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 786e6deb 2021-07-21 op #include "compat.h"
18 786e6deb 2021-07-21 op 
19 748b15fa 2021-03-12 op #include <ctype.h>
20 748b15fa 2021-03-12 op #include <fnmatch.h>
21 748b15fa 2021-03-12 op #include <string.h>
22 748b15fa 2021-03-12 op 
23 395b9f4e 2021-07-12 op #include "parser.h"
24 395b9f4e 2021-07-12 op #include "telescope.h"
25 9d65b1d9 2022-01-11 op #include "utils.h"
26 395b9f4e 2021-07-12 op 
27 748b15fa 2021-03-12 op static int	check_for_utf8(char*);
28 748b15fa 2021-03-12 op 
29 0d877486 2022-02-24 op static const struct parser_table {
30 fd1c80ce 2024-06-14 op 	const char		*mediatype;
31 fd1c80ce 2024-06-14 op 	const struct parser	*parser;
32 748b15fa 2021-03-12 op } ptable[] = {
33 c1d27b0e 2024-06-14 op 	{ "text/gemini",	&gemtext_parser },
34 c1d27b0e 2024-06-14 op 	{ "text/x-patch",	&textpatch_parser },
35 c1d27b0e 2024-06-14 op 	{ "text/x-diff",	&textpatch_parser },
36 c1d27b0e 2024-06-14 op 	{ "application/x-patch",&textpatch_parser },
37 c1d27b0e 2024-06-14 op 	{ "text/*",		&textplain_parser },
38 748b15fa 2021-03-12 op 	{ NULL, NULL}
39 748b15fa 2021-03-12 op };
40 748b15fa 2021-03-12 op 
41 748b15fa 2021-03-12 op static int
42 748b15fa 2021-03-12 op check_for_utf8(char *b)
43 748b15fa 2021-03-12 op {
44 748b15fa 2021-03-12 op 	for (;;) {
45 748b15fa 2021-03-12 op 		while (*b != '\0' && isspace(*b))
46 748b15fa 2021-03-12 op 			b++;
47 748b15fa 2021-03-12 op 		if (*b == '\0')
48 748b15fa 2021-03-12 op 			break;
49 d89eb764 2022-04-24 op 		if (strncmp(b, "charset=", 8) != 0) {
50 748b15fa 2021-03-12 op 			while (*b != '\0' && *b != ';')
51 748b15fa 2021-03-12 op 				b++;
52 748b15fa 2021-03-12 op 			if (*b == '\0')
53 748b15fa 2021-03-12 op 				break;
54 748b15fa 2021-03-12 op 			b++;
55 748b15fa 2021-03-12 op 			continue;
56 748b15fa 2021-03-12 op 		}
57 748b15fa 2021-03-12 op 
58 748b15fa 2021-03-12 op 		/* is charset= */
59 748b15fa 2021-03-12 op 		b += strlen("charset=");
60 748b15fa 2021-03-12 op 		/* TODO: improve the matching */
61 d89eb764 2022-04-24 op 		return !strncmp(b, "ASCII", 5) || !strncmp(b, "ascii", 5) ||
62 d89eb764 2022-04-24 op 			!strncmp(b, "UTF-8", 5) || !strncmp(b, "utf-8", 5);
63 748b15fa 2021-03-12 op 	}
64 748b15fa 2021-03-12 op 
65 748b15fa 2021-03-12 op 	return 1;
66 748b15fa 2021-03-12 op }
67 748b15fa 2021-03-12 op 
68 748b15fa 2021-03-12 op int
69 748b15fa 2021-03-12 op setup_parser_for(struct tab *tab)
70 748b15fa 2021-03-12 op {
71 0d877486 2022-02-24 op 	char *b, buf[GEMINI_URL_LEN] = {0};
72 0d877486 2022-02-24 op 	const struct parser_table *t;
73 748b15fa 2021-03-12 op 
74 95a8c791 2021-08-26 op 	memcpy(buf, tab->meta, sizeof(tab->meta));
75 748b15fa 2021-03-12 op 
76 748b15fa 2021-03-12 op 	for (b = buf; *b != ';' && *b != '\0'; ++b)
77 748b15fa 2021-03-12 op 		;
78 748b15fa 2021-03-12 op 
79 748b15fa 2021-03-12 op 	if (*b == ';') {
80 748b15fa 2021-03-12 op 		*b = '\0';
81 748b15fa 2021-03-12 op 		++b;
82 748b15fa 2021-03-12 op 	}
83 748b15fa 2021-03-12 op 
84 748b15fa 2021-03-12 op 	if (!check_for_utf8(b))
85 748b15fa 2021-03-12 op 		return 0;
86 748b15fa 2021-03-12 op 
87 748b15fa 2021-03-12 op 	for (t = ptable; t->mediatype != NULL; ++t) {
88 748b15fa 2021-03-12 op 		if (!fnmatch(t->mediatype, buf, 0)) {
89 c1d27b0e 2024-06-14 op 			parser_init(&tab->buffer, t->parser);
90 748b15fa 2021-03-12 op 			return 1;
91 748b15fa 2021-03-12 op 		}
92 748b15fa 2021-03-12 op 	}
93 748b15fa 2021-03-12 op 
94 748b15fa 2021-03-12 op 	return 0;
95 748b15fa 2021-03-12 op }