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 5e11c00c 2021-03-02 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 a5c3e03d 2021-03-02 op static int gemtext_free(struct parser*);
34 5e11c00c 2021-03-02 op static int parse_text(struct parser*, enum line_type, const char*, size_t);
35 5e11c00c 2021-03-02 op static int parse_link(struct parser*, enum line_type, const char*, size_t);
36 5e11c00c 2021-03-02 op static int parse_title(struct parser*, enum line_type, const char*, size_t);
37 5e11c00c 2021-03-02 op static int parse_item(struct parser*, enum line_type, const char*, size_t);
38 5e11c00c 2021-03-02 op static int parse_quote(struct parser*, enum line_type, const char*, size_t);
39 5e11c00c 2021-03-02 op static int parse_pre_start(struct parser*, enum line_type, const char*, size_t);
40 5e11c00c 2021-03-02 op static int parse_pre_cnt(struct parser*, enum line_type, const char*, size_t);
41 5e11c00c 2021-03-02 op static int parse_pre_end(struct parser*, enum line_type, const char*, size_t);
43 5e11c00c 2021-03-02 op typedef int (parselinefn)(struct parser*, enum line_type, const char*, size_t);
45 5e11c00c 2021-03-02 op static parselinefn *parsers[] = {
46 5e11c00c 2021-03-02 op parse_text, /* LINE_TEXT */
47 5e11c00c 2021-03-02 op parse_link, /* LINE_LINK */
48 5e11c00c 2021-03-02 op parse_title, /* LINE_TITLE_1 */
49 5e11c00c 2021-03-02 op parse_title, /* LINE_TITLE_2 */
50 5e11c00c 2021-03-02 op parse_title, /* LINE_TITLE_3 */
51 5e11c00c 2021-03-02 op parse_item, /* LINE_ITEM */
52 5e11c00c 2021-03-02 op parse_quote, /* LINE_QUOTE */
53 5e11c00c 2021-03-02 op parse_pre_start, /* LINE_PRE_START */
54 5e11c00c 2021-03-02 op parse_pre_cnt, /* LINE_PRE_CONTENT */
55 5e11c00c 2021-03-02 op parse_pre_end, /* LINE_PRE_END */
59 5e11c00c 2021-03-02 op gemtext_initparser(struct parser *p)
61 5e11c00c 2021-03-02 op memset(p, 0, sizeof(*p));
63 5e11c00c 2021-03-02 op p->parse = &gemtext_parse;
64 5e11c00c 2021-03-02 op p->free = &gemtext_free;
67 5e11c00c 2021-03-02 op static inline int
68 5e11c00c 2021-03-02 op emit_line(struct parser *p, enum line_type type, char *line, char *alt)
70 5e11c00c 2021-03-02 op struct line *l;
72 5e11c00c 2021-03-02 op if ((l = calloc(1, sizeof(*l))) == NULL)
75 5e11c00c 2021-03-02 op l->type = type;
76 5e11c00c 2021-03-02 op l->line = line;
79 5e11c00c 2021-03-02 op if (TAILQ_EMPTY(&p->head))
80 5e11c00c 2021-03-02 op TAILQ_INSERT_HEAD(&p->head, l, lines);
82 5e11c00c 2021-03-02 op TAILQ_INSERT_TAIL(&p->head, l, lines);
88 5e11c00c 2021-03-02 op parse_text(struct parser *p, enum line_type t, const char *buf, size_t len)
92 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
94 5e11c00c 2021-03-02 op memcpy(l, buf, len);
95 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
99 5e11c00c 2021-03-02 op parse_link(struct parser *p, enum line_type t, const char *buf, size_t len)
102 5e11c00c 2021-03-02 op const char *url_start;
104 5e11c00c 2021-03-02 op if (len <= 2)
105 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
109 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
114 5e11c00c 2021-03-02 op if (len == 0)
115 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
117 5e11c00c 2021-03-02 op url_start = buf;
118 5e11c00c 2021-03-02 op while (len > 0 && !isspace(buf[0])) {
123 5e11c00c 2021-03-02 op if ((u = calloc(1, buf - url_start + 1)) == NULL)
125 5e11c00c 2021-03-02 op memcpy(u, url_start, buf - url_start);
127 5e11c00c 2021-03-02 op if (len == 0)
128 f250a75b 2021-03-06 op goto nolabel;
130 f250a75b 2021-03-06 op while (len > 0 && isspace(buf[0])) {
