1 1ac119fb 2024-01-23 op /*
  2 1ac119fb 2024-01-23 op  * Copyright (c) 2021, 2022 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 /*
 18 1ac119fb 2024-01-23 op  * A streaming gemtext parser.
 19 1ac119fb 2024-01-23 op  *
 20 1ac119fb 2024-01-23 op  * TODO:
 21 1ac119fb 2024-01-23 op  *  - handle NULs
 22 1ac119fb 2024-01-23 op  *  - UTF8
 23 1ac119fb 2024-01-23 op  */
 24 1ac119fb 2024-01-23 op 
 25 1ac119fb 2024-01-23 op #include "compat.h"
 26 1ac119fb 2024-01-23 op 
 27 1ac119fb 2024-01-23 op #include <ctype.h>
 28 1ac119fb 2024-01-23 op #include <string.h>
 29 1ac119fb 2024-01-23 op #include <stdlib.h>
 30 1ac119fb 2024-01-23 op 
 31 1ac119fb 2024-01-23 op #include "defaults.h"
 32 1ac119fb 2024-01-23 op #include "parser.h"
 33 1ac119fb 2024-01-23 op #include "utf8.h"
 34 1ac119fb 2024-01-23 op 
 35 1ac119fb 2024-01-23 op static int	gemtext_parse(struct parser *, const char *, size_t);
 36 1ac119fb 2024-01-23 op static int	gemtext_foreach_line(struct parser *, const char *, size_t);
 37 1ac119fb 2024-01-23 op static int	gemtext_free(struct parser *);
 38 1ac119fb 2024-01-23 op static int	gemtext_serialize(struct parser *, FILE *);
 39 1ac119fb 2024-01-23 op 
 40 1ac119fb 2024-01-23 op static int	parse_text(struct parser*, enum line_type, const char*, size_t);
 41 1ac119fb 2024-01-23 op static int	parse_link(struct parser*, enum line_type, const char*, size_t);
 42 1ac119fb 2024-01-23 op static int	parse_title(struct parser*, enum line_type, const char*, size_t);
 43 1ac119fb 2024-01-23 op static int	parse_item(struct parser*, enum line_type, const char*, size_t);
 44 1ac119fb 2024-01-23 op static int	parse_quote(struct parser*, enum line_type, const char*, size_t);
 45 1ac119fb 2024-01-23 op static int	parse_pre_start(struct parser*, enum line_type, const char*, size_t);
 46 1ac119fb 2024-01-23 op static int	parse_pre_cnt(struct parser*, enum line_type, const char*, size_t);
 47 1ac119fb 2024-01-23 op static int	parse_pre_end(struct parser*, enum line_type, const char*, size_t);
 48 1ac119fb 2024-01-23 op static void	search_title(struct parser*, enum line_type);
 49 1ac119fb 2024-01-23 op 
 50 1ac119fb 2024-01-23 op typedef int (parselinefn)(struct parser*, enum line_type, const char*, size_t);
 51 1ac119fb 2024-01-23 op 
 52 1ac119fb 2024-01-23 op static parselinefn *parsers[] = {
 53 1ac119fb 2024-01-23 op 	[LINE_TEXT]		= parse_text,
 54 1ac119fb 2024-01-23 op 	[LINE_LINK]		= parse_link,
 55 1ac119fb 2024-01-23 op 	[LINE_TITLE_1]		= parse_title,
 56 1ac119fb 2024-01-23 op 	[LINE_TITLE_2]		= parse_title,
 57 1ac119fb 2024-01-23 op 	[LINE_TITLE_3]		= parse_title,
 58 1ac119fb 2024-01-23 op 	[LINE_ITEM]		= parse_item,
 59 1ac119fb 2024-01-23 op 	[LINE_QUOTE]		= parse_quote,
