1 d5af38cc 2022-01-10 op /*
  2 d5af38cc 2022-01-10 op  * Copyright (c) 2022 Omar Polo <op@omarpolo.com>
  3 d5af38cc 2022-01-10 op  *
  4 d5af38cc 2022-01-10 op  * Permission to use, copy, modify, and distribute this software for any
  5 d5af38cc 2022-01-10 op  * purpose with or without fee is hereby granted, provided that the above
  6 d5af38cc 2022-01-10 op  * copyright notice and this permission notice appear in all copies.
  7 d5af38cc 2022-01-10 op  *
  8 d5af38cc 2022-01-10 op  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  9 d5af38cc 2022-01-10 op  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
 10 d5af38cc 2022-01-10 op  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
 11 d5af38cc 2022-01-10 op  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
 12 d5af38cc 2022-01-10 op  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
 13 d5af38cc 2022-01-10 op  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
 14 d5af38cc 2022-01-10 op  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 15 d5af38cc 2022-01-10 op  */
 16 d5af38cc 2022-01-10 op 
 17 d5af38cc 2022-01-10 op #include "compat.h"
 18 d5af38cc 2022-01-10 op 
 19 d5af38cc 2022-01-10 op #include <stdlib.h>
 20 d5af38cc 2022-01-10 op #include <stdint.h>
 21 d5af38cc 2022-01-10 op #include <string.h>
 22 d5af38cc 2022-01-10 op 
 23 d5af38cc 2022-01-10 op #include "telescope.h"
 24 d5af38cc 2022-01-10 op #include "mcache.h"
 25 8f3c9af8 2022-01-11 op #include "parser.h"
 26 9d65b1d9 2022-01-11 op #include "utils.h"
 27 d5af38cc 2022-01-10 op 
 28 946d3439 2022-01-12 op static struct timeval tv = { 5 * 60, 0 };
 29 4b4c3458 2022-01-11 op static struct event timerev;
 30 4b4c3458 2022-01-11 op 
 31 a96922a1 2022-01-11 op static struct ohash	h;
 32 a96922a1 2022-01-11 op static size_t		npages;
 33 a96922a1 2022-01-11 op static size_t		tot;
 34 d5af38cc 2022-01-10 op 
 35 d5af38cc 2022-01-10 op struct mcache_entry {
 36 4b4c3458 2022-01-11 op 	time_t		 ts;
 37 fe84b750 2022-01-19 op 	parserfn	 parser;
 38 d5af38cc 2022-01-10 op 	int		 trust;
 39 d5af38cc 2022-01-10 op 	struct evbuffer	*evb;
 40 d5af38cc 2022-01-10 op 	char		 url[];
 41 d5af38cc 2022-01-10 op };
 42 d5af38cc 2022-01-10 op 
 43 ffcd827c 2022-01-11 op static void
 44 ffcd827c 2022-01-11 op mcache_free_entry(const char *url)
 45 ffcd827c 2022-01-11 op {
 46 ffcd827c 2022-01-11 op 	struct mcache_entry	*e;
 47 ffcd827c 2022-01-11 op 	unsigned int		 slot;
 48 ffcd827c 2022-01-11 op 
 49 a96922a1 2022-01-11 op 	slot = ohash_qlookup(&h, url);
 50 a96922a1 2022-01-11 op 	if ((e = ohash_remove(&h, slot)) == NULL)
 51 ffcd827c 2022-01-11 op 		return;
 52 ffcd827c 2022-01-11 op 
 53 a96922a1 2022-01-11 op 	npages--;
 54 a96922a1 2022-01-11 op 	tot -= EVBUFFER_LENGTH(e->evb);
 55 befa807e 2022-01-11 op 
 56 ffcd827c 2022-01-11 op 	evbuffer_free(e->evb);
 57 ffcd827c 2022-01-11 op 	free(e);
 58 ffcd827c 2022-01-11 op }
 59 ffcd827c 2022-01-11 op 
 60 4b4c3458 2022-01-11 op static void
 61 4b4c3458 2022-01-11 op clean_old_entries(int fd, short ev, void *data)
 62 4b4c3458 2022-01-11 op {
 63 4b4c3458 2022-01-11 op 	struct mcache_entry	*e;
 64 4b4c3458 2022-01-11 op 	unsigned int		 i;
 65 4b4c3458 2022-01-11 op 	time_t			 treshold;
 66 4b4c3458 2022-01-11 op 
 67 946d3439 2022-01-12 op 	/* delete pages older than an hour */
 68 946d3439 2022-01-12 op 	treshold = time(NULL) - 60 * 60;
 69 4b4c3458 2022-01-11 op 
 70 a96922a1 2022-01-11 op 	for (e = ohash_first(&h, &i); e != NULL; e = ohash_next(&h, &i))
 71 4b4c3458 2022-01-11 op 		if (e->ts < treshold)
 72 4b4c3458 2022-01-11 op 			mcache_free_entry(e->url);
 73 4b4c3458 2022-01-11 op 
 74 4b4c3458 2022-01-11 op 	evtimer_add(&timerev, &tv);
 75 4b4c3458 2022-01-11 op }
 76 4b4c3458 2022-01-11 op 
 77 d5af38cc 2022-01-10 op void
 78 d5af38cc 2022-01-10 op mcache_init(void)
 79 d5af38cc 2022-01-10 op {
 80 d5af38cc 2022-01-10 op 	struct ohash_info info = {
