2 75a8a1ec 2022-02-08 op * Copyright (c) 2021 Omar Polo <op@omarpolo.com>
4 75a8a1ec 2022-02-08 op * Permission to use, copy, modify, and distribute this software for any
5 75a8a1ec 2022-02-08 op * purpose with or without fee is hereby granted, provided that the above
6 75a8a1ec 2022-02-08 op * copyright notice and this permission notice appear in all copies.
8 75a8a1ec 2022-02-08 op * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 75a8a1ec 2022-02-08 op * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 75a8a1ec 2022-02-08 op * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 75a8a1ec 2022-02-08 op * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 75a8a1ec 2022-02-08 op * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 75a8a1ec 2022-02-08 op * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 75a8a1ec 2022-02-08 op * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 75a8a1ec 2022-02-08 op #include "compat.h"
19 75a8a1ec 2022-02-08 op #include <stdio.h>
20 75a8a1ec 2022-02-08 op #include <stdlib.h>
21 75a8a1ec 2022-02-08 op #include <string.h>
23 75a8a1ec 2022-02-08 op #include "parser.h"
24 75a8a1ec 2022-02-08 op #include "utils.h"
26 75a8a1ec 2022-02-08 op struct gm_selector {
28 75a8a1ec 2022-02-08 op const char *ds;
29 75a8a1ec 2022-02-08 op const char *selector;
30 75a8a1ec 2022-02-08 op const char *addr;
31 75a8a1ec 2022-02-08 op const char *port;
34 75a8a1ec 2022-02-08 op static void gm_parse_selector(char *, struct gm_selector *);
36 75a8a1ec 2022-02-08 op static int gm_parse(struct parser *, const char *, size_t);
37 75a8a1ec 2022-02-08 op static int gm_foreach_line(struct parser *, const char *, size_t);
38 75a8a1ec 2022-02-08 op static int gm_free(struct parser *);
39 0110411e 2022-04-13 op static int gm_serialize(struct parser *, FILE *);
42 75a8a1ec 2022-02-08 op gophermap_initparser(struct parser *p)
44 75a8a1ec 2022-02-08 op memset(p, 0, sizeof(*p));
46 75a8a1ec 2022-02-08 op p->name = "gophermap";
47 75a8a1ec 2022-02-08 op p->parse = &gm_parse;
48 75a8a1ec 2022-02-08 op p->free = &gm_free;
49 75a8a1ec 2022-02-08 op p->serialize = &gm_serialize;
51 75a8a1ec 2022-02-08 op TAILQ_INIT(&p->head);
55 75a8a1ec 2022-02-08 op gm_parse_selector(char *line, struct gm_selector *s)
57 75a8a1ec 2022-02-08 op s->type = *line++;
59 75a8a1ec 2022-02-08 op s->selector = "";
63 75a8a1ec 2022-02-08 op if ((line = strchr(line, '\t')) == NULL)
65 75a8a1ec 2022-02-08 op *line++ = '\0';
66 75a8a1ec 2022-02-08 op s->selector = line;
68 75a8a1ec 2022-02-08 op if ((line = strchr(line, '\t')) == NULL)
70 75a8a1ec 2022-02-08 op *line++ = '\0';
71 75a8a1ec 2022-02-08 op s->addr = line;
73 75a8a1ec 2022-02-08 op if ((line = strchr(line, '\t')) == NULL)
75 75a8a1ec 2022-02-08 op *line++ = '\0';
76 75a8a1ec 2022-02-08 op s->port = line;
80 75a8a1ec 2022-02-08 op gm_parse(struct parser *p, const char *buf, size_t size)
82 75a8a1ec 2022-02-08 op return parser_foreach_line(p, buf, size, gm_foreach_line);
