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

/* SPDX-License-Identifier: LGPL-2.1-or-later */

#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 sz = 0;
        char quote = 0;                 /* 0 or ' or " */
        bool backslash = false;         /* whether we've just seen a backslash */
        char c;
        int r;

        assert(p);
        assert(ret);
        assert(!FLAGS_SET(flags, EXTRACT_KEEP_QUOTE | EXTRACT_UNQUOTE));

        /* 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, 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) {
                                if (!(flags & EXTRACT_RETAIN_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, sz+1))
                                return -ENOMEM;
                        break;
                }
        }

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

                        if (c == 0) {
                                if ((flags & EXTRACT_UNESCAPE_RELAX) &&
                                    (quote == 0 || flags & EXTRACT_RELAX)) {
                                        /* If we find an unquoted trailing backslash and we're in
                                         * EXTRACT_UNESCAPE_RELAX mode, keep it verbatim in the
                                         * output.
                                         *
                                         * Unbalanced quotes will only be allowed in EXTRACT_RELAX
                                         * mode, EXTRACT_UNESCAPE_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_MAX, &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) || **p == '\\'))
                                        /* An escaped separator char or the escape char itself */
                                        s[sz++] = c;
                                else if (flags & EXTRACT_UNESCAPE_RELAX) {
                                        s[sz++] = '\\';
                                        s[sz++] = c;
                                } else
                                        return -EINVAL;
                        } else
                                s[sz++] = c;

                        backslash = false;

                } else if (quote != 0) {     /* 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;
                                        if (flags & EXTRACT_UNQUOTE)
                                                break;
                                } else if (c == '\\' && !(flags & EXTRACT_RETAIN_ESCAPE)) {
                                        backslash = true;
                                        break;
                                }

                                if (!GREEDY_REALLOC(s, sz+2))
                                        return -ENOMEM;

                                s[sz++] = c;

                                if (quote == 0)
                                        break;
                        }

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

                                }

                                if (!GREEDY_REALLOC(s, sz+2))
                                        return -ENOMEM;

                                s[sz++] = c;

                                if (quote != 0)
                                        break;
                        }
                }
        }

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_UNESCAPE_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_UNESCAPE_RELAX)) {