 60 1ac119fb 2024-01-23 op 	[LINE_PRE_START]	= parse_pre_start,
 61 1ac119fb 2024-01-23 op 	[LINE_PRE_CONTENT]	= parse_pre_cnt,
 62 1ac119fb 2024-01-23 op 	[LINE_PRE_END]		= parse_pre_end,
 63 1ac119fb 2024-01-23 op };
 64 1ac119fb 2024-01-23 op 
 65 1ac119fb 2024-01-23 op void
 66 1ac119fb 2024-01-23 op gemtext_initparser(struct parser *p)
 67 1ac119fb 2024-01-23 op {
 68 1ac119fb 2024-01-23 op 	memset(p, 0, sizeof(*p));
 69 1ac119fb 2024-01-23 op 
 70 1ac119fb 2024-01-23 op 	p->name = "text/gemini";
 71 1ac119fb 2024-01-23 op 	p->parse = &gemtext_parse;
 72 1ac119fb 2024-01-23 op 	p->free  = &gemtext_free;
 73 1ac119fb 2024-01-23 op 	p->serialize = &gemtext_serialize;
 74 1ac119fb 2024-01-23 op 
 75 1ac119fb 2024-01-23 op 	TAILQ_INIT(&p->head);
 76 1ac119fb 2024-01-23 op }
 77 1ac119fb 2024-01-23 op 
 78 1ac119fb 2024-01-23 op static inline int
 79 1ac119fb 2024-01-23 op emit_line(struct parser *p, enum line_type type, char *line, char *alt)
 80 1ac119fb 2024-01-23 op {
 81 1ac119fb 2024-01-23 op 	struct line *l;
 82 1ac119fb 2024-01-23 op 
 83 1ac119fb 2024-01-23 op 	if ((l = calloc(1, sizeof(*l))) == NULL)
 84 1ac119fb 2024-01-23 op 		return 0;
 85 1ac119fb 2024-01-23 op 
 86 1ac119fb 2024-01-23 op 	l->type = type;
 87 1ac119fb 2024-01-23 op 	l->line = line;
 88 1ac119fb 2024-01-23 op 	l->alt = alt;
 89 1ac119fb 2024-01-23 op 
 90 1ac119fb 2024-01-23 op 	switch (l->type) {
 91 1ac119fb 2024-01-23 op 	case LINE_PRE_START:
 92 1ac119fb 2024-01-23 op 	case LINE_PRE_END:
 93 1ac119fb 2024-01-23 op 		if (hide_pre_context)
 94 1ac119fb 2024-01-23 op 			l->flags = L_HIDDEN;
 95 1ac119fb 2024-01-23 op 		if (l->type == LINE_PRE_END &&
 96 1ac119fb 2024-01-23 op 		    hide_pre_closing_line)
 97 1ac119fb 2024-01-23 op 			l->flags = L_HIDDEN;
 98 1ac119fb 2024-01-23 op 		break;
 99 1ac119fb 2024-01-23 op 	case LINE_PRE_CONTENT:
100 1ac119fb 2024-01-23 op 		if (hide_pre_blocks)
101 1ac119fb 2024-01-23 op 			l->flags = L_HIDDEN;
102 1ac119fb 2024-01-23 op 		break;
103 1ac119fb 2024-01-23 op 	case LINE_LINK:
104 1ac119fb 2024-01-23 op 		if (emojify_link &&
105 1ac119fb 2024-01-23 op 		    !emojied_line(line, (const char **)&l->data))
106 1ac119fb 2024-01-23 op 			l->data = NULL;
107 1ac119fb 2024-01-23 op 		break;
108 1ac119fb 2024-01-23 op 	default:
109 1ac119fb 2024-01-23 op 		break;
110 1ac119fb 2024-01-23 op 	}
111 1ac119fb 2024-01-23 op 
112 1ac119fb 2024-01-23 op 	TAILQ_INSERT_TAIL(&p->head, l, lines);
113 1ac119fb 2024-01-23 op 
114 1ac119fb 2024-01-23 op 	return 1;
115 1ac119fb 2024-01-23 op }
116 1ac119fb 2024-01-23 op 
117 1ac119fb 2024-01-23 op static int
118 1ac119fb 2024-01-23 op parse_text(struct parser *p, enum line_type t, const char *buf, size_t len)
