2 5e11c00c 2021-03-02 op * Copyright (c) 2021 Omar Polo <op@omarpolo.com>
4 5e11c00c 2021-03-02 op * Permission to use, copy, modify, and distribute this software for any
5 5e11c00c 2021-03-02 op * purpose with or without fee is hereby granted, provided that the above
6 5e11c00c 2021-03-02 op * copyright notice and this permission notice appear in all copies.
8 5e11c00c 2021-03-02 op * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 5e11c00c 2021-03-02 op * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 5e11c00c 2021-03-02 op * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 5e11c00c 2021-03-02 op * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 5e11c00c 2021-03-02 op * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 5e11c00c 2021-03-02 op * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 5e11c00c 2021-03-02 op * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18 5e11c00c 2021-03-02 op * A streaming gemtext parser.
21 5e11c00c 2021-03-02 op * - handle NULs
25 a4ec04c5 2021-05-13 op #include "telescope.h"
27 5e11c00c 2021-03-02 op #include <ctype.h>
28 5e11c00c 2021-03-02 op #include <string.h>
29 5e11c00c 2021-03-02 op #include <stdlib.h>
31 5e11c00c 2021-03-02 op static int gemtext_parse(struct parser*, const char*, size_t);
32 a5845bb5 2021-03-20 op static int gemtext_foreach_line(struct parser*, const char*, size_t);
33 a5c3e03d 2021-03-02 op static int gemtext_free(struct parser*);
35 5e11c00c 2021-03-02 op static int parse_text(struct parser*, enum line_type, const char*, size_t);
36 5e11c00c 2021-03-02 op static int parse_link(struct parser*, enum line_type, const char*, size_t);
37 5e11c00c 2021-03-02 op static int parse_title(struct parser*, enum line_type, const char*, size_t);
38 5e11c00c 2021-03-02 op static int parse_item(struct parser*, enum line_type, const char*, size_t);
39 5e11c00c 2021-03-02 op static int parse_quote(struct parser*, enum line_type, const char*, size_t);
40 5e11c00c 2021-03-02 op static int parse_pre_start(struct parser*, enum line_type, const char*, size_t);
41 5e11c00c 2021-03-02 op static int parse_pre_cnt(struct parser*, enum line_type, const char*, size_t);
42 5e11c00c 2021-03-02 op static int parse_pre_end(struct parser*, enum line_type, const char*, size_t);
43 48e3f26d 2021-05-13 op static void search_title(struct parser*, enum line_type);
45 5e11c00c 2021-03-02 op typedef int (parselinefn)(struct parser*, enum line_type, const char*, size_t);
47 5e11c00c 2021-03-02 op static parselinefn *parsers[] = {
48 5e11c00c 2021-03-02 op parse_text, /* LINE_TEXT */
49 5e11c00c 2021-03-02 op parse_link, /* LINE_LINK */
50 5e11c00c 2021-03-02 op parse_title, /* LINE_TITLE_1 */
51 5e11c00c 2021-03-02 op parse_title, /* LINE_TITLE_2 */
52 5e11c00c 2021-03-02 op parse_title, /* LINE_TITLE_3 */
53 5e11c00c 2021-03-02 op parse_item, /* LINE_ITEM */
54 5e11c00c 2021-03-02 op parse_quote, /* LINE_QUOTE */
55 5e11c00c 2021-03-02 op parse_pre_start, /* LINE_PRE_START */
56 5e11c00c 2021-03-02 op parse_pre_cnt, /* LINE_PRE_CONTENT */
57 5e11c00c 2021-03-02 op parse_pre_end, /* LINE_PRE_END */
61 5e11c00c 2021-03-02 op gemtext_initparser(struct parser *p)
63 5e11c00c 2021-03-02 op memset(p, 0, sizeof(*p));
65 fc43eadd 2021-03-12 op p->name = "text/gemini";
66 5e11c00c 2021-03-02 op p->parse = &gemtext_parse;
67 5e11c00c 2021-03-02 op p->free = &gemtext_free;
70 5e11c00c 2021-03-02 op static inline int
71 5e11c00c 2021-03-02 op emit_line(struct parser *p, enum line_type type, char *line, char *alt)
73 5e11c00c 2021-03-02 op struct line *l;
75 5e11c00c 2021-03-02 op if ((l = calloc(1, sizeof(*l))) == NULL)
78 5e11c00c 2021-03-02 op l->type = type;
79 5e11c00c 2021-03-02 op l->line = line;
82 5e11c00c 2021-03-02 op if (TAILQ_EMPTY(&p->head))
83 5e11c00c 2021-03-02 op TAILQ_INSERT_HEAD(&p->head, l, lines);
85 5e11c00c 2021-03-02 op TAILQ_INSERT_TAIL(&p->head, l, lines);
