1 1ac119fb 2024-01-23 op /*
  2 1ac119fb 2024-01-23 op  * Copyright (c) 2021 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 1ac119fb 2024-01-23 op #include "utils.h"
 25 e5e04904 2024-02-06 op 
 26 e5e04904 2024-02-06 op #ifndef LINE_MAX
 27 e5e04904 2024-02-06 op #define LINE_MAX 2048
 28 e5e04904 2024-02-06 op #endif
 29 1ac119fb 2024-01-23 op 
 30 1ac119fb 2024-01-23 op struct gm_selector {
 31 1ac119fb 2024-01-23 op 	char		 type;
 32 1ac119fb 2024-01-23 op 	const char	*ds;
 33 1ac119fb 2024-01-23 op 	const char	*selector;
 34 1ac119fb 2024-01-23 op 	const char	*addr;
 35 1ac119fb 2024-01-23 op 	const char	*port;
 36 1ac119fb 2024-01-23 op };
 37 1ac119fb 2024-01-23 op 
 38 1ac119fb 2024-01-23 op static void	gm_parse_selector(char *, struct gm_selector *);
 39 1ac119fb 2024-01-23 op 
 40 1ac119fb 2024-01-23 op static int	gm_parse(struct parser *, const char *, size_t);
 41 1ac119fb 2024-01-23 op static int	gm_foreach_line(struct parser *, const char *, size_t);
 42 1ac119fb 2024-01-23 op static int	gm_free(struct parser *);
 43 1ac119fb 2024-01-23 op static int	gm_serialize(struct parser *, FILE *);
 44 1ac119fb 2024-01-23 op 
 45 1ac119fb 2024-01-23 op void
 46 1ac119fb 2024-01-23 op gophermap_initparser(struct parser *p)
 47 1ac119fb 2024-01-23 op {
 48 1ac119fb 2024-01-23 op 	memset(p, 0, sizeof(*p));
 49 1ac119fb 2024-01-23 op 
 50 1ac119fb 2024-01-23 op 	p->name = "gophermap";
 51 1ac119fb 2024-01-23 op 	p->parse = &gm_parse;
 52 1ac119fb 2024-01-23 op 	p->free = &gm_free;
 53 1ac119fb 2024-01-23 op 	p->serialize = &gm_serialize;
 54 1ac119fb 2024-01-23 op 
 55 1ac119fb 2024-01-23 op 	TAILQ_INIT(&p->head);
 56 1ac119fb 2024-01-23 op }
 57 1ac119fb 2024-01-23 op 
 58 1ac119fb 2024-01-23 op static void
 59 1ac119fb 2024-01-23 op gm_parse_selector(char *line, struct gm_selector *s)
 60 1ac119fb 2024-01-23 op {
 61 1ac119fb 2024-01-23 op 	s->type = *line++;
 62 1ac119fb 2024-01-23 op 	s->ds = line;
 63 1ac119fb 2024-01-23 op 	s->selector = "";
 64 1ac119fb 2024-01-23 op 	s->addr = "";
 65 1ac119fb 2024-01-23 op 	s->port = "";
 66 1ac119fb 2024-01-23 op 
 67 1ac119fb 2024-01-23 op 	if ((line = strchr(line, '\t')) == NULL)
 68 1ac119fb 2024-01-23 op 		return;
 69 1ac119fb 2024-01-23 op 	*line++ = '\0';
 70 1ac119fb 2024-01-23 op 	s->selector = line;
 71 1ac119fb 2024-01-23 op 
 72 1ac119fb 2024-01-23 op 	if ((line = strchr(line, '\t')) == NULL)
 73 1ac119fb 2024-01-23 op 		return;
 74 1ac119fb 2024-01-23 op 	*line++ = '\0';
 75 1ac119fb 2024-01-23 op 	s->addr = line;
 76 1ac119fb 2024-01-23 op 
 77 1ac119fb 2024-01-23 op 	if ((line = strchr(line, '\t')) == NULL)
