1 1ac119fb 2024-01-23 op /*
  2 1ac119fb 2024-01-23 op  * Copyright (c) 2021 Omar Polo <op@omarpolo.com>
  3 1ac119fb 2024-01-23 op  *
  4 1ac119fb 2024-01-23 op  * Permission to use, copy, modify, and distribute this software for any
  5 1ac119fb 2024-01-23 op  * purpose with or without fee is hereby granted, provided that the above
  6 1ac119fb 2024-01-23 op  * copyright notice and this permission notice appear in all copies.
  7 1ac119fb 2024-01-23 op  *
  8 1ac119fb 2024-01-23 op  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  9 1ac119fb 2024-01-23 op  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
 10 1ac119fb 2024-01-23 op  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
 11 1ac119fb 2024-01-23 op  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
 12 1ac119fb 2024-01-23 op  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
 13 1ac119fb 2024-01-23 op  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
 14 1ac119fb 2024-01-23 op  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 15 1ac119fb 2024-01-23 op  */
 16 1ac119fb 2024-01-23 op 
 17 1ac119fb 2024-01-23 op #include "compat.h"
 18 1ac119fb 2024-01-23 op 
 19 1ac119fb 2024-01-23 op #include <stdlib.h>
 20 1ac119fb 2024-01-23 op #include <string.h>
 21 1ac119fb 2024-01-23 op 
 22 1ac119fb 2024-01-23 op #include "hist.h"
 23 1ac119fb 2024-01-23 op #include "parser.h"
 24 1ac119fb 2024-01-23 op #include "telescope.h"
 25 1ac119fb 2024-01-23 op 
 26 1ac119fb 2024-01-23 op void
 27 1ac119fb 2024-01-23 op parser_init(struct tab *tab, parserfn fn)
 28 1ac119fb 2024-01-23 op {
 29 1ac119fb 2024-01-23 op 	erase_buffer(&tab->buffer);
 30 1ac119fb 2024-01-23 op 	fn(&tab->buffer.page);
 31 1ac119fb 2024-01-23 op 	tab->buffer.page.init = fn;
 32 1ac119fb 2024-01-23 op }
 33 1ac119fb 2024-01-23 op 
 34 1ac119fb 2024-01-23 op int
 35 1ac119fb 2024-01-23 op parser_parse(struct tab *tab, const char *chunk, size_t len)
 36 1ac119fb 2024-01-23 op {
 37 1ac119fb 2024-01-23 op 	return tab->buffer.page.parse(&tab->buffer.page, chunk, len);
 38 1ac119fb 2024-01-23 op }
 39 1ac119fb 2024-01-23 op 
 40 1ac119fb 2024-01-23 op int
 41 1ac119fb 2024-01-23 op parser_parsef(struct tab *tab, const char *fmt, ...)
 42 1ac119fb 2024-01-23 op {
 43 1ac119fb 2024-01-23 op 	char *s;
 44 1ac119fb 2024-01-23 op 	va_list ap;
 45 1ac119fb 2024-01-23 op 	int r;
 46 1ac119fb 2024-01-23 op 
 47 1ac119fb 2024-01-23 op 	va_start(ap, fmt);
 48 1ac119fb 2024-01-23 op 	r = vasprintf(&s, fmt, ap);
 49 1ac119fb 2024-01-23 op 	va_end(ap);
 50 1ac119fb 2024-01-23 op 
 51 1ac119fb 2024-01-23 op 	if (r == -1)
 52 1ac119fb 2024-01-23 op 		return 0;
 53 1ac119fb 2024-01-23 op 
 54 1ac119fb 2024-01-23 op 	r = parser_parse(tab, s, strlen(s));
 55 1ac119fb 2024-01-23 op 	free(s);
 56 1ac119fb 2024-01-23 op 	return r;
 57 1ac119fb 2024-01-23 op }
 58 1ac119fb 2024-01-23 op 
 59 1ac119fb 2024-01-23 op int
 60 1ac119fb 2024-01-23 op parser_free(struct tab *tab)
 61 1ac119fb 2024-01-23 op {
 62 1ac119fb 2024-01-23 op 	int	 r;
 63 1ac119fb 2024-01-23 op 	char	*tilde, *slash;
 64 1ac119fb 2024-01-23 op 
 65 1ac119fb 2024-01-23 op 	r = tab->buffer.page.free(&tab->buffer.page);
 66 1ac119fb 2024-01-23 op 
 67 1ac119fb 2024-01-23 op 	if (*tab->buffer.page.title != '\0')
 68 1ac119fb 2024-01-23 op 		return r;
 69 1ac119fb 2024-01-23 op 
 70 1ac119fb 2024-01-23 op 	/*
 71 1ac119fb 2024-01-23 op 	 * heuristic: see if there is a "tilde user" and use that as
 72 1ac119fb 2024-01-23 op 	 * page title, using the full domain name as fallback.
