2 069c9a92 2022-01-19 op * Copyright (c) 2021, 2022 Omar Polo <op@omarpolo.com>
4 70d237df 2021-07-25 op * Permission to use, copy, modify, and distribute this software for any
5 70d237df 2021-07-25 op * purpose with or without fee is hereby granted, provided that the above
6 70d237df 2021-07-25 op * copyright notice and this permission notice appear in all copies.
8 70d237df 2021-07-25 op * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 70d237df 2021-07-25 op * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 70d237df 2021-07-25 op * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 70d237df 2021-07-25 op * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 70d237df 2021-07-25 op * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 70d237df 2021-07-25 op * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 70d237df 2021-07-25 op * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18 70d237df 2021-07-25 op * A streaming gemtext parser.
21 70d237df 2021-07-25 op * - handle NULs
25 70d237df 2021-07-25 op #include "compat.h"
27 70d237df 2021-07-25 op #include <ctype.h>
28 70d237df 2021-07-25 op #include <string.h>
29 70d237df 2021-07-25 op #include <stdlib.h>
31 70d237df 2021-07-25 op #include "defaults.h"
32 70d237df 2021-07-25 op #include "parser.h"
33 70d237df 2021-07-25 op #include "utf8.h"
35 069c9a92 2022-01-19 op static int gemtext_parse(struct parser *, const char *, size_t);
36 069c9a92 2022-01-19 op static int gemtext_foreach_line(struct parser *, const char *, size_t);
37 069c9a92 2022-01-19 op static int gemtext_free(struct parser *);
38 069c9a92 2022-01-19 op static int gemtext_serialize(struct parser *, struct evbuffer *);
40 70d237df 2021-07-25 op static int parse_text(struct parser*, enum line_type, const char*, size_t);
41 70d237df 2021-07-25 op static int parse_link(struct parser*, enum line_type, const char*, size_t);
42 70d237df 2021-07-25 op static int parse_title(struct parser*, enum line_type, const char*, size_t);
43 70d237df 2021-07-25 op static int parse_item(struct parser*, enum line_type, const char*, size_t);
44 70d237df 2021-07-25 op static int parse_quote(struct parser*, enum line_type, const char*, size_t);
45 70d237df 2021-07-25 op static int parse_pre_start(struct parser*, enum line_type, const char*, size_t);
46 70d237df 2021-07-25 op static int parse_pre_cnt(struct parser*, enum line_type, const char*, size_t);
47 70d237df 2021-07-25 op static int parse_pre_end(struct parser*, enum line_type, const char*, size_t);
48 70d237df 2021-07-25 op static void search_title(struct parser*, enum line_type);
50 70d237df 2021-07-25 op typedef int (parselinefn)(struct parser*, enum line_type, const char*, size_t);
52 70d237df 2021-07-25 op static parselinefn *parsers[] = {
53 70d237df 2021-07-25 op [LINE_TEXT] = parse_text,
54 70d237df 2021-07-25 op [LINE_LINK] = parse_link,
55 70d237df 2021-07-25 op [LINE_TITLE_1] = parse_title,
56 70d237df 2021-07-25 op [LINE_TITLE_2] = parse_title,
57 70d237df 2021-07-25 op [LINE_TITLE_3] = parse_title,
58 70d237df 2021-07-25 op [LINE_ITEM] = parse_item,
59 70d237df 2021-07-25 op [LINE_QUOTE] = parse_quote,
60 70d237df 2021-07-25 op [LINE_PRE_START] = parse_pre_start,
61 70d237df 2021-07-25 op [LINE_PRE_CONTENT] = parse_pre_cnt,
62 70d237df 2021-07-25 op [LINE_PRE_END] = parse_pre_end,
66 70d237df 2021-07-25 op gemtext_initparser(struct parser *p)
68 70d237df 2021-07-25 op memset(p, 0, sizeof(*p));
70 70d237df 2021-07-25 op p->name = "text/gemini";
71 70d237df 2021-07-25 op p->parse = &gemtext_parse;
72 70d237df 2021-07-25 op p->free = &gemtext_free;
73 069c9a92 2022-01-19 op p->serialize = &gemtext_serialize;
