[thirdparty/systemd.git] / src / basic / escape.c

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

#include <stdlib.h>
#include <string.h>

#include "alloc-util.h"
#include "escape.h"
#include "hexdecoct.h"
#include "string-util.h"
#include "strv.h"
#include "utf8.h"

int cescape_char(char c, char *buf) {
        char *buf_old = buf;

        /* Needs space for 4 characters in the buffer */

        switch (c) {

                case '\a':
                        *(buf++) = '\\';
                        *(buf++) = 'a';
                        break;
                case '\b':
                        *(buf++) = '\\';
                        *(buf++) = 'b';
                        break;
                case '\f':
                        *(buf++) = '\\';
                        *(buf++) = 'f';
                        break;
                case '\n':
                        *(buf++) = '\\';
                        *(buf++) = 'n';
                        break;
                case '\r':
                        *(buf++) = '\\';
                        *(buf++) = 'r';
                        break;
                case '\t':
                        *(buf++) = '\\';
                        *(buf++) = 't';
                        break;
                case '\v':
                        *(buf++) = '\\';
                        *(buf++) = 'v';
                        break;
                case '\\':
                        *(buf++) = '\\';
                        *(buf++) = '\\';
                        break;
                case '"':
                        *(buf++) = '\\';
                        *(buf++) = '"';
                        break;
                case '\'':
                        *(buf++) = '\\';
                        *(buf++) = '\'';
                        break;

                default:
                        /* For special chars we prefer octal over
                         * hexadecimal encoding, simply because glib's
                         * g_strescape() does the same */
                        if ((c < ' ') || (c >= 127)) {
                                *(buf++) = '\\';
                                *(buf++) = octchar((unsigned char) c >> 6);
                                *(buf++) = octchar((unsigned char) c >> 3);
                                *(buf++) = octchar((unsigned char) c);
                        } else
                                *(buf++) = c;
                        break;
        }

        return buf - buf_old;
}

char* cescape_length(const char *s, size_t n) {
        const char *f;
        char *r, *t;

        /* Does C style string escaping. May be reversed with cunescape(). */

        assert(s || n == 0);

        if (n == SIZE_MAX)
                n = strlen(s);

        if (n > (SIZE_MAX - 1) / 4)
                return NULL;

        r = new(char, n*4 + 1);
        if (!r)
                return NULL;

        for (f = s, t = r; f < s + n; f++)
                t += cescape_char(*f, t);

        *t = 0;

        return r;
}

int cunescape_one(const char *p, size_t length, char32_t *ret, bool *eight_bit, bool accept_nul) {
        int r = 1;

        assert(p);
        assert(ret);

        /* Unescapes C style. Returns the unescaped character in ret.
         * Sets *eight_bit to true if the escaped sequence either fits in
         * one byte in UTF-8 or is a non-unicode literal byte and should
         * instead be copied directly.
         */

        if (length != SIZE_MAX && length < 1)
                return -EINVAL;

        switch (p[0]) {

        case 'a':
                *ret = '\a';
                break;
        case 'b':
                *ret = '\b';
                break;
        case 'f':
                *ret = '\f';
                break;
        case 'n':
                *ret = '\n';
                break;
        case 'r':
                *ret = '\r';
                break;
        case 't':
                *ret = '\t';
                break;
        case 'v':
                *ret = '\v';
                break;
        case '\\':
                *ret = '\\';
                break;
        case '"':
                *ret = '"';
                break;
        case '\'':
                *ret = '\'';
                break;

        case 's':
                /* This is an extension of the XDG syntax files */
                *ret = ' ';
                break;

        case 'x': {
                /* hexadecimal encoding */
                int a, b;

                if (length != SIZE_MAX && length < 3)
                        return -EINVAL;

                a = unhexchar(p[1]);
                if (a < 0)
                        return -EINVAL;

                b = unhexchar(p[2]);
                if (b < 0)
                        return -EINVAL;

                /* Don't allow NUL bytes */
                if (a == 0 && b == 0 && !accept_nul)
                        return -EINVAL;

                *ret = (a << 4U) | b;
                *eight_bit = true;
                r = 3;
                break;
        }

        case 'u': {
                /* C++11 style 16-bit unicode */

                int a[4];
                size_t i;
                uint32_t c;

                if (length != SIZE_MAX && length < 5)
                        return -EINVAL;

                for (i = 0; i < 4; i++) {
                        a[i] = unhexchar(p[1 + i]);
                        if (a[i] < 0)
                                return a[i];
                }

                c = ((uint32_t) a[0] << 12U) | ((uint32_t) a[1] << 8U) | ((uint32_t) a[2] << 4U) | (uint32_t) a[3];

                /* Don't allow 0 chars */
                if (c == 0 && !accept_nul)
                        return -EINVAL;

                *ret = c;
                r = 5;
                break;
        }

        case 'U': {
                /* C++11 style 32-bit unicode */

                int a[8];
                size_t i;
                char32_t c;

                if (length != SIZE_MAX && length < 9)
                        return -EINVAL;

                for (i = 0; i < 8; i++) {
                        a[i] = unhexchar(p[1 + i]);
                        if (a[i] < 0)
                                return a[i];
                }

                c = ((uint32_t) a[0] << 28U) | ((uint32_t) a[1] << 24U) | ((uint32_t) a[2] << 20U) | ((uint32_t) a[3] << 16U) |
                    ((uint32_t) a[4] << 12U) | ((uint32_t) a[5] <<  8U) | ((uint32_t) a[6] <<  4U) |  (uint32_t) a[7];

                /* Don't allow 0 chars */
                if (c == 0 && !accept_nul)
                        return -EINVAL;

                /* Don't allow invalid code points */
                if (!unichar_is_valid(c))
                        return -EINVAL;

                *ret = c;
                r = 9