119 1ac119fb 2024-01-23 op {
120 1ac119fb 2024-01-23 op 	char *l;
121 1ac119fb 2024-01-23 op 
122 1ac119fb 2024-01-23 op 	if ((l = calloc(1, len+1)) == NULL)
123 1ac119fb 2024-01-23 op 		return 0;
124 1ac119fb 2024-01-23 op 	memcpy(l, buf, len);
125 1ac119fb 2024-01-23 op 	return emit_line(p, t, l, NULL);
126 1ac119fb 2024-01-23 op }
127 1ac119fb 2024-01-23 op 
128 1ac119fb 2024-01-23 op static int
129 1ac119fb 2024-01-23 op parse_link(struct parser *p, enum line_type t, const char *buf, size_t len)
130 1ac119fb 2024-01-23 op {
131 1ac119fb 2024-01-23 op 	char *l, *u;
132 1ac119fb 2024-01-23 op 	const char *url_start;
133 1ac119fb 2024-01-23 op 
134 1ac119fb 2024-01-23 op 	if (len <= 2)
135 1ac119fb 2024-01-23 op 		return emit_line(p, LINE_TEXT, NULL, NULL);
136 1ac119fb 2024-01-23 op 	buf += 2;
137 1ac119fb 2024-01-23 op 	len -= 2;
138 1ac119fb 2024-01-23 op 
139 1ac119fb 2024-01-23 op 	while (len > 0 && isspace(buf[0])) {
140 1ac119fb 2024-01-23 op 		buf++;
141 1ac119fb 2024-01-23 op 		len--;
142 1ac119fb 2024-01-23 op 	}
143 1ac119fb 2024-01-23 op 
144 1ac119fb 2024-01-23 op 	if (len == 0)
145 1ac119fb 2024-01-23 op 		return emit_line(p, LINE_TEXT, NULL, NULL);
146 1ac119fb 2024-01-23 op 
147 1ac119fb 2024-01-23 op 	url_start = buf;
148 1ac119fb 2024-01-23 op 	while (len > 0 && !isspace(buf[0])) {
149 1ac119fb 2024-01-23 op 		buf++;
150 1ac119fb 2024-01-23 op 		len--;
151 1ac119fb 2024-01-23 op 	}
152 1ac119fb 2024-01-23 op 
153 1ac119fb 2024-01-23 op 	if ((u = calloc(1, buf - url_start + 1)) == NULL)
154 1ac119fb 2024-01-23 op 		return 0;
155 1ac119fb 2024-01-23 op 	memcpy(u, url_start, buf - url_start);
156 1ac119fb 2024-01-23 op 
157 1ac119fb 2024-01-23 op 	if (len == 0)
158 1ac119fb 2024-01-23 op 		goto nolabel;
159 1ac119fb 2024-01-23 op 
160 1ac119fb 2024-01-23 op 	while (len > 0 && isspace(buf[0])) {
161 1ac119fb 2024-01-23 op 		buf++;
162 1ac119fb 2024-01-23 op 		len--;
163 1ac119fb 2024-01-23 op 	}
164 1ac119fb 2024-01-23 op 
165 1ac119fb 2024-01-23 op 	if (len == 0)
166 1ac119fb 2024-01-23 op 		goto nolabel;
167 1ac119fb 2024-01-23 op 
168 1ac119fb 2024-01-23 op 	if ((l = calloc(1, len + 1)) == NULL)
169 1ac119fb 2024-01-23 op 		return 0;
170 1ac119fb 2024-01-23 op 
171 1ac119fb 2024-01-23 op 	memcpy(l, buf, len);
172 1ac119fb 2024-01-23 op 	return emit_line(p, t, l, u);
173 1ac119fb 2024-01-23 op 
174 1ac119fb 2024-01-23 op nolabel:
175 1ac119fb 2024-01-23 op 	if ((l = strdup(u)) == NULL)
176 1ac119fb 2024-01-23 op 		return 0;
177 1ac119fb 2024-01-23 op 	return emit_line(p, t, l, u);
178 1ac119fb 2024-01-23 op }
179 1ac119fb 2024-01-23 op 
180 1ac119fb 2024-01-23 op static int
181 1ac119fb 2024-01-23 op parse_title(struct parser *p, enum line_type t, const char *buf, size_t len)