 81 d5af38cc 2022-01-10 op 		.key_offset = offsetof(struct mcache_entry, url),
 82 d5af38cc 2022-01-10 op 		.calloc = hash_calloc,
 83 d5af38cc 2022-01-10 op 		.free = hash_free,
 84 d5af38cc 2022-01-10 op 		.alloc = hash_alloc,
 85 d5af38cc 2022-01-10 op 	};
 86 d5af38cc 2022-01-10 op 
 87 a96922a1 2022-01-11 op 	ohash_init(&h, 5, &info);
 88 4b4c3458 2022-01-11 op 
 89 4b4c3458 2022-01-11 op 	evtimer_set(&timerev, clean_old_entries, NULL);
 90 d5af38cc 2022-01-10 op }
 91 d5af38cc 2022-01-10 op 
 92 d5af38cc 2022-01-10 op int
 93 8f3c9af8 2022-01-11 op mcache_tab(struct tab *tab)
 94 d5af38cc 2022-01-10 op {
 95 d5af38cc 2022-01-10 op 	struct mcache_entry	*e;
 96 d5af38cc 2022-01-10 op 	unsigned int		 slot;
 97 d5af38cc 2022-01-10 op 	size_t			 l, len;
 98 8f3c9af8 2022-01-11 op 	const char		*url;
 99 d5af38cc 2022-01-10 op 
100 8f3c9af8 2022-01-11 op 	url = tab->hist_cur->h;
101 d5af38cc 2022-01-10 op 	l = strlen(url);
102 d5af38cc 2022-01-10 op 	len = sizeof(*e) + l + 1;
103 d5af38cc 2022-01-10 op 
104 d5af38cc 2022-01-10 op 	if ((e = calloc(1, len)) == NULL)
105 d5af38cc 2022-01-10 op 		return -1;
106 4b4c3458 2022-01-11 op 	e->ts = time(NULL);
107 fe84b750 2022-01-19 op 	e->parser = tab->buffer.page.init;
108 8f3c9af8 2022-01-11 op 	e->trust = tab->trust;
109 d5af38cc 2022-01-10 op 	memcpy(e->url, url, l);
110 d5af38cc 2022-01-10 op 
111 d5af38cc 2022-01-10 op 	if ((e->evb = evbuffer_new()) == NULL)
112 d5af38cc 2022-01-10 op 		goto err;
113 d5af38cc 2022-01-10 op 
114 fe84b750 2022-01-19 op 	if (!parser_serialize(tab, e->evb))
115 fe84b750 2022-01-19 op 		goto err;
116 d5af38cc 2022-01-10 op 
117 ffcd827c 2022-01-11 op 	/* free any previously cached copies of this page */
118 ffcd827c 2022-01-11 op 	mcache_free_entry(url);
119 ffcd827c 2022-01-11 op 
120 a96922a1 2022-01-11 op 	slot = ohash_qlookup(&h, url);
121 a96922a1 2022-01-11 op 	ohash_insert(&h, slot, e);
122 befa807e 2022-01-11 op 
123 a96922a1 2022-01-11 op 	npages++;
124 a96922a1 2022-01-11 op 	tot += EVBUFFER_LENGTH(e->evb);
125 befa807e 2022-01-11 op 
126 4b4c3458 2022-01-11 op 	if (!evtimer_pending(&timerev, NULL))
127 4b4c3458 2022-01-11 op 		evtimer_add(&timerev, &tv);
128 4b4c3458 2022-01-11 op 
129 d5af38cc 2022-01-10 op 	return 0;
130 d5af38cc 2022-01-10 op 
131 d5af38cc 2022-01-10 op err:
132 d5af38cc 2022-01-10 op 	if (e->evb != NULL)
133 d5af38cc 2022-01-10 op 		evbuffer_free(e->evb);
134 d5af38cc 2022-01-10 op 	free(e);
135 d5af38cc 2022-01-10 op 	return -1;
136 d5af38cc 2022-01-10 op }
137 d5af38cc 2022-01-10 op 
138 d5af38cc 2022-01-10 op int
139 8f3c9af8 2022-01-11 op mcache_lookup(const char *url, struct tab *tab)
140 d5af38cc 2022-01-10 op {
141 d5af38cc 2022-01-10 op 	struct mcache_entry	*e;
142 d5af38cc 2022-01-10 op 	unsigned int		 slot;
143 d5af38cc 2022-01-10 op 
144 a96922a1 2022-01-11 op 	slot = ohash_qlookup(&h, url);
145 a96922a1 2022-01-11 op 	if ((e = ohash_find(&h, slot)) == NULL)
146 d5af38cc 2022-01-10 op 		return 0;
147 d5af38cc 2022-01-10 op 
148 fe84b750 2022-01-19 op 	parser_init(tab, e->parser);
149 8f3c9af8 2022-01-11 op 	if (!parser_parse(tab, EVBUFFER_DATA(e->evb), EVBUFFER_LENGTH(e->evb)))
150 8f3c9af8 2022-01-11 op 		goto err;
151 8f3c9af8 2022-01-11 op 	if (!parser_free(tab))
152 8f3c9af8 2022-01-11 op 		goto err;
153 8f3c9af8 2022-01-11 op 
154 8f3c9af8 2022-01-11 op 	tab->trust = e->trust;
155 d5af38cc 2022-01-10 op 	return 1;
156 8f3c9af8 2022-01-11 op 
157 8f3c9af8 2022-01-11 op err:
158 8f3c9af8 2022-01-11 op 	parser_free(tab);
159 8f3c9af8 2022-01-11 op 	erase_buffer(&tab->buffer);
160 8f3c9af8 2022-01-11 op 	return 0;
161 d5af38cc 2022-01-10 op }
162 befa807e 2022-01-11 op 
163 befa807e 2022-01-11 op void
164 a96922a1 2022-01-11 op mcache_info(size_t *r_npages, size_t *r_tot)
165 befa807e 2022-01-11 op {
166 a96922a1 2022-01-11 op 	*r_npages = npages;
167 a96922a1 2022-01-11 op 	*r_tot = tot;
168 befa807e 2022-01-11 op }