2 191b4ca2 2021-03-12 op * Copyright (c) 2021 Omar Polo <op@omarpolo.com>
4 191b4ca2 2021-03-12 op * Permission to use, copy, modify, and distribute this software for any
5 191b4ca2 2021-03-12 op * purpose with or without fee is hereby granted, provided that the above
6 191b4ca2 2021-03-12 op * copyright notice and this permission notice appear in all copies.
8 191b4ca2 2021-03-12 op * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 191b4ca2 2021-03-12 op * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 191b4ca2 2021-03-12 op * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 191b4ca2 2021-03-12 op * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 191b4ca2 2021-03-12 op * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 191b4ca2 2021-03-12 op * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 191b4ca2 2021-03-12 op * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 395b9f4e 2021-07-12 op #include "parser.h"
18 191b4ca2 2021-03-12 op #include "telescope.h"
20 191b4ca2 2021-03-12 op #include <stdlib.h>
21 191b4ca2 2021-03-12 op #include <string.h>
24 8c1b0837 2021-07-25 op parser_init(struct tab *tab, parserfn fn)
26 8c1b0837 2021-07-25 op fn(&tab->buffer.page);
30 8c1b0837 2021-07-25 op parser_parse(struct tab *tab, const char *chunk, size_t len)
32 8c1b0837 2021-07-25 op return tab->buffer.page.parse(&tab->buffer.page, chunk, len);
36 8c1b0837 2021-07-25 op parser_free(struct tab *tab)
40 8c1b0837 2021-07-25 op r = tab->buffer.page.free(&tab->buffer.page);
42 8c1b0837 2021-07-25 op /* fallback to the host as title if nothing else */
43 8c1b0837 2021-07-25 op if (*tab->buffer.page.title == '\0')
44 8c1b0837 2021-07-25 op strlcpy(tab->buffer.page.title, tab->uri.host,
45 8c1b0837 2021-07-25 op sizeof(tab->buffer.page.title));
51 191b4ca2 2021-03-12 op parser_append(struct parser *p, const char *buf, size_t len)
53 191b4ca2 2021-03-12 op size_t newlen;
56 191b4ca2 2021-03-12 op newlen = len + p->len;
57 191b4ca2 2021-03-12 op if ((t = calloc(1, newlen)) == NULL)
59 191b4ca2 2021-03-12 op memcpy(t, p->buf, p->len);
60 191b4ca2 2021-03-12 op memcpy(t + p->len, buf, len);
63 191b4ca2 2021-03-12 op p->len = newlen;
68 191b4ca2 2021-03-12 op parser_set_buf(struct parser *p, const char *buf, size_t len)
72 191b4ca2 2021-03-12 op if (len == 0) {
75 20107775 2021-03-19 op p->buf = NULL;
80 be2ee49f 2021-05-14 op * p->buf and buf can (and probably almost always will)
84 20107775 2021-03-19 op if ((tmp = calloc(1, len)) == NULL)
86 20107775 2021-03-19 op memcpy(tmp, buf, len);
94 a5845bb5 2021-03-20 op parser_foreach_line(struct parser *p, const char *buf, size_t size,
95 a5845bb5 2021-03-20 op parsechunkfn fn)
98 31aa9f59 2021-03-20 op unsigned int ch;
99 31aa9f59 2021-03-20 op size_t i, l, len;
101 31aa9f59 2021-03-20 op if (!parser_append(p, buf, size))
104 31aa9f59 2021-03-20 op len = p->len;
106 27dce34f 2021-07-06 op if (!(p->flags & PARSER_IN_BODY) && len < 3)
109 27dce34f 2021-07-06 op if (!(p->flags & PARSER_IN_BODY)) {
110 27dce34f 2021-07-06 op p->flags |= PARSER_IN_BODY;
113 c0b634dd 2021-07-06 op * drop the BOM: only UTF-8 is supported, and there
114 c0b634dd 2021-07-06 op * it's useless; some editors may still add one
117 c0b634dd 2021-07-06 op if (memmem(b, len, "\xEF\xBB\xBF", 3) == b) {
123 31aa9f59 2021-03-20 op /* drop every "funny" ASCII character */
124 31aa9f59 2021-03-20 op for (i = 0; i < len; ) {
126 31aa9f59 2021-03-20 op if ((ch >= ' ' || ch == '\n' || ch == '\t')
127 31aa9f59 2021-03-20 op && ch != 127) { /* del */
131 31aa9f59 2021-03-20 op memmove(&b[i], &b[i+1], len - i);
135 a5845bb5 2021-03-20 op while (len > 0) {
136 a5845bb5 2021-03-20 op if ((e = memmem((char*)b, len, "\n", 1)) == NULL)
140 a5845bb5 2021-03-20 op if (!fn(p, b, l))
146 a5845bb5 2021-03-20 op if (len > 0) {
147 a5845bb5 2021-03-20 op /* skip \n */
153 a5845bb5 2021-03-20 op return parser_set_buf(p, b, len);