[thirdparty/systemd.git] / src / basic / extract-word.c

/* SPDX-License-Identifier: LGPL-2.1+ */

#include <errno.h>
#include <stdarg.h>
#include <stdbool.h>
#include <stddef.h>
#include <stdint.h>
#include <stdlib.h>
#include <syslog.h>

#include "alloc-util.h"
#include "escape.h"
#include "extract-word.h"
#include "log.h"
#include "macro.h"
#include "string-util.h"
#include "strv.h"
#include "utf8.h"

int extract_first_word(const char **p, char **ret, const char *separators, ExtractFlags flags) {
        _cleanup_free_ char *s = NULL;
        size_t allocated = 0, sz = 0;
        char c;
        int r;

        char quote = 0;                 /* 0 or ' or " */
        bool backslash = false;         /* whether we've just seen a backslash */

        assert(p);
        assert(ret);

        /* Bail early if called after last value or with no input */
        if (!*p)
                goto finish;
        c = **p;

        if (!separators)
                separators = WHITESPACE;

        /* Parses the first word of a string, and returns it in
         * *ret. Removes all quotes in the process. When parsing fails
         * (because of an uneven number of quotes or similar), leaves
         * the pointer *p at the first invalid character. */

        if (flags & EXTRACT_DONT_COALESCE_SEPARATORS)
                if (!GREEDY_REALLOC(s, allocated, sz+1))
                        return -ENOMEM;

        for (;; (*p)++, c = **p) {
                if (c == 0)
                        goto finish_force_terminate;
                else if (strchr(separators, c)) {
                        if (flags & EXTRACT_DONT_COALESCE_SEPARATORS) {
                                (*p)++;
                                goto finish_force_next;
                        }
                } else {
                        /* We found a non-blank character, so we will always
                         * want to return a string (even if it is empty),
                         * allocate it here. */
                        if (!GREEDY_REALLOC(s, allocated, sz+1))
                                return -ENOMEM;
                        break;
                }
        }

        for (;; (*p)++, c = **p) {
                if (backslash) {
                        if (!GREEDY_REALLOC(s, allocated, sz+7))
                                return -ENOMEM;

                        if (c == 0) {
                                if ((flags & EXTRACT_CUNESCAPE_RELAX) &&
                                    (!quote || flags & EXTRACT_RELAX)) {
                                        /* If we find an unquoted trailing backslash and we're in
                                         * EXTRACT_CUNESCAPE_RELAX mode, keep it verbatim in the
                                         * output.
                                         *
                                         * Unbalanced quotes will only be allowed in EXTRACT_RELAX
                                         * mode, EXTRACT_CUNESCAPE_RELAX mode does not allow them.
                                         */
                                        s[sz++] = '\\';
                                        goto finish_force_terminate;
                                }
                                if (flags & EXTRACT_RELAX)
                                        goto finish_force_terminate;
                                return -EINVAL;
                        }

                        if (flags & (EXTRACT_CUNESCAPE|EXTRACT_UNESCAPE_SEPARATORS)) {
                                bool eight_bit = false;
                                char32_t u;

                                if ((flags & EXTRACT_CUNESCAPE) &&
                                    (r = cunescape_one(*p, (size_t) -1, &u, &eight_bit, false)) >= 0) {
                                        /* A valid escaped sequence */
                                        assert(r >= 1);

                                        (*p) += r - 1;

                                        if (eight_bit)
                                                s[sz++] = u;
                                        else
                                                sz += utf8_encode_unichar(s + sz, u);
                                } else if ((flags & EXTRACT_UNESCAPE_SEPARATORS) &&
                                           strchr(separators, **p))
                                        /* An escaped separator char */
                                        s[sz++] = c;
                                else if (flags & EXTRACT_CUNESCAPE_RELAX) {
                                        s[sz++] = '\\';
                                        s[sz++] = c;
                                } else
                                        return -EINVAL;
                        } else
                                s[sz++] = c;

                        backslash = false;

