above
 6 748b15fa 2021-03-12 op  * copyright notice and this permission notice appear in all copies.
 7 748b15fa 2021-03-12 op  *
 8 748b15fa 2021-03-12 op  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
 9 748b15fa 2021-03-12 op  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 748b15fa 2021-03-12 op  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 748b15fa 2021-03-12 op  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 748b15fa 2021-03-12 op  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 748b15fa 2021-03-12 op  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 748b15fa 2021-03-12 op  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15 748b15fa 2021-03-12 op  */
16 748b15fa 2021-03-12 op 
17 748b15fa 2021-03-12 op #include <ctype.h>
18 748b15fa 2021-03-12 op #include <fnmatch.h>
19 748b15fa 2021-03-12 op #include <string.h>
20 748b15fa 2021-03-12 op 
21 395b9f4e 2021-07-12 op #include "parser.h"
22 395b9f4e 2021-07-12 op #include "telescope.h"
23 395b9f4e 2021-07-12 op 
24 748b15fa 2021-03-12 op static int	check_for_utf8(char*);
25 748b15fa 2021-03-12 op 
26 748b15fa 2021-03-12 op static struct parser_table {
27 748b15fa 2021-03-12 op 	const char	*mediatype;
28 748b15fa 2021-03-12 op 	void		(*parserinit)(struct parser*);
29 748b15fa 2021-03-12 op } ptable[] = {
30 c49d61bc 2021-03-12 op 	{ "text/gemini",	gemtext_initparser },
31 c49d61bc 2021-03-12 op 	{ "text/*",		textplain_initparser },
32 748b15fa 2021-03-12 op 	{ NULL, NULL}
33 748b15fa 2021-03-12 op };
34 748b15fa 2021-03-12 op 
35 748b15fa 2021-03-12 op static int
36 748b15fa 2021-03-12 op check_for_utf8(char *b)
37 748b15fa 2021-03-12 op {
38 748b15fa 2021-03-12 op 	for (;;) {
39 748b15fa 2021-03-12 op 		while (*b != '\0' && isspace(*b))
40 748b15fa 2021-03-12 op 			b++;
41 748b15fa 2021-03-12 op 		if (*b == '\0')
42 748b15fa 2021-03-12 op 			break;
43 748b15fa 2021-03-12 op 		if (!has_prefix(b, "charset=")) {
44 748b15fa 2021-03-12 op 			while (*b != '\0' && *b != ';')
45 748b15fa 2021-03-12 op 				b++;
46 748b15fa 2021-03-12 op 			if (*b == '\0')
47 748b15fa 2021-03-12 op 				break;
48 748b15fa 2021-03-12 op 			b++;
49 748b15fa 2021-03-12 op 			continue;
50 748b15fa 2021-03-12 op 		}
51 748b15fa 2021-03-12 op 
52 748b15fa 2021-03-12 op 		/* is charset= */
53 748b15fa 2021-03-12 op 		b += strlen("charset=");
54 748b15fa 2021-03-12 op 		/* TODO: improve the matching */
55 748b15fa 2021-03-12 op 		return has_prefix(b, "ASCII") || has_prefix(b, "ascii") ||
56 748b15fa 2021-03-12 op 			has_prefix(b, "UTF-8") || has_prefix(b, "utf-8");
57 748b15fa 2021-03-12 op 	}
58 748b15fa 2021-03-12 op 
59 748b15fa 2021-03-12 op 	return 1;
60 748b15fa 2021-03-12 op }
61 748b15fa 2021-03-12 op 
62 748b15fa 2021-03-12 op int
63 748b15fa 2021-03-12 op setup_parser_for(struct tab *tab)
64 748b15fa 2021-03-12 op {
65 748b15fa 2021-03-12 op 	char			*b, buf[GEMINI_URL_LEN] = {0};
66 748b15fa 2021-03-12 op 	struct parser_table	*t;
67 748b15fa 2021-03-12 op 
68 748b15fa 2021-03-12 op         memcpy(buf, tab->meta, sizeof(tab->meta));
69 748b15fa 2021-03-12 op 
70 748b15fa 2021-03-12 op 	for (b = buf; *b != ';' && *b != '\0'; ++b)
71 748b15fa 2021-03-12 op 		;
72 748b15fa 2021-03-12 op 
73 748b15fa 2021-03-12 op 	if (*b == ';') {
74 748b15fa 2021-03-12 op 		*b = '\0';
75 748b15fa 2021-03-12 op 		++b;
76 748b15fa 2021-03-12 op 	}
77 748b15fa 2021-03-12 op 
78 748b15fa 2021-03-12 op 	if (!check_for_utf8(b))
79 748b15fa 2021-03-12 op 		return 0;
80 748b15fa 2021-03-12 op 
81 748b15fa 2021-03-12 op 	for (t = ptable; t->mediatype != NULL; ++t) {
82 748b15fa 2021-03-12 op 		if (!fnmatch(t->mediatype, buf, 0)) {
83 46f6e974 2021-05-17 op 			t->parserinit(&tab->buffer.page);
84 748b15fa 2021-03-12 op 			return 1;
85 748b15fa 2021-03-12 op 		}
86 748b15fa 2021-03-12 op 	}
87 748b15fa 2021-03-12 op 
88 748b15fa 2021-03-12 op 	return 0;
89 748b15fa 2021-03-12 op }