 78 1ac119fb 2024-01-23 op 		return;
 79 1ac119fb 2024-01-23 op 	*line++ = '\0';
 80 1ac119fb 2024-01-23 op 	s->port = line;
 81 1ac119fb 2024-01-23 op }
 82 1ac119fb 2024-01-23 op 
 83 1ac119fb 2024-01-23 op static int
 84 1ac119fb 2024-01-23 op gm_parse(struct parser *p, const char *buf, size_t size)
 85 1ac119fb 2024-01-23 op {
 86 1ac119fb 2024-01-23 op 	return parser_foreach_line(p, buf, size, gm_foreach_line);
 87 1ac119fb 2024-01-23 op }
 88 1ac119fb 2024-01-23 op 
 89 1ac119fb 2024-01-23 op static inline int
 90 1ac119fb 2024-01-23 op emit_line(struct parser *p, enum line_type type, struct gm_selector *s)
 91 1ac119fb 2024-01-23 op {
 92 1ac119fb 2024-01-23 op 	struct line *l;
 93 1ac119fb 2024-01-23 op 	char buf[LINE_MAX], b[2] = {0};
 94 1ac119fb 2024-01-23 op 
 95 1ac119fb 2024-01-23 op 	if ((l = calloc(1, sizeof(*l))) == NULL)
 96 1ac119fb 2024-01-23 op 		goto err;
 97 1ac119fb 2024-01-23 op 
 98 1ac119fb 2024-01-23 op 	if ((l->line = strdup(s->ds)) == NULL)
 99 1ac119fb 2024-01-23 op 		goto err;
100 1ac119fb 2024-01-23 op 
101 1ac119fb 2024-01-23 op 	switch (l->type = type) {
102 1ac119fb 2024-01-23 op 	case LINE_LINK:
103 1ac119fb 2024-01-23 op 		if (s->type == 'h' && !strncmp(s->selector, "URL:", 4)) {
104 1ac119fb 2024-01-23 op 			strlcpy(buf, s->selector+4, sizeof(buf));
105 1ac119fb 2024-01-23 op 		} else {
106 1ac119fb 2024-01-23 op 			strlcpy(buf, "gopher://", sizeof(buf));
107 1ac119fb 2024-01-23 op 			strlcat(buf, s->addr, sizeof(buf));
108 1ac119fb 2024-01-23 op 			strlcat(buf, ":", sizeof(buf));
109 1ac119fb 2024-01-23 op 			strlcat(buf, s->port, sizeof(buf));
110 1ac119fb 2024-01-23 op 			strlcat(buf, "/", sizeof(buf));
111 1ac119fb 2024-01-23 op 			b[0] = s->type;
112 1ac119fb 2024-01-23 op 			strlcat(buf, b, sizeof(buf));
113 1ac119fb 2024-01-23 op 			if (*s->selector != '/')
114 1ac119fb 2024-01-23 op 				strlcat(buf, "/", sizeof(buf));
115 1ac119fb 2024-01-23 op 			strlcat(buf, s->selector, sizeof(buf));
116 1ac119fb 2024-01-23 op 		}
117 1ac119fb 2024-01-23 op 
118 1ac119fb 2024-01-23 op 		if ((l->alt = strdup(buf)) == NULL)
119 1ac119fb 2024-01-23 op 			goto err;
120 1ac119fb 2024-01-23 op 		break;
121 1ac119fb 2024-01-23 op 
122 1ac119fb 2024-01-23 op 	default:
123 1ac119fb 2024-01-23 op 		break;
124 1ac119fb 2024-01-23 op 	}
125 1ac119fb 2024-01-23 op 
126 1ac119fb 2024-01-23 op 	TAILQ_INSERT_TAIL(&p->head, l, lines);
127 1ac119fb 2024-01-23 op 
128 1ac119fb 2024-01-23 op 	return 1;
129 1ac119fb 2024-01-23 op 
130 1ac119fb 2024-01-23 op err:
131 1ac119fb 2024-01-23 op 	if (l != NULL) {
132 1ac119fb 2024-01-23 op 		free(l->line);