135 5e11c00c 2021-03-02 op if (len == 0)
136 f250a75b 2021-03-06 op goto nolabel;
138 5e11c00c 2021-03-02 op if ((l = calloc(1, len + 1)) == NULL)
141 5e11c00c 2021-03-02 op memcpy(l, buf, len);
142 f250a75b 2021-03-06 op return emit_line(p, t, l, u);
145 f250a75b 2021-03-06 op if ((l = strdup(u)) == NULL)
147 f250a75b 2021-03-06 op return emit_line(p, t, l, u);
151 5e11c00c 2021-03-02 op parse_title(struct parser *p, enum line_type t, const char *buf, size_t len)
156 5e11c00c 2021-03-02 op case LINE_TITLE_1:
157 5e11c00c 2021-03-02 op if (len <= 1)
158 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
162 5e11c00c 2021-03-02 op case LINE_TITLE_2:
163 5e11c00c 2021-03-02 op if (len <= 2)
164 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
168 5e11c00c 2021-03-02 op case LINE_TITLE_3:
169 5e11c00c 2021-03-02 op if (len <= 3)
170 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
175 5e11c00c 2021-03-02 op /* unreachable */
179 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
184 5e11c00c 2021-03-02 op if (len == 0)
185 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
187 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
189 5e11c00c 2021-03-02 op memcpy(l, buf, len);
190 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
194 5e11c00c 2021-03-02 op parse_item(struct parser *p, enum line_type t, const char *buf, size_t len)
198 5e11c00c 2021-03-02 op if (len == 1)
199 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
204 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
209 5e11c00c 2021-03-02 op if (len == 0)
210 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
212 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
214 5e11c00c 2021-03-02 op memcpy(l, buf, len);
215 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
219 5e11c00c 2021-03-02 op parse_quote(struct parser *p, enum line_type t, const char *buf, size_t len)
223 5e11c00c 2021-03-02 op if (len == 1)
224 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
229 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
234 5e11c00c 2021-03-02 op if (len == 0)
235 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
237 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
239 5e11c00c 2021-03-02 op memcpy(l, buf, len);
240 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
244 5e11c00c 2021-03-02 op parse_pre_start(struct parser *p, enum line_type t, const char *buf, size_t len)
248 5e11c00c 2021-03-02 op if (len <= 3)
249 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
254 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
259 5e11c00c 2021-03-02 op if (len == 0)
260 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
262 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
265 5e11c00c 2021-03-02 op memcpy(l, buf, len);
266 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, l);
270 5e11c00c 2021-03-02 op parse_pre_cnt(struct parser *p, enum line_type t, const char *buf, size_t len)
274 5e11c00c 2021-03-02 op if (len == 0)
275 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
277 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
279 5e11c00c 2021-03-02 op memcpy(l, buf, len);
280 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
284 5e11c00c 2021-03-02 op parse_pre_end(struct parser *p, enum line_type t, const char *buf, size_t len)
286 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
289 5e11c00c 2021-03-02 op static inline enum line_type
290 5e11c00c 2021-03-02 op detect_line_type(const char *buf, size_t len, int in_pre)
294 5e11c00c 2021-03-02 op if (len == 0)
