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 5f2dc0bd 2021-07-13 op #include "defaults.h"
26 395b9f4e 2021-07-12 op #include "parser.h"
27 a4ec04c5 2021-05-13 op #include "telescope.h"
29 5e11c00c 2021-03-02 op #include <ctype.h>
30 5e11c00c 2021-03-02 op #include <string.h>
31 5e11c00c 2021-03-02 op #include <stdlib.h>
33 5e11c00c 2021-03-02 op static int gemtext_parse(struct parser*, const char*, size_t);
34 a5845bb5 2021-03-20 op static int gemtext_foreach_line(struct parser*, const char*, size_t);
35 a5c3e03d 2021-03-02 op static int gemtext_free(struct parser*);
37 5e11c00c 2021-03-02 op static int parse_text(struct parser*, enum line_type, const char*, size_t);
38 5e11c00c 2021-03-02 op static int parse_link(struct parser*, enum line_type, const char*, size_t);
39 5e11c00c 2021-03-02 op static int parse_title(struct parser*, enum line_type, const char*, size_t);
40 5e11c00c 2021-03-02 op static int parse_item(struct parser*, enum line_type, const char*, size_t);
41 5e11c00c 2021-03-02 op static int parse_quote(struct parser*, enum line_type, const char*, size_t);
42 5e11c00c 2021-03-02 op static int parse_pre_start(struct parser*, enum line_type, const char*, size_t);
43 5e11c00c 2021-03-02 op static int parse_pre_cnt(struct parser*, enum line_type, const char*, size_t);
44 5e11c00c 2021-03-02 op static int parse_pre_end(struct parser*, enum line_type, const char*, size_t);
45 48e3f26d 2021-05-13 op static void search_title(struct parser*, enum line_type);
47 5e11c00c 2021-03-02 op typedef int (parselinefn)(struct parser*, enum line_type, const char*, size_t);
49 5e11c00c 2021-03-02 op static parselinefn *parsers[] = {
50 3902f7b9 2021-07-14 op [LINE_TEXT] = parse_text,
51 3902f7b9 2021-07-14 op [LINE_LINK] = parse_link,
52 3902f7b9 2021-07-14 op [LINE_TITLE_1] = parse_title,
53 6b80b8fe 2021-07-14 op [LINE_TITLE_2] = parse_title,
54 3902f7b9 2021-07-14 op [LINE_TITLE_3] = parse_title,
55 6b80b8fe 2021-07-14 op [LINE_ITEM] = parse_item,
56 3902f7b9 2021-07-14 op [LINE_QUOTE] = parse_quote,
57 3902f7b9 2021-07-14 op [LINE_PRE_START] = parse_pre_start,
58 6b80b8fe 2021-07-14 op [LINE_PRE_CONTENT] = parse_pre_cnt,
59 3902f7b9 2021-07-14 op [LINE_PRE_END] = parse_pre_end,
63 5e11c00c 2021-03-02 op gemtext_initparser(struct parser *p)
65 5e11c00c 2021-03-02 op memset(p, 0, sizeof(*p));
67 fc43eadd 2021-03-12 op p->name = "text/gemini";
68 5e11c00c 2021-03-02 op p->parse = &gemtext_parse;
69 5e11c00c 2021-03-02 op p->free = &gemtext_free;
72 5e11c00c 2021-03-02 op static inline int
73 5e11c00c 2021-03-02 op emit_line(struct parser *p, enum line_type type, char *line, char *alt)
75 5e11c00c 2021-03-02 op struct line *l;
77 5e11c00c 2021-03-02 op if ((l = calloc(1, sizeof(*l))) == NULL)
80 5e11c00c 2021-03-02 op l->type = type;
81 5e11c00c 2021-03-02 op l->line = line;
84 b1379f34 2021-07-05 op switch (l->type) {
85 b1379f34 2021-07-05 op case LINE_PRE_START:
86 b1379f34 2021-07-05 op case LINE_PRE_END:
87 b1379f34 2021-07-05 op if (hide_pre_context)
88 b1379f34 2021-07-05 op l->flags = L_HIDDEN;
90 b1379f34 2021-07-05 op case LINE_PRE_CONTENT:
91 b1379f34 2021-07-05 op if (hide_pre_blocks)
92 b1379f34 2021-07-05 op l->flags = L_HIDDEN;
99 5e11c00c 2021-03-02 op if (TAILQ_EMPTY(&p->head))
100 5e11c00c 2021-03-02 op TAILQ_INSERT_HEAD(&p->head, l, lines);
102 5e11c00c 2021-03-02 op TAILQ_INSERT_TAIL(&p->head, l, lines);
108 5e11c00c 2021-03-02 op parse_text(struct parser *p, enum line_type t, const char *buf, size_t len)
112 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
114 5e11c00c 2021-03-02 op memcpy(l, buf, len);
115 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
119 5e11c00c 2021-03-02 op parse_link(struct parser *p, enum line_type t, const char *buf, size_t len)