133 1ac119fb 2024-01-23 op 		free(l->alt);
134 1ac119fb 2024-01-23 op 		free(l);
135 1ac119fb 2024-01-23 op 	}
136 1ac119fb 2024-01-23 op 	return 0;
137 1ac119fb 2024-01-23 op }
138 1ac119fb 2024-01-23 op 
139 1ac119fb 2024-01-23 op static int
140 1ac119fb 2024-01-23 op gm_foreach_line(struct parser *p, const char *line, size_t linelen)
141 1ac119fb 2024-01-23 op {
142 1ac119fb 2024-01-23 op 	char buf[LINE_MAX] = {0};
143 1ac119fb 2024-01-23 op 	struct gm_selector s = {0};
144 1ac119fb 2024-01-23 op 
145 1ac119fb 2024-01-23 op 	memcpy(buf, line, MIN(sizeof(buf)-1, linelen));
146 1ac119fb 2024-01-23 op 	gm_parse_selector(buf, &s);
147 1ac119fb 2024-01-23 op 
148 1ac119fb 2024-01-23 op 	switch (s.type) {
149 1ac119fb 2024-01-23 op 	case '0':	/* text file */
150 1ac119fb 2024-01-23 op 	case '1':	/* gopher submenu */
151 1ac119fb 2024-01-23 op 	case '2':	/* CCSO nameserver */
152 1ac119fb 2024-01-23 op 	case '4':	/* binhex-encoded file */
153 1ac119fb 2024-01-23 op 	case '5':	/* DOS file */
154 1ac119fb 2024-01-23 op 	case '6':	/* uuencoded file */
155 1ac119fb 2024-01-23 op 	case '7':	/* full-text search */
156 1ac119fb 2024-01-23 op 	case '8':	/* telnet */
157 1ac119fb 2024-01-23 op 	case '9':	/* binary file */
158 1ac119fb 2024-01-23 op 	case '+':	/* mirror or alternate server */
159 1ac119fb 2024-01-23 op 	case 'g':	/* gif */
160 1ac119fb 2024-01-23 op 	case 'I':	/* image */
161 1ac119fb 2024-01-23 op 	case 'T':	/* telnet 3270 */
162 1ac119fb 2024-01-23 op 	case ':':	/* gopher+: bitmap image */
163 1ac119fb 2024-01-23 op 	case ';':	/* gopher+: movie file */
164 1ac119fb 2024-01-23 op 	case 'd':	/* non-canonical: doc */
165 1ac119fb 2024-01-23 op 	case 'h':	/* non-canonical: html file */
166 1ac119fb 2024-01-23 op 	case 's':	/* non-canonical: sound file */
167 1ac119fb 2024-01-23 op 		if (!emit_line(p, LINE_LINK, &s))
168 1ac119fb 2024-01-23 op 			return 0;
169 1ac119fb 2024-01-23 op 		break;
170 1ac119fb 2024-01-23 op 
171 1ac119fb 2024-01-23 op 	case 'i':	/* non-canonical: message */
172 1ac119fb 2024-01-23 op 		if (!emit_line(p, LINE_TEXT, &s))
173 1ac119fb 2024-01-23 op 			return 0;
174 1ac119fb 2024-01-23 op 		break;
175 1ac119fb 2024-01-23 op 
176 1ac119fb 2024-01-23 op 	case '3':	/* error code */
177 1ac119fb 2024-01-23 op 		if (!emit_line(p, LINE_QUOTE, &s))
178 1ac119fb 2024-01-23 op 			return 0;
179 1ac119fb 2024-01-23 op 		break;
180 1ac119fb 2024-01-23 op 	}
181 1ac119fb 2024-01-23 op 
182 1ac119fb 2024-01-23 op 	return 1;
183 1ac119fb 2024-01-23 op }
184 1ac119fb 2024-01-23 op 
185 1ac119fb 2024-01-23 op static int