295 5e11c00c 2021-03-02 op return LINE_TEXT;
297 5e11c00c 2021-03-02 op if (in_pre) {
298 5e11c00c 2021-03-02 op if (len >= 3 &&
299 5e11c00c 2021-03-02 op buf[0] == '`' && buf[1] == '`' && buf[2] == '`')
300 5e11c00c 2021-03-02 op return LINE_PRE_END;
302 5e11c00c 2021-03-02 op return LINE_PRE_CONTENT;
305 5e11c00c 2021-03-02 op switch (*buf) {
306 5e11c00c 2021-03-02 op case '*': return LINE_ITEM;
307 5e11c00c 2021-03-02 op case '>': return LINE_QUOTE;
309 5e11c00c 2021-03-02 op if (len >= 1 && buf[1] == '>')
310 5e11c00c 2021-03-02 op return LINE_LINK;
313 5e11c00c 2021-03-02 op if (len == 1)
314 5e11c00c 2021-03-02 op return LINE_TEXT;
315 5e11c00c 2021-03-02 op if (buf[1] != '#')
316 5e11c00c 2021-03-02 op return LINE_TITLE_1;
317 5e11c00c 2021-03-02 op if (len == 2)
318 5e11c00c 2021-03-02 op return LINE_TEXT;
319 5e11c00c 2021-03-02 op if (buf[2] != '#')
320 5e11c00c 2021-03-02 op return LINE_TITLE_2;
321 5e11c00c 2021-03-02 op if (len == 3)
322 5e11c00c 2021-03-02 op return LINE_TEXT;
323 5e11c00c 2021-03-02 op return LINE_TITLE_3;
326 5e11c00c 2021-03-02 op return LINE_TEXT;
327 5e11c00c 2021-03-02 op if (buf[0] == '`' && buf[1] == '`' && buf[2] == '`')
328 5e11c00c 2021-03-02 op return LINE_PRE_START;
332 5e11c00c 2021-03-02 op return LINE_TEXT;
335 5e11c00c 2021-03-02 op static inline int
336 5e11c00c 2021-03-02 op append(struct parser *p, const char *buf, size_t len)
338 5e11c00c 2021-03-02 op size_t newlen;
341 5e11c00c 2021-03-02 op newlen = len + p->len;
342 5e11c00c 2021-03-02 op if ((t = calloc(1, newlen)) == NULL)
344 805283d1 2021-03-06 op memcpy(t, p->buf, p->len);
345 805283d1 2021-03-06 op memcpy(t + p->len, buf, len);
346 5e11c00c 2021-03-02 op free(p->buf);
348 5e11c00c 2021-03-02 op p->len = newlen;
352 5e11c00c 2021-03-02 op static inline int
353 5e11c00c 2021-03-02 op set_buf(struct parser *p, const char *buf, size_t len)
355 5e11c00c 2021-03-02 op free(p->buf);
356 5e11c00c 2021-03-02 op p->buf = NULL;
358 805283d1 2021-03-06 op if (len == 0) {
363 5e11c00c 2021-03-02 op if ((p->buf = calloc(1, len)) == NULL)
365 5e11c00c 2021-03-02 op memcpy(p->buf, buf, len);
366 db1e2315 2021-03-02 op p->len = len;
371 5e11c00c 2021-03-02 op gemtext_parse(struct parser *p, const char *buf, size_t size)
373 5e11c00c 2021-03-02 op const char *b, *e;
374 5e11c00c 2021-03-02 op enum line_type t;
375 5e11c00c 2021-03-02 op size_t len, l;
377 5e11c00c 2021-03-02 op if (p->len == 0) {
381 5e11c00c 2021-03-02 op if (!append(p, buf, size))
384 5e11c00c 2021-03-02 op len = p->len;
387 5e11c00c 2021-03-02 op while (len > 0) {
388 5e11c00c 2021-03-02 op if ((e = telescope_strnchr((char*)b, '\n', len)) == NULL)
391 5e11c00c 2021-03-02 op t = detect_line_type(b, l, p->flags);
392 5e11c00c 2021-03-02 op if (t == LINE_PRE_START)
393 5e11c00c 2021-03-02 op p->flags = 1;
394 5e11c00c 2021-03-02 op if (t == LINE_PRE_END)
395 5e11c00c 2021-03-02 op p->flags = 0;
396 5e11c00c 2021-03-02 op if (!parsers[t](p, t, b, l))
402 5e11c00c 2021-03-02 op if (len > 0) {
403 5e11c00c 2021-03-02 op /* skip \n */
409 5e11c00c 2021-03-02 op return set_buf(p, b, len);
413 5e11c00c 2021-03-02 op gemtext_free(struct parser *p)
415 a5c3e03d 2021-03-02 op enum line_type t;
417 a5c3e03d 2021-03-02 op /* flush the buffer */
418 a5c3e03d 2021-03-02 op if (p->len != 0) {
419 a5c3e03d 2021-03-02 op t = detect_line_type(p->buf, p->len, p->flags);
420 a5c3e03d 2021-03-02 op if (!parsers[t](p, t, p->buf, p->len))
422 a5c3e03d 2021-03-02 op if (p->flags && !emit_line(p, LINE_PRE_END, NULL, NULL))
426 5e11c00c 2021-03-02 op free(p->buf);