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 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);
44 5e11c00c 2021-03-02 op typedef int (parselinefn)(struct parser*, enum line_type, const char*, size_t);
46 5e11c00c 2021-03-02 op static parselinefn *parsers[] = {
47 5e11c00c 2021-03-02 op parse_text, /* LINE_TEXT */
48 5e11c00c 2021-03-02 op parse_link, /* LINE_LINK */
49 5e11c00c 2021-03-02 op parse_title, /* LINE_TITLE_1 */
50 5e11c00c 2021-03-02 op parse_title, /* LINE_TITLE_2 */
51 5e11c00c 2021-03-02 op parse_title, /* LINE_TITLE_3 */
52 5e11c00c 2021-03-02 op parse_item, /* LINE_ITEM */
53 5e11c00c 2021-03-02 op parse_quote, /* LINE_QUOTE */
54 5e11c00c 2021-03-02 op parse_pre_start, /* LINE_PRE_START */
55 5e11c00c 2021-03-02 op parse_pre_cnt, /* LINE_PRE_CONTENT */
56 5e11c00c 2021-03-02 op parse_pre_end, /* LINE_PRE_END */
60 5e11c00c 2021-03-02 op gemtext_initparser(struct parser *p)
62 5e11c00c 2021-03-02 op memset(p, 0, sizeof(*p));
64 fc43eadd 2021-03-12 op p->name = "text/gemini";
65 5e11c00c 2021-03-02 op p->parse = &gemtext_parse;
66 5e11c00c 2021-03-02 op p->free = &gemtext_free;
69 5e11c00c 2021-03-02 op static inline int
70 5e11c00c 2021-03-02 op emit_line(struct parser *p, enum line_type type, char *line, char *alt)
72 5e11c00c 2021-03-02 op struct line *l;
74 5e11c00c 2021-03-02 op if ((l = calloc(1, sizeof(*l))) == NULL)
77 5e11c00c 2021-03-02 op l->type = type;
78 5e11c00c 2021-03-02 op l->line = line;
81 5e11c00c 2021-03-02 op if (TAILQ_EMPTY(&p->head))
82 5e11c00c 2021-03-02 op TAILQ_INSERT_HEAD(&p->head, l, lines);
84 5e11c00c 2021-03-02 op TAILQ_INSERT_TAIL(&p->head, l, lines);
90 5e11c00c 2021-03-02 op parse_text(struct parser *p, enum line_type t, const char *buf, size_t len)
94 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
96 5e11c00c 2021-03-02 op memcpy(l, buf, len);
97 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
101 5e11c00c 2021-03-02 op parse_link(struct parser *p, enum line_type t, const char *buf, size_t len)
104 5e11c00c 2021-03-02 op const char *url_start;
106 5e11c00c 2021-03-02 op if (len <= 2)
107 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
111 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
116 5e11c00c 2021-03-02 op if (len == 0)
117 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
119 5e11c00c 2021-03-02 op url_start = buf;
120 5e11c00c 2021-03-02 op while (len > 0 && !isspace(buf[0])) {
125 5e11c00c 2021-03-02 op if ((u = calloc(1, buf - url_start + 1)) == NULL)
127 5e11c00c 2021-03-02 op memcpy(u, url_start, buf - url_start);
129 5e11c00c 2021-03-02 op if (len == 0)
130 f250a75b 2021-03-06 op goto nolabel;
132 f250a75b 2021-03-06 op while (len > 0 && isspace(buf[0])) {
137 5e11c00c 2021-03-02 op if (len == 0)
138 f250a75b 2021-03-06 op goto nolabel;
140 5e11c00c 2021-03-02 op if ((l = calloc(1, len + 1)) == NULL)
143 5e11c00c 2021-03-02 op memcpy(l, buf, len);
144 f250a75b 2021-03-06 op return emit_line(p, t, l, u);
147 f250a75b 2021-03-06 op if ((l = strdup(u)) == NULL)
149 f250a75b 2021-03-06 op return emit_line(p, t, l, u);
153 5e11c00c 2021-03-02 op parse_title(struct parser *p, enum line_type t, const char *buf, size_t len)
158 5e11c00c 2021-03-02 op case LINE_TITLE_1:
159 5e11c00c 2021-03-02 op if (len <= 1)
160 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
164 5e11c00c 2021-03-02 op case LINE_TITLE_2:
165 5e11c00c 2021-03-02 op if (len <= 2)
166 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
170 5e11c00c 2021-03-02 op case LINE_TITLE_3:
171 5e11c00c 2021-03-02 op if (len <= 3)
172 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
177 5e11c00c 2021-03-02 op /* unreachable */
181 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
186 5e11c00c 2021-03-02 op if (len == 0)
187 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
189 dc5df781 2021-03-13 op if (t == LINE_TITLE_1 && *p->title == '\0')
190 dc5df781 2021-03-13 op strncpy(p->title, buf, MIN(sizeof(p->title)-1, len));
192 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
194 5e11c00c 2021-03-02 op memcpy(l, buf, len);
195 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
199 5e11c00c 2021-03-02 op parse_item(struct parser *p, enum line_type t, const char *buf, size_t len)
203 5e11c00c 2021-03-02 op if (len == 1)
204 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
209 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
214 5e11c00c 2021-03-02 op if (len == 0)