91 5e11c00c 2021-03-02 op parse_text(struct parser *p, enum line_type t, const char *buf, size_t len)
95 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
97 5e11c00c 2021-03-02 op memcpy(l, buf, len);
98 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
102 5e11c00c 2021-03-02 op parse_link(struct parser *p, enum line_type t, const char *buf, size_t len)
105 5e11c00c 2021-03-02 op const char *url_start;
107 5e11c00c 2021-03-02 op if (len <= 2)
108 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
112 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
117 5e11c00c 2021-03-02 op if (len == 0)
118 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
120 5e11c00c 2021-03-02 op url_start = buf;
121 5e11c00c 2021-03-02 op while (len > 0 && !isspace(buf[0])) {
126 5e11c00c 2021-03-02 op if ((u = calloc(1, buf - url_start + 1)) == NULL)
128 5e11c00c 2021-03-02 op memcpy(u, url_start, buf - url_start);
130 5e11c00c 2021-03-02 op if (len == 0)
131 f250a75b 2021-03-06 op goto nolabel;
133 f250a75b 2021-03-06 op while (len > 0 && isspace(buf[0])) {
138 5e11c00c 2021-03-02 op if (len == 0)
139 f250a75b 2021-03-06 op goto nolabel;
141 5e11c00c 2021-03-02 op if ((l = calloc(1, len + 1)) == NULL)
144 5e11c00c 2021-03-02 op memcpy(l, buf, len);
145 f250a75b 2021-03-06 op return emit_line(p, t, l, u);
148 f250a75b 2021-03-06 op if ((l = strdup(u)) == NULL)
150 f250a75b 2021-03-06 op return emit_line(p, t, l, u);
154 5e11c00c 2021-03-02 op parse_title(struct parser *p, enum line_type t, const char *buf, size_t len)
159 5e11c00c 2021-03-02 op case LINE_TITLE_1:
160 5e11c00c 2021-03-02 op if (len <= 1)
161 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
165 5e11c00c 2021-03-02 op case LINE_TITLE_2:
166 5e11c00c 2021-03-02 op if (len <= 2)
167 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
171 5e11c00c 2021-03-02 op case LINE_TITLE_3:
172 5e11c00c 2021-03-02 op if (len <= 3)
173 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
178 5e11c00c 2021-03-02 op /* unreachable */
182 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
187 5e11c00c 2021-03-02 op if (len == 0)
188 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
190 dc5df781 2021-03-13 op if (t == LINE_TITLE_1 && *p->title == '\0')
191 dc5df781 2021-03-13 op strncpy(p->title, buf, MIN(sizeof(p->title)-1, len));
193 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
195 5e11c00c 2021-03-02 op memcpy(l, buf, len);
196 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
200 5e11c00c 2021-03-02 op parse_item(struct parser *p, enum line_type t, const char *buf, size_t len)
204 5e11c00c 2021-03-02 op if (len == 1)
205 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
210 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
215 5e11c00c 2021-03-02 op if (len == 0)
216 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
218 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
220 5e11c00c 2021-03-02 op memcpy(l, buf, len);
221 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
225 5e11c00c 2021-03-02 op parse_quote(struct parser *p, enum line_type t, const char *buf, size_t len)
229 5e11c00c 2021-03-02 op if (len == 1)
230 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
235 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
240 5e11c00c 2021-03-02 op if (len == 0)
241 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
243 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
245 5e11c00c 2021-03-02 op memcpy(l, buf, len);
246 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
250 5e11c00c 2021-03-02 op parse_pre_start(struct parser *p, enum line_type t, const char *buf, size_t len)
254 5e11c00c 2021-03-02 op if (len <= 3)
255 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
260 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
265 5e11c00c 2021-03-02 op if (len == 0)
266 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
