busybox/libbb/parse_config.c

/* vi: set sw=4 ts=4: */
/*
 * config file parser helper
 *
 * Copyright (C) 2008 by Vladimir Dronnikov <dronnikov@gmail.com>
 *
 * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
 * Also for use in uClibc (http://uclibc.org/) licensed under LGPLv2.1 or later.
 */

#include "libbb.h"

#if defined ENABLE_PARSE && ENABLE_PARSE
int parse_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
int parse_main(int argc UNUSED_PARAM, char **argv)
{
	const char *delims = "# \t";
	unsigned flags = PARSE_NORMAL;
	int mintokens = 0, ntokens = 128;

	opt_complementary = "-1:n+:m+:f+";
	getopt32(argv, "n:m:d:f:", &ntokens, &mintokens, &delims, &flags);
	//argc -= optind;
	argv += optind;
	while (*argv) {
		parser_t *p = config_open(*argv);
		if (p) {
			int n;
			char **t = xmalloc(sizeof(char *) * ntokens);
			while ((n = config_read(p, t, ntokens, mintokens, delims, flags)) != 0) {
				for (int i = 0; i < n; ++i)
					printf("[%s]", t[i]);
				puts("");
			}
			config_close(p);
		}
		argv++;
	}
	return EXIT_SUCCESS;
}
#endif

/*

Typical usage:

----- CUT -----
	char *t[3];	// tokens placeholder
	parser_t *p = config_open(filename);
	if (p) {
		// parse line-by-line
		while (config_read(p, t, 3, 0, delimiters, flags)) { // 1..3 tokens
			// use tokens
			bb_error_msg("TOKENS: [%s][%s][%s]", t[0], t[1], t[2]);
		}
		...
		// free parser
		config_close(p);
	}
----- CUT -----

*/

parser_t* FAST_FUNC config_open2(const char *filename, FILE* FAST_FUNC (*fopen_func)(const char *path))
{
	FILE* fp;
	parser_t *parser;

	fp = fopen_func(filename);
	if (!fp)
		return NULL;
	parser = xzalloc(sizeof(*parser));
	parser->fp = fp;
	return parser;
}

parser_t* FAST_FUNC config_open(const char *filename)
{
	return config_open2(filename, fopen_or_warn_stdin);
}

static void config_free_data(parser_t *parser)
{
	free(parser->line);
	parser->line = NULL;
	if (PARSE_KEEP_COPY) { /* compile-time constant */
		free(parser->data);
		parser->data = NULL;
	}
}

void FAST_FUNC config_close(parser_t *parser)
{
	if (parser) {
		config_free_data(parser);
		fclose(parser->fp);
		free(parser);
	}
}

/*
0. If parser is NULL return 0.
1. Read a line from config file. If nothing to read then return 0.
   Handle continuation character. Advance lineno for each physical line.
   Discard everything past comment character.
2. if PARSE_TRIM is set (default), remove leading and trailing delimiters.
3. If resulting line is empty goto 1.
4. Look for first delimiter. If !PARSE_COLLAPSE or !PARSE_TRIM is set then
   remember the token as empty.
5. Else (default) if number of seen tokens is equal to max number of tokens
   (token is the last one) and PARSE_GREEDY is set then the remainder
   of the line is the last token.
   Else (token is not last or PARSE_GREEDY is not set) just replace
   first delimiter with '\0' thus delimiting the token.
6. Advance line pointer past the end of token. If number of seen tokens
   is less than required number of tokens then goto 4.
7. Check the number of seen tokens is not less the min number of tokens.
   Complain or die otherwise depending on PARSE_MIN_DIE.
8. Return the number of seen tokens.

mintokens > 0 make config_read() print error message if less than mintokens
(but more than 0) are found. Empty lines are always skipped (not warned about).
*/
#undef config_read
int FAST_FUNC config_read(parser_t *parser, char **tokens, unsigned flags, const char *delims)
{
	char *line;
	int ntokens, mintokens;
	int t, len;

	ntokens = (uint8_t)flags;
	mintokens = (uint8_t)(flags >> 8);

	if (parser == NULL)
		return 0;

again:
	memset(tokens, 0, sizeof(tokens[0]) * ntokens);
	config_free_data(parser);

	/* Read one line (handling continuations with backslash) */
	line = bb_get_chunk_with_continuation(parser->fp, &len, &parser->lineno);
	if (line == NULL)
		return 0;
	parser->line = line;

	/* Strip trailing line-feed if any */
	if (len && line[len-1] == '\n')
		line[len-1] = '\0';