122 5e11c00c 2021-03-02 op const char *url_start;
124 5e11c00c 2021-03-02 op if (len <= 2)
125 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
129 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
134 5e11c00c 2021-03-02 op if (len == 0)
135 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
137 5e11c00c 2021-03-02 op url_start = buf;
138 5e11c00c 2021-03-02 op while (len > 0 && !isspace(buf[0])) {
143 5e11c00c 2021-03-02 op if ((u = calloc(1, buf - url_start + 1)) == NULL)
145 5e11c00c 2021-03-02 op memcpy(u, url_start, buf - url_start);
147 5e11c00c 2021-03-02 op if (len == 0)
148 f250a75b 2021-03-06 op goto nolabel;
150 f250a75b 2021-03-06 op while (len > 0 && isspace(buf[0])) {
155 5e11c00c 2021-03-02 op if (len == 0)
156 f250a75b 2021-03-06 op goto nolabel;
158 5e11c00c 2021-03-02 op if ((l = calloc(1, len + 1)) == NULL)
161 5e11c00c 2021-03-02 op memcpy(l, buf, len);
162 f250a75b 2021-03-06 op return emit_line(p, t, l, u);
165 f250a75b 2021-03-06 op if ((l = strdup(u)) == NULL)
167 f250a75b 2021-03-06 op return emit_line(p, t, l, u);
171 5e11c00c 2021-03-02 op parse_title(struct parser *p, enum line_type t, const char *buf, size_t len)
176 5e11c00c 2021-03-02 op case LINE_TITLE_1:
177 5e11c00c 2021-03-02 op if (len <= 1)
178 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
182 5e11c00c 2021-03-02 op case LINE_TITLE_2:
183 5e11c00c 2021-03-02 op if (len <= 2)
184 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
188 5e11c00c 2021-03-02 op case LINE_TITLE_3:
189 5e11c00c 2021-03-02 op if (len <= 3)
190 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
195 5e11c00c 2021-03-02 op /* unreachable */
199 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
204 5e11c00c 2021-03-02 op if (len == 0)
205 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
207 dc5df781 2021-03-13 op if (t == LINE_TITLE_1 && *p->title == '\0')
208 dc5df781 2021-03-13 op strncpy(p->title, buf, MIN(sizeof(p->title)-1, len));
210 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
212 5e11c00c 2021-03-02 op memcpy(l, buf, len);
213 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
217 5e11c00c 2021-03-02 op parse_item(struct parser *p, enum line_type t, const char *buf, size_t len)
221 5e11c00c 2021-03-02 op if (len == 1)
222 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
227 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
232 5e11c00c 2021-03-02 op if (len == 0)
233 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
235 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
237 5e11c00c 2021-03-02 op memcpy(l, buf, len);
238 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
242 5e11c00c 2021-03-02 op parse_quote(struct parser *p, enum line_type t, const char *buf, size_t len)
246 5e11c00c 2021-03-02 op if (len == 1)
247 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
252 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
257 5e11c00c 2021-03-02 op if (len == 0)
258 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
260 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
262 5e11c00c 2021-03-02 op memcpy(l, buf, len);
263 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
267 5e11c00c 2021-03-02 op parse_pre_start(struct parser *p, enum line_type t, const char *buf, size_t len)
271 5e11c00c 2021-03-02 op if (len <= 3)
272 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
277 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
282 5e11c00c 2021-03-02 op if (len == 0)
283 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
285 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
288 5e11c00c 2021-03-02 op memcpy(l, buf, len);
289 e2060f05 2021-03-17 op return emit_line(p, t, l, NULL);
293 5e11c00c 2021-03-02 op parse_pre_cnt(struct parser *p, enum line_type t, const char *buf, size_t len)
