2 70d237df 2021-07-25 op * Copyright (c) 2021 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 70d237df 2021-07-25 op static int gemtext_parse(struct parser*, const char*, size_t);
36 70d237df 2021-07-25 op static int gemtext_foreach_line(struct parser*, const char*, size_t);
37 70d237df 2021-07-25 op static int gemtext_free(struct parser*);
39 70d237df 2021-07-25 op static int parse_text(struct parser*, enum line_type, const char*, size_t);
40 70d237df 2021-07-25 op static int parse_link(struct parser*, enum line_type, const char*, size_t);
41 70d237df 2021-07-25 op static int parse_title(struct parser*, enum line_type, const char*, size_t);
42 70d237df 2021-07-25 op static int parse_item(struct parser*, enum line_type, const char*, size_t);
43 70d237df 2021-07-25 op static int parse_quote(struct parser*, enum line_type, const char*, size_t);
44 70d237df 2021-07-25 op static int parse_pre_start(struct parser*, enum line_type, const char*, size_t);
45 70d237df 2021-07-25 op static int parse_pre_cnt(struct parser*, enum line_type, const char*, size_t);
46 70d237df 2021-07-25 op static int parse_pre_end(struct parser*, enum line_type, const char*, size_t);
47 70d237df 2021-07-25 op static void search_title(struct parser*, enum line_type);
49 70d237df 2021-07-25 op typedef int (parselinefn)(struct parser*, enum line_type, const char*, size_t);
51 70d237df 2021-07-25 op static parselinefn *parsers[] = {
52 70d237df 2021-07-25 op [LINE_TEXT] = parse_text,
53 70d237df 2021-07-25 op [LINE_LINK] = parse_link,
54 70d237df 2021-07-25 op [LINE_TITLE_1] = parse_title,
55 70d237df 2021-07-25 op [LINE_TITLE_2] = parse_title,
56 70d237df 2021-07-25 op [LINE_TITLE_3] = parse_title,
57 70d237df 2021-07-25 op [LINE_ITEM] = parse_item,
58 70d237df 2021-07-25 op [LINE_QUOTE] = parse_quote,
59 70d237df 2021-07-25 op [LINE_PRE_START] = parse_pre_start,
60 70d237df 2021-07-25 op [LINE_PRE_CONTENT] = parse_pre_cnt,
61 70d237df 2021-07-25 op [LINE_PRE_END] = parse_pre_end,
65 70d237df 2021-07-25 op gemtext_initparser(struct parser *p)
67 70d237df 2021-07-25 op memset(p, 0, sizeof(*p));
69 70d237df 2021-07-25 op p->name = "text/gemini";
70 70d237df 2021-07-25 op p->parse = &gemtext_parse;
71 70d237df 2021-07-25 op p->free = &gemtext_free;
74 70d237df 2021-07-25 op static inline int
75 70d237df 2021-07-25 op emit_line(struct parser *p, enum line_type type, char *line, char *alt)
77 70d237df 2021-07-25 op struct line *l;
79 70d237df 2021-07-25 op if ((l = calloc(1, sizeof(*l))) == NULL)
82 70d237df 2021-07-25 op l->type = type;
83 70d237df 2021-07-25 op l->line = line;
86 70d237df 2021-07-25 op switch (l->type) {
87 70d237df 2021-07-25 op case LINE_PRE_START:
88 70d237df 2021-07-25 op case LINE_PRE_END:
89 70d237df 2021-07-25 op if (hide_pre_context)
90 70d237df 2021-07-25 op l->flags = L_HIDDEN;
91 70d237df 2021-07-25 op if (l->type == LINE_PRE_END &&
92 70d237df 2021-07-25 op hide_pre_closing_line)
93 70d237df 2021-07-25 op l->flags = L_HIDDEN;
95 70d237df 2021-07-25 op case LINE_PRE_CONTENT:
96 70d237df 2021-07-25 op if (hide_pre_blocks)
97 70d237df 2021-07-25 op l->flags = L_HIDDEN;
99 70d237df 2021-07-25 op case LINE_LINK:
100 70d237df 2021-07-25 op if (emojify_link &&
101 70d237df 2021-07-25 op !emojied_line(line, (const char **)&l->data))
102 70d237df 2021-07-25 op l->data = NULL;
108 70d237df 2021-07-25 op if (TAILQ_EMPTY(&p->head))
109 70d237df 2021-07-25 op TAILQ_INSERT_HEAD(&p->head, l, lines);
111 70d237df 2021-07-25 op TAILQ_INSERT_TAIL(&p->head, l, lines);
117 70d237df 2021-07-25 op parse_text(struct parser *p, enum line_type t, const char *buf, size_t len)
121 70d237df 2021-07-25 op if ((l = calloc(1, len+1)) == NULL)
123 70d237df 2021-07-25 op memcpy(l, buf, len);
124 70d237df 2021-07-25 op return emit_line(p, t, l, NULL);
128 70d237df 2021-07-25 op parse_link(struct parser *p, enum line_type t, const char *buf, size_t len)
