1 1ac119fb 2024-01-23 op /*
  2 62c0f697 2024-02-21 op  * Copyright (c) 2021, 2024 Omar Polo <op@omarpolo.com>
  3 1ac119fb 2024-01-23 op  *
  4 1ac119fb 2024-01-23 op  * Permission to use, copy, modify, and distribute this software for any
  5 1ac119fb 2024-01-23 op  * purpose with or without fee is hereby granted, provided that the above
  6 1ac119fb 2024-01-23 op  * copyright notice and this permission notice appear in all copies.
  7 1ac119fb 2024-01-23 op  *
  8 1ac119fb 2024-01-23 op  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  9 1ac119fb 2024-01-23 op  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
 10 1ac119fb 2024-01-23 op  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
 11 1ac119fb 2024-01-23 op  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
 12 1ac119fb 2024-01-23 op  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
 13 1ac119fb 2024-01-23 op  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
 14 1ac119fb 2024-01-23 op  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 15 1ac119fb 2024-01-23 op  */
 16 1ac119fb 2024-01-23 op 
 17 1ac119fb 2024-01-23 op #include "compat.h"
 18 1ac119fb 2024-01-23 op 
 19 1ac119fb 2024-01-23 op #include <stdio.h>
 20 1ac119fb 2024-01-23 op #include <stdlib.h>
 21 1ac119fb 2024-01-23 op #include <string.h>
 22 1ac119fb 2024-01-23 op 
 23 1ac119fb 2024-01-23 op #include "parser.h"
 24 62c0f697 2024-02-21 op #include "iri.h"
 25 1ac119fb 2024-01-23 op #include "utils.h"
 26 e5e04904 2024-02-06 op 
 27 e5e04904 2024-02-06 op #ifndef LINE_MAX
 28 e5e04904 2024-02-06 op #define LINE_MAX 2048
 29 e5e04904 2024-02-06 op #endif
 30 1ac119fb 2024-01-23 op 
 31 1ac119fb 2024-01-23 op struct gm_selector {
 32 1ac119fb 2024-01-23 op 	char		 type;
 33 1ac119fb 2024-01-23 op 	const char	*ds;
 34 1ac119fb 2024-01-23 op 	const char	*selector;
 35 1ac119fb 2024-01-23 op 	const char	*addr;
 36 1ac119fb 2024-01-23 op 	const char	*port;
 37 1ac119fb 2024-01-23 op };
 38 1ac119fb 2024-01-23 op 
 39 1ac119fb 2024-01-23 op static void	gm_parse_selector(char *, struct gm_selector *);
 40 1ac119fb 2024-01-23 op 
 41 1ac119fb 2024-01-23 op static int	gm_parse(struct parser *, const char *, size_t);
 42 1ac119fb 2024-01-23 op static int	gm_foreach_line(struct parser *, const char *, size_t);
 43 1ac119fb 2024-01-23 op static int	gm_free(struct parser *);
 44 1ac119fb 2024-01-23 op static int	gm_serialize(struct parser *, FILE *);
 45 1ac119fb 2024-01-23 op 
 46 1ac119fb 2024-01-23 op void
 47 1ac119fb 2024-01-23 op gophermap_initparser(struct parser *p)
 48 1ac119fb 2024-01-23 op {
 49 1ac119fb 2024-01-23 op 	memset(p, 0, sizeof(*p));
 50 1ac119fb 2024-01-23 op 
 51 1ac119fb 2024-01-23 op 	p->name = "gophermap";
 52 1ac119fb 2024-01-23 op 	p->parse = &gm_parse;
 53 1ac119fb 2024-01-23 op 	p->free = &gm_free;
 54 1ac119fb 2024-01-23 op 	p->serialize = &gm_serialize;
 55 1ac119fb 2024-01-23 op 