182 1ac119fb 2024-01-23 op {
183 1ac119fb 2024-01-23 op 	char *l;
184 1ac119fb 2024-01-23 op 
185 1ac119fb 2024-01-23 op 	switch (t) {
186 1ac119fb 2024-01-23 op 	case LINE_TITLE_1:
187 1ac119fb 2024-01-23 op 		if (len <= 1)
188 1ac119fb 2024-01-23 op 			return emit_line(p, t, NULL, NULL);
189 1ac119fb 2024-01-23 op 		buf++;
190 1ac119fb 2024-01-23 op 		len--;
191 1ac119fb 2024-01-23 op 		break;
192 1ac119fb 2024-01-23 op 	case LINE_TITLE_2:
193 1ac119fb 2024-01-23 op 		if (len <= 2)
194 1ac119fb 2024-01-23 op 			return emit_line(p, t, NULL, NULL);
195 1ac119fb 2024-01-23 op 		buf += 2;
196 1ac119fb 2024-01-23 op 		len -= 2;
197 1ac119fb 2024-01-23 op 		break;
198 1ac119fb 2024-01-23 op 	case LINE_TITLE_3:
199 1ac119fb 2024-01-23 op 		if (len <= 3)
200 1ac119fb 2024-01-23 op 			return emit_line(p, t, NULL, NULL);
201 1ac119fb 2024-01-23 op 		buf += 3;
202 1ac119fb 2024-01-23 op 		len -= 3;
203 1ac119fb 2024-01-23 op 		break;
204 1ac119fb 2024-01-23 op 	default:
205 1ac119fb 2024-01-23 op 		/* unreachable */
206 1ac119fb 2024-01-23 op 		abort();
207 1ac119fb 2024-01-23 op 	}
208 1ac119fb 2024-01-23 op 
209 1ac119fb 2024-01-23 op 	while (len > 0 && isspace(buf[0])) {
210 1ac119fb 2024-01-23 op 		buf++;
211 1ac119fb 2024-01-23 op 		len--;
212 1ac119fb 2024-01-23 op 	}
213 1ac119fb 2024-01-23 op 
214 1ac119fb 2024-01-23 op 	if (len == 0)
215 1ac119fb 2024-01-23 op 		return emit_line(p, t, NULL, NULL);
216 1ac119fb 2024-01-23 op 
217 1ac119fb 2024-01-23 op 	if (t == LINE_TITLE_1 && *p->title == '\0')
218 1ac119fb 2024-01-23 op 		strncpy(p->title, buf, MIN(sizeof(p->title)-1, len));
219 1ac119fb 2024-01-23 op 
220 1ac119fb 2024-01-23 op 	if ((l = calloc(1, len+1)) == NULL)
221 1ac119fb 2024-01-23 op 		return 0;
222 1ac119fb 2024-01-23 op 	memcpy(l, buf, len);
223 1ac119fb 2024-01-23 op 	return emit_line(p, t, l, NULL);
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 parse_item(struct parser *p, enum line_type t, const char *buf, size_t len)
228 1ac119fb 2024-01-23 op {
229 1ac119fb 2024-01-23 op 	char *l;
230 1ac119fb 2024-01-23 op 
231 1ac119fb 2024-01-23 op 	if (len == 1)
232 1ac119fb 2024-01-23 op 		return emit_line(p, t, NULL, NULL);
233 1ac119fb 2024-01-23 op 
234 1ac119fb 2024-01-23 op 	buf++;
235 1ac119fb 2024-01-23 op 	len--;
236 1ac119fb 2024-01-23 op 
237 1ac119fb 2024-01-23 op 	while (len > 0 && isspace(buf[0])) {
238 1ac119fb 2024-01-23 op 		buf++;
239 1ac119fb 2024-01-23 op 		len--;
240 1ac119fb 2024-01-23 op 	}
241 1ac119fb 2024-01-23 op 
242 1ac119fb 2024-01-23 op 	if (len == 0)
243 1ac119fb 2024-01-23 op 		return emit_line(p, t, NULL, NULL);
244 1ac119fb 2024-01-23 op 
245 1ac119fb 2024-01-23 op 	if ((l = calloc(1, len+1)) == NULL)
246 1ac119fb 2024-01-23 op 		return 0;