297 5e11c00c 2021-03-02 op if (len == 0)
298 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
300 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
302 5e11c00c 2021-03-02 op memcpy(l, buf, len);
303 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
307 5e11c00c 2021-03-02 op parse_pre_end(struct parser *p, enum line_type t, const char *buf, size_t len)
309 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
312 5e11c00c 2021-03-02 op static inline enum line_type
313 5e11c00c 2021-03-02 op detect_line_type(const char *buf, size_t len, int in_pre)
315 5e11c00c 2021-03-02 op if (in_pre) {
316 5e11c00c 2021-03-02 op if (len >= 3 &&
317 5e11c00c 2021-03-02 op buf[0] == '`' && buf[1] == '`' && buf[2] == '`')
318 5e11c00c 2021-03-02 op return LINE_PRE_END;
320 5e11c00c 2021-03-02 op return LINE_PRE_CONTENT;
323 1575cb8a 2021-07-07 op if (len == 0)
324 1575cb8a 2021-07-07 op return LINE_TEXT;
326 5e11c00c 2021-03-02 op switch (*buf) {
327 5e11c00c 2021-03-02 op case '*': return LINE_ITEM;
328 5e11c00c 2021-03-02 op case '>': return LINE_QUOTE;
330 5e11c00c 2021-03-02 op if (len >= 1 && buf[1] == '>')
331 5e11c00c 2021-03-02 op return LINE_LINK;
334 5e11c00c 2021-03-02 op if (len == 1)
335 5e11c00c 2021-03-02 op return LINE_TEXT;
336 5e11c00c 2021-03-02 op if (buf[1] != '#')
337 5e11c00c 2021-03-02 op return LINE_TITLE_1;
338 5e11c00c 2021-03-02 op if (len == 2)
339 5e11c00c 2021-03-02 op return LINE_TEXT;
340 5e11c00c 2021-03-02 op if (buf[2] != '#')
341 5e11c00c 2021-03-02 op return LINE_TITLE_2;
342 5e11c00c 2021-03-02 op if (len == 3)
343 5e11c00c 2021-03-02 op return LINE_TEXT;
344 5e11c00c 2021-03-02 op return LINE_TITLE_3;
347 5e11c00c 2021-03-02 op return LINE_TEXT;
348 5e11c00c 2021-03-02 op if (buf[0] == '`' && buf[1] == '`' && buf[2] == '`')
349 5e11c00c 2021-03-02 op return LINE_PRE_START;
353 5e11c00c 2021-03-02 op return LINE_TEXT;
357 5e11c00c 2021-03-02 op gemtext_parse(struct parser *p, const char *buf, size_t size)
359 a5845bb5 2021-03-20 op return parser_foreach_line(p, buf, size, gemtext_foreach_line);
363 a5845bb5 2021-03-20 op gemtext_foreach_line(struct parser *p, const char *line, size_t linelen)
365 a5845bb5 2021-03-20 op enum line_type t;
367 27dce34f 2021-07-06 op t = detect_line_type(line, linelen, p->flags & PARSER_IN_PRE);
368 a5845bb5 2021-03-20 op if (t == LINE_PRE_START)
369 27dce34f 2021-07-06 op p->flags ^= PARSER_IN_PRE;
370 a5845bb5 2021-03-20 op if (t == LINE_PRE_END)
371 27dce34f 2021-07-06 op p->flags ^= PARSER_IN_PRE;
372 a5845bb5 2021-03-20 op return parsers[t](p, t, line, linelen);
376 5e11c00c 2021-03-02 op gemtext_free(struct parser *p)
378 a5c3e03d 2021-03-02 op enum line_type t;
380 a5c3e03d 2021-03-02 op /* flush the buffer */
381 a5c3e03d 2021-03-02 op if (p->len != 0) {
382 27dce34f 2021-07-06 op t = detect_line_type(p->buf, p->len, p->flags & PARSER_IN_PRE);
383 a5c3e03d 2021-03-02 op if (!parsers[t](p, t, p->buf, p->len))
385 27dce34f 2021-07-06 op if ((p->flags & PARSER_IN_PRE) &&
386 27dce34f 2021-07-06 op !emit_line(p, LINE_PRE_END, NULL, NULL))
390 5e11c00c 2021-03-02 op free(p->buf);
393 48e3f26d 2021-05-13 op * use the first level 2 or 3 header as page title if none
396 48e3f26d 2021-05-13 op if (*p->title == '\0')
397 48e3f26d 2021-05-13 op search_title(p, LINE_TITLE_2);
398 48e3f26d 2021-05-13 op if (*p->title == '\0')
399 48e3f26d 2021-05-13 op search_title(p, LINE_TITLE_3);
405 48e3f26d 2021-05-13 op search_title(struct parser *p, enum line_type level)
407 48e3f26d 2021-05-13 op struct line *l;
409 48e3f26d 2021-05-13 op TAILQ_FOREACH(l, &p->head, lines) {
410 48e3f26d 2021-05-13 op if (l->type == level) {
411 48e3f26d 2021-05-13 op if (l->line == NULL)
413 48e3f26d 2021-05-13 op strlcpy(p->title, l->line, sizeof(p->title));