 56 1ac119fb 2024-01-23 op 	TAILQ_INIT(&p->head);
 57 1ac119fb 2024-01-23 op }
 58 1ac119fb 2024-01-23 op 
 59 1ac119fb 2024-01-23 op static void
 60 1ac119fb 2024-01-23 op gm_parse_selector(char *line, struct gm_selector *s)
 61 1ac119fb 2024-01-23 op {
 62 1ac119fb 2024-01-23 op 	s->type = *line++;
 63 1ac119fb 2024-01-23 op 	s->ds = line;
 64 1ac119fb 2024-01-23 op 	s->selector = "";
 65 1ac119fb 2024-01-23 op 	s->addr = "";
 66 1ac119fb 2024-01-23 op 	s->port = "";
 67 1ac119fb 2024-01-23 op 
 68 1ac119fb 2024-01-23 op 	if ((line = strchr(line, '\t')) == NULL)
 69 1ac119fb 2024-01-23 op 		return;
 70 1ac119fb 2024-01-23 op 	*line++ = '\0';
 71 1ac119fb 2024-01-23 op 	s->selector = line;
 72 1ac119fb 2024-01-23 op 
 73 1ac119fb 2024-01-23 op 	if ((line = strchr(line, '\t')) == NULL)
 74 1ac119fb 2024-01-23 op 		return;
 75 1ac119fb 2024-01-23 op 	*line++ = '\0';
 76 1ac119fb 2024-01-23 op 	s->addr = line;
 77 1ac119fb 2024-01-23 op 
 78 1ac119fb 2024-01-23 op 	if ((line = strchr(line, '\t')) == NULL)
 79 1ac119fb 2024-01-23 op 		return;
 80 1ac119fb 2024-01-23 op 	*line++ = '\0';
 81 1ac119fb 2024-01-23 op 	s->port = line;
 82 1ac119fb 2024-01-23 op }
 83 1ac119fb 2024-01-23 op 
 84 1ac119fb 2024-01-23 op static int
 85 1ac119fb 2024-01-23 op gm_parse(struct parser *p, const char *buf, size_t size)
 86 1ac119fb 2024-01-23 op {
 87 1ac119fb 2024-01-23 op 	return parser_foreach_line(p, buf, size, gm_foreach_line);
 88 62c0f697 2024-02-21 op }
 89 62c0f697 2024-02-21 op 
 90 62c0f697 2024-02-21 op static int
 91 62c0f697 2024-02-21 op selector2uri(struct gm_selector *s, char *buf, size_t len)
 92 62c0f697 2024-02-21 op {
 93 62c0f697 2024-02-21 op 	int	 r;
 94 62c0f697 2024-02-21 op 
 95 62c0f697 2024-02-21 op 	r = snprintf(buf, len, "gopher://%s:%s/%c%s",
 96 62c0f697 2024-02-21 op 	    s->addr, s->port, s->type, *s->selector != '/' ? "/" : "");
 97 62c0f697 2024-02-21 op 	if (r < 0 || (size_t)r >= len)
 98 62c0f697 2024-02-21 op 		return (-1);
 99 62c0f697 2024-02-21 op 
100 62c0f697 2024-02-21 op 	buf += r;
101 62c0f697 2024-02-21 op 	len -= r;
102 62c0f697 2024-02-21 op 	return (iri_urlescape(s->selector, buf, len));
103 1ac119fb 2024-01-23 op }
104 1ac119fb 2024-01-23 op 
105 1ac119fb 2024-01-23 op static inline int
106 1ac119fb 2024-01-23 op emit_line(struct parser *p, enum line_type type, struct gm_selector *s)
107 1ac119fb 2024-01-23 op {
108 1ac119fb 2024-01-23 op 	struct line *l;
109 62c0f697 2024-02-21 op 	char buf[LINE_MAX];
110 1ac119fb 2024-01-23 op 
111 1ac119fb 2024-01-23 op 	if ((l = calloc(1, sizeof(*l))) == NULL)
112 1ac119fb 2024-01-23 op 		goto err;
113 1ac119fb 2024-01-23 op 
114 1ac119fb 2024-01-23 op 	if ((l->line = strdup(s->ds)) == NULL)
115 1ac119fb 2024-01-23 op 		goto err;
116 1ac119fb 2024-01-23 op 
117 1ac119fb 2024-01-23 op 	switch (l->type = type) {
118 1ac119fb 2024-01-23 op 	case LINE_LINK:
119 1ac119fb 2024-01-23 op 		if (s->type == 'h' && !strncmp(s->selector, "URL:", 4)) {
120 1ac119fb 2024-01-23 op 			strlcpy(buf, s->selector+4, sizeof(buf));
121 62c0f697 2024-02-21 op 		} else if (selector2uri(s, buf, sizeof(buf)) == -1)
122 62c0f697 2024-02-21 op 			goto err;
123 1ac119fb 2024-01-23 op 
124 1ac119fb 2024-01-23 op 		if ((l->alt = strdup(buf)) == NULL)
125 1ac119fb 2024-01-23 op 			goto err;
126 1ac119fb 2024-01-23 op 		break;
127 1ac119fb 2024-01-23 op 
128 1ac119fb 2024-01-23 op 	default:
129 1ac119fb 2024-01-23 op 		break;
130 1ac119fb 2024-01-23 op 	}
131 1ac119fb 2024-01-23 op 
132 1ac119fb 2024-01-23 op 	TAILQ_INSERT_TAIL(&p->head, l, lines);
133 1ac119fb 2024-01-23 op 
134 1ac119fb 2024-01-23 op 	return 1;
135 1ac119fb 2024-01-23 op 
136 1ac119fb 2024-01-23 op err:
137 1ac119fb 2024-01-23 op 	if (l != NULL) {
138 1ac119fb 2024-01-23 op 		free(l->line);
139 1ac119fb 2024-01-23 op 		free(l->alt);
140 1ac119fb 2024-01-23 op 		free(l);
141 1ac119fb 2024-01-23 op 	}
142 1ac119fb 2024-01-23 op 	return 0;
143 1ac119fb 2024-01-23 op }
144 1ac119fb 2024-01-23 op 
145 1ac119fb 2024-01-23 op static int
146 1ac119fb 2024-01-23 op gm_foreach_line(struct parser *p, const char *line, size_t linelen)
147 1ac119fb 2024-01-23 op {
148 1ac119fb 2024-01-23 op 	char buf[LINE_MAX] = {0};
149 1ac119fb 2024-01-23 op 	struct gm_selector s = {0};
150 1ac119fb 2024-01-23 op 
151 1ac119fb 2024-01-23 op 	memcpy(buf, line, MIN(sizeof(buf)-1, linelen));
152 1ac119fb 2024-01-23 op 	gm_parse_selector(buf, &s);
153 1ac119fb 2024-01-23 op 
154 1ac119fb 2024-01-23 op 	switch (s.type) {
155 1ac119fb 2024-01-23 op 	case '0':	/* text file */
156 1ac119fb 2024-01-23 op 	case '1':	/* gopher submenu */
157 1ac119fb 2024-01-23 op 	case '2':	/* CCSO nameserver */
158 1ac119fb 2024-01-23 op 	case '4':	/* binhex-encoded file */
159 1ac119fb 2024-01-23 op 	case '5':	/* DOS file */
160 1ac119fb 2024-01-23 op 	case '6':	/* uuencoded file */
161 1ac119fb 2024-01-23 op 	case '7':	/* full-text search */
162 1ac119fb 2024-01-23 op 	case '8':	/* telnet */
163 1ac119fb 2024-01-23 op 	case '9':	/* binary file */
164 1ac119fb 2024-01-23 op 	case '+':	/* mirror or alternate server */
165 1ac119fb 2024-01-23 op 	case 'g':	/* gif */
166 1ac119fb 2024-01-23 op 	case 'I':	/* image */
167 1ac119fb 2024-01-23 op 	case 'T':	/* telnet 3270 */
168 1ac119fb 2024-01-23 op 	case ':':	/* gopher+: bitmap image */
169 1ac119fb 2024-01-23 op 	case ';':	/* gopher+: movie file */
170 1ac119fb 2024-01-23 op 	case 'd':	/* non-canonical: doc */
171 1ac119fb 2024-01-23 op 	case 'h':	/* non-canonical: html file */
172 1ac119fb 2024-01-23 op 	case 's':	/* non-canonical: sound file */
173 1ac119fb 2024-01-23 op 		if (!emit_line(p, LINE_LINK, &s))
174 1ac119fb 2024-01-23 op 			return 0;
175 1ac119fb 2024-01-23 op 		break;
176 1ac119fb 2024-01-23 op 
177 1ac119fb 2024-01-23 op 	case 'i':	/* non-canonical: message */