247 1ac119fb 2024-01-23 op 	memcpy(l, buf, len);
248 1ac119fb 2024-01-23 op 	return emit_line(p, t, l, NULL);
249 1ac119fb 2024-01-23 op }
250 1ac119fb 2024-01-23 op 
251 1ac119fb 2024-01-23 op static int
252 1ac119fb 2024-01-23 op parse_quote(struct parser *p, enum line_type t, const char *buf, size_t len)
253 1ac119fb 2024-01-23 op {
254 1ac119fb 2024-01-23 op 	char *l;
255 1ac119fb 2024-01-23 op 
256 1ac119fb 2024-01-23 op 	if (len == 1)
257 1ac119fb 2024-01-23 op 		return emit_line(p, t, NULL, NULL);
258 1ac119fb 2024-01-23 op 
259 1ac119fb 2024-01-23 op 	buf++;
260 1ac119fb 2024-01-23 op 	len--;
261 1ac119fb 2024-01-23 op 
262 1ac119fb 2024-01-23 op 	while (len > 0 && isspace(buf[0])) {
263 1ac119fb 2024-01-23 op 		buf++;
264 1ac119fb 2024-01-23 op 		len--;
265 1ac119fb 2024-01-23 op 	}
266 1ac119fb 2024-01-23 op 
267 1ac119fb 2024-01-23 op 	if (len == 0)
268 1ac119fb 2024-01-23 op 		return emit_line(p, t, NULL, NULL);
269 1ac119fb 2024-01-23 op 
270 1ac119fb 2024-01-23 op 	if ((l = calloc(1, len+1)) == NULL)
271 1ac119fb 2024-01-23 op 		return 0;
272 1ac119fb 2024-01-23 op 	memcpy(l, buf, len);
273 1ac119fb 2024-01-23 op 	return emit_line(p, t, l, NULL);
274 1ac119fb 2024-01-23 op }
275 1ac119fb 2024-01-23 op 
276 1ac119fb 2024-01-23 op static int
277 1ac119fb 2024-01-23 op parse_pre_start(struct parser *p, enum line_type t, const char *buf, size_t len)
278 1ac119fb 2024-01-23 op {
279 1ac119fb 2024-01-23 op 	char *l;
280 1ac119fb 2024-01-23 op 
281 1ac119fb 2024-01-23 op 	if (len <= 3)
282 1ac119fb 2024-01-23 op 		return emit_line(p, t, NULL, NULL);
283 1ac119fb 2024-01-23 op 
284 1ac119fb 2024-01-23 op 	buf += 3;
285 1ac119fb 2024-01-23 op 	len -= 3;
286 1ac119fb 2024-01-23 op 
287 1ac119fb 2024-01-23 op 	while (len > 0 && isspace(buf[0])) {
288 1ac119fb 2024-01-23 op 		buf++;
289 1ac119fb 2024-01-23 op 		len--;
290 1ac119fb 2024-01-23 op 	}
291 1ac119fb 2024-01-23 op 
292 1ac119fb 2024-01-23 op 	if (len == 0)
293 1ac119fb 2024-01-23 op 		return emit_line(p, t, NULL, NULL);
294 1ac119fb 2024-01-23 op 
295 1ac119fb 2024-01-23 op 	if ((l = calloc(1, len+1)) == NULL)
296 1ac119fb 2024-01-23 op 		return 0;
297 1ac119fb 2024-01-23 op 
298 1ac119fb 2024-01-23 op 	memcpy(l, buf, len);
299 1ac119fb 2024-01-23 op 	return emit_line(p, t, l, NULL);
300 1ac119fb 2024-01-23 op }
301 1ac119fb 2024-01-23 op 
302 1ac119fb 2024-01-23 op static int
303 1ac119fb 2024-01-23 op parse_pre_cnt(struct parser *p, enum line_type t, const char *buf, size_t len)
304 1ac119fb 2024-01-23 op {
305 1ac119fb 2024-01-23 op 	char *l;
306 1ac119fb 2024-01-23 op 
307 1ac119fb 2024-01-23 op 	if (len == 0)
308 1ac119fb 2024-01-23 op 		return emit_line(p, t, NULL, NULL);
309 1ac119fb 2024-01-23 op 