75 78894e73 2021-08-12 op TAILQ_INIT(&p->head);
78 70d237df 2021-07-25 op static inline int
79 70d237df 2021-07-25 op emit_line(struct parser *p, enum line_type type, char *line, char *alt)
81 70d237df 2021-07-25 op struct line *l;
83 70d237df 2021-07-25 op if ((l = calloc(1, sizeof(*l))) == NULL)
86 70d237df 2021-07-25 op l->type = type;
87 70d237df 2021-07-25 op l->line = line;
90 70d237df 2021-07-25 op switch (l->type) {
91 70d237df 2021-07-25 op case LINE_PRE_START:
92 70d237df 2021-07-25 op case LINE_PRE_END:
93 70d237df 2021-07-25 op if (hide_pre_context)
94 70d237df 2021-07-25 op l->flags = L_HIDDEN;
95 70d237df 2021-07-25 op if (l->type == LINE_PRE_END &&
96 70d237df 2021-07-25 op hide_pre_closing_line)
97 70d237df 2021-07-25 op l->flags = L_HIDDEN;
99 70d237df 2021-07-25 op case LINE_PRE_CONTENT:
100 70d237df 2021-07-25 op if (hide_pre_blocks)
101 70d237df 2021-07-25 op l->flags = L_HIDDEN;
103 70d237df 2021-07-25 op case LINE_LINK:
104 70d237df 2021-07-25 op if (emojify_link &&
105 70d237df 2021-07-25 op !emojied_line(line, (const char **)&l->data))
106 70d237df 2021-07-25 op l->data = NULL;
112 32ac17a4 2021-08-12 op TAILQ_INSERT_TAIL(&p->head, l, lines);
118 70d237df 2021-07-25 op parse_text(struct parser *p, enum line_type t, const char *buf, size_t len)
122 70d237df 2021-07-25 op if ((l = calloc(1, len+1)) == NULL)
124 70d237df 2021-07-25 op memcpy(l, buf, len);
125 70d237df 2021-07-25 op return emit_line(p, t, l, NULL);
129 70d237df 2021-07-25 op parse_link(struct parser *p, enum line_type t, const char *buf, size_t len)
132 70d237df 2021-07-25 op const char *url_start;
134 70d237df 2021-07-25 op if (len <= 2)
135 1ebb7ceb 2022-01-17 op return emit_line(p, LINE_TEXT, NULL, NULL);
139 70d237df 2021-07-25 op while (len > 0 && isspace(buf[0])) {
144 70d237df 2021-07-25 op if (len == 0)
145 1ebb7ceb 2022-01-17 op return emit_line(p, LINE_TEXT, NULL, NULL);
147 70d237df 2021-07-25 op url_start = buf;
148 70d237df 2021-07-25 op while (len > 0 && !isspace(buf[0])) {
153 70d237df 2021-07-25 op if ((u = calloc(1, buf - url_start + 1)) == NULL)
155 70d237df 2021-07-25 op memcpy(u, url_start, buf - url_start);
157 70d237df 2021-07-25 op if (len == 0)
158 70d237df 2021-07-25 op goto nolabel;
160 70d237df 2021-07-25 op while (len > 0 && isspace(buf[0])) {
165 70d237df 2021-07-25 op if (len == 0)
166 70d237df 2021-07-25 op goto nolabel;
168 70d237df 2021-07-25 op if ((l = calloc(1, len + 1)) == NULL)
171 70d237df 2021-07-25 op memcpy(l, buf, len);
172 70d237df 2021-07-25 op return emit_line(p, t, l, u);
175 70d237df 2021-07-25 op if ((l = strdup(u)) == NULL)
177 70d237df 2021-07-25 op return emit_line(p, t, l, u);
181 70d237df 2021-07-25 op parse_title(struct parser *p, enum line_type t, const char *buf, size_t len)
186 70d237df 2021-07-25 op case LINE_TITLE_1:
187 70d237df 2021-07-25 op if (len <= 1)
188 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
192 70d237df 2021-07-25 op case LINE_TITLE_2:
193 70d237df 2021-07-25 op if (len <= 2)
194 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
198 70d237df 2021-07-25 op case LINE_TITLE_3:
199 70d237df 2021-07-25 op if (len <= 3)
200 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
205 70d237df 2021-07-25 op /* unreachable */
209 70d237df 2021-07-25 op while (len > 0 && isspace(buf[0])) {
214 70d237df 2021-07-25 op if (len == 0)
215 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
217 70d237df 2021-07-25 op if (t == LINE_TITLE_1 && *p->title == '\0')
218 70d237df 2021-07-25 op strncpy(p->title, buf, MIN(sizeof(p->title)-1, len));
220 70d237df 2021-07-25 op if ((l = calloc(1, len+1)) == NULL)