 73 1ac119fb 2024-01-23 op 	 */
 74 1ac119fb 2024-01-23 op 	if ((tilde = strstr(hist_cur(tab->hist), "/~")) != NULL) {
 75 1ac119fb 2024-01-23 op 		strlcpy(tab->buffer.page.title, tilde+1,
 76 1ac119fb 2024-01-23 op 		    sizeof(tab->buffer.page.title));
 77 1ac119fb 2024-01-23 op 
 78 1ac119fb 2024-01-23 op 		if ((slash = strchr(tab->buffer.page.title, '/')) != NULL)
 79 1ac119fb 2024-01-23 op 			*slash = '\0';
 80 1ac119fb 2024-01-23 op 	} else
 81 1ac119fb 2024-01-23 op 		strlcpy(tab->buffer.page.title, tab->iri.iri_host,
 82 1ac119fb 2024-01-23 op 		    sizeof(tab->buffer.page.title));
 83 1ac119fb 2024-01-23 op 
 84 1ac119fb 2024-01-23 op 	return r;
 85 1ac119fb 2024-01-23 op }
 86 1ac119fb 2024-01-23 op 
 87 1ac119fb 2024-01-23 op int
 88 1ac119fb 2024-01-23 op parser_serialize(struct tab *tab, FILE *fp)
 89 1ac119fb 2024-01-23 op {
 90 1ac119fb 2024-01-23 op 	struct line	*line;
 91 1ac119fb 2024-01-23 op 	const char	*text;
 92 1ac119fb 2024-01-23 op 	int		 r;
 93 1ac119fb 2024-01-23 op 
 94 1ac119fb 2024-01-23 op 	if (tab->buffer.page.serialize != NULL)
 95 1ac119fb 2024-01-23 op 		return tab->buffer.page.serialize(&tab->buffer.page, fp);
 96 1ac119fb 2024-01-23 op 
 97 1ac119fb 2024-01-23 op 	/* a default implementation good enough for plain text */
 98 1ac119fb 2024-01-23 op 	TAILQ_FOREACH(line, &tab->buffer.page.head, lines) {
 99 1ac119fb 2024-01-23 op 		if ((text = line->line) == NULL)
100 1ac119fb 2024-01-23 op 			text = "";
101 1ac119fb 2024-01-23 op 
102 1ac119fb 2024-01-23 op 		r = fprintf(fp, "%s\n", text);
103 1ac119fb 2024-01-23 op 		if (r == -1)
104 1ac119fb 2024-01-23 op 			return 0;
105 1ac119fb 2024-01-23 op 	}
106 1ac119fb 2024-01-23 op 
107 1ac119fb 2024-01-23 op 	return 1;
108 1ac119fb 2024-01-23 op }
109 1ac119fb 2024-01-23 op 
110 1ac119fb 2024-01-23 op int
111 1ac119fb 2024-01-23 op parser_append(struct parser *p, const char *buf, size_t len)
112 1ac119fb 2024-01-23 op {
113 1ac119fb 2024-01-23 op 	size_t newlen;
114 1ac119fb 2024-01-23 op 	char *t;
115 1ac119fb 2024-01-23 op 
116 1ac119fb 2024-01-23 op 	newlen = len + p->len;
117 1ac119fb 2024-01-23 op 	if ((t = calloc(1, newlen)) == NULL)
118 1ac119fb 2024-01-23 op 		return 0;
119 1ac119fb 2024-01-23 op 	memcpy(t, p->buf, p->len);
120 1ac119fb 2024-01-23 op 	memcpy(t + p->len, buf, len);
121 1ac119fb 2024-01-23 op 	free(p->buf);
122 1ac119fb 2024-01-23 op 	p->buf = t;
123 1ac119fb 2024-01-23 op 	p->len = newlen;
124 1ac119fb 2024-01-23 op 	return 1;
125 1ac119fb 2024-01-23 op }
126 1ac119fb 2024-01-23 op 
127 1ac119fb 2024-01-23 op int
128 1ac119fb 2024-01-23 op parser_set_buf(struct parser *p, const char *buf, size_t len)
129 1ac119fb 2024-01-23 op {
130 1ac119fb 2024-01-23 op 	char *tmp;
131 1ac119fb 2024-01-23 op 
132 1ac119fb 2024-01-23 op 	if (len == 0) {
133 1ac119fb 2024-01-23 op 		p->len = 0;
134 1ac119fb 2024-01-23 op 		free(p->buf);
135 1ac119fb 2024-01-23 op 		p->buf = NULL;
136 1ac119fb 2024-01-23 op 		return 1;
137 1ac119fb 2024-01-23 op 	}
138 1ac119fb 2024-01-23 op 
139 1ac119fb 2024-01-23 op 	/*
140 1ac119fb 2024-01-23 op 	 * p->buf and buf can (and probably almost always will)
141 1ac119fb 2024-01-23 op 	 * overlap!