                /* Retry it with EXTRACT_UNESCAPE_RELAX. */
                *p = save;
                r = extract_first_word(p, ret, separators, flags|EXTRACT_UNESCAPE_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
db9ecf05	1	/* SPDX-License-Identifier: LGPL-2.1-or-later */
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;
319a4f4b	22	size_t sz = 0;
84ac7bea LP	23	char quote = 0; /* 0 or ' or " */
84ac7bea LP	24	bool backslash = false; /* whether we've just seen a backslash */
fa67d9c0 LP	25	char c;
fa67d9c0 LP	26	int r;
84ac7bea LP	27
	28	assert(p);
	29	assert(ret);
1104d114	30	assert(!FLAGS_SET(flags, EXTRACT_KEEP_QUOTE \| EXTRACT_UNQUOTE));
84ac7bea	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	45	if (flags & EXTRACT_DONT_COALESCE_SEPARATORS)
319a4f4b	46	if (!GREEDY_REALLOC(s, sz+1))
b85e1c25	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) {
70b6ee61 LB	54	if (!(flags & EXTRACT_RETAIN_SEPARATORS))
70b6ee61 LB	55	(*p)++;
b85e1c25	56	goto finish_force_next;
93de9eb7	57	}
b85e1c25	58	} else {
84ac7bea LP	59	/* We found a non-blank character, so we will always
	60	* want to return a string (even if it is empty),
	61	* allocate it here. */
319a4f4b	62	if (!GREEDY_REALLOC(s, sz+1))
84ac7bea	63	return -ENOMEM;
b85e1c25	64	break;
84ac7bea	65	}
b85e1c25 FB	66	}
b85e1c25 FB	67
313cefa1	68	for (;; (p)++, c = *p) {
84ac7bea	69	if (backslash) {
319a4f4b	70	if (!GREEDY_REALLOC(s, sz+7))
84ac7bea LP	71	return -ENOMEM;
	72
	73	if (c == 0) {
3141089f	74	if ((flags & EXTRACT_UNESCAPE_RELAX) &&
fa67d9c0	75	(quote == 0 \|\| flags & EXTRACT_RELAX)) {
84ac7bea	76	/* If we find an unquoted trailing backslash and we're in
3141089f	77	* EXTRACT_UNESCAPE_RELAX mode, keep it verbatim in the
84ac7bea LP	78	* output.
	79	*
	80	* Unbalanced quotes will only be allowed in EXTRACT_RELAX
3141089f	81	* mode, EXTRACT_UNESCAPE_RELAX mode does not allow them.
84ac7bea LP	82	*/
	83	s[sz++] = '\\';
	84	goto finish_force_terminate;
	85	}
	86	if (flags & EXTRACT_RELAX)
	87	goto finish_force_terminate;
	88	return -EINVAL;
	89	}
	90
1e198efc	91	if (flags & (EXTRACT_CUNESCAPE\|EXTRACT_UNESCAPE_SEPARATORS)) {
3565e095	92	bool eight_bit = false;
c932fb71	93	char32_t u;
84ac7bea	94
1e198efc	95	if ((flags & EXTRACT_CUNESCAPE) &&
f5fbe71d	96	(r = cunescape_one(*p, SIZE_MAX, &u, &eight_bit, false)) >= 0) {
1e198efc ZJS	97	/* A valid escaped sequence */
	98	assert(r >= 1);
	99
3ff13c29 FB	100	(*p) += r - 1;
3ff13c29 FB	101
3565e095 ZJS	102	if (eight_bit)
3565e095 ZJS	103	s[sz++] = u;
3ff13c29	104	else
3565e095	105	sz += utf8_encode_unichar(s + sz, u);
1e198efc	106	} else if ((flags & EXTRACT_UNESCAPE_SEPARATORS) &&
76c4e48e ZJS	107	(strchr(separators, p) \|\| p == '\\'))
76c4e48e ZJS	108	/* An escaped separator char or the escape char itself */
1e198efc	109	s[sz++] = c;
3141089f	110	else if (flags & EXTRACT_UNESCAPE_RELAX) {
1e198efc ZJS	111	s[sz++] = '\\';
	112	s[sz++] = c;
	113	} else
	114	return -EINVAL;
84ac7bea LP	115	} else
	116	s[sz++] = c;
	117
84ac7bea LP	118	backslash = false;
84ac7bea LP	119
fa67d9c0	120	} else if (quote != 0) { /* inside either single or double quotes */
313cefa1	121	for (;; (p)++, c = *p) {
27fc921b FB	122	if (c == 0) {
	123	if (flags & EXTRACT_RELAX)
	124	goto finish_force_terminate;
	125	return -EINVAL;
	126	} else if (c == quote) { /* found the end quote */
	127	quote = 0;
1104d114 YW	128	if (flags & EXTRACT_UNQUOTE)
1104d114 YW	129	break;
c89f52ac	130	} else if (c == '\\' && !(flags & EXTRACT_RETAIN_ESCAPE)) {
27fc921b FB	131	backslash = true;
27fc921b FB	132	break;
27fc921b	133	}
1104d114 YW	134
	135	if (!GREEDY_REALLOC(s, sz+2))
	136	return -ENOMEM;
	137
	138	s[sz++] = c;
	139
	140	if (quote == 0)
	141	break;
84ac7bea LP	142	}
84ac7bea LP	143
84ac7bea	144	} else {
313cefa1	145	for (;; (p)++, c = *p) {
27fc921b FB	146	if (c == 0)
27fc921b FB	147	goto finish_force_terminate;
1104d114	148	else if (IN_SET(c, '\'', '"') && (flags & (EXTRACT_KEEP_QUOTE \| EXTRACT_UNQUOTE))) {
27fc921b	149	quote = c;
1104d114 YW	150	if (flags & EXTRACT_UNQUOTE)
1104d114 YW	151	break;
c89f52ac	152	} else if (c == '\\' && !(flags & EXTRACT_RETAIN_ESCAPE)) {
