op public repos

Blame

Date:: Wed Jan 19 23:56:36 2022 UTC
Message:: parsers: add serialize function Add a parser_serialize function and an optional function `serialize' for parsers. The default implementation just prints the text line, which is suitable for parser_textplain and parser_textpatch; gemtext and gophermap uses their own specific unparse functions. This is intended for an incoming cache refactoring: we'll cache the page serialized in the correct format, not once it has been translated to gemtext. This will allow in the future to provide commands such as ``save page to disk'' or interface with an on-disk cache.
Actions:: History | Blob | Raw File
  1 191b4ca2 2021-03-12 op /*
  2 191b4ca2 2021-03-12 op  * Copyright (c) 2021 Omar Polo <op@omarpolo.com>
  3 191b4ca2 2021-03-12 op  *
  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.
  7 191b4ca2 2021-03-12 op  *
  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.
 15 191b4ca2 2021-03-12 op  */
 16 191b4ca2 2021-03-12 op 
 17 a467be63 2021-08-14 op #include "compat.h"
 18 191b4ca2 2021-03-12 op 
 19 191b4ca2 2021-03-12 op #include <stdlib.h>
 20 191b4ca2 2021-03-12 op #include <string.h>
 21 191b4ca2 2021-03-12 op 
 22 a467be63 2021-08-14 op #include "parser.h"
 23 a467be63 2021-08-14 op #include "telescope.h"
 24 a467be63 2021-08-14 op #include "ui.h"
 25 a467be63 2021-08-14 op 
 26 1fce2e75 2021-08-14 op /*
 27 1fce2e75 2021-08-14 op  * Load a text/gemini page given the string page.  Always returns 0.
 28 1fce2e75 2021-08-14 op  */
 29 1fce2e75 2021-08-14 op int
 30 1fce2e75 2021-08-14 op load_page_from_str(struct tab *tab, const char *page)
 31 1fce2e75 2021-08-14 op {
 32 1fce2e75 2021-08-14 op 	parser_init(tab, gemtext_initparser);
 33 1fce2e75 2021-08-14 op 	if (!tab->buffer.page.parse(&tab->buffer.page, page, strlen(page)))
 34 1fce2e75 2021-08-14 op 		abort();
 35 1fce2e75 2021-08-14 op 	if (!tab->buffer.page.free(&tab->buffer.page))
 36 1fce2e75 2021-08-14 op 		abort();
 37 1fce2e75 2021-08-14 op 	ui_on_tab_refresh(tab);
 38 1fce2e75 2021-08-14 op 	ui_on_tab_loaded(tab);
 39 1fce2e75 2021-08-14 op 	return 0;
 40 1fce2e75 2021-08-14 op }
 41 1fce2e75 2021-08-14 op 
 42 8c1b0837 2021-07-25 op void
 43 8c1b0837 2021-07-25 op parser_init(struct tab *tab, parserfn fn)
 44 8c1b0837 2021-07-25 op {
 45 9eca3d7c 2021-11-02 op 	erase_buffer(&tab->buffer);
 46 8c1b0837 2021-07-25 op 	fn(&tab->buffer.page);
 47 5a21e0eb 2022-01-19 op 	tab->buffer.page.init = fn;
 48 8c1b0837 2021-07-25 op }
 49 8c1b0837 2021-07-25 op 
 50 191b4ca2 2021-03-12 op int
 51 8c1b0837 2021-07-25 op parser_parse(struct tab *tab, const char *chunk, size_t len)
 52 8c1b0837 2021-07-25 op {
 53 8c1b0837 2021-07-25 op 	return tab->buffer.page.parse(&tab->buffer.page, chunk, len);
 54 8c1b0837 2021-07-25 op }
 55 8c1b0837 2021-07-25 op 
 56 8c1b0837 2021-07-25 op int
 57 8c1b0837 2021-07-25 op parser_free(struct tab *tab)
 58 8c1b0837 2021-07-25 op {
 59 2fd6d754 2021-07-30 op 	int	 r;
 60 2fd6d754 2021-07-30 op 	char	*tilde, *slash;
 61 8c1b0837 2021-07-25 op 
 62 8c1b0837 2021-07-25 op 	r = tab->buffer.page.free(&tab->buffer.page);
 63 8c1b0837 2021-07-25 op 
 64 2fd6d754 2021-07-30 op 	if (*tab->buffer.page.title != '\0')
 65 2fd6d754 2021-07-30 op 		return r;
 66 2fd6d754 2021-07-30 op 
 67 dafcab64 2021-08-26 op 	/*
 68 dafcab64 2021-08-26 op 	 * heuristic: see if there is a "tilde user" and use that as
 69 dafcab64 2021-08-26 op 	 * page title, using the full domain name as fallback.