142 1ac119fb 2024-01-23 op 	 */
143 1ac119fb 2024-01-23 op 
144 1ac119fb 2024-01-23 op 	if ((tmp = calloc(1, len)) == NULL)
145 1ac119fb 2024-01-23 op 		return 0;
146 1ac119fb 2024-01-23 op 	memcpy(tmp, buf, len);
147 1ac119fb 2024-01-23 op 	free(p->buf);
148 1ac119fb 2024-01-23 op 	p->buf = tmp;
149 1ac119fb 2024-01-23 op 	p->len = len;
150 1ac119fb 2024-01-23 op 	return 1;
151 1ac119fb 2024-01-23 op }
152 1ac119fb 2024-01-23 op 
153 1ac119fb 2024-01-23 op int
154 1ac119fb 2024-01-23 op parser_foreach_line(struct parser *p, const char *buf, size_t size,
155 1ac119fb 2024-01-23 op     parsechunkfn fn)
156 1ac119fb 2024-01-23 op {
157 1ac119fb 2024-01-23 op 	char		*b, *e;
158 1ac119fb 2024-01-23 op 	unsigned int	 ch;
159 1ac119fb 2024-01-23 op 	size_t		 i, l, len;
160 1ac119fb 2024-01-23 op 
161 1ac119fb 2024-01-23 op 	if (!parser_append(p, buf, size))
162 1ac119fb 2024-01-23 op 		return 0;
163 1ac119fb 2024-01-23 op 	b = p->buf;
164 1ac119fb 2024-01-23 op 	len = p->len;
165 1ac119fb 2024-01-23 op 
166 1ac119fb 2024-01-23 op 	if (!(p->flags & PARSER_IN_BODY) && len < 3)
167 1ac119fb 2024-01-23 op 		return 1;
168 1ac119fb 2024-01-23 op 
169 1ac119fb 2024-01-23 op 	if (!(p->flags & PARSER_IN_BODY)) {
170 1ac119fb 2024-01-23 op 		p->flags |= PARSER_IN_BODY;
171 1ac119fb 2024-01-23 op 
172 1ac119fb 2024-01-23 op 		/*
173 1ac119fb 2024-01-23 op 		 * drop the BOM: only UTF-8 is supported, and there
174 1ac119fb 2024-01-23 op 		 * it's useless; some editors may still add one
175 1ac119fb 2024-01-23 op 		 * though.
176 1ac119fb 2024-01-23 op 		 */
177 1ac119fb 2024-01-23 op 		if (memmem(b, len, "\xEF\xBB\xBF", 3) == b) {
178 1ac119fb 2024-01-23 op 			b += 3;
179 1ac119fb 2024-01-23 op 			len -= 3;
180 1ac119fb 2024-01-23 op 		}
181 1ac119fb 2024-01-23 op 	}
182 1ac119fb 2024-01-23 op 
183 1ac119fb 2024-01-23 op 	/* drop every "funny" ASCII character */
184 1ac119fb 2024-01-23 op 	for (i = 0; i < len; ) {
185 1ac119fb 2024-01-23 op 		ch = b[i];
186 1ac119fb 2024-01-23 op 		if ((ch >= ' ' || ch == '\n' || ch == '\t')
187 1ac119fb 2024-01-23 op 		    && ch != 127) { /* del */
188 1ac119fb 2024-01-23 op 			++i;
189 1ac119fb 2024-01-23 op 			continue;
190 1ac119fb 2024-01-23 op 		}
191 1ac119fb 2024-01-23 op 		memmove(&b[i], &b[i+1], len - i - 1);
192 1ac119fb 2024-01-23 op 		len--;
193 1ac119fb 2024-01-23 op 	}
194 1ac119fb 2024-01-23 op 
195 1ac119fb 2024-01-23 op 	while (len > 0) {
196 1ac119fb 2024-01-23 op 		if ((e = memmem((char*)b, len, "\n", 1)) == NULL)
197 1ac119fb 2024-01-23 op 			break;
198 1ac119fb 2024-01-23 op 		l = e - b;
199 1ac119fb 2024-01-23 op 
200 1ac119fb 2024-01-23 op 		if (!fn(p, b, l))
201 1ac119fb 2024-01-23 op 			return 0;
202 1ac119fb 2024-01-23 op 
203 1ac119fb 2024-01-23 op 		len -= l;
204 1ac119fb 2024-01-23 op 		b += l;
205 1ac119fb 2024-01-23 op 
206 1ac119fb 2024-01-23 op 		if (len > 0) {
207 1ac119fb 2024-01-23 op 			/* skip \n */
208 1ac119fb 2024-01-23 op 			len--;
209 1ac119fb 2024-01-23 op 			b++;
210 1ac119fb 2024-01-23 op 		}
211 1ac119fb 2024-01-23 op 	}
212 1ac119fb 2024-01-23 op 
213 1ac119fb 2024-01-23 op 	return parser_set_buf(p, b, len);
214 1ac119fb 2024-01-23 op }