186 1ac119fb 2024-01-23 op gm_free(struct parser *p)
187 1ac119fb 2024-01-23 op {
188 1ac119fb 2024-01-23 op 	/* flush the buffer */
189 1ac119fb 2024-01-23 op 	if (p->len != 0)
190 1ac119fb 2024-01-23 op 		gm_foreach_line(p, p->buf, p->len);
191 1ac119fb 2024-01-23 op 
192 1ac119fb 2024-01-23 op 	free(p->buf);
193 1ac119fb 2024-01-23 op 
194 1ac119fb 2024-01-23 op 	return 1;
195 1ac119fb 2024-01-23 op }
196 1ac119fb 2024-01-23 op 
197 1ac119fb 2024-01-23 op static inline const char *
198 1ac119fb 2024-01-23 op gopher_skip_selector(const char *path, int *ret_type)
199 1ac119fb 2024-01-23 op {
200 1ac119fb 2024-01-23 op 	*ret_type = 0;
201 1ac119fb 2024-01-23 op 
202 1ac119fb 2024-01-23 op 	if (!strcmp(path, "/") || *path == '\0') {
203 1ac119fb 2024-01-23 op 		*ret_type = '1';
204 1ac119fb 2024-01-23 op 		return path;
205 1ac119fb 2024-01-23 op 	}
206 1ac119fb 2024-01-23 op 
207 1ac119fb 2024-01-23 op 	if (*path != '/')
208 1ac119fb 2024-01-23 op 		return path;
209 1ac119fb 2024-01-23 op 	path++;
210 1ac119fb 2024-01-23 op 
211 1ac119fb 2024-01-23 op 	switch (*ret_type = *path) {
212 1ac119fb 2024-01-23 op 	case '0':
213 1ac119fb 2024-01-23 op 	case '1':
214 1ac119fb 2024-01-23 op 	case '7':
215 1ac119fb 2024-01-23 op 		break;
216 1ac119fb 2024-01-23 op 
217 1ac119fb 2024-01-23 op 	default:
218 1ac119fb 2024-01-23 op 		*ret_type = 0;
219 1ac119fb 2024-01-23 op 		path -= 1;
220 1ac119fb 2024-01-23 op 		return path;
221 1ac119fb 2024-01-23 op 	}
222 1ac119fb 2024-01-23 op 
223 1ac119fb 2024-01-23 op 	return ++path;
224 1ac119fb 2024-01-23 op }
225 1ac119fb 2024-01-23 op 
226 1ac119fb 2024-01-23 op static int
227 1ac119fb 2024-01-23 op serialize_link(struct line *line, const char *text, FILE *fp)
228 1ac119fb 2024-01-23 op {
229 1ac119fb 2024-01-23 op 	size_t		 portlen = 0;
230 1ac119fb 2024-01-23 op 	int		 type;
231 1ac119fb 2024-01-23 op 	const char	*uri, *endhost, *port, *path, *colon;
232 1ac119fb 2024-01-23 op 
233 1ac119fb 2024-01-23 op 	if ((uri = line->alt) == NULL)
234 1ac119fb 2024-01-23 op 		return -1;
235 1ac119fb 2024-01-23 op 
236 1ac119fb 2024-01-23 op 	if (strncmp(uri, "gopher://", 9) != 0)
237 1ac119fb 2024-01-23 op 		return fprintf(fp, "h%s\tURL:%s\terror.host\t1\n",
238 1ac119fb 2024-01-23 op 		    text, line->alt);
239 1ac119fb 2024-01-23 op 
240 1ac119fb 2024-01-23 op 	uri += 9; /* skip gopher:// */
241 1ac119fb 2024-01-23 op 
242 1ac119fb 2024-01-23 op 	path = strchr(uri, '/');
243 1ac119fb 2024-01-23 op 	colon = strchr(uri, ':');
244 1ac119fb 2024-01-23 op 
245 1ac119fb 2024-01-23 op 	if (path != NULL && colon > path)
246 1ac119fb 2024-01-23 op 		colon = NULL;