 70 dafcab64 2021-08-26 op 	 */
 71 2fd6d754 2021-07-30 op 	if ((tilde = strstr(tab->hist_cur->h, "/~")) != NULL) {
 72 2fd6d754 2021-07-30 op 		strlcpy(tab->buffer.page.title, tilde+1,
 73 2fd6d754 2021-07-30 op 		    sizeof(tab->buffer.page.title));
 74 2fd6d754 2021-07-30 op 
 75 2fd6d754 2021-07-30 op 		if ((slash = strchr(tab->buffer.page.title, '/')) != NULL)
 76 2fd6d754 2021-07-30 op 			*slash = '\0';
 77 2fd6d754 2021-07-30 op 	} else
 78 8c1b0837 2021-07-25 op 		strlcpy(tab->buffer.page.title, tab->uri.host,
 79 8c1b0837 2021-07-25 op 		    sizeof(tab->buffer.page.title));
 80 8c1b0837 2021-07-25 op 
 81 8c1b0837 2021-07-25 op 	return r;
 82 8c1b0837 2021-07-25 op }
 83 8c1b0837 2021-07-25 op 
 84 8c1b0837 2021-07-25 op int
 85 069c9a92 2022-01-19 op parser_serialize(struct tab *tab, struct evbuffer *evb)
 86 069c9a92 2022-01-19 op {
 87 069c9a92 2022-01-19 op 	struct line	*line;
 88 069c9a92 2022-01-19 op 	const char	*text;
 89 069c9a92 2022-01-19 op 	int		 r;
 90 069c9a92 2022-01-19 op 
 91 069c9a92 2022-01-19 op 	if (tab->buffer.page.serialize != NULL)
 92 069c9a92 2022-01-19 op 		return tab->buffer.page.serialize(&tab->buffer.page, evb);
 93 069c9a92 2022-01-19 op 
 94 069c9a92 2022-01-19 op 	/* a default implementation good enough for plain text */
 95 069c9a92 2022-01-19 op 	TAILQ_FOREACH(line, &tab->buffer.page.head, lines) {
 96 069c9a92 2022-01-19 op 		if ((text = line->line) == NULL)
 97 069c9a92 2022-01-19 op 			text = "";
 98 069c9a92 2022-01-19 op 
 99 069c9a92 2022-01-19 op 		r = evbuffer_add_printf(evb, "%s\n", text);
100 069c9a92 2022-01-19 op 		if (r == -1)
101 069c9a92 2022-01-19 op 			return 0;
102 069c9a92 2022-01-19 op 	}
103 069c9a92 2022-01-19 op 
104 069c9a92 2022-01-19 op 	return 1;
105 069c9a92 2022-01-19 op }
106 069c9a92 2022-01-19 op 
107 069c9a92 2022-01-19 op int
108 191b4ca2 2021-03-12 op parser_append(struct parser *p, const char *buf, size_t len)
109 191b4ca2 2021-03-12 op {
110 191b4ca2 2021-03-12 op 	size_t newlen;
111 191b4ca2 2021-03-12 op 	char *t;
112 191b4ca2 2021-03-12 op 
113 191b4ca2 2021-03-12 op 	newlen = len + p->len;
114 191b4ca2 2021-03-12 op 	if ((t = calloc(1, newlen)) == NULL)
115 191b4ca2 2021-03-12 op 		return 0;
116 191b4ca2 2021-03-12 op 	memcpy(t, p->buf, p->len);
117 191b4ca2 2021-03-12 op 	memcpy(t + p->len, buf, len);
118 191b4ca2 2021-03-12 op 	free(p->buf);
119 191b4ca2 2021-03-12 op 	p->buf = t;
120 191b4ca2 2021-03-12 op 	p->len = newlen;
121 191b4ca2 2021-03-12 op 	return 1;
122 191b4ca2 2021-03-12 op }
123 191b4ca2 2021-03-12 op 
124 191b4ca2 2021-03-12 op int
125 191b4ca2 2021-03-12 op parser_set_buf(struct parser *p, const char *buf, size_t len)
126 191b4ca2 2021-03-12 op {
127 20107775 2021-03-19 op 	char *tmp;
128 191b4ca2 2021-03-12 op 
129 191b4ca2 2021-03-12 op 	if (len == 0) {
130 191b4ca2 2021-03-12 op 		p->len = 0;
131 20107775 2021-03-19 op 		free(p->buf);
132 20107775 2021-03-19 op 		p->buf = NULL;
133 191b4ca2 2021-03-12 op 		return 1;
134 191b4ca2 2021-03-12 op 	}
135 191b4ca2 2021-03-12 op 
136 be2ee49f 2021-05-14 op 	/*
137 be2ee49f 2021-05-14 op 	 * p->buf and buf can (and probably almost always will)
138 be2ee49f 2021-05-14 op 	 * overlap!