131 70d237df 2021-07-25 op const char *url_start;
133 70d237df 2021-07-25 op if (len <= 2)
134 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
138 70d237df 2021-07-25 op while (len > 0 && isspace(buf[0])) {
143 70d237df 2021-07-25 op if (len == 0)
144 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
146 70d237df 2021-07-25 op url_start = buf;
147 70d237df 2021-07-25 op while (len > 0 && !isspace(buf[0])) {
152 70d237df 2021-07-25 op if ((u = calloc(1, buf - url_start + 1)) == NULL)
154 70d237df 2021-07-25 op memcpy(u, url_start, buf - url_start);
156 70d237df 2021-07-25 op if (len == 0)
157 70d237df 2021-07-25 op goto nolabel;
159 70d237df 2021-07-25 op while (len > 0 && isspace(buf[0])) {
164 70d237df 2021-07-25 op if (len == 0)
165 70d237df 2021-07-25 op goto nolabel;
167 70d237df 2021-07-25 op if ((l = calloc(1, len + 1)) == NULL)
170 70d237df 2021-07-25 op memcpy(l, buf, len);
171 70d237df 2021-07-25 op return emit_line(p, t, l, u);
174 70d237df 2021-07-25 op if ((l = strdup(u)) == NULL)
176 70d237df 2021-07-25 op return emit_line(p, t, l, u);
180 70d237df 2021-07-25 op parse_title(struct parser *p, enum line_type t, const char *buf, size_t len)
185 70d237df 2021-07-25 op case LINE_TITLE_1:
186 70d237df 2021-07-25 op if (len <= 1)
187 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
191 70d237df 2021-07-25 op case LINE_TITLE_2:
192 70d237df 2021-07-25 op if (len <= 2)
193 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
197 70d237df 2021-07-25 op case LINE_TITLE_3:
198 70d237df 2021-07-25 op if (len <= 3)
199 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
204 70d237df 2021-07-25 op /* unreachable */
208 70d237df 2021-07-25 op while (len > 0 && isspace(buf[0])) {
213 70d237df 2021-07-25 op if (len == 0)
214 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
216 70d237df 2021-07-25 op if (t == LINE_TITLE_1 && *p->title == '\0')
217 70d237df 2021-07-25 op strncpy(p->title, buf, MIN(sizeof(p->title)-1, len));
219 70d237df 2021-07-25 op if ((l = calloc(1, len+1)) == NULL)
221 70d237df 2021-07-25 op memcpy(l, buf, len);
222 70d237df 2021-07-25 op return emit_line(p, t, l, NULL);
226 70d237df 2021-07-25 op parse_item(struct parser *p, enum line_type t, const char *buf, size_t len)
230 70d237df 2021-07-25 op if (len == 1)
231 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
236 70d237df 2021-07-25 op while (len > 0 && isspace(buf[0])) {
241 70d237df 2021-07-25 op if (len == 0)
242 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
244 70d237df 2021-07-25 op if ((l = calloc(1, len+1)) == NULL)
246 70d237df 2021-07-25 op memcpy(l, buf, len);
247 70d237df 2021-07-25 op return emit_line(p, t, l, NULL);
251 70d237df 2021-07-25 op parse_quote(struct parser *p, enum line_type t, const char *buf, size_t len)
255 70d237df 2021-07-25 op if (len == 1)
256 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
261 70d237df 2021-07-25 op while (len > 0 && isspace(buf[0])) {
266 70d237df 2021-07-25 op if (len == 0)
267 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
269 70d237df 2021-07-25 op if ((l = calloc(1, len+1)) == NULL)
271 70d237df 2021-07-25 op memcpy(l, buf, len);
272 70d237df 2021-07-25 op return emit_line(p, t, l, NULL);
276 70d237df 2021-07-25 op parse_pre_start(struct parser *p, enum line_type t, const char *buf, size_t len)
280 70d237df 2021-07-25 op if (len <= 3)
281 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
286 70d237df 2021-07-25 op while (len > 0 && isspace(buf[0])) {
291 70d237df 2021-07-25 op if (len == 0)
292 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
294 70d237df 2021-07-25 op if ((l = calloc(1, len+1)) == NULL)
297 70d237df 2021-07-25 op memcpy(l, buf, len);
298 70d237df 2021-07-25 op return emit_line(p, t, l, NULL);