268 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
271 5e11c00c 2021-03-02 op memcpy(l, buf, len);
272 e2060f05 2021-03-17 op return emit_line(p, t, l, NULL);
276 5e11c00c 2021-03-02 op parse_pre_cnt(struct parser *p, enum line_type t, const char *buf, size_t len)
280 5e11c00c 2021-03-02 op if (len == 0)
281 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
283 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
285 5e11c00c 2021-03-02 op memcpy(l, buf, len);
286 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
290 5e11c00c 2021-03-02 op parse_pre_end(struct parser *p, enum line_type t, const char *buf, size_t len)
292 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
295 5e11c00c 2021-03-02 op static inline enum line_type
296 5e11c00c 2021-03-02 op detect_line_type(const char *buf, size_t len, int in_pre)
298 5e11c00c 2021-03-02 op if (len == 0)
299 5e11c00c 2021-03-02 op return LINE_TEXT;
301 5e11c00c 2021-03-02 op if (in_pre) {
302 5e11c00c 2021-03-02 op if (len >= 3 &&
303 5e11c00c 2021-03-02 op buf[0] == '`' && buf[1] == '`' && buf[2] == '`')
304 5e11c00c 2021-03-02 op return LINE_PRE_END;
306 5e11c00c 2021-03-02 op return LINE_PRE_CONTENT;
309 5e11c00c 2021-03-02 op switch (*buf) {
310 5e11c00c 2021-03-02 op case '*': return LINE_ITEM;
311 5e11c00c 2021-03-02 op case '>': return LINE_QUOTE;
313 5e11c00c 2021-03-02 op if (len >= 1 && buf[1] == '>')
314 5e11c00c 2021-03-02 op return LINE_LINK;
317 5e11c00c 2021-03-02 op if (len == 1)
318 5e11c00c 2021-03-02 op return LINE_TEXT;
319 5e11c00c 2021-03-02 op if (buf[1] != '#')
320 5e11c00c 2021-03-02 op return LINE_TITLE_1;
321 5e11c00c 2021-03-02 op if (len == 2)
322 5e11c00c 2021-03-02 op return LINE_TEXT;
323 5e11c00c 2021-03-02 op if (buf[2] != '#')
324 5e11c00c 2021-03-02 op return LINE_TITLE_2;
325 5e11c00c 2021-03-02 op if (len == 3)
326 5e11c00c 2021-03-02 op return LINE_TEXT;
327 5e11c00c 2021-03-02 op return LINE_TITLE_3;
330 5e11c00c 2021-03-02 op return LINE_TEXT;
331 5e11c00c 2021-03-02 op if (buf[0] == '`' && buf[1] == '`' && buf[2] == '`')
332 5e11c00c 2021-03-02 op return LINE_PRE_START;
336 5e11c00c 2021-03-02 op return LINE_TEXT;
340 5e11c00c 2021-03-02 op gemtext_parse(struct parser *p, const char *buf, size_t size)
342 a5845bb5 2021-03-20 op return parser_foreach_line(p, buf, size, gemtext_foreach_line);
346 a5845bb5 2021-03-20 op gemtext_foreach_line(struct parser *p, const char *line, size_t linelen)
348 a5845bb5 2021-03-20 op enum line_type t;
350 a5845bb5 2021-03-20 op t = detect_line_type(line, linelen, p->flags);
351 a5845bb5 2021-03-20 op if (t == LINE_PRE_START)
352 a5845bb5 2021-03-20 op p->flags = 1;
353 a5845bb5 2021-03-20 op if (t == LINE_PRE_END)
354 a5845bb5 2021-03-20 op p->flags = 0;
355 a5845bb5 2021-03-20 op return parsers[t](p, t, line, linelen);
359 5e11c00c 2021-03-02 op gemtext_free(struct parser *p)
361 a5c3e03d 2021-03-02 op enum line_type t;
363 a5c3e03d 2021-03-02 op /* flush the buffer */
364 a5c3e03d 2021-03-02 op if (p->len != 0) {
365 a5c3e03d 2021-03-02 op t = detect_line_type(p->buf, p->len, p->flags);
366 a5c3e03d 2021-03-02 op if (!parsers[t](p, t, p->buf, p->len))
368 a5c3e03d 2021-03-02 op if (p->flags && !emit_line(p, LINE_PRE_END, NULL, NULL))
372 5e11c00c 2021-03-02 op free(p->buf);
375 48e3f26d 2021-05-13 op * use the first level 2 or 3 header as page title if none
378 48e3f26d 2021-05-13 op if (*p->title == '\0')
379 48e3f26d 2021-05-13 op search_title(p, LINE_TITLE_2);
380 48e3f26d 2021-05-13 op if (*p->title == '\0')
381 48e3f26d 2021-05-13 op search_title(p, LINE_TITLE_3);
387 48e3f26d 2021-05-13 op search_title(struct parser *p, enum line_type level)
389 48e3f26d 2021-05-13 op struct line *l;
391 48e3f26d 2021-05-13 op TAILQ_FOREACH(l, &p->head, lines) {
392 48e3f26d 2021-05-13 op if (l->type == level) {
393 48e3f26d 2021-05-13 op if (l->line == NULL)
395 48e3f26d 2021-05-13 op strlcpy(p->title, l->line, sizeof(p->title));