310 1ac119fb 2024-01-23 op 	if ((l = calloc(1, len+1)) == NULL)
311 1ac119fb 2024-01-23 op 		return 0;
312 1ac119fb 2024-01-23 op 	memcpy(l, buf, len);
313 1ac119fb 2024-01-23 op 	return emit_line(p, t, l, NULL);
314 1ac119fb 2024-01-23 op }
315 1ac119fb 2024-01-23 op 
316 1ac119fb 2024-01-23 op static int
317 1ac119fb 2024-01-23 op parse_pre_end(struct parser *p, enum line_type t, const char *buf, size_t len)
318 1ac119fb 2024-01-23 op {
319 1ac119fb 2024-01-23 op 	return emit_line(p, t, NULL, NULL);
320 1ac119fb 2024-01-23 op }
321 1ac119fb 2024-01-23 op 
322 1ac119fb 2024-01-23 op static inline enum line_type
323 1ac119fb 2024-01-23 op detect_line_type(const char *buf, size_t len, int in_pre)
324 1ac119fb 2024-01-23 op {
325 1ac119fb 2024-01-23 op 	if (in_pre) {
326 1ac119fb 2024-01-23 op 		if (len >= 3 &&
327 1ac119fb 2024-01-23 op 		    buf[0] == '`' && buf[1] == '`' && buf[2] == '`')
328 1ac119fb 2024-01-23 op 			return LINE_PRE_END;
329 1ac119fb 2024-01-23 op 		else
330 1ac119fb 2024-01-23 op 			return LINE_PRE_CONTENT;
331 1ac119fb 2024-01-23 op 	}
332 1ac119fb 2024-01-23 op 
333 1ac119fb 2024-01-23 op 	if (len == 0)
334 1ac119fb 2024-01-23 op 		return LINE_TEXT;
335 1ac119fb 2024-01-23 op 
336 1ac119fb 2024-01-23 op 	switch (*buf) {
337 1ac119fb 2024-01-23 op 	case '*':
338 1ac119fb 2024-01-23 op 		if (len > 1 && buf[1] == ' ')
339 1ac119fb 2024-01-23 op 			return LINE_ITEM;
340 1ac119fb 2024-01-23 op 		break;
341 1ac119fb 2024-01-23 op 	case '>': return LINE_QUOTE;
342 1ac119fb 2024-01-23 op 	case '=':
343 1ac119fb 2024-01-23 op 		if (len >= 1 && buf[1] == '>')
344 1ac119fb 2024-01-23 op 			return LINE_LINK;
345 1ac119fb 2024-01-23 op 		break;
346 1ac119fb 2024-01-23 op 	case '#':
347 1ac119fb 2024-01-23 op 		if (len == 1)
348 1ac119fb 2024-01-23 op 			return LINE_TEXT;
349 1ac119fb 2024-01-23 op 		if (buf[1] != '#')
350 1ac119fb 2024-01-23 op 			return LINE_TITLE_1;
351 1ac119fb 2024-01-23 op 		if (len == 2)
352 1ac119fb 2024-01-23 op 			return LINE_TEXT;
353 1ac119fb 2024-01-23 op 		if (buf[2] != '#')
354 1ac119fb 2024-01-23 op 			return LINE_TITLE_2;
355 1ac119fb 2024-01-23 op 		if (len == 3)
356 1ac119fb 2024-01-23 op 			return LINE_TEXT;
357 1ac119fb 2024-01-23 op 		return LINE_TITLE_3;
358 1ac119fb 2024-01-23 op 	case '`':
359 1ac119fb 2024-01-23 op 		if (len < 3)
360 1ac119fb 2024-01-23 op 			return LINE_TEXT;
361 1ac119fb 2024-01-23 op 		if (buf[0] == '`' && buf[1] == '`' && buf[2] == '`')
362 1ac119fb 2024-01-23 op 			return LINE_PRE_START;
363 1ac119fb 2024-01-23 op 		break;
364 1ac119fb 2024-01-23 op 	}
365 1ac119fb 2024-01-23 op 
366 1ac119fb 2024-01-23 op 	return LINE_TEXT;
367 1ac119fb 2024-01-23 op }
368 1ac119fb 2024-01-23 op 
369 1ac119fb 2024-01-23 op static int