215 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
217 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
219 5e11c00c 2021-03-02 op memcpy(l, buf, len);
220 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
224 5e11c00c 2021-03-02 op parse_quote(struct parser *p, enum line_type t, const char *buf, size_t len)
228 5e11c00c 2021-03-02 op if (len == 1)
229 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
234 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
239 5e11c00c 2021-03-02 op if (len == 0)
240 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
242 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
244 5e11c00c 2021-03-02 op memcpy(l, buf, len);
245 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
249 5e11c00c 2021-03-02 op parse_pre_start(struct parser *p, enum line_type t, const char *buf, size_t len)
253 5e11c00c 2021-03-02 op if (len <= 3)
254 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
259 5e11c00c 2021-03-02 op while (len > 0 && isspace(buf[0])) {
264 5e11c00c 2021-03-02 op if (len == 0)
265 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
267 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
270 5e11c00c 2021-03-02 op memcpy(l, buf, len);
271 e2060f05 2021-03-17 op return emit_line(p, t, l, NULL);
275 5e11c00c 2021-03-02 op parse_pre_cnt(struct parser *p, enum line_type t, const char *buf, size_t len)
279 5e11c00c 2021-03-02 op if (len == 0)
280 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
282 5e11c00c 2021-03-02 op if ((l = calloc(1, len+1)) == NULL)
284 5e11c00c 2021-03-02 op memcpy(l, buf, len);
285 5e11c00c 2021-03-02 op return emit_line(p, t, l, NULL);
289 5e11c00c 2021-03-02 op parse_pre_end(struct parser *p, enum line_type t, const char *buf, size_t len)
291 5e11c00c 2021-03-02 op return emit_line(p, t, NULL, NULL);
294 5e11c00c 2021-03-02 op static inline enum line_type
295 5e11c00c 2021-03-02 op detect_line_type(const char *buf, size_t len, int in_pre)
297 5e11c00c 2021-03-02 op if (len == 0)
298 5e11c00c 2021-03-02 op return LINE_TEXT;
300 5e11c00c 2021-03-02 op if (in_pre) {
301 5e11c00c 2021-03-02 op if (len >= 3 &&
302 5e11c00c 2021-03-02 op buf[0] == '`' && buf[1] == '`' && buf[2] == '`')
303 5e11c00c 2021-03-02 op return LINE_PRE_END;
305 5e11c00c 2021-03-02 op return LINE_PRE_CONTENT;
308 5e11c00c 2021-03-02 op switch (*buf) {
309 5e11c00c 2021-03-02 op case '*': return LINE_ITEM;
310 5e11c00c 2021-03-02 op case '>': return LINE_QUOTE;
312 5e11c00c 2021-03-02 op if (len >= 1 && buf[1] == '>')
313 5e11c00c 2021-03-02 op return LINE_LINK;
316 5e11c00c 2021-03-02 op if (len == 1)
317 5e11c00c 2021-03-02 op return LINE_TEXT;
318 5e11c00c 2021-03-02 op if (buf[1] != '#')
319 5e11c00c 2021-03-02 op return LINE_TITLE_1;
320 5e11c00c 2021-03-02 op if (len == 2)
321 5e11c00c 2021-03-02 op return LINE_TEXT;
322 5e11c00c 2021-03-02 op if (buf[2] != '#')
323 5e11c00c 2021-03-02 op return LINE_TITLE_2;
324 5e11c00c 2021-03-02 op if (len == 3)
325 5e11c00c 2021-03-02 op return LINE_TEXT;
326 5e11c00c 2021-03-02 op return LINE_TITLE_3;
329 5e11c00c 2021-03-02 op return LINE_TEXT;
330 5e11c00c 2021-03-02 op if (buf[0] == '`' && buf[1] == '`' && buf[2] == '`')
331 5e11c00c 2021-03-02 op return LINE_PRE_START;
335 5e11c00c 2021-03-02 op return LINE_TEXT;
339 5e11c00c 2021-03-02 op gemtext_parse(struct parser *p, const char *buf, size_t size)
341 a5845bb5 2021-03-20 op return parser_foreach_line(p, buf, size, gemtext_foreach_line);
345 a5845bb5 2021-03-20 op gemtext_foreach_line(struct parser *p, const char *line, size_t linelen)
347 a5845bb5 2021-03-20 op enum line_type t;
349 a5845bb5 2021-03-20 op t = detect_line_type(line, linelen, p->flags);
350 a5845bb5 2021-03-20 op if (t == LINE_PRE_START)
351 a5845bb5 2021-03-20 op p->flags = 1;
352 a5845bb5 2021-03-20 op if (t == LINE_PRE_END)
353 a5845bb5 2021-03-20 op p->flags = 0;
354 a5845bb5 2021-03-20 op return parsers[t](p, t, line, linelen);
358 5e11c00c 2021-03-02 op gemtext_free(struct parser *p)
360 a5c3e03d 2021-03-02 op enum line_type t;
362 a5c3e03d 2021-03-02 op /* flush the buffer */
363 a5c3e03d 2021-03-02 op if (p->len != 0) {
364 a5c3e03d 2021-03-02 op t = detect_line_type(p->buf, p->len, p->flags);
365 a5c3e03d 2021-03-02 op if (!parsers[t](p, t, p->buf, p->len))
367 a5c3e03d 2021-03-02 op if (p->flags && !emit_line(p, LINE_PRE_END, NULL, NULL))
371 5e11c00c 2021-03-02 op free(p->buf);