222 70d237df 2021-07-25 op memcpy(l, buf, len);
223 70d237df 2021-07-25 op return emit_line(p, t, l, NULL);
227 70d237df 2021-07-25 op parse_item(struct parser *p, enum line_type t, const char *buf, size_t len)
231 70d237df 2021-07-25 op if (len == 1)
232 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
237 70d237df 2021-07-25 op while (len > 0 && isspace(buf[0])) {
242 70d237df 2021-07-25 op if (len == 0)
243 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
245 70d237df 2021-07-25 op if ((l = calloc(1, len+1)) == NULL)
247 70d237df 2021-07-25 op memcpy(l, buf, len);
248 70d237df 2021-07-25 op return emit_line(p, t, l, NULL);
252 70d237df 2021-07-25 op parse_quote(struct parser *p, enum line_type t, const char *buf, size_t len)
256 70d237df 2021-07-25 op if (len == 1)
257 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
262 70d237df 2021-07-25 op while (len > 0 && isspace(buf[0])) {
267 70d237df 2021-07-25 op if (len == 0)
268 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
270 70d237df 2021-07-25 op if ((l = calloc(1, len+1)) == NULL)
272 70d237df 2021-07-25 op memcpy(l, buf, len);
273 70d237df 2021-07-25 op return emit_line(p, t, l, NULL);
277 70d237df 2021-07-25 op parse_pre_start(struct parser *p, enum line_type t, const char *buf, size_t len)
281 70d237df 2021-07-25 op if (len <= 3)
282 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
287 70d237df 2021-07-25 op while (len > 0 && isspace(buf[0])) {
292 70d237df 2021-07-25 op if (len == 0)
293 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
295 70d237df 2021-07-25 op if ((l = calloc(1, len+1)) == NULL)
298 70d237df 2021-07-25 op memcpy(l, buf, len);
299 70d237df 2021-07-25 op return emit_line(p, t, l, NULL);
303 70d237df 2021-07-25 op parse_pre_cnt(struct parser *p, enum line_type t, const char *buf, size_t len)
307 70d237df 2021-07-25 op if (len == 0)
308 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
310 70d237df 2021-07-25 op if ((l = calloc(1, len+1)) == NULL)
312 70d237df 2021-07-25 op memcpy(l, buf, len);
313 70d237df 2021-07-25 op return emit_line(p, t, l, NULL);
317 70d237df 2021-07-25 op parse_pre_end(struct parser *p, enum line_type t, const char *buf, size_t len)
319 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
322 70d237df 2021-07-25 op static inline enum line_type
323 70d237df 2021-07-25 op detect_line_type(const char *buf, size_t len, int in_pre)
325 70d237df 2021-07-25 op if (in_pre) {
326 70d237df 2021-07-25 op if (len >= 3 &&
327 70d237df 2021-07-25 op buf[0] == '`' && buf[1] == '`' && buf[2] == '`')
328 70d237df 2021-07-25 op return LINE_PRE_END;
330 70d237df 2021-07-25 op return LINE_PRE_CONTENT;
333 70d237df 2021-07-25 op if (len == 0)
334 70d237df 2021-07-25 op return LINE_TEXT;
336 70d237df 2021-07-25 op switch (*buf) {
338 db056a13 2021-10-26 op if (len > 1 && buf[1] == ' ')
339 db056a13 2021-10-26 op return LINE_ITEM;
341 70d237df 2021-07-25 op case '>': return LINE_QUOTE;
343 70d237df 2021-07-25 op if (len >= 1 && buf[1] == '>')
344 70d237df 2021-07-25 op return LINE_LINK;
347 70d237df 2021-07-25 op if (len == 1)
348 70d237df 2021-07-25 op return LINE_TEXT;
349 70d237df 2021-07-25 op if (buf[1] != '#')
350 70d237df 2021-07-25 op return LINE_TITLE_1;
351 70d237df 2021-07-25 op if (len == 2)
352 70d237df 2021-07-25 op return LINE_TEXT;
353 70d237df 2021-07-25 op if (buf[2] != '#')
354 70d237df 2021-07-25 op return LINE_TITLE_2;
355 70d237df 2021-07-25 op if (len == 3)
356 70d237df 2021-07-25 op return LINE_TEXT;
357 70d237df 2021-07-25 op return LINE_TITLE_3;
360 70d237df 2021-07-25 op return LINE_TEXT;
361 70d237df 2021-07-25 op if (buf[0] == '`' && buf[1] == '`' && buf[2] == '`')
