*
* Copyright (C) 2008 by Vladimir Dronnikov <dronnikov@gmail.com>
*
- * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
+ * Licensed under GPLv2 or later, see file LICENSE in this source tree.
+ * Also for use in uClibc (http://uclibc.org/) licensed under LGPLv2.1 or later.
*/
+/* Uncomment to enable test applet */
+////config:config PARSE
+////config: bool "Uniform config file parser debugging applet: parse"
+////config: default n
+////config: help
+////config: Typical usage of parse API:
+////config: char *t[3];
+////config: parser_t *p = config_open(filename);
+////config: while (config_read(p, t, 3, 0, delimiters, flags)) { // 1..3 tokens
+////config: bb_error_msg("TOKENS: '%s''%s''%s'", t[0], t[1], t[2]);
+////config: }
+////config: config_close(p);
+
+////applet:IF_PARSE(APPLET(parse, BB_DIR_USR_BIN, BB_SUID_DROP))
+
+//kbuild:lib-y += parse_config.o
+
+//usage:#define parse_trivial_usage
+//usage: "[-x] [-n MAXTOKENS] [-m MINTOKENS] [-d DELIMS] [-f FLAGS] FILE..."
+//usage:#define parse_full_usage "\n\n"
+//usage: " -x Suppress output (for benchmarking)"
+
#include "libbb.h"
-#if ENABLE_PARSE
+#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 = 0;
+ char **t;
+ unsigned flags = PARSE_NORMAL;
int mintokens = 0, ntokens = 128;
+ unsigned noout;
+
opt_complementary = "-1:n+:m+:f+";
- getopt32(argv, "n:m:d:f:", &ntokens, &mintokens, &delims, &flags);
+ noout = 1 & getopt32(argv, "xn:m:d:f:", &ntokens, &mintokens, &delims, &flags);
//argc -= optind;
argv += optind;
+
+ t = xmalloc(sizeof(t[0]) * ntokens);
while (*argv) {
+ int n;
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) {
+ while ((n = config_read(p, t, ntokens, mintokens, delims, flags)) != 0) {
+ if (!noout) {
for (int i = 0; i < n; ++i)
printf("[%s]", t[i]);
puts("");
}
- config_close(p);
}
+ 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)
{
- parser_t *parser = xzalloc(sizeof(parser_t));
- /* empty file configures nothing */
- parser->fp = fopen_or_warn_stdin(filename);
- if (parser->fp)
- return parser;
- if (ENABLE_FEATURE_CLEAN_UP)
- free(parser);
- return NULL;
+ return config_open2(filename, fopen_or_warn_stdin);
}
-static void config_free_data(parser_t *const parser)
+void FAST_FUNC config_close(parser_t *parser)
{
- free(parser->line);
- parser->line = NULL;
- USE_FEATURE_PARSE_COPY(
- free(parser->data);
- parser->data = NULL;
- )
+ if (parser) {
+ if (PARSE_KEEP_COPY) /* compile-time constant */
+ free(parser->data);
+ fclose(parser->fp);
+ free(parser->line);
+ free(parser->nline);
+ free(parser);
+ }
}
-void FAST_FUNC config_close(parser_t *parser)
+/* This function reads an entire line from a text file,
+ * up to a newline, exclusive.
+ * Trailing '\' is recognized as line continuation.
+ * Returns -1 if EOF/error.
+ */
+static int get_line_with_continuation(parser_t *parser)
{
- config_free_data(parser);
- fclose(parser->fp);
+ ssize_t len, nlen;
+ char *line;
+
+ len = getline(&parser->line, &parser->line_alloc, parser->fp);
+ if (len <= 0)
+ return len;
+
+ line = parser->line;
+ for (;;) {
+ parser->lineno++;
+ if (line[len - 1] == '\n')
+ len--;
+ if (len == 0 || line[len - 1] != '\\')
+ break;
+ len--;
+
+ nlen = getline(&parser->nline, &parser->nline_alloc, parser->fp);
+ if (nlen <= 0)
+ break;
+
+ if (parser->line_alloc < len + nlen + 1) {
+ parser->line_alloc = len + nlen + 1;
+ line = parser->line = xrealloc(line, parser->line_alloc);
+ }
+ memcpy(&line[len], parser->nline, nlen);
+ len += nlen;
+ }
+
+ line[len] = '\0';
+ return len;
}
+
/*
-1. Read a line from config file. If nothing to read then bail out returning 0.
- Handle continuation character. Advance lineno for each physical line. Cut comments.
-2. if PARSE_DONT_TRIM is not set (default) skip leading and cut trailing delimiters, if any.