                } else if (quote) {     /* inside either single or double quotes */
                        for (;; (*p)++, c = **p) {
                                if (c == 0) {
                                        if (flags & EXTRACT_RELAX)
                                                goto finish_force_terminate;
                                        return -EINVAL;
                                } else if (c == quote) {        /* found the end quote */
                                        quote = 0;
                                        break;
                                } else if (c == '\\' && !(flags & EXTRACT_RETAIN_ESCAPE)) {
                                        backslash = true;
                                        break;
                                } else {
                                        if (!GREEDY_REALLOC(s, allocated, sz+2))
                                                return -ENOMEM;

                                        s[sz++] = c;
                                }
                        }

                } else {
                        for (;; (*p)++, c = **p) {
                                if (c == 0)
                                        goto finish_force_terminate;
                                else if (IN_SET(c, '\'', '"') && (flags & EXTRACT_UNQUOTE)) {
                                        quote = c;
                                        break;
                                } else if (c == '\\' && !(flags & EXTRACT_RETAIN_ESCAPE)) {
                                        backslash = true;
                                        break;
                                } else if (strchr(separators, c)) {
                                        if (flags & EXTRACT_DONT_COALESCE_SEPARATORS) {
                                                (*p)++;
                                                goto finish_force_next;
                                        }
                                        /* Skip additional coalesced separators. */
                                        for (;; (*p)++, c = **p) {
                                                if (c == 0)
                                                        goto finish_force_terminate;
                                                if (!strchr(separators, c))
                                                        break;
                                        }
                                        goto finish;

                                } else {
                                        if (!GREEDY_REALLOC(s, allocated, sz+2))
                                                return -ENOMEM;

                                        s[sz++] = c;
                                }
                        }
                }
        }

finish_force_terminate:
        *p = NULL;
finish:
        if (!s) {
                *p = NULL;
                *ret = NULL;
                return 0;
        }

finish_force_next:
        s[sz] = 0;
        *ret = TAKE_PTR(s);

        return 1;
}

int extract_first_word_and_warn(
                const char **p,
                char **ret,
                const char *separators,
                ExtractFlags flags,
                const char *unit,
                const char *filename,
                unsigned line,
                const char *rvalue) {

        /* Try to unquote it, if it fails, warn about it and try again
         * but this time using EXTRACT_CUNESCAPE_RELAX to keep the
         * backslashes verbatim in invalid escape sequences. */

        const char *save;
        int r;

        save = *p;
        r = extract_first_word(p, ret, separators, flags);
        if (r >= 0)
                return r;

        if (r == -EINVAL && !(flags & EXTRACT_CUNESCAPE_RELAX)) {

                /* Retry it with EXTRACT_CUNESCAPE_RELAX. */
                *p = save;
                r = extract_first_word(p, ret, separators, flags|EXTRACT_CUNESCAPE_RELAX);
                if (r >= 0) {
                        /* It worked this time, hence it must have been an invalid escape sequence. */
                        log_syntax(unit, LOG_WARNING, filename, line, EINVAL, "Ignoring unknown escape sequences: \"%s\"", *ret);
                        return r;
                }

                /* If it's still EINVAL; then it must be unbalanced quoting, report this. */
                if (r == -EINVAL)
                        return log_syntax(unit, LOG_ERR, filename, line, r, "Unbalanced quoting, ignoring: \"%s\"", rvalue);
        }

        /* Can be any error, report it */
        return log_syntax(unit, LOG_ERR, filename, line, r, "Unable to decode word \"%s\", ignoring: %m", rvalue);
}

/* We pass ExtractFlags as unsigned int (to avoid undefined behaviour when passing
 * an object that undergoes default argument promotion as an argument to va_start).
 * Let's make sure that ExtractFlags fits into an unsigned int. */
assert_cc(sizeof(enum ExtractFlags) <= sizeof(unsigned));

int extract_many_words(const char **p, const char *separators, unsigned flags, ...) {
        va_list ap;
        char **l;
        int n = 0, i, c, r;

        /* Parses a number of words from a string, stripping any
         * quotes if necessary. */

        assert(p);

        /* Count how many words are expected */
        va_start(ap, flags);
        for (;;) {
                if (!va_arg(ap, char **))
                        break;
                n++;
        }
        va_end(ap);

        if (n <= 0)
                return 0;

