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 a467be63 2021-08-14 op #include "compat.h"
19 191b4ca2 2021-03-12 op #include <stdlib.h>
20 191b4ca2 2021-03-12 op #include <string.h>
22 a467be63 2021-08-14 op #include "parser.h"
23 a467be63 2021-08-14 op #include "telescope.h"
26 8c1b0837 2021-07-25 op parser_init(struct tab *tab, parserfn fn)
28 9eca3d7c 2021-11-02 op erase_buffer(&tab->buffer);
29 8c1b0837 2021-07-25 op fn(&tab->buffer.page);
30 5a21e0eb 2022-01-19 op tab->buffer.page.init = fn;
34 8c1b0837 2021-07-25 op parser_parse(struct tab *tab, const char *chunk, size_t len)
36 8c1b0837 2021-07-25 op return tab->buffer.page.parse(&tab->buffer.page, chunk, len);
40 8c1b0837 2021-07-25 op parser_free(struct tab *tab)
43 2fd6d754 2021-07-30 op char *tilde, *slash;
45 8c1b0837 2021-07-25 op r = tab->buffer.page.free(&tab->buffer.page);
47 2fd6d754 2021-07-30 op if (*tab->buffer.page.title != '\0')
51 dafcab64 2021-08-26 op * heuristic: see if there is a "tilde user" and use that as
52 dafcab64 2021-08-26 op * page title, using the full domain name as fallback.
54 2fd6d754 2021-07-30 op if ((tilde = strstr(tab->hist_cur->h, "/~")) != NULL) {
55 2fd6d754 2021-07-30 op strlcpy(tab->buffer.page.title, tilde+1,
56 2fd6d754 2021-07-30 op sizeof(tab->buffer.page.title));
58 2fd6d754 2021-07-30 op if ((slash = strchr(tab->buffer.page.title, '/')) != NULL)
59 2fd6d754 2021-07-30 op *slash = '\0';
61 8c1b0837 2021-07-25 op strlcpy(tab->buffer.page.title, tab->uri.host,
62 8c1b0837 2021-07-25 op sizeof(tab->buffer.page.title));
68 069c9a92 2022-01-19 op parser_serialize(struct tab *tab, struct evbuffer *evb)
70 069c9a92 2022-01-19 op struct line *line;
71 069c9a92 2022-01-19 op const char *text;
74 069c9a92 2022-01-19 op if (tab->buffer.page.serialize != NULL)
75 069c9a92 2022-01-19 op return tab->buffer.page.serialize(&tab->buffer.page, evb);
77 069c9a92 2022-01-19 op /* a default implementation good enough for plain text */
78 069c9a92 2022-01-19 op TAILQ_FOREACH(line, &tab->buffer.page.head, lines) {
79 069c9a92 2022-01-19 op if ((text = line->line) == NULL)
82 069c9a92 2022-01-19 op r = evbuffer_add_printf(evb, "%s\n", text);
91 191b4ca2 2021-03-12 op parser_append(struct parser *p, const char *buf, size_t len)
93 191b4ca2 2021-03-12 op size_t newlen;
96 191b4ca2 2021-03-12 op newlen = len + p->len;
97 191b4ca2 2021-03-12 op if ((t = calloc(1, newlen)) == NULL)
99 191b4ca2 2021-03-12 op memcpy(t, p->buf, p->len);
100 191b4ca2 2021-03-12 op memcpy(t + p->len, buf, len);
101 191b4ca2 2021-03-12 op free(p->buf);
103 191b4ca2 2021-03-12 op p->len = newlen;
108 191b4ca2 2021-03-12 op parser_set_buf(struct parser *p, const char *buf, size_t len)
112 191b4ca2 2021-03-12 op if (len == 0) {
114 20107775 2021-03-19 op free(p->buf);
115 20107775 2021-03-19 op p->buf = NULL;
120 be2ee49f 2021-05-14 op * p->buf and buf can (and probably almost always will)
124 20107775 2021-03-19 op if ((tmp = calloc(1, len)) == NULL)
126 20107775 2021-03-19 op memcpy(tmp, buf, len);
127 20107775 2021-03-19 op free(p->buf);
128 20107775 2021-03-19 op p->buf = tmp;
129 191b4ca2 2021-03-12 op p->len = len;
134 a5845bb5 2021-03-20 op parser_foreach_line(struct parser *p, const char *buf, size_t size,
135 a5845bb5 2021-03-20 op parsechunkfn fn)
138 31aa9f59 2021-03-20 op unsigned int ch;
139 31aa9f59 2021-03-20 op size_t i, l, len;
141 31aa9f59 2021-03-20 op if (!parser_append(p, buf, size))
144 31aa9f59 2021-03-20 op len = p->len;
146 27dce34f 2021-07-06 op if (!(p->flags & PARSER_IN_BODY) && len < 3)
149 27dce34f 2021-07-06 op if (!(p->flags & PARSER_IN_BODY)) {
150 95a8c791 2021-08-26 op p->flags |= PARSER_IN_BODY;
153 c0b634dd 2021-07-06 op * drop the BOM: only UTF-8 is supported, and there
154 c0b634dd 2021-07-06 op * it's useless; some editors may still add one
157 c0b634dd 2021-07-06 op if (memmem(b, len, "\xEF\xBB\xBF", 3) == b) {
163 31aa9f59 2021-03-20 op /* drop every "funny" ASCII character */
164 31aa9f59 2021-03-20 op for (i = 0; i < len; ) {
166 31aa9f59 2021-03-20 op if ((ch >= ' ' || ch == '\n' || ch == '\t')
167 31aa9f59 2021-03-20 op && ch != 127) { /* del */
171 08b9feb5 2021-07-26 op memmove(&b[i], &b[i+1], len - i - 1);
175 a5845bb5 2021-03-20 op while (len > 0) {
176 a5845bb5 2021-03-20 op if ((e = memmem((char*)b, len, "\n", 1)) == NULL)
180 a5845bb5 2021-03-20 op if (!fn(p, b, l))
186 a5845bb5 2021-03-20 op if (len > 0) {
187 a5845bb5 2021-03-20 op /* skip \n */
193 a5845bb5 2021-03-20 op return parser_set_buf(p, b, len);