27fc921b FB	153	backslash = true;
	154	break;
	155	} else if (strchr(separators, c)) {
	156	if (flags & EXTRACT_DONT_COALESCE_SEPARATORS) {
70b6ee61 LB	157	if (!(flags & EXTRACT_RETAIN_SEPARATORS))
70b6ee61 LB	158	(*p)++;
27fc921b FB	159	goto finish_force_next;
27fc921b FB	160	}
70b6ee61 LB	161	if (!(flags & EXTRACT_RETAIN_SEPARATORS))
	162	/* Skip additional coalesced separators. */
	163	for (;; (p)++, c = *p) {
	164	if (c == 0)
	165	goto finish_force_terminate;
	166	if (!strchr(separators, c))
	167	break;
	168	}
0247447e FB	169	goto finish;
0247447e FB	170
27fc921b	171	}
1104d114 YW	172
	173	if (!GREEDY_REALLOC(s, sz+2))
	174	return -ENOMEM;
	175
	176	s[sz++] = c;
	177
	178	if (quote != 0)
	179	break;
84ac7bea LP	180	}
84ac7bea LP	181	}
84ac7bea LP	182	}
	183
	184	finish_force_terminate:
	185	*p = NULL;
	186	finish:
	187	if (!s) {
	188	*p = NULL;
	189	*ret = NULL;
	190	return 0;
	191	}
	192
	193	finish_force_next:
	194	s[sz] = 0;
ae2a15bc	195	*ret = TAKE_PTR(s);
84ac7bea LP	196
	197	return 1;
	198	}
	199
	200	int extract_first_word_and_warn(
	201	const char **p,
	202	char **ret,
	203	const char *separators,
	204	ExtractFlags flags,
	205	const char *unit,
	206	const char *filename,
	207	unsigned line,
	208	const char *rvalue) {
	209
dea7b6b0	210	/* Try to unquote it, if it fails, warn about it and try again
3141089f	211	* but this time using EXTRACT_UNESCAPE_RELAX to keep the
dea7b6b0 LP	212	* backslashes verbatim in invalid escape sequences. */
dea7b6b0 LP	213
84ac7bea LP	214	const char *save;
	215	int r;
	216
	217	save = *p;
	218	r = extract_first_word(p, ret, separators, flags);
dea7b6b0 LP	219	if (r >= 0)
	220	return r;
	221
3141089f	222	if (r == -EINVAL && !(flags & EXTRACT_UNESCAPE_RELAX)) {
84ac7bea	223
3141089f	224	/* Retry it with EXTRACT_UNESCAPE_RELAX. */
84ac7bea	225	*p = save;
3141089f	226	r = extract_first_word(p, ret, separators, flags\|EXTRACT_UNESCAPE_RELAX);
dea7b6b0	227	if (r >= 0) {
330785f5 ZJS	228	/* It worked this time, hence it must have been an invalid escape sequence. */
330785f5 ZJS	229	log_syntax(unit, LOG_WARNING, filename, line, EINVAL, "Ignoring unknown escape sequences: \"%s\"", *ret);
dea7b6b0 LP	230	return r;
	231	}
	232
	233	/* If it's still EINVAL; then it must be unbalanced quoting, report this. */
	234	if (r == -EINVAL)
	235	return log_syntax(unit, LOG_ERR, filename, line, r, "Unbalanced quoting, ignoring: \"%s\"", rvalue);
84ac7bea LP	236	}
84ac7bea LP	237
dea7b6b0 LP	238	/* Can be any error, report it */
dea7b6b0 LP	239	return log_syntax(unit, LOG_ERR, filename, line, r, "Unable to decode word \"%s\", ignoring: %m", rvalue);
84ac7bea LP	240	}
84ac7bea LP	241
e4d85dbb ZJS	242	/* We pass ExtractFlags as unsigned int (to avoid undefined behaviour when passing
	243	* an object that undergoes default argument promotion as an argument to va_start).
	244	* Let's make sure that ExtractFlags fits into an unsigned int. */
	245	assert_cc(sizeof(enum ExtractFlags) <= sizeof(unsigned));
	246
	247	int extract_many_words(const char *p, const char separators, unsigned flags, ...) {
84ac7bea LP	248	va_list ap;
	249	char **l;
	250	int n = 0, i, c, r;
	251
	252	/* Parses a number of words from a string, stripping any
	253	* quotes if necessary. */
	254
	255	assert(p);
	256
	257	/* Count how many words are expected */
	258	va_start(ap, flags);
	259	for (;;) {
	260	if (!va_arg(ap, char **))
	261	break;
	262	n++;
	263	}
	264	va_end(ap);
	265
	266	if (n <= 0)
	267	return 0;
	268
	269	/* Read all words into a temporary array */
	270	l = newa0(char*, n);
	271	for (c = 0; c < n; c++) {
	272
	273	r = extract_first_word(p, &l[c], separators, flags);
	274	if (r < 0) {
	275	int j;
	276
	277	for (j = 0; j < c; j++)
	278	free(l[j]);
	279
	280	return r;
	281	}
	282
	283	if (r == 0)
	284	break;
	285	}
	286
	287	/* If we managed to parse all words, return them in the passed
	288	* in parameters */
	289	va_start(ap, flags);
	290	for (i = 0; i < n; i++) {
	291	char **v;
	292
	293	v = va_arg(ap, char **);
	294	assert(v);
	295
	296	*v = l[i];
	297	}
	298	va_end(ap);
	299
	300	return c;
	301	}