362 70d237df 2021-07-25 op return LINE_PRE_START;
366 70d237df 2021-07-25 op return LINE_TEXT;
370 70d237df 2021-07-25 op gemtext_parse(struct parser *p, const char *buf, size_t size)
372 70d237df 2021-07-25 op return parser_foreach_line(p, buf, size, gemtext_foreach_line);
376 70d237df 2021-07-25 op gemtext_foreach_line(struct parser *p, const char *line, size_t linelen)
378 70d237df 2021-07-25 op enum line_type t;
380 70d237df 2021-07-25 op t = detect_line_type(line, linelen, p->flags & PARSER_IN_PRE);
381 70d237df 2021-07-25 op if (t == LINE_PRE_START)
382 70d237df 2021-07-25 op p->flags ^= PARSER_IN_PRE;
383 70d237df 2021-07-25 op if (t == LINE_PRE_END)
384 70d237df 2021-07-25 op p->flags ^= PARSER_IN_PRE;
385 70d237df 2021-07-25 op return parsers[t](p, t, line, linelen);
389 70d237df 2021-07-25 op gemtext_free(struct parser *p)
391 70d237df 2021-07-25 op enum line_type t;
393 70d237df 2021-07-25 op /* flush the buffer */
394 70d237df 2021-07-25 op if (p->len != 0) {
395 70d237df 2021-07-25 op t = detect_line_type(p->buf, p->len, p->flags & PARSER_IN_PRE);
396 70d237df 2021-07-25 op if (!parsers[t](p, t, p->buf, p->len))
398 70d237df 2021-07-25 op if ((p->flags & PARSER_IN_PRE) &&
399 70d237df 2021-07-25 op !emit_line(p, LINE_PRE_END, NULL, NULL))
403 70d237df 2021-07-25 op free(p->buf);
406 70d237df 2021-07-25 op * use the first level 2 or 3 header as page title if none
409 70d237df 2021-07-25 op if (*p->title == '\0')
410 70d237df 2021-07-25 op search_title(p, LINE_TITLE_2);
411 70d237df 2021-07-25 op if (*p->title == '\0')
412 70d237df 2021-07-25 op search_title(p, LINE_TITLE_3);
418 70d237df 2021-07-25 op search_title(struct parser *p, enum line_type level)
420 70d237df 2021-07-25 op struct line *l;
422 70d237df 2021-07-25 op TAILQ_FOREACH(l, &p->head, lines) {
423 70d237df 2021-07-25 op if (l->type == level) {
424 70d237df 2021-07-25 op if (l->line == NULL)
426 70d237df 2021-07-25 op strlcpy(p->title, l->line, sizeof(p->title));
432 069c9a92 2022-01-19 op static const char *gemtext_prefixes[] = {
433 069c9a92 2022-01-19 op [LINE_TEXT] = "",
434 069c9a92 2022-01-19 op [LINE_TITLE_1] = "# ",
435 069c9a92 2022-01-19 op [LINE_TITLE_2] = "## ",
436 069c9a92 2022-01-19 op [LINE_TITLE_3] = "### ",
437 069c9a92 2022-01-19 op [LINE_ITEM] = "* ",
438 069c9a92 2022-01-19 op [LINE_QUOTE] = "> ",
439 069c9a92 2022-01-19 op [LINE_PRE_START] = "``` ",
440 069c9a92 2022-01-19 op [LINE_PRE_CONTENT] = "",
441 069c9a92 2022-01-19 op [LINE_PRE_END] = "```",
445 069c9a92 2022-01-19 op gemtext_serialize(struct parser *p, struct evbuffer *evb)
447 069c9a92 2022-01-19 op struct line *line;
448 069c9a92 2022-01-19 op const char *text;
449 069c9a92 2022-01-19 op const char *alt;
452 069c9a92 2022-01-19 op TAILQ_FOREACH(line, &p->head, lines) {
453 069c9a92 2022-01-19 op if ((text = line->line) == NULL)
456 069c9a92 2022-01-19 op if ((alt = line->alt) == NULL)
459 069c9a92 2022-01-19 op switch (line->type) {
460 069c9a92 2022-01-19 op case LINE_TEXT:
461 069c9a92 2022-01-19 op case LINE_TITLE_1:
462 069c9a92 2022-01-19 op case LINE_TITLE_2:
463 069c9a92 2022-01-19 op case LINE_TITLE_3:
464 069c9a92 2022-01-19 op case LINE_ITEM:
465 069c9a92 2022-01-19 op case LINE_QUOTE:
466 069c9a92 2022-01-19 op case LINE_PRE_START:
467 069c9a92 2022-01-19 op case LINE_PRE_CONTENT:
468 069c9a92 2022-01-19 op case LINE_PRE_END:
469 069c9a92 2022-01-19 op r = evbuffer_add_printf(evb, "%s%s\n",
470 069c9a92 2022-01-19 op gemtext_prefixes[line->type], text);
473 069c9a92 2022-01-19 op case LINE_LINK:
474 069c9a92 2022-01-19 op r = evbuffer_add_printf(evb, "=> %s %s\n",
479 069c9a92 2022-01-19 op /* not reached */