302 70d237df 2021-07-25 op parse_pre_cnt(struct parser *p, enum line_type t, const char *buf, size_t len)
306 70d237df 2021-07-25 op if (len == 0)
307 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
309 70d237df 2021-07-25 op if ((l = calloc(1, len+1)) == NULL)
311 70d237df 2021-07-25 op memcpy(l, buf, len);
312 70d237df 2021-07-25 op return emit_line(p, t, l, NULL);
316 70d237df 2021-07-25 op parse_pre_end(struct parser *p, enum line_type t, const char *buf, size_t len)
318 70d237df 2021-07-25 op return emit_line(p, t, NULL, NULL);
321 70d237df 2021-07-25 op static inline enum line_type
322 70d237df 2021-07-25 op detect_line_type(const char *buf, size_t len, int in_pre)
324 70d237df 2021-07-25 op if (in_pre) {
325 70d237df 2021-07-25 op if (len >= 3 &&
326 70d237df 2021-07-25 op buf[0] == '`' && buf[1] == '`' && buf[2] == '`')
327 70d237df 2021-07-25 op return LINE_PRE_END;
329 70d237df 2021-07-25 op return LINE_PRE_CONTENT;
332 70d237df 2021-07-25 op if (len == 0)
333 70d237df 2021-07-25 op return LINE_TEXT;
335 70d237df 2021-07-25 op switch (*buf) {
336 70d237df 2021-07-25 op case '*': return LINE_ITEM;
337 70d237df 2021-07-25 op case '>': return LINE_QUOTE;
339 70d237df 2021-07-25 op if (len >= 1 && buf[1] == '>')
340 70d237df 2021-07-25 op return LINE_LINK;
343 70d237df 2021-07-25 op if (len == 1)
344 70d237df 2021-07-25 op return LINE_TEXT;
345 70d237df 2021-07-25 op if (buf[1] != '#')
346 70d237df 2021-07-25 op return LINE_TITLE_1;
347 70d237df 2021-07-25 op if (len == 2)
348 70d237df 2021-07-25 op return LINE_TEXT;
349 70d237df 2021-07-25 op if (buf[2] != '#')
350 70d237df 2021-07-25 op return LINE_TITLE_2;
351 70d237df 2021-07-25 op if (len == 3)
352 70d237df 2021-07-25 op return LINE_TEXT;
353 70d237df 2021-07-25 op return LINE_TITLE_3;
356 70d237df 2021-07-25 op return LINE_TEXT;
357 70d237df 2021-07-25 op if (buf[0] == '`' && buf[1] == '`' && buf[2] == '`')
358 70d237df 2021-07-25 op return LINE_PRE_START;
362 70d237df 2021-07-25 op return LINE_TEXT;
366 70d237df 2021-07-25 op gemtext_parse(struct parser *p, const char *buf, size_t size)
368 70d237df 2021-07-25 op return parser_foreach_line(p, buf, size, gemtext_foreach_line);
372 70d237df 2021-07-25 op gemtext_foreach_line(struct parser *p, const char *line, size_t linelen)
374 70d237df 2021-07-25 op enum line_type t;
376 70d237df 2021-07-25 op t = detect_line_type(line, linelen, p->flags & PARSER_IN_PRE);
377 70d237df 2021-07-25 op if (t == LINE_PRE_START)
378 70d237df 2021-07-25 op p->flags ^= PARSER_IN_PRE;
379 70d237df 2021-07-25 op if (t == LINE_PRE_END)
380 70d237df 2021-07-25 op p->flags ^= PARSER_IN_PRE;
381 70d237df 2021-07-25 op return parsers[t](p, t, line, linelen);
385 70d237df 2021-07-25 op gemtext_free(struct parser *p)
387 70d237df 2021-07-25 op enum line_type t;
389 70d237df 2021-07-25 op /* flush the buffer */
390 70d237df 2021-07-25 op if (p->len != 0) {
391 70d237df 2021-07-25 op t = detect_line_type(p->buf, p->len, p->flags & PARSER_IN_PRE);
392 70d237df 2021-07-25 op if (!parsers[t](p, t, p->buf, p->len))
394 70d237df 2021-07-25 op if ((p->flags & PARSER_IN_PRE) &&
395 70d237df 2021-07-25 op !emit_line(p, LINE_PRE_END, NULL, NULL))
399 70d237df 2021-07-25 op free(p->buf);
402 70d237df 2021-07-25 op * use the first level 2 or 3 header as page title if none
405 70d237df 2021-07-25 op if (*p->title == '\0')
406 70d237df 2021-07-25 op search_title(p, LINE_TITLE_2);
407 70d237df 2021-07-25 op if (*p->title == '\0')
408 70d237df 2021-07-25 op search_title(p, LINE_TITLE_3);
414 70d237df 2021-07-25 op search_title(struct parser *p, enum line_type level)
416 70d237df 2021-07-25 op struct line *l;
418 70d237df 2021-07-25 op TAILQ_FOREACH(l, &p->head, lines) {
419 70d237df 2021-07-25 op if (l->type == level) {
420 70d237df 2021-07-25 op if (l->line == NULL)
422 70d237df 2021-07-25 op strlcpy(p->title, l->line, sizeof(p->title));