+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_DONT_REDUCE or PARSE_DONT_TRIM is set then pin empty token.
-5. Else (default) if number of seen tokens is equal to max number of tokens (token is the last one)
- and PARSE_LAST_IS_GREEDY is set then pin the remainder of the line as the last token.
- Else (token is not last or PARSE_LAST_IS_GREEDY is not set) just replace first delimiter with '\0'
- thus delimiting token and pin it.
-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. Control the number of seen tokens is not less the min number of tokens. Die if condition is not met.
+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() exit with error message if less than mintokens
+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, *q;
- char comment = *delims++;
- int ii;
- int ntokens = flags & 0xFF;
- int mintokens = (flags & 0xFF00) >> 8;
-
- /*
- // N.B. this could only be used in read-in-one-go version, or when tokens use xstrdup(). TODO
- if (!parser->lineno || !(flags & PARSE_DONT_NULL))
- */
- memset(tokens, 0, sizeof(tokens[0]) * ntokens);
- config_free_data(parser);
-
- while (1) {
-//TODO: speed up xmalloc_fgetline by internally using fgets, not fgetc
- line = xmalloc_fgetline(parser->fp);
- if (!line)
- return 0;
+ char *line;
+ int ntokens, mintokens;
+ int t;
- parser->lineno++;
- // handle continuations. Tito's code stolen :)
- while (1) {
- ii = strlen(line);
- if (!ii)
- goto next_line;
- if (line[ii - 1] != '\\')
- break;
- // multi-line object
- line[--ii] = '\0';
-//TODO: add xmalloc_fgetline-like iface but with appending to existing str
- q = xmalloc_fgetline(parser->fp);
- if (q) {
- parser->lineno++;
- line = xasprintf("%s%s", line, q);
- free(q);
- }
- }
- // comments mean EOLs
- if (comment) {
- q = strchrnul(line, comment);
- *q = '\0';
- ii = q - line;
- }
- // skip leading and trailing delimiters
- if (!(flags & PARSE_DONT_TRIM)) {
- // skip leading
- int n = strspn(line, delims);
- if (n) {
- ii -= n;
- strcpy(line, line + n);
- }
- // cut trailing
- if (ii) {
- while (strchr(delims, line[--ii]))
- continue;
- line[++ii] = '\0';
- }
- }
- // if something still remains -> return it
- if (ii)
- break;
+ if (!parser)
+ return 0;
- next_line:
- /* skip empty line */
- free(line);
- }
+ ntokens = (uint8_t)flags;
+ mintokens = (uint8_t)(flags >> 8);
- // non-empty line found, parse and return the number of tokens
+ again:
+ memset(tokens, 0, sizeof(tokens[0]) * ntokens);
- // store line
- parser->line = line = xrealloc(line, ii + 1);
- USE_FEATURE_PARSE_COPY(
+ /* Read one line (handling continuations with backslash) */
+ if (get_line_with_continuation(parser) < 0)
+ return 0;
+
+ line = parser->line;
+
+ /* 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) {
+ free(parser->data);
parser->data = xstrdup(line);
- )
-
- /* now split line to tokens */
- ntokens--; // now it's max allowed token no
- // N.B, non-empty remainder is also a token,
- // so if ntokens <= 1, we just return the whole line
- // N.B. if PARSE_LAST_IS_GREEDY is set the remainder of the line is stuck to the last token
- for (ii = 0; *line && ii <= ntokens; ) {
- //bb_info_msg("L[%s]", line);
- // get next token
- // at the last token and need greedy token ->
- if ((flags & PARSE_LAST_IS_GREEDY) && (ii == ntokens)) {
- // ... don't cut the line
- q = line + strlen(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 {
- // vanilla token. cut the line at the first delim
- q = line + strcspn(line, delims);
- *q++ = '\0';
+ /* Combining, find comment char if any */
+ line = strchrnul(line, PARSE_EOL_COMMENTS ? delims[0] : '\0');
+
+ /* Trim any extra delimiters from the end */
+ if (flags & PARSE_TRIM) {
+ while (strchr(delims + 1, line[-1]) != NULL)
+ line--;
+ }
}
- // pin token
- if ((flags & (PARSE_DONT_REDUCE|PARSE_DONT_TRIM)) || *line) {
- //bb_info_msg("N[%d] T[%s]", ii, line);
- tokens[ii++] = 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) {
+ strcpy_and_process_escape_sequences(tokens[t], tokens[t]);
}
- line = q;
+#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();
+ goto again;
}
- if (ii < mintokens)
- bb_error_msg_and_die("bad line %u: %d tokens found, %d needed",
- parser->lineno, ii, mintokens);
-
- return ii;
+ return t;
}