        /* Read all words into a temporary array */
        l = newa0(char*, n);
        for (c = 0; c < n; c++) {

                r = extract_first_word(p, &l[c], separators, flags);
                if (r < 0) {
                        int j;

                        for (j = 0; j < c; j++)
                                free(l[j]);

                        return r;
                }

                if (r == 0)
                        break;
        }

        /* If we managed to parse all words, return them in the passed
         * in parameters */
        va_start(ap, flags);
        for (i = 0; i < n; i++) {
                char **v;

                v = va_arg(ap, char **);
                assert(v);

                *v = l[i];
        }
        va_end(ap);

        return c;
}
Commit	Line	Data
53e1b683	1	/* SPDX-License-Identifier: LGPL-2.1+ */
84ac7bea	2
11c3a366 TA	3	#include <errno.h>
	4	#include <stdarg.h>
	5	#include <stdbool.h>
	6	#include <stddef.h>
	7	#include <stdint.h>
	8	#include <stdlib.h>
11c3a366 TA	9	#include <syslog.h>
11c3a366 TA	10
b5efdb8a	11	#include "alloc-util.h"
4f5dd394	12	#include "escape.h"
b11d6a7b	13	#include "extract-word.h"
11c3a366 TA	14	#include "log.h"
11c3a366 TA	15	#include "macro.h"
b11d6a7b	16	#include "string-util.h"
46bf625a	17	#include "strv.h"
84ac7bea	18	#include "utf8.h"
84ac7bea LP	19
	20	int extract_first_word(const char p, char ret, const char *separators, ExtractFlags flags) {
	21	_cleanup_free_ char *s = NULL;
	22	size_t allocated = 0, sz = 0;
b85e1c25	23	char c;
84ac7bea LP	24	int r;
	25
	26	char quote = 0; /* 0 or ' or " */
	27	bool backslash = false; /* whether we've just seen a backslash */
84ac7bea LP	28
	29	assert(p);
	30	assert(ret);
	31
84ac7bea LP	32	/* Bail early if called after last value or with no input */
84ac7bea LP	33	if (!*p)
c58bd76a	34	goto finish;
93de9eb7	35	c = **p;
84ac7bea	36
8372da44 FB	37	if (!separators)
	38	separators = WHITESPACE;
	39
84ac7bea LP	40	/* Parses the first word of a string, and returns it in
	41	* *ret. Removes all quotes in the process. When parsing fails
	42	* (because of an uneven number of quotes or similar), leaves
	43	* the pointer p at the first invalid character. /
	44
b85e1c25 FB	45	if (flags & EXTRACT_DONT_COALESCE_SEPARATORS)
	46	if (!GREEDY_REALLOC(s, allocated, sz+1))
	47	return -ENOMEM;
84ac7bea	48
313cefa1	49	for (;; (p)++, c = *p) {
b85e1c25 FB	50	if (c == 0)
	51	goto finish_force_terminate;
	52	else if (strchr(separators, c)) {
93de9eb7	53	if (flags & EXTRACT_DONT_COALESCE_SEPARATORS) {
313cefa1	54	(*p)++;
b85e1c25	55	goto finish_force_next;
93de9eb7	56	}
b85e1c25	57	} else {
84ac7bea LP	58	/* We found a non-blank character, so we will always
	59	* want to return a string (even if it is empty),
	60	* allocate it here. */
	61	if (!GREEDY_REALLOC(s, allocated, sz+1))
	62	return -ENOMEM;
b85e1c25	63	break;
84ac7bea	64	}
b85e1c25 FB	65	}
b85e1c25 FB	66
313cefa1	67	for (;; (p)++, c = *p) {
84ac7bea LP	68	if (backslash) {
	69	if (!GREEDY_REALLOC(s, allocated, sz+7))
	70	return -ENOMEM;
	71
	72	if (c == 0) {
	73	if ((flags & EXTRACT_CUNESCAPE_RELAX) &&
