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 62c0f697 2024-02-21 op #include "iri.h"
 24 c1d27b0e 2024-06-14 op #include "parser.h"
 25 c1d27b0e 2024-06-14 op #include "telescope.h"
 26 1ac119fb 2024-01-23 op #include "utils.h"
 27 3d89457c 2024-06-18 thomas.ad #include "xwrapper.h"
 28 e5e04904 2024-02-06 op 
 29 e5e04904 2024-02-06 op #ifndef LINE_MAX
 30 e5e04904 2024-02-06 op #define LINE_MAX 2048
 31 e5e04904 2024-02-06 op #endif
 32 1ac119fb 2024-01-23 op 
 33 1ac119fb 2024-01-23 op struct gm_selector {
 34 1ac119fb 2024-01-23 op 	char		 type;
 35 1ac119fb 2024-01-23 op 	const char	*ds;
 36 1ac119fb 2024-01-23 op 	const char	*selector;
 37 1ac119fb 2024-01-23 op 	const char	*addr;
 38 1ac119fb 2024-01-23 op 	const char	*port;
 39 1ac119fb 2024-01-23 op };
 40 1ac119fb 2024-01-23 op 
 41 1ac119fb 2024-01-23 op static void	gm_parse_selector(char *, struct gm_selector *);
 42 1ac119fb 2024-01-23 op 
 43 c1d27b0e 2024-06-14 op static int	gm_parse_line(struct buffer *, const char *, size_t);
 44 c1d27b0e 2024-06-14 op static int	gm_serialize(struct buffer *, FILE *);
 45 1ac119fb 2024-01-23 op 
 46 fd1c80ce 2024-06-14 op const struct parser gophermap_parser = {
 47 c1d27b0e 2024-06-14 op 	.name = "gophermap",
 48 c1d27b0e 2024-06-14 op 	.parseline = &gm_parse_line,
 49 c1d27b0e 2024-06-14 op 	.serialize = &gm_serialize,
 50 c1d27b0e 2024-06-14 op };
 51 1ac119fb 2024-01-23 op 
 52 1ac119fb 2024-01-23 op static void
 53 1ac119fb 2024-01-23 op gm_parse_selector(char *line, struct gm_selector *s)
 54 1ac119fb 2024-01-23 op {
 55 1ac119fb 2024-01-23 op 	s->type = *line++;
 56 1ac119fb 2024-01-23 op 	s->ds = line;
 57 1ac119fb 2024-01-23 op 	s->selector = "";
 58 1ac119fb 2024-01-23 op 	s->addr = "";
 59 1ac119fb 2024-01-23 op 	s->port = "";
 60 1ac119fb 2024-01-23 op 
 61 1ac119fb 2024-01-23 op 	if ((line = strchr(line, '\t')) == NULL)
 62 1ac119fb 2024-01-23 op 		return;
 63 1ac119fb 2024-01-23 op 	*line++ = '\0';
 64 1ac119fb 2024-01-23 op 	s->selector = line;
 65 1ac119fb 2024-01-23 op 
 66 1ac119fb 2024-01-23 op 	if ((line = strchr(line, '\t')) == NULL)
 67 1ac119fb 2024-01-23 op 		return;
 68 1ac119fb 2024-01-23 op 	*line++ = '\0';
 69 1ac119fb 2024-01-23 op 	s->addr = line;
 70 1ac119fb 2024-01-23 op 
 71 1ac119fb 2024-01-23 op 	if ((line = strchr(line, '\t')) == NULL)
 72 1ac119fb 2024-01-23 op 		return;
 73 1ac119fb 2024-01-23 op 	*line++ = '\0';
 74 1ac119fb 2024-01-23 op 	s->port = line;
 75 1ac119fb 2024-01-23 op }
 76 1ac119fb 2024-01-23 op 
 77 1ac119fb 2024-01-23 op static int
 78 62c0f697 2024-02-21 op selector2uri(struct gm_selector *s, char *buf, size_t len)
 79 62c0f697 2024-02-21 op {
 80 62c0f697 2024-02-21 op 	int	 r;
 81 62c0f697 2024-02-21 op 
 82 62c0f697 2024-02-21 op 	r = snprintf(buf, len, "gopher://%s:%s/%c%s",
 83 62c0f697 2024-02-21 op 	    s->addr, s->port, s->type, *s->selector != '/' ? "/" : "");
 84 62c0f697 2024-02-21 op 	if (r < 0 || (size_t)r >= len)
 85 62c0f697 2024-02-21 op 		return (-1);