	/* Skip token in the start of line? */
	if (flags & PARSE_TRIM)
		line += strspn(line, delims + 1);

	if (line[0] == '\0' || line[0] == delims[0])
		goto again;

	if (flags & PARSE_KEEP_COPY)
		parser->data = xstrdup(line);

	/* Tokenize the line */
	t = 0;
	do {
		/* Pin token */
		tokens[t] = line;

		/* Combine remaining arguments? */
		if ((t != (ntokens-1)) || !(flags & PARSE_GREEDY)) {
			/* Vanilla token, find next delimiter */
			line += strcspn(line, delims[0] ? delims : delims + 1);
		} else {
			/* Combining, find comment char if any */
			line = strchrnul(line, delims[0]);

			/* Trim any extra delimiters from the end */
			if (flags & PARSE_TRIM) {
				while (strchr(delims + 1, line[-1]) != NULL)
					line--;
			}
		}

		/* Token not terminated? */
		if (*line == delims[0])
			*line = '\0';
		else if (*line != '\0')
			*line++ = '\0';

#if 0 /* unused so far */
		if (flags & PARSE_ESCAPE) {
			const char *from;
			char *to;

			from = to = tokens[t];
			while (*from) {
				if (*from == '\\') {
					from++;
					*to++ = bb_process_escape_sequence(&from);
				} else {
					*to++ = *from++;
				}
			}
			*to = '\0';
		}
#endif
		/* Skip possible delimiters */
		if (flags & PARSE_COLLAPSE)
			line += strspn(line, delims + 1);

		t++;
	} while (*line && *line != delims[0] && t < ntokens);

	if (t < mintokens) {
		bb_error_msg("bad line %u: %d tokens found, %d needed",
				parser->lineno, t, mintokens);
		if (flags & PARSE_MIN_DIE)
			xfunc_die();
		if (flags & PARSE_KEEP_COPY)
			free(parser->data);
		goto again;
	}

