Blame


1 75a8a1ec 2022-02-08 op /*
2 75a8a1ec 2022-02-08 op * Copyright (c) 2021 Omar Polo <op@omarpolo.com>
3 75a8a1ec 2022-02-08 op *
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.
7 75a8a1ec 2022-02-08 op *
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.
15 75a8a1ec 2022-02-08 op */
16 75a8a1ec 2022-02-08 op
17 75a8a1ec 2022-02-08 op #include "compat.h"
18 75a8a1ec 2022-02-08 op
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>
22 75a8a1ec 2022-02-08 op
23 75a8a1ec 2022-02-08 op #include "parser.h"
24 75a8a1ec 2022-02-08 op #include "utils.h"
25 75a8a1ec 2022-02-08 op
26 75a8a1ec 2022-02-08 op struct gm_selector {
27 75a8a1ec 2022-02-08 op char type;
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;
32 75a8a1ec 2022-02-08 op };
33 75a8a1ec 2022-02-08 op
34 75a8a1ec 2022-02-08 op static void gm_parse_selector(char *, struct gm_selector *);
35 75a8a1ec 2022-02-08 op
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 *);
40 75a8a1ec 2022-02-08 op
41 75a8a1ec 2022-02-08 op void
42 75a8a1ec 2022-02-08 op gophermap_initparser(struct parser *p)
43 75a8a1ec 2022-02-08 op {
44 75a8a1ec 2022-02-08 op memset(p, 0, sizeof(*p));
45 75a8a1ec 2022-02-08 op
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;
50 75a8a1ec 2022-02-08 op
51 75a8a1ec 2022-02-08 op TAILQ_INIT(&p->head);
52 75a8a1ec 2022-02-08 op }
53 75a8a1ec 2022-02-08 op
54 75a8a1ec 2022-02-08 op static void
55 75a8a1ec 2022-02-08 op gm_parse_selector(char *line, struct gm_selector *s)
56 75a8a1ec 2022-02-08 op {
57 75a8a1ec 2022-02-08 op s->type = *line++;
58 75a8a1ec 2022-02-08 op s->ds = line;
59 75a8a1ec 2022-02-08 op s->selector = "";
60 75a8a1ec 2022-02-08 op s->addr = "";
61 75a8a1ec 2022-02-08 op s->port = "";
62 75a8a1ec 2022-02-08 op
63 75a8a1ec 2022-02-08 op if ((line = strchr(line, '\t')) == NULL)
64 75a8a1ec 2022-02-08 op return;
65 75a8a1ec 2022-02-08 op *line++ = '\0';
66 75a8a1ec 2022-02-08 op s->selector = line;
67 75a8a1ec 2022-02-08 op
68 75a8a1ec 2022-02-08 op if ((line = strchr(line, '\t')) == NULL)
69 75a8a1ec 2022-02-08 op return;
70 75a8a1ec 2022-02-08 op *line++ = '\0';
71 75a8a1ec 2022-02-08 op s->addr = line;
72 75a8a1ec 2022-02-08 op
73 75a8a1ec 2022-02-08 op if ((line = strchr(line, '\t')) == NULL)
74 75a8a1ec 2022-02-08 op return;
75 75a8a1ec 2022-02-08 op *line++ = '\0';
76 75a8a1ec 2022-02-08 op s->port = line;
77 75a8a1ec 2022-02-08 op }
78 75a8a1ec 2022-02-08 op
79 75a8a1ec 2022-02-08 op static int
80 75a8a1ec 2022-02-08 op gm_parse(struct parser *p, const char *buf, size_t size)
81 75a8a1ec 2022-02-08 op {
82 75a8a1ec 2022-02-08 op return parser_foreach_line(p, buf, size, gm_foreach_line);
83 75a8a1ec 2022-02-08 op }
84 75a8a1ec 2022-02-08 op
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)
87 75a8a1ec 2022-02-08 op {
88 75a8a1ec 2022-02-08 op struct line *l;
89 75a8a1ec 2022-02-08 op char buf[LINE_MAX], b[2] = {0};
90 75a8a1ec 2022-02-08 op
91 75a8a1ec 2022-02-08 op if ((l = calloc(1, sizeof(*l))) == NULL)
92 75a8a1ec 2022-02-08 op goto err;
93 75a8a1ec 2022-02-08 op
94 75a8a1ec 2022-02-08 op if ((l->line = strdup(s->ds)) == NULL)
95 75a8a1ec 2022-02-08 op goto err;
96 75a8a1ec 2022-02-08 op