 86 62c0f697 2024-02-21 op 
 87 62c0f697 2024-02-21 op 	buf += r;
 88 62c0f697 2024-02-21 op 	len -= r;
 89 62c0f697 2024-02-21 op 	return (iri_urlescape(s->selector, buf, len));
 90 1ac119fb 2024-01-23 op }
 91 1ac119fb 2024-01-23 op 
 92 1ac119fb 2024-01-23 op static inline int
 93 c1d27b0e 2024-06-14 op emit_line(struct buffer *b, enum line_type type, struct gm_selector *s)
 94 1ac119fb 2024-01-23 op {
 95 1ac119fb 2024-01-23 op 	struct line *l;
 96 62c0f697 2024-02-21 op 	char buf[LINE_MAX];
 97 1ac119fb 2024-01-23 op 
 98 3d89457c 2024-06-18 thomas.ad 	l = xcalloc(1, sizeof(*l));
 99 1ac119fb 2024-01-23 op 
100 3d89457c 2024-06-18 thomas.ad 	l->line = xstrdup(s->ds);
101 1ac119fb 2024-01-23 op 
102 1ac119fb 2024-01-23 op 	switch (l->type = type) {
103 1ac119fb 2024-01-23 op 	case LINE_LINK:
104 1ac119fb 2024-01-23 op 		if (s->type == 'h' && !strncmp(s->selector, "URL:", 4)) {
105 1ac119fb 2024-01-23 op 			strlcpy(buf, s->selector+4, sizeof(buf));
106 62c0f697 2024-02-21 op 		} else if (selector2uri(s, buf, sizeof(buf)) == -1)
107 62c0f697 2024-02-21 op 			goto err;
108 1ac119fb 2024-01-23 op 
109 3d89457c 2024-06-18 thomas.ad 		l->alt = xstrdup(buf);
110 1ac119fb 2024-01-23 op 		break;
111 1ac119fb 2024-01-23 op 
112 1ac119fb 2024-01-23 op 	default:
113 1ac119fb 2024-01-23 op 		break;
114 1ac119fb 2024-01-23 op 	}
115 1ac119fb 2024-01-23 op 
116 c1d27b0e 2024-06-14 op 	TAILQ_INSERT_TAIL(&b->head, l, lines);
117 1ac119fb 2024-01-23 op 
118 1ac119fb 2024-01-23 op 	return 1;
119 1ac119fb 2024-01-23 op 
120 1ac119fb 2024-01-23 op err:
121 1ac119fb 2024-01-23 op 	if (l != NULL) {
122 1ac119fb 2024-01-23 op 		free(l->line);
123 1ac119fb 2024-01-23 op 		free(l->alt);
124 1ac119fb 2024-01-23 op 		free(l);
125 1ac119fb 2024-01-23 op 	}
126 1ac119fb 2024-01-23 op 	return 0;
127 1ac119fb 2024-01-23 op }
128 1ac119fb 2024-01-23 op 
129 1ac119fb 2024-01-23 op static int
130 c1d27b0e 2024-06-14 op gm_parse_line(struct buffer *b, const char *line, size_t linelen)
131 1ac119fb 2024-01-23 op {
132 1ac119fb 2024-01-23 op 	char buf[LINE_MAX] = {0};
133 1ac119fb 2024-01-23 op 	struct gm_selector s = {0};
134 1ac119fb 2024-01-23 op 
135 1ac119fb 2024-01-23 op 	memcpy(buf, line, MIN(sizeof(buf)-1, linelen));
136 1ac119fb 2024-01-23 op 	gm_parse_selector(buf, &s);
137 1ac119fb 2024-01-23 op 
138 1ac119fb 2024-01-23 op 	switch (s.type) {
139 1ac119fb 2024-01-23 op 	case '0':	/* text file */
140 1ac119fb 2024-01-23 op 	case '1':	/* gopher submenu */
141 1ac119fb 2024-01-23 op 	case '2':	/* CCSO nameserver */
142 1ac119fb 2024-01-23 op 	case '4':	/* binhex-encoded file */
143 1ac119fb 2024-01-23 op 	case '5':	/* DOS file */
144 1ac119fb 2024-01-23 op 	case '6':	/* uuencoded file */
145 1ac119fb 2024-01-23 op 	case '7':	/* full-text search */
146 1ac119fb 2024-01-23 op 	case '8':	/* telnet */
147 1ac119fb 2024-01-23 op 	case '9':	/* binary file */
148 1ac119fb 2024-01-23 op 	case '+':	/* mirror or alternate server */