247 1ac119fb 2024-01-23 op 
248 1ac119fb 2024-01-23 op 	if ((endhost = colon) == NULL &&
249 1ac119fb 2024-01-23 op 	    (endhost = path) == NULL)
250 1ac119fb 2024-01-23 op 		endhost = strchr(uri, '\0');
251 1ac119fb 2024-01-23 op 
252 1ac119fb 2024-01-23 op 	if (colon != NULL) {
253 1ac119fb 2024-01-23 op 		for (port = colon+1; *port && *port != '/'; ++port)
254 1ac119fb 2024-01-23 op 			++portlen;
255 1ac119fb 2024-01-23 op 		port = colon+1;
256 1ac119fb 2024-01-23 op 	} else {
257 1ac119fb 2024-01-23 op 		port = "70";
258 1ac119fb 2024-01-23 op 		portlen = 2;
259 1ac119fb 2024-01-23 op 	}
260 1ac119fb 2024-01-23 op 
261 1ac119fb 2024-01-23 op 	if (path == NULL) {
262 1ac119fb 2024-01-23 op 		type = '1';
263 1ac119fb 2024-01-23 op 		path = "";
264 1ac119fb 2024-01-23 op 	} else
265 1ac119fb 2024-01-23 op 		path = gopher_skip_selector(path, &type);
266 1ac119fb 2024-01-23 op 
267 1ac119fb 2024-01-23 op 	return fprintf(fp, "%c%s\t%s\t%.*s\t%.*s\n", type, text,
268 1ac119fb 2024-01-23 op 	    path, (int)(endhost - uri), uri, (int)portlen, port);
269 1ac119fb 2024-01-23 op }
270 1ac119fb 2024-01-23 op 
271 1ac119fb 2024-01-23 op static int
272 1ac119fb 2024-01-23 op gm_serialize(struct parser *p, FILE *fp)
273 1ac119fb 2024-01-23 op {
274 1ac119fb 2024-01-23 op 	struct line	*line;
275 1ac119fb 2024-01-23 op 	const char	*text;
276 1ac119fb 2024-01-23 op 	int		 r;
277 1ac119fb 2024-01-23 op 
278 1ac119fb 2024-01-23 op 	TAILQ_FOREACH(line, &p->head, lines) {
279 1ac119fb 2024-01-23 op 		if ((text = line->line) == NULL)
280 1ac119fb 2024-01-23 op 			text = "";
281 1ac119fb 2024-01-23 op 
282 1ac119fb 2024-01-23 op 		switch (line->type) {
283 1ac119fb 2024-01-23 op 		case LINE_LINK:
284 1ac119fb 2024-01-23 op 			r = serialize_link(line, text, fp);
285 1ac119fb 2024-01-23 op 			break;
286 1ac119fb 2024-01-23 op 
287 1ac119fb 2024-01-23 op 		case LINE_TEXT:
288 1ac119fb 2024-01-23 op 			r = fprintf(fp, "i%s\t\terror.host\t1\n", text);
289 1ac119fb 2024-01-23 op 			break;
290 1ac119fb 2024-01-23 op 
291 1ac119fb 2024-01-23 op 		case LINE_QUOTE:
292 1ac119fb 2024-01-23 op 			r = fprintf(fp, "3%s\t\terror.host\t1\n", text);
293 1ac119fb 2024-01-23 op 			break;
294 1ac119fb 2024-01-23 op 
295 1ac119fb 2024-01-23 op 		default:
296 1ac119fb 2024-01-23 op 			/* unreachable */
297 1ac119fb 2024-01-23 op 			abort();
298 1ac119fb 2024-01-23 op 		}
299 1ac119fb 2024-01-23 op 
300 1ac119fb 2024-01-23 op 		if (r == -1)
301 1ac119fb 2024-01-23 op 			return 0;
302 1ac119fb 2024-01-23 op 	}
303 1ac119fb 2024-01-23 op 
304 1ac119fb 2024-01-23 op 	return 1;
305 1ac119fb 2024-01-23 op }