85 75a8a1ec 2022-02-08 op static inline int
86 75a8a1ec 2022-02-08 op emit_line(struct parser *p, enum line_type type, struct gm_selector *s)
88 75a8a1ec 2022-02-08 op struct line *l;
89 75a8a1ec 2022-02-08 op char buf[LINE_MAX], b[2] = {0};
91 75a8a1ec 2022-02-08 op if ((l = calloc(1, sizeof(*l))) == NULL)
94 75a8a1ec 2022-02-08 op if ((l->line = strdup(s->ds)) == NULL)
97 75a8a1ec 2022-02-08 op switch (l->type = type) {
98 75a8a1ec 2022-02-08 op case LINE_LINK:
99 d89eb764 2022-04-24 op if (s->type == 'h' && !strncmp(s->selector, "URL:", 4)) {
100 75a8a1ec 2022-02-08 op strlcpy(buf, s->selector+4, sizeof(buf));
102 75a8a1ec 2022-02-08 op strlcpy(buf, "gopher://", sizeof(buf));
103 75a8a1ec 2022-02-08 op strlcat(buf, s->addr, sizeof(buf));
104 75a8a1ec 2022-02-08 op strlcat(buf, ":", sizeof(buf));
105 75a8a1ec 2022-02-08 op strlcat(buf, s->port, sizeof(buf));
106 75a8a1ec 2022-02-08 op strlcat(buf, "/", sizeof(buf));
107 75a8a1ec 2022-02-08 op b[0] = s->type;
108 75a8a1ec 2022-02-08 op strlcat(buf, b, sizeof(buf));
109 75a8a1ec 2022-02-08 op if (*s->selector != '/')
110 75a8a1ec 2022-02-08 op strlcat(buf, "/", sizeof(buf));
111 75a8a1ec 2022-02-08 op strlcat(buf, s->selector, sizeof(buf));
114 75a8a1ec 2022-02-08 op if ((l->alt = strdup(buf)) == NULL)
122 75a8a1ec 2022-02-08 op TAILQ_INSERT_TAIL(&p->head, l, lines);
127 75a8a1ec 2022-02-08 op if (l != NULL) {
128 75a8a1ec 2022-02-08 op free(l->line);
129 75a8a1ec 2022-02-08 op free(l->alt);
136 75a8a1ec 2022-02-08 op gm_foreach_line(struct parser *p, const char *line, size_t linelen)
138 75a8a1ec 2022-02-08 op char buf[LINE_MAX] = {0};
139 75a8a1ec 2022-02-08 op struct gm_selector s = {0};
141 75a8a1ec 2022-02-08 op memcpy(buf, line, MIN(sizeof(buf)-1, linelen));
142 75a8a1ec 2022-02-08 op gm_parse_selector(buf, &s);
144 75a8a1ec 2022-02-08 op switch (s.type) {
145 75a8a1ec 2022-02-08 op case '0': /* text file */
146 75a8a1ec 2022-02-08 op case '1': /* gopher submenu */
147 75a8a1ec 2022-02-08 op case '2': /* CCSO nameserver */
148 75a8a1ec 2022-02-08 op case '4': /* binhex-encoded file */
149 75a8a1ec 2022-02-08 op case '5': /* DOS file */
150 75a8a1ec 2022-02-08 op case '6': /* uuencoded file */
151 75a8a1ec 2022-02-08 op case '7': /* full-text search */
152 75a8a1ec 2022-02-08 op case '8': /* telnet */
153 75a8a1ec 2022-02-08 op case '9': /* binary file */
154 75a8a1ec 2022-02-08 op case '+': /* mirror or alternate server */
155 75a8a1ec 2022-02-08 op case 'g': /* gif */
156 75a8a1ec 2022-02-08 op case 'I': /* image */
157 75a8a1ec 2022-02-08 op case 'T': /* telnet 3270 */
158 75a8a1ec 2022-02-08 op case ':': /* gopher+: bitmap image */
159 75a8a1ec 2022-02-08 op case ';': /* gopher+: movie file */
160 75a8a1ec 2022-02-08 op case 'd': /* non-canonical: doc */
161 75a8a1ec 2022-02-08 op case 'h': /* non-canonical: html file */
162 75a8a1ec 2022-02-08 op case 's': /* non-canonical: sound file */
163 75a8a1ec 2022-02-08 op if (!emit_line(p, LINE_LINK, &s))
169 75a8a1ec 2022-02-08 op case 'i': /* non-canonical: message */