149 1ac119fb 2024-01-23 op 	case 'g':	/* gif */
150 1ac119fb 2024-01-23 op 	case 'I':	/* image */
151 1ac119fb 2024-01-23 op 	case 'T':	/* telnet 3270 */
152 1ac119fb 2024-01-23 op 	case ':':	/* gopher+: bitmap image */
153 1ac119fb 2024-01-23 op 	case ';':	/* gopher+: movie file */
154 1ac119fb 2024-01-23 op 	case 'd':	/* non-canonical: doc */
155 1ac119fb 2024-01-23 op 	case 'h':	/* non-canonical: html file */
156 1ac119fb 2024-01-23 op 	case 's':	/* non-canonical: sound file */
157 c1d27b0e 2024-06-14 op 		if (!emit_line(b, LINE_LINK, &s))
158 1ac119fb 2024-01-23 op 			return 0;
159 1ac119fb 2024-01-23 op 		break;
160 1ac119fb 2024-01-23 op 
161 1ac119fb 2024-01-23 op 	case 'i':	/* non-canonical: message */
162 c1d27b0e 2024-06-14 op 		if (!emit_line(b, LINE_TEXT, &s))
163 1ac119fb 2024-01-23 op 			return 0;
164 1ac119fb 2024-01-23 op 		break;
165 1ac119fb 2024-01-23 op 
166 1ac119fb 2024-01-23 op 	case '3':	/* error code */
167 c1d27b0e 2024-06-14 op 		if (!emit_line(b, LINE_QUOTE, &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 
172 1ac119fb 2024-01-23 op 	return 1;
173 1ac119fb 2024-01-23 op }
174 1ac119fb 2024-01-23 op 
175 1ac119fb 2024-01-23 op static inline const char *
176 1ac119fb 2024-01-23 op gopher_skip_selector(const char *path, int *ret_type)
177 1ac119fb 2024-01-23 op {
178 1ac119fb 2024-01-23 op 	*ret_type = 0;
179 1ac119fb 2024-01-23 op 
180 1ac119fb 2024-01-23 op 	if (!strcmp(path, "/") || *path == '\0') {
181 1ac119fb 2024-01-23 op 		*ret_type = '1';
182 1ac119fb 2024-01-23 op 		return path;
183 1ac119fb 2024-01-23 op 	}
184 1ac119fb 2024-01-23 op 
185 1ac119fb 2024-01-23 op 	if (*path != '/')
186 1ac119fb 2024-01-23 op 		return path;
187 1ac119fb 2024-01-23 op 	path++;
188 1ac119fb 2024-01-23 op 
189 1ac119fb 2024-01-23 op 	switch (*ret_type = *path) {
190 1ac119fb 2024-01-23 op 	case '0':
191 1ac119fb 2024-01-23 op 	case '1':
192 1ac119fb 2024-01-23 op 	case '7':
193 1ac119fb 2024-01-23 op 		break;
194 1ac119fb 2024-01-23 op 
195 1ac119fb 2024-01-23 op 	default:
196 1ac119fb 2024-01-23 op 		*ret_type = 0;
197 1ac119fb 2024-01-23 op 		path -= 1;
198 1ac119fb 2024-01-23 op 		return path;
199 1ac119fb 2024-01-23 op 	}
200 1ac119fb 2024-01-23 op 
201 1ac119fb 2024-01-23 op 	return ++path;
202 1ac119fb 2024-01-23 op }
203 1ac119fb 2024-01-23 op 
204 1ac119fb 2024-01-23 op static int
205 1ac119fb 2024-01-23 op serialize_link(struct line *line, const char *text, FILE *fp)
206 1ac119fb 2024-01-23 op {
207 1ac119fb 2024-01-23 op 	size_t		 portlen = 0;
208 1ac119fb 2024-01-23 op 	int		 type;
209 1ac119fb 2024-01-23 op 	const char	*uri, *endhost, *port, *path, *colon;
210 1ac119fb 2024-01-23 op 
211 1ac119fb 2024-01-23 op 	if ((uri = line->alt) == NULL)
212 1ac119fb 2024-01-23 op 		return -1;
213 1ac119fb 2024-01-23 op 
214 1ac119fb 2024-01-23 op 	if (strncmp(uri, "gopher://", 9) != 0)
215 1ac119fb 2024-01-23 op 		return fprintf(fp, "h%s\tURL:%s\terror.host\t1\n",
216 1ac119fb 2024-01-23 op 		    text, line->alt);