178 1ac119fb 2024-01-23 op 		if (!emit_line(p, LINE_TEXT, &s))
179 1ac119fb 2024-01-23 op 			return 0;
180 1ac119fb 2024-01-23 op 		break;
181 1ac119fb 2024-01-23 op 
182 1ac119fb 2024-01-23 op 	case '3':	/* error code */
183 1ac119fb 2024-01-23 op 		if (!emit_line(p, LINE_QUOTE, &s))
184 1ac119fb 2024-01-23 op 			return 0;
185 1ac119fb 2024-01-23 op 		break;
186 1ac119fb 2024-01-23 op 	}
187 1ac119fb 2024-01-23 op 
188 1ac119fb 2024-01-23 op 	return 1;
189 1ac119fb 2024-01-23 op }
190 1ac119fb 2024-01-23 op 
191 1ac119fb 2024-01-23 op static int
192 1ac119fb 2024-01-23 op gm_free(struct parser *p)
193 1ac119fb 2024-01-23 op {
194 1ac119fb 2024-01-23 op 	/* flush the buffer */
195 1ac119fb 2024-01-23 op 	if (p->len != 0)
196 1ac119fb 2024-01-23 op 		gm_foreach_line(p, p->buf, p->len);
197 1ac119fb 2024-01-23 op 
198 1ac119fb 2024-01-23 op 	free(p->buf);
199 1ac119fb 2024-01-23 op 
200 1ac119fb 2024-01-23 op 	return 1;
201 1ac119fb 2024-01-23 op }
202 1ac119fb 2024-01-23 op 
203 1ac119fb 2024-01-23 op static inline const char *
204 1ac119fb 2024-01-23 op gopher_skip_selector(const char *path, int *ret_type)
205 1ac119fb 2024-01-23 op {
206 1ac119fb 2024-01-23 op 	*ret_type = 0;
207 1ac119fb 2024-01-23 op 
208 1ac119fb 2024-01-23 op 	if (!strcmp(path, "/") || *path == '\0') {
209 1ac119fb 2024-01-23 op 		*ret_type = '1';
210 1ac119fb 2024-01-23 op 		return path;
211 1ac119fb 2024-01-23 op 	}
212 1ac119fb 2024-01-23 op 
213 1ac119fb 2024-01-23 op 	if (*path != '/')
214 1ac119fb 2024-01-23 op 		return path;
215 1ac119fb 2024-01-23 op 	path++;
216 1ac119fb 2024-01-23 op 
217 1ac119fb 2024-01-23 op 	switch (*ret_type = *path) {
218 1ac119fb 2024-01-23 op 	case '0':
219 1ac119fb 2024-01-23 op 	case '1':
220 1ac119fb 2024-01-23 op 	case '7':
221 1ac119fb 2024-01-23 op 		break;
222 1ac119fb 2024-01-23 op 
223 1ac119fb 2024-01-23 op 	default:
224 1ac119fb 2024-01-23 op 		*ret_type = 0;
225 1ac119fb 2024-01-23 op 		path -= 1;
226 1ac119fb 2024-01-23 op 		return path;
227 1ac119fb 2024-01-23 op 	}
228 1ac119fb 2024-01-23 op 
229 1ac119fb 2024-01-23 op 	return ++path;
230 1ac119fb 2024-01-23 op }
231 1ac119fb 2024-01-23 op 
232 1ac119fb 2024-01-23 op static int
233 1ac119fb 2024-01-23 op serialize_link(struct line *line, const char *text, FILE *fp)
234 1ac119fb 2024-01-23 op {
235 1ac119fb 2024-01-23 op 	size_t		 portlen = 0;
236 1ac119fb 2024-01-23 op 	int		 type;
237 1ac119fb 2024-01-23 op 	const char	*uri, *endhost, *port, *path, *colon;
238 1ac119fb 2024-01-23 op 
239 1ac119fb 2024-01-23 op 	if ((uri = line->alt) == NULL)
240 1ac119fb 2024-01-23 op 		return -1;
241 1ac119fb 2024-01-23 op 
242 1ac119fb 2024-01-23 op 	if (strncmp(uri, "gopher://", 9) != 0)
243 1ac119fb 2024-01-23 op 		return fprintf(fp, "h%s\tURL:%s\terror.host\t1\n",
244 1ac119fb 2024-01-23 op 		    text, line->alt);
245 1ac119fb 2024-01-23 op 
246 1ac119fb 2024-01-23 op 	uri += 9; /* skip gopher:// */