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));
101 75a8a1ec 2022-02-08 op } else {
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));
112 75a8a1ec 2022-02-08 op }
113 75a8a1ec 2022-02-08 op
114 75a8a1ec 2022-02-08 op if ((l->alt = strdup(buf)) == NULL)
115 75a8a1ec 2022-02-08 op goto err;
116 75a8a1ec 2022-02-08 op break;
117 75a8a1ec 2022-02-08 op
118 75a8a1ec 2022-02-08 op default:
119 75a8a1ec 2022-02-08 op break;
120 75a8a1ec 2022-02-08 op }
121 75a8a1ec 2022-02-08 op
122 75a8a1ec 2022-02-08 op TAILQ_INSERT_TAIL(&p->head, l, lines);
123 75a8a1ec 2022-02-08 op
124 75a8a1ec 2022-02-08 op return 1;
125 75a8a1ec 2022-02-08 op
126 75a8a1ec 2022-02-08 op err:
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);
130 75a8a1ec 2022-02-08 op free(l);
131 75a8a1ec 2022-02-08 op }
132 75a8a1ec 2022-02-08 op return 0;
133 75a8a1ec 2022-02-08 op }
134 75a8a1ec 2022-02-08 op
135 75a8a1ec 2022-02-08 op static int
136 75a8a1ec 2022-02-08 op gm_foreach_line(struct parser *p, const char *line, size_t linelen)
137 75a8a1ec 2022-02-08 op {
138 75a8a1ec 2022-02-08 op char buf[LINE_MAX] = {0};
139 75a8a1ec 2022-02-08 op struct gm_selector s = {0};
140 75a8a1ec 2022-02-08 op
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);
143 75a8a1ec 2022-02-08 op
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))
164 75a8a1ec 2022-02-08 op return 0;
165 75a8a1ec 2022-02-08 op break;
166 75a8a1ec 2022-02-08 op
167 75a8a1ec 2022-02-08 op break;
168 75a8a1ec 2022-02-08 op
169 75a8a1ec 2022-02-08 op case 'i': /* non-canonical: message */
170 75a8a1ec 2022-02-08 op if (!emit_line(p, LINE_TEXT, &s))
171 75a8a1ec 2022-02-08 op return 0;
172 75a8a1ec 2022-02-08 op break;
173 75a8a1ec 2022-02-08 op
174 75a8a1ec 2022-02-08 op case '3': /* error code */
175 75a8a1ec 2022-02-08 op if (!emit_line(p, LINE_QUOTE, &s))
176 75a8a1ec 2022-02-08 op return 0;
177 75a8a1ec 2022-02-08 op break;
178 75a8a1ec 2022-02-08 op }
179 75a8a1ec 2022-02-08 op
180 75a8a1ec 2022-02-08 op return 1;
181 75a8a1ec 2022-02-08 op }
182 75a8a1ec 2022-02-08 op
183 75a8a1ec 2022-02-08 op static int
184 75a8a1ec 2022-02-08 op gm_free(struct parser *p)
185 75a8a1ec 2022-02-08 op {
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);
189 75a8a1ec 2022-02-08 op
190 75a8a1ec 2022-02-08 op free(p->buf);
191 75a8a1ec 2022-02-08 op
192 75a8a1ec 2022-02-08 op return 1;
193 75a8a1ec 2022-02-08 op }
194 75a8a1ec 2022-02-08 op
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)
197 75a8a1ec 2022-02-08 op {
198 75a8a1ec 2022-02-08 op *ret_type = 0;
199 75a8a1ec 2022-02-08 op
200 75a8a1ec 2022-02-08 op if (!strcmp(path, "/") || *path == '\0') {
201 75a8a1ec 2022-02-08 op *ret_type = '1';
202 75a8a1ec 2022-02-08 op return path;
203 75a8a1ec 2022-02-08 op }
204 75a8a1ec 2022-02-08 op
205 75a8a1ec 2022-02-08 op if (*path != '/')
206 75a8a1ec 2022-02-08 op return path;
207 75a8a1ec 2022-02-08 op path++;
208 75a8a1ec 2022-02-08 op
209 75a8a1ec 2022-02-08 op switch (*ret_type = *path) {
210 75a8a1ec 2022-02-08 op case '0':
211 75a8a1ec 2022-02-08 op case '1':
212 75a8a1ec 2022-02-08 op case '7':
213 75a8a1ec 2022-02-08 op break;
214 75a8a1ec 2022-02-08 op
215 75a8a1ec 2022-02-08 op default:
216 75a8a1ec 2022-02-08 op *ret_type = 0;
217 75a8a1ec 2022-02-08 op path -= 1;
218 75a8a1ec 2022-02-08 op return path;
219 75a8a1ec 2022-02-08 op }
220 75a8a1ec 2022-02-08 op
221 75a8a1ec 2022-02-08 op return ++path;
222 75a8a1ec 2022-02-08 op }
223 75a8a1ec 2022-02-08 op
224 75a8a1ec 2022-02-08 op static int
225 0110411e 2022-04-13 op serialize_link(struct line *line, const char *text, FILE *fp)
226 75a8a1ec 2022-02-08 op {
227 75a8a1ec 2022-02-08 op size_t portlen = 0;
228 75a8a1ec 2022-02-08 op int type;
229 75a8a1ec 2022-02-08 op const char *uri, *endhost, *port, *path, *colon;
230 75a8a1ec 2022-02-08 op
231 75a8a1ec 2022-02-08 op if ((uri = line->alt) == NULL)
232 75a8a1ec 2022-02-08 op return -1;
233 75a8a1ec 2022-02-08 op
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);
237 75a8a1ec 2022-02-08 op
238 75a8a1ec 2022-02-08 op uri += 9; /* skip gopher:// */
239 75a8a1ec 2022-02-08 op
240 75a8a1ec 2022-02-08 op path = strchr(uri, '/');
241 75a8a1ec 2022-02-08 op colon = strchr(uri, ':');
242 75a8a1ec 2022-02-08 op
243 75a8a1ec 2022-02-08 op if (path != NULL && colon > path)
244 75a8a1ec 2022-02-08 op colon = NULL;
245 75a8a1ec 2022-02-08 op
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');
249 75a8a1ec 2022-02-08 op
250 75a8a1ec 2022-02-08 op if (colon != NULL) {
251 75a8a1ec 2022-02-08 op for (port = colon+1; *port && *port != '/'; ++port)
252 75a8a1ec 2022-02-08 op ++portlen;
253 75a8a1ec 2022-02-08 op port = colon+1;
254 75a8a1ec 2022-02-08 op } else {
255 75a8a1ec 2022-02-08 op port = "70";
256 75a8a1ec 2022-02-08 op portlen = 2;
257 75a8a1ec 2022-02-08 op }
258 75a8a1ec 2022-02-08 op
259 75a8a1ec 2022-02-08 op if (path == NULL) {
260 75a8a1ec 2022-02-08 op type = '1';
261 75a8a1ec 2022-02-08 op path = "";
262 75a8a1ec 2022-02-08 op } else
263 75a8a1ec 2022-02-08 op path = gopher_skip_selector(path, &type);
264 75a8a1ec 2022-02-08 op
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);
267 75a8a1ec 2022-02-08 op }
268 75a8a1ec 2022-02-08 op
269 75a8a1ec 2022-02-08 op static int
270 0110411e 2022-04-13 op gm_serialize(struct parser *p, FILE *fp)
271 75a8a1ec 2022-02-08 op {
272 75a8a1ec 2022-02-08 op struct line *line;
273 75a8a1ec 2022-02-08 op const char *text;
274 75a8a1ec 2022-02-08 op int r;
275 75a8a1ec 2022-02-08 op
276 75a8a1ec 2022-02-08 op TAILQ_FOREACH(line, &p->head, lines) {
277 75a8a1ec 2022-02-08 op if ((text = line->line) == NULL)
278 75a8a1ec 2022-02-08 op text = "";
279 75a8a1ec 2022-02-08 op
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);
283 75a8a1ec 2022-02-08 op break;
284 75a8a1ec 2022-02-08 op
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);
287 75a8a1ec 2022-02-08 op break;
288 75a8a1ec 2022-02-08 op
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);
291 75a8a1ec 2022-02-08 op break;
292 75a8a1ec 2022-02-08 op
293 75a8a1ec 2022-02-08 op default:
294 75a8a1ec 2022-02-08 op /* unreachable */
295 75a8a1ec 2022-02-08 op abort();
296 75a8a1ec 2022-02-08 op }
297 75a8a1ec 2022-02-08 op
298 75a8a1ec 2022-02-08 op if (r == -1)
299 75a8a1ec 2022-02-08 op return 0;
300 75a8a1ec 2022-02-08 op }
301 75a8a1ec 2022-02-08 op
302 75a8a1ec 2022-02-08 op return 1;
303 75a8a1ec 2022-02-08 op }