217 1ac119fb 2024-01-23 op 
218 1ac119fb 2024-01-23 op 	uri += 9; /* skip gopher:// */
219 1ac119fb 2024-01-23 op 
220 1ac119fb 2024-01-23 op 	path = strchr(uri, '/');
221 1ac119fb 2024-01-23 op 	colon = strchr(uri, ':');
222 1ac119fb 2024-01-23 op 
223 1ac119fb 2024-01-23 op 	if (path != NULL && colon > path)
224 1ac119fb 2024-01-23 op 		colon = NULL;
225 1ac119fb 2024-01-23 op 
226 1ac119fb 2024-01-23 op 	if ((endhost = colon) == NULL &&
227 1ac119fb 2024-01-23 op 	    (endhost = path) == NULL)
228 1ac119fb 2024-01-23 op 		endhost = strchr(uri, '\0');
229 1ac119fb 2024-01-23 op 
230 1ac119fb 2024-01-23 op 	if (colon != NULL) {
231 1ac119fb 2024-01-23 op 		for (port = colon+1; *port && *port != '/'; ++port)
232 1ac119fb 2024-01-23 op 			++portlen;
233 1ac119fb 2024-01-23 op 		port = colon+1;
234 1ac119fb 2024-01-23 op 	} else {
235 1ac119fb 2024-01-23 op 		port = "70";
236 1ac119fb 2024-01-23 op 		portlen = 2;
237 1ac119fb 2024-01-23 op 	}
238 1ac119fb 2024-01-23 op 
239 1ac119fb 2024-01-23 op 	if (path == NULL) {
240 1ac119fb 2024-01-23 op 		type = '1';
241 1ac119fb 2024-01-23 op 		path = "";
242 1ac119fb 2024-01-23 op 	} else
243 1ac119fb 2024-01-23 op 		path = gopher_skip_selector(path, &type);
244 1ac119fb 2024-01-23 op 
245 1ac119fb 2024-01-23 op 	return fprintf(fp, "%c%s\t%s\t%.*s\t%.*s\n", type, text,
246 1ac119fb 2024-01-23 op 	    path, (int)(endhost - uri), uri, (int)portlen, port);
247 1ac119fb 2024-01-23 op }
248 1ac119fb 2024-01-23 op 
249 1ac119fb 2024-01-23 op static int
250 c1d27b0e 2024-06-14 op gm_serialize(struct buffer *b, FILE *fp)
251 1ac119fb 2024-01-23 op {
252 1ac119fb 2024-01-23 op 	struct line	*line;
253 1ac119fb 2024-01-23 op 	const char	*text;
254 1ac119fb 2024-01-23 op 	int		 r;
255 1ac119fb 2024-01-23 op 
256 c1d27b0e 2024-06-14 op 	TAILQ_FOREACH(line, &b->head, lines) {
257 1ac119fb 2024-01-23 op 		if ((text = line->line) == NULL)
258 1ac119fb 2024-01-23 op 			text = "";
259 1ac119fb 2024-01-23 op 
260 1ac119fb 2024-01-23 op 		switch (line->type) {
261 1ac119fb 2024-01-23 op 		case LINE_LINK:
262 1ac119fb 2024-01-23 op 			r = serialize_link(line, text, fp);
263 1ac119fb 2024-01-23 op 			break;
264 1ac119fb 2024-01-23 op 
265 1ac119fb 2024-01-23 op 		case LINE_TEXT:
266 1ac119fb 2024-01-23 op 			r = fprintf(fp, "i%s\t\terror.host\t1\n", text);
267 1ac119fb 2024-01-23 op 			break;
268 1ac119fb 2024-01-23 op 
269 1ac119fb 2024-01-23 op 		case LINE_QUOTE:
270 1ac119fb 2024-01-23 op 			r = fprintf(fp, "3%s\t\terror.host\t1\n", text);
271 1ac119fb 2024-01-23 op 			break;
272 1ac119fb 2024-01-23 op 
273 1ac119fb 2024-01-23 op 		default:
274 1ac119fb 2024-01-23 op 			/* unreachable */
275 1ac119fb 2024-01-23 op 			abort();
276 1ac119fb 2024-01-23 op 		}
277 1ac119fb 2024-01-23 op 
278 1ac119fb 2024-01-23 op 		if (r == -1)
279 1ac119fb 2024-01-23 op 			return 0;
280 1ac119fb 2024-01-23 op 	}
281 1ac119fb 2024-01-23 op 
282 1ac119fb 2024-01-23 op 	return 1;
283 1ac119fb 2024-01-23 op }