247 1ac119fb 2024-01-23 op 
248 1ac119fb 2024-01-23 op 	path = strchr(uri, '/');
249 1ac119fb 2024-01-23 op 	colon = strchr(uri, ':');
250 1ac119fb 2024-01-23 op 
251 1ac119fb 2024-01-23 op 	if (path != NULL && colon > path)
252 1ac119fb 2024-01-23 op 		colon = NULL;
253 1ac119fb 2024-01-23 op 
254 1ac119fb 2024-01-23 op 	if ((endhost = colon) == NULL &&
255 1ac119fb 2024-01-23 op 	    (endhost = path) == NULL)
256 1ac119fb 2024-01-23 op 		endhost = strchr(uri, '\0');
257 1ac119fb 2024-01-23 op 
258 1ac119fb 2024-01-23 op 	if (colon != NULL) {
259 1ac119fb 2024-01-23 op 		for (port = colon+1; *port && *port != '/'; ++port)
260 1ac119fb 2024-01-23 op 			++portlen;
261 1ac119fb 2024-01-23 op 		port = colon+1;
262 1ac119fb 2024-01-23 op 	} else {
263 1ac119fb 2024-01-23 op 		port = "70";
264 1ac119fb 2024-01-23 op 		portlen = 2;
265 1ac119fb 2024-01-23 op 	}
266 1ac119fb 2024-01-23 op 
267 1ac119fb 2024-01-23 op 	if (path == NULL) {
268 1ac119fb 2024-01-23 op 		type = '1';
269 1ac119fb 2024-01-23 op 		path = "";
270 1ac119fb 2024-01-23 op 	} else
271 1ac119fb 2024-01-23 op 		path = gopher_skip_selector(path, &type);
272 1ac119fb 2024-01-23 op 
273 1ac119fb 2024-01-23 op 	return fprintf(fp, "%c%s\t%s\t%.*s\t%.*s\n", type, text,
274 1ac119fb 2024-01-23 op 	    path, (int)(endhost - uri), uri, (int)portlen, port);
275 1ac119fb 2024-01-23 op }
276 1ac119fb 2024-01-23 op 
277 1ac119fb 2024-01-23 op static int
278 1ac119fb 2024-01-23 op gm_serialize(struct parser *p, FILE *fp)
279 1ac119fb 2024-01-23 op {
280 1ac119fb 2024-01-23 op 	struct line	*line;
281 1ac119fb 2024-01-23 op 	const char	*text;
282 1ac119fb 2024-01-23 op 	int		 r;
283 1ac119fb 2024-01-23 op 
284 1ac119fb 2024-01-23 op 	TAILQ_FOREACH(line, &p->head, lines) {
285 1ac119fb 2024-01-23 op 		if ((text = line->line) == NULL)
286 1ac119fb 2024-01-23 op 			text = "";
287 1ac119fb 2024-01-23 op 
288 1ac119fb 2024-01-23 op 		switch (line->type) {
289 1ac119fb 2024-01-23 op 		case LINE_LINK:
290 1ac119fb 2024-01-23 op 			r = serialize_link(line, text, fp);
291 1ac119fb 2024-01-23 op 			break;
292 1ac119fb 2024-01-23 op 
293 1ac119fb 2024-01-23 op 		case LINE_TEXT:
294 1ac119fb 2024-01-23 op 			r = fprintf(fp, "i%s\t\terror.host\t1\n", text);
295 1ac119fb 2024-01-23 op 			break;
296 1ac119fb 2024-01-23 op 
297 1ac119fb 2024-01-23 op 		case LINE_QUOTE:
298 1ac119fb 2024-01-23 op 			r = fprintf(fp, "3%s\t\terror.host\t1\n", text);
299 1ac119fb 2024-01-23 op 			break;
300 1ac119fb 2024-01-23 op 
301 1ac119fb 2024-01-23 op 		default:
302 1ac119fb 2024-01-23 op 			/* unreachable */
303 1ac119fb 2024-01-23 op 			abort();
304 1ac119fb 2024-01-23 op 		}
305 1ac119fb 2024-01-23 op 
306 1ac119fb 2024-01-23 op 		if (r == -1)
307 1ac119fb 2024-01-23 op 			return 0;
308 1ac119fb 2024-01-23 op 	}
309 1ac119fb 2024-01-23 op 
310 1ac119fb 2024-01-23 op 	return 1;
311 1ac119fb 2024-01-23 op }