170 75a8a1ec 2022-02-08 op if (!emit_line(p, LINE_TEXT, &s))
174 75a8a1ec 2022-02-08 op case '3': /* error code */
175 75a8a1ec 2022-02-08 op if (!emit_line(p, LINE_QUOTE, &s))
184 75a8a1ec 2022-02-08 op gm_free(struct parser *p)
186 75a8a1ec 2022-02-08 op /* flush the buffer */
187 75a8a1ec 2022-02-08 op if (p->len != 0)
188 75a8a1ec 2022-02-08 op gm_foreach_line(p, p->buf, p->len);
190 75a8a1ec 2022-02-08 op free(p->buf);
195 75a8a1ec 2022-02-08 op static inline const char *
196 75a8a1ec 2022-02-08 op gopher_skip_selector(const char *path, int *ret_type)
198 75a8a1ec 2022-02-08 op *ret_type = 0;
200 75a8a1ec 2022-02-08 op if (!strcmp(path, "/") || *path == '\0') {
201 75a8a1ec 2022-02-08 op *ret_type = '1';
205 75a8a1ec 2022-02-08 op if (*path != '/')
209 75a8a1ec 2022-02-08 op switch (*ret_type = *path) {
216 75a8a1ec 2022-02-08 op *ret_type = 0;
221 75a8a1ec 2022-02-08 op return ++path;
225 0110411e 2022-04-13 op serialize_link(struct line *line, const char *text, FILE *fp)
227 75a8a1ec 2022-02-08 op size_t portlen = 0;
229 75a8a1ec 2022-02-08 op const char *uri, *endhost, *port, *path, *colon;
231 75a8a1ec 2022-02-08 op if ((uri = line->alt) == NULL)
234 d89eb764 2022-04-24 op if (strncmp(uri, "gopher://", 9) != 0)
235 0110411e 2022-04-13 op return fprintf(fp, "h%s\tURL:%s\terror.host\t1\n",
236 75a8a1ec 2022-02-08 op text, line->alt);
238 75a8a1ec 2022-02-08 op uri += 9; /* skip gopher:// */
240 75a8a1ec 2022-02-08 op path = strchr(uri, '/');
241 75a8a1ec 2022-02-08 op colon = strchr(uri, ':');
243 75a8a1ec 2022-02-08 op if (path != NULL && colon > path)
244 75a8a1ec 2022-02-08 op colon = NULL;
246 75a8a1ec 2022-02-08 op if ((endhost = colon) == NULL &&
247 75a8a1ec 2022-02-08 op (endhost = path) == NULL)
248 7ef6ceef 2022-05-19 op endhost = strchr(uri, '\0');
250 75a8a1ec 2022-02-08 op if (colon != NULL) {
251 75a8a1ec 2022-02-08 op for (port = colon+1; *port && *port != '/'; ++port)
253 75a8a1ec 2022-02-08 op port = colon+1;
259 75a8a1ec 2022-02-08 op if (path == NULL) {
263 75a8a1ec 2022-02-08 op path = gopher_skip_selector(path, &type);
265 0110411e 2022-04-13 op return fprintf(fp, "%c%s\t%s\t%.*s\t%.*s\n", type, text,
266 75a8a1ec 2022-02-08 op path, (int)(endhost - uri), uri, (int)portlen, port);
270 0110411e 2022-04-13 op gm_serialize(struct parser *p, FILE *fp)
272 75a8a1ec 2022-02-08 op struct line *line;
273 75a8a1ec 2022-02-08 op const char *text;
276 75a8a1ec 2022-02-08 op TAILQ_FOREACH(line, &p->head, lines) {
277 75a8a1ec 2022-02-08 op if ((text = line->line) == NULL)
280 75a8a1ec 2022-02-08 op switch (line->type) {
281 75a8a1ec 2022-02-08 op case LINE_LINK:
282 0110411e 2022-04-13 op r = serialize_link(line, text, fp);
285 75a8a1ec 2022-02-08 op case LINE_TEXT:
286 0110411e 2022-04-13 op r = fprintf(fp, "i%s\t\terror.host\t1\n", text);
289 75a8a1ec 2022-02-08 op case LINE_QUOTE:
290 0110411e 2022-04-13 op r = fprintf(fp, "3%s\t\terror.host\t1\n", text);
294 75a8a1ec 2022-02-08 op /* unreachable */