	return t;
}
1	/* vi: set sw=4 ts=4: */
2	/*
3	* config file parser helper
4	*
5	* Copyright (C) 2008 by Vladimir Dronnikov <dronnikov@gmail.com>
6	*
7	* Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
8	* Also for use in uClibc (http://uclibc.org/) licensed under LGPLv2.1 or later.
9	*/
10
11	#include "libbb.h"
12
13	#if defined ENABLE_PARSE && ENABLE_PARSE
14	int parse_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
15	int parse_main(int argc UNUSED_PARAM, char **argv)
16	{
17	const char *delims = "# \t";
18	unsigned flags = PARSE_NORMAL;
19	int mintokens = 0, ntokens = 128;
20
21	opt_complementary = "-1:n+:m+:f+";
22	getopt32(argv, "n:m:d:f:", &ntokens, &mintokens, &delims, &flags);
23	//argc -= optind;
24	argv += optind;
25	while (*argv) {
26	parser_t p = config_open(argv);
27	if (p) {
28	int n;
29	char *t = xmalloc(sizeof(char ) * ntokens);
30	while ((n = config_read(p, t, ntokens, mintokens, delims, flags)) != 0) {
31	for (int i = 0; i < n; ++i)
32	printf("[%s]", t[i]);
33	puts("");
34	}
35	config_close(p);
36	}
37	argv++;
38	}
39	return EXIT_SUCCESS;
40	}
41	#endif
42
43	/*
44
45	Typical usage:
46
47	----- CUT -----
48	char *t[3]; // tokens placeholder
49	parser_t *p = config_open(filename);
50	if (p) {
51	// parse line-by-line
52	while (config_read(p, t, 3, 0, delimiters, flags)) { // 1..3 tokens
53	// use tokens
54	bb_error_msg("TOKENS: [%s][%s][%s]", t[0], t[1], t[2]);
55	}
56	...
57	// free parser
58	config_close(p);
59	}
60	----- CUT -----
61
62	*/
63
64	parser_t* FAST_FUNC config_open2(const char filename, FILE FAST_FUNC (fopen_func)(const char path))
65	{
66	FILE* fp;
67	parser_t *parser;
68
69	fp = fopen_func(filename);
70	if (!fp)
71	return NULL;
72	parser = xzalloc(sizeof(*parser));
73	parser->fp = fp;
74	return parser;
75	}
76
77	parser_t* FAST_FUNC config_open(const char *filename)
78	{
79	return config_open2(filename, fopen_or_warn_stdin);
80	}
81
82	static void config_free_data(parser_t *parser)
83	{
84	free(parser->line);
85	parser->line = NULL;
86	if (PARSE_KEEP_COPY) { /* compile-time constant */
87	free(parser->data);
88	parser->data = NULL;
89	}
90	}
91
92	void FAST_FUNC config_close(parser_t *parser)
93	{
94	if (parser) {
95	config_free_data(parser);
96	fclose(parser->fp);
97	free(parser);
98	}
99	}
100
101	/*
102	0. If parser is NULL return 0.
103	1. Read a line from config file. If nothing to read then return 0.
104	Handle continuation character. Advance lineno for each physical line.
105	Discard everything past comment character.
106	2. if PARSE_TRIM is set (default), remove leading and trailing delimiters.
107	3. If resulting line is empty goto 1.
108	4. Look for first delimiter. If !PARSE_COLLAPSE or !PARSE_TRIM is set then
109	remember the token as empty.
110	5. Else (default) if number of seen tokens is equal to max number of tokens
111	(token is the last one) and PARSE_GREEDY is set then the remainder
112	of the line is the last token.
113	Else (token is not last or PARSE_GREEDY is not set) just replace
114	first delimiter with '\0' thus delimiting the token.
115	6. Advance line pointer past the end of token. If number of seen tokens
116	is less than required number of tokens then goto 4.
117	7. Check the number of seen tokens is not less the min number of tokens.
118	Complain or die otherwise depending on PARSE_MIN_DIE.
119	8. Return the number of seen tokens.
120
121	mintokens > 0 make config_read() print error message if less than mintokens
122	(but more than 0) are found. Empty lines are always skipped (not warned about).
123	*/
124	#undef config_read
125	int FAST_FUNC config_read(parser_t parser, char tokens, unsigned flags, const char delims)
126	{
127	char *line;
128	int ntokens, mintokens;
129	int t, len;
130
131	ntokens = (uint8_t)flags;
132	mintokens = (uint8_t)(flags >> 8);
133
134	if (parser == NULL)
135	return 0;
136
137	again:
138	memset(tokens, 0, sizeof(tokens[0]) * ntokens);
139	config_free_data(parser);
140
141	/* Read one line (handling continuations with backslash) */
142	line = bb_get_chunk_with_continuation(parser->fp, &len, &parser->lineno);
143	if (line == NULL)
144	return 0;
145	parser->line = line;
146
147	/* Strip trailing line-feed if any */
148	if (len && line[len-1] == '\n')
149	line[len-1] = '\0';
150
151	/* Skip token in the start of line? */
152	if (flags & PARSE_TRIM)
153	line += strspn(line, delims + 1);
154
155	if (line[0] == '\0' \|\| line[0] == delims[0])
156	goto again;
157
158	if (flags & PARSE_KEEP_COPY)
159	parser->data = xstrdup(line);
160
161	/* Tokenize the line */
162	t = 0;
163	do {
164	/* Pin token */
165	tokens[t] = line;
166
167	/* Combine remaining arguments? */
168	if ((t != (ntokens-1)) \|\| !(flags & PARSE_GREEDY)) {
169	/* Vanilla token, find next delimiter */
170	line += strcspn(line, delims[0] ? delims : delims + 1);
171	} else {
172	/* Combining, find comment char if any */
173	line = strchrnul(line, delims[0]);
174
175	/* Trim any extra delimiters from the end */
176	if (flags & PARSE_TRIM) {
177	while (strchr(delims + 1, line[-1]) != NULL)
178	line--;
179	}
180	}
181
182	/* Token not terminated? */
183	if (*line == delims[0])
184	*line = '\0';
185	else if (*line != '\0')
186	*line++ = '\0';
187
188	#if 0 /* unused so far */
189	if (flags & PARSE_ESCAPE) {
190	const char *from;
191	char *to;
192
193	from = to = tokens[t];
194	while (*from) {
195	if (*from == '\\') {
196	from++;
197	*to++ = bb_process_escape_sequence(&from);
198	} else {
199	to++ = from++;
200	}
201	}
202	*to = '\0';
203	}
204	#endif
205	/* Skip possible delimiters */
206	if (flags & PARSE_COLLAPSE)
207	line += strspn(line, delims + 1);
208
209	t++;
210	} while (line && line != delims[0] && t < ntokens);
211
212	if (t < mintokens) {
213	bb_error_msg("bad line %u: %d tokens found, %d needed",
214	parser->lineno, t, mintokens);
215	if (flags & PARSE_MIN_DIE)
216	xfunc_die();
217	if (flags & PARSE_KEEP_COPY)
218	free(parser->data);
219	goto again;
220	}
221
222	return t;
223	}