370 1ac119fb 2024-01-23 op gemtext_parse(struct parser *p, const char *buf, size_t size)
371 1ac119fb 2024-01-23 op {
372 1ac119fb 2024-01-23 op 	return parser_foreach_line(p, buf, size, gemtext_foreach_line);
373 1ac119fb 2024-01-23 op }
374 1ac119fb 2024-01-23 op 
375 1ac119fb 2024-01-23 op static int
376 1ac119fb 2024-01-23 op gemtext_foreach_line(struct parser *p, const char *line, size_t linelen)
377 1ac119fb 2024-01-23 op {
378 1ac119fb 2024-01-23 op 	enum line_type t;
379 1ac119fb 2024-01-23 op 
380 1ac119fb 2024-01-23 op 	t = detect_line_type(line, linelen, p->flags & PARSER_IN_PRE);
381 1ac119fb 2024-01-23 op 	if (t == LINE_PRE_START)
382 1ac119fb 2024-01-23 op 		p->flags ^= PARSER_IN_PRE;
383 1ac119fb 2024-01-23 op 	if (t == LINE_PRE_END)
384 1ac119fb 2024-01-23 op 		p->flags ^= PARSER_IN_PRE;
385 1ac119fb 2024-01-23 op 	return parsers[t](p, t, line, linelen);
386 1ac119fb 2024-01-23 op }
387 1ac119fb 2024-01-23 op 
388 1ac119fb 2024-01-23 op static int
389 1ac119fb 2024-01-23 op gemtext_free(struct parser *p)
390 1ac119fb 2024-01-23 op {
391 1ac119fb 2024-01-23 op 	enum line_type	t;
392 1ac119fb 2024-01-23 op 
393 1ac119fb 2024-01-23 op 	/* flush the buffer */
394 1ac119fb 2024-01-23 op 	if (p->len != 0) {
395 1ac119fb 2024-01-23 op 		t = detect_line_type(p->buf, p->len, p->flags & PARSER_IN_PRE);
396 1ac119fb 2024-01-23 op 		if (!parsers[t](p, t, p->buf, p->len))
397 1ac119fb 2024-01-23 op 			return 0;
398 1ac119fb 2024-01-23 op 		if ((p->flags & PARSER_IN_PRE) &&
399 1ac119fb 2024-01-23 op 		    !emit_line(p, LINE_PRE_END, NULL, NULL))
400 1ac119fb 2024-01-23 op 			return 0;
401 1ac119fb 2024-01-23 op 	}
402 1ac119fb 2024-01-23 op 
403 1ac119fb 2024-01-23 op 	free(p->buf);
404 1ac119fb 2024-01-23 op 
405 1ac119fb 2024-01-23 op 	/*
406 1ac119fb 2024-01-23 op 	 * use the first level 2 or 3 header as page title if none
407 1ac119fb 2024-01-23 op 	 * found yet.
408 1ac119fb 2024-01-23 op 	 */
409 1ac119fb 2024-01-23 op 	if (*p->title == '\0')
410 1ac119fb 2024-01-23 op 		search_title(p, LINE_TITLE_2);
411 1ac119fb 2024-01-23 op 	if (*p->title == '\0')
412 1ac119fb 2024-01-23 op 		search_title(p, LINE_TITLE_3);
413 1ac119fb 2024-01-23 op 
414 1ac119fb 2024-01-23 op 	return 1;
415 1ac119fb 2024-01-23 op }
416 1ac119fb 2024-01-23 op 
417 1ac119fb 2024-01-23 op static void
418 1ac119fb 2024-01-23 op search_title(struct parser *p, enum line_type level)
419 1ac119fb 2024-01-23 op {
420 1ac119fb 2024-01-23 op 	struct line *l;
421 1ac119fb 2024-01-23 op 
422 1ac119fb 2024-01-23 op 	TAILQ_FOREACH(l, &p->head, lines) {
423 1ac119fb 2024-01-23 op 		if (l->type == level) {
424 1ac119fb 2024-01-23 op 			if (l->line == NULL)
425 1ac119fb 2024-01-23 op 				continue;