139 be2ee49f 2021-05-14 op 	 */
140 20107775 2021-03-19 op 
141 20107775 2021-03-19 op 	if ((tmp = calloc(1, len)) == NULL)
142 191b4ca2 2021-03-12 op 		return 0;
143 20107775 2021-03-19 op 	memcpy(tmp, buf, len);
144 20107775 2021-03-19 op 	free(p->buf);
145 20107775 2021-03-19 op 	p->buf = tmp;
146 191b4ca2 2021-03-12 op 	p->len = len;
147 191b4ca2 2021-03-12 op 	return 1;
148 191b4ca2 2021-03-12 op }
149 a5845bb5 2021-03-20 op 
150 a5845bb5 2021-03-20 op int
151 a5845bb5 2021-03-20 op parser_foreach_line(struct parser *p, const char *buf, size_t size,
152 a5845bb5 2021-03-20 op     parsechunkfn fn)
153 a5845bb5 2021-03-20 op {
154 31aa9f59 2021-03-20 op 	char		*b, *e;
155 31aa9f59 2021-03-20 op 	unsigned int	 ch;
156 31aa9f59 2021-03-20 op 	size_t		 i, l, len;
157 a5845bb5 2021-03-20 op 
158 31aa9f59 2021-03-20 op 	if (!parser_append(p, buf, size))
159 31aa9f59 2021-03-20 op 		return 0;
160 31aa9f59 2021-03-20 op 	b = p->buf;
161 31aa9f59 2021-03-20 op 	len = p->len;
162 31aa9f59 2021-03-20 op 
163 27dce34f 2021-07-06 op 	if (!(p->flags & PARSER_IN_BODY) && len < 3)
164 c0b634dd 2021-07-06 op 		return 1;
165 c0b634dd 2021-07-06 op 
166 27dce34f 2021-07-06 op 	if (!(p->flags & PARSER_IN_BODY)) {
167 95a8c791 2021-08-26 op 		p->flags |= PARSER_IN_BODY;
168 c0b634dd 2021-07-06 op 
169 c0b634dd 2021-07-06 op 		/*
170 c0b634dd 2021-07-06 op 		 * drop the BOM: only UTF-8 is supported, and there
171 c0b634dd 2021-07-06 op 		 * it's useless; some editors may still add one
172 c0b634dd 2021-07-06 op 		 * though.
173 c0b634dd 2021-07-06 op 		 */
174 c0b634dd 2021-07-06 op 		if (memmem(b, len, "\xEF\xBB\xBF", 3) == b) {
175 c0b634dd 2021-07-06 op 			b += 3;
176 c0b634dd 2021-07-06 op 			len -= 3;
177 c0b634dd 2021-07-06 op 		}
178 c0b634dd 2021-07-06 op 	}
179 c0b634dd 2021-07-06 op 
180 31aa9f59 2021-03-20 op 	/* drop every "funny" ASCII character */
181 31aa9f59 2021-03-20 op 	for (i = 0; i < len; ) {
182 31aa9f59 2021-03-20 op 		ch = b[i];
183 31aa9f59 2021-03-20 op 		if ((ch >= ' ' || ch == '\n' || ch == '\t')
184 31aa9f59 2021-03-20 op 		    && ch != 127) { /* del */
185 31aa9f59 2021-03-20 op 			++i;
186 31aa9f59 2021-03-20 op 			continue;
187 31aa9f59 2021-03-20 op 		}
188 08b9feb5 2021-07-26 op 		memmove(&b[i], &b[i+1], len - i - 1);
189 31aa9f59 2021-03-20 op 		len--;
190 a5845bb5 2021-03-20 op 	}
191 a5845bb5 2021-03-20 op 
192 a5845bb5 2021-03-20 op 	while (len > 0) {
193 a5845bb5 2021-03-20 op 		if ((e = memmem((char*)b, len, "\n", 1)) == NULL)
194 a5845bb5 2021-03-20 op 			break;
195 a5845bb5 2021-03-20 op 		l = e - b;
196 a5845bb5 2021-03-20 op 
197 a5845bb5 2021-03-20 op 		if (!fn(p, b, l))
198 a5845bb5 2021-03-20 op 			return 0;
199 a5845bb5 2021-03-20 op 
200 a5845bb5 2021-03-20 op 		len -= l;
201 a5845bb5 2021-03-20 op 		b += l;
202 a5845bb5 2021-03-20 op 
203 a5845bb5 2021-03-20 op 		if (len > 0) {
204 a5845bb5 2021-03-20 op 			/* skip \n */
205 a5845bb5 2021-03-20 op 			len--;
206 a5845bb5 2021-03-20 op 			b++;
207 a5845bb5 2021-03-20 op 		}
208 a5845bb5 2021-03-20 op 	}
209 a5845bb5 2021-03-20 op 
210 a5845bb5 2021-03-20 op 	return parser_set_buf(p, b, len);
211 a5845bb5 2021-03-20 op }