	74	(!quote \|\| flags & EXTRACT_RELAX)) {
	75	/* If we find an unquoted trailing backslash and we're in
	76	* EXTRACT_CUNESCAPE_RELAX mode, keep it verbatim in the
	77	* output.
	78	*
	79	* Unbalanced quotes will only be allowed in EXTRACT_RELAX
	80	* mode, EXTRACT_CUNESCAPE_RELAX mode does not allow them.
	81	*/
	82	s[sz++] = '\\';
	83	goto finish_force_terminate;
	84	}
	85	if (flags & EXTRACT_RELAX)
	86	goto finish_force_terminate;
	87	return -EINVAL;
	88	}
	89
1e198efc	90	if (flags & (EXTRACT_CUNESCAPE\|EXTRACT_UNESCAPE_SEPARATORS)) {
3565e095	91	bool eight_bit = false;
c932fb71	92	char32_t u;
84ac7bea	93
1e198efc ZJS	94	if ((flags & EXTRACT_CUNESCAPE) &&
	95	(r = cunescape_one(*p, (size_t) -1, &u, &eight_bit, false)) >= 0) {
	96	/* A valid escaped sequence */
	97	assert(r >= 1);
	98
3ff13c29 FB	99	(*p) += r - 1;
3ff13c29 FB	100
3565e095 ZJS	101	if (eight_bit)
3565e095 ZJS	102	s[sz++] = u;
3ff13c29	103	else
3565e095	104	sz += utf8_encode_unichar(s + sz, u);
1e198efc ZJS	105	} else if ((flags & EXTRACT_UNESCAPE_SEPARATORS) &&
	106	strchr(separators, **p))
	107	/* An escaped separator char */
	108	s[sz++] = c;
	109	else if (flags & EXTRACT_CUNESCAPE_RELAX) {
	110	s[sz++] = '\\';
	111	s[sz++] = c;
	112	} else
	113	return -EINVAL;
84ac7bea LP	114	} else
	115	s[sz++] = c;
	116
84ac7bea LP	117	backslash = false;
	118
	119	} else if (quote) { /* inside either single or double quotes */
313cefa1	120	for (;; (p)++, c = *p) {
27fc921b FB	121	if (c == 0) {
	122	if (flags & EXTRACT_RELAX)
	123	goto finish_force_terminate;
	124	return -EINVAL;
	125	} else if (c == quote) { /* found the end quote */
	126	quote = 0;
	127	break;
c89f52ac	128	} else if (c == '\\' && !(flags & EXTRACT_RETAIN_ESCAPE)) {
27fc921b FB	129	backslash = true;
	130	break;
	131	} else {
	132	if (!GREEDY_REALLOC(s, allocated, sz+2))
	133	return -ENOMEM;
84ac7bea	134
27fc921b FB	135	s[sz++] = c;
27fc921b FB	136	}
84ac7bea LP	137	}
84ac7bea LP	138
84ac7bea	139	} else {
313cefa1	140	for (;; (p)++, c = *p) {
27fc921b FB	141	if (c == 0)
27fc921b FB	142	goto finish_force_terminate;
4ec85141	143	else if (IN_SET(c, '\'', '"') && (flags & EXTRACT_UNQUOTE)) {
27fc921b FB	144	quote = c;
27fc921b FB	145	break;
c89f52ac	146	} else if (c == '\\' && !(flags & EXTRACT_RETAIN_ESCAPE)) {
27fc921b FB	147	backslash = true;
	148	break;
	149	} else if (strchr(separators, c)) {
	150	if (flags & EXTRACT_DONT_COALESCE_SEPARATORS) {
313cefa1	151	(*p)++;
27fc921b FB	152	goto finish_force_next;
27fc921b FB	153	}
0247447e	154	/* Skip additional coalesced separators. */
313cefa1	155	for (;; (p)++, c = *p) {
0247447e FB	156	if (c == 0)
	157	goto finish_force_terminate;
	158	if (!strchr(separators, c))
	159	break;
	160	}
	161	goto finish;
	162
27fc921b FB	163	} else {
	164	if (!GREEDY_REALLOC(s, allocated, sz+2))
	165	return -ENOMEM;
84ac7bea	166
27fc921b FB	167	s[sz++] = c;
27fc921b FB	168	}
84ac7bea LP	169	}
84ac7bea LP	170	}
84ac7bea LP	171	}
	172
	173	finish_force_terminate:
	174	*p = NULL;
	175	finish:
	176	if (!s) {
	177	*p = NULL;
	178	*ret = NULL;
	179	return 0;
	180	}
	181
	182	finish_force_next:
	183	s[sz] = 0;
ae2a15bc	184	*ret = TAKE_PTR(s);
84ac7bea LP	185
	186	return 1;
	187	}
	188
	189	int extract_first_word_and_warn(
	190	const char **p,
	191	char **ret,
	192	const char *separators,
	193	ExtractFlags flags,
	194	const char *unit,
	195	const char *filename,
	196	unsigned line,
	197	const char *rvalue) {
	198
dea7b6b0 LP	199	/* Try to unquote it, if it fails, warn about it and try again
	200	* but this time using EXTRACT_CUNESCAPE_RELAX to keep the
	201	* backslashes verbatim in invalid escape sequences. */
	202
84ac7bea LP	203	const char *save;
	204	int r;
	205
	206	save = *p;
	207	r = extract_first_word(p, ret, separators, flags);
dea7b6b0 LP	208	if (r >= 0)
	209	return r;
	210
	211	if (r == -EINVAL && !(flags & EXTRACT_CUNESCAPE_RELAX)) {
84ac7bea LP	212
	213	/* Retry it with EXTRACT_CUNESCAPE_RELAX. */
	214	*p = save;
	215	r = extract_first_word(p, ret, separators, flags\|EXTRACT_CUNESCAPE_RELAX);
dea7b6b0	216	if (r >= 0) {
330785f5 ZJS	217	/* It worked this time, hence it must have been an invalid escape sequence. */
330785f5 ZJS	218	log_syntax(unit, LOG_WARNING, filename, line, EINVAL, "Ignoring unknown escape sequences: \"%s\"", *ret);
dea7b6b0 LP	219	return r;
	220	}
	221
	222	/* If it's still EINVAL; then it must be unbalanced quoting, report this. */
	223	if (r == -EINVAL)
	224	return log_syntax(unit, LOG_ERR, filename, line, r, "Unbalanced quoting, ignoring: \"%s\"", rvalue);
84ac7bea LP	225	}
84ac7bea LP	226
dea7b6b0 LP	227	/* Can be any error, report it */
dea7b6b0 LP	228	return log_syntax(unit, LOG_ERR, filename, line, r, "Unable to decode word \"%s\", ignoring: %m", rvalue);
84ac7bea LP	229	}
84ac7bea LP	230
e4d85dbb ZJS	231	/* We pass ExtractFlags as unsigned int (to avoid undefined behaviour when passing
	232	* an object that undergoes default argument promotion as an argument to va_start).
	233	* Let's make sure that ExtractFlags fits into an unsigned int. */
	234	assert_cc(sizeof(enum ExtractFlags) <= sizeof(unsigned));
	235
	236	int extract_many_words(const char *p, const char separators, unsigned flags, ...) {
84ac7bea LP	237	va_list ap;
	238	char **l;
	239	int n = 0, i, c, r;
	240
	241	/* Parses a number of words from a string, stripping any
	242	* quotes if necessary. */
	243
	244	assert(p);
	245
	246	/* Count how many words are expected */
	247	va_start(ap, flags);
	248	for (;;) {
	249	if (!va_arg(ap, char **))
	250	break;
	251	n++;
	252	}
	253	va_end(ap);
	254
	255	if (n <= 0)
	256	return 0;
	257
	258	/* Read all words into a temporary array */
	259	l = newa0(char*, n);
	260	for (c = 0; c < n; c++) {
	261
	262	r = extract_first_word(p, &l[c], separators, flags);
	263	if (r < 0) {
	264	int j;
	265
	266	for (j = 0; j < c; j++)
	267	free(l[j]);
	268
	269	return r;
	270	}
	271
	272	if (r == 0)
	273	break;
	274	}
	275
	276	/* If we managed to parse all words, return them in the passed
	277	* in parameters */
	278	va_start(ap, flags);
	279	for (i = 0; i < n; i++) {
	280	char **v;
	281
	282	v = va_arg(ap, char **);
	283	assert(v);
	284
	285	*v = l[i];
	286	}
	287	va_end(ap);
	288
	289	return c;
	290	}