426 1ac119fb 2024-01-23 op 			strlcpy(p->title, l->line, sizeof(p->title));
427 1ac119fb 2024-01-23 op 			break;
428 1ac119fb 2024-01-23 op 		}
429 1ac119fb 2024-01-23 op 	}
430 1ac119fb 2024-01-23 op }
431 1ac119fb 2024-01-23 op 
432 1ac119fb 2024-01-23 op static const char *gemtext_prefixes[] = {
433 1ac119fb 2024-01-23 op 	[LINE_TEXT] = "",
434 1ac119fb 2024-01-23 op 	[LINE_TITLE_1] = "# ",
435 1ac119fb 2024-01-23 op 	[LINE_TITLE_2] = "## ",
436 1ac119fb 2024-01-23 op 	[LINE_TITLE_3] = "### ",
437 1ac119fb 2024-01-23 op 	[LINE_ITEM] = "* ",
438 1ac119fb 2024-01-23 op 	[LINE_QUOTE] = "> ",
439 1ac119fb 2024-01-23 op 	[LINE_PRE_START] = "``` ",
440 1ac119fb 2024-01-23 op 	[LINE_PRE_CONTENT] = "",
441 1ac119fb 2024-01-23 op 	[LINE_PRE_END] = "```",
442 1ac119fb 2024-01-23 op };
443 1ac119fb 2024-01-23 op 
444 1ac119fb 2024-01-23 op static int
445 1ac119fb 2024-01-23 op gemtext_serialize(struct parser *p, FILE *fp)
446 1ac119fb 2024-01-23 op {
447 1ac119fb 2024-01-23 op 	struct line	*line;
448 1ac119fb 2024-01-23 op 	const char	*text;
449 1ac119fb 2024-01-23 op 	const char	*alt;
450 1ac119fb 2024-01-23 op 	int		 r;
451 1ac119fb 2024-01-23 op 
452 1ac119fb 2024-01-23 op 	TAILQ_FOREACH(line, &p->head, lines) {
453 1ac119fb 2024-01-23 op 		if ((text = line->line) == NULL)
454 1ac119fb 2024-01-23 op 			text = "";
455 1ac119fb 2024-01-23 op 
456 1ac119fb 2024-01-23 op 		if ((alt = line->alt) == NULL)
457 1ac119fb 2024-01-23 op 			alt = "";
458 1ac119fb 2024-01-23 op 
459 1ac119fb 2024-01-23 op 		switch (line->type) {
460 1ac119fb 2024-01-23 op 		case LINE_TEXT:
461 1ac119fb 2024-01-23 op 		case LINE_TITLE_1:
462 1ac119fb 2024-01-23 op 		case LINE_TITLE_2:
463 1ac119fb 2024-01-23 op 		case LINE_TITLE_3:
464 1ac119fb 2024-01-23 op 		case LINE_ITEM:
465 1ac119fb 2024-01-23 op 		case LINE_QUOTE:
466 1ac119fb 2024-01-23 op 		case LINE_PRE_START:
467 1ac119fb 2024-01-23 op 		case LINE_PRE_CONTENT:
468 1ac119fb 2024-01-23 op 		case LINE_PRE_END:
469 1ac119fb 2024-01-23 op 			r = fprintf(fp, "%s%s\n", gemtext_prefixes[line->type],
470 1ac119fb 2024-01-23 op 			    text);
471 1ac119fb 2024-01-23 op 			break;
472 1ac119fb 2024-01-23 op 
473 1ac119fb 2024-01-23 op 		case LINE_LINK:
474 1ac119fb 2024-01-23 op 			r = fprintf(fp, "=> %s %s\n", alt, text);
475 1ac119fb 2024-01-23 op 			break;
476 1ac119fb 2024-01-23 op 
477 1ac119fb 2024-01-23 op 		default:
478 1ac119fb 2024-01-23 op 			/* not reached */
479 1ac119fb 2024-01-23 op 			abort();
480 1ac119fb 2024-01-23 op 		}
481 1ac119fb 2024-01-23 op 
482 1ac119fb 2024-01-23 op 		if (r == -1)
483 1ac119fb 2024-01-23 op 			return 0;
484 1ac119fb 2024-01-23 op 	}
485 1ac119fb 2024-01-23 op 
486 1ac119fb 2024-01-23 op 	return 1;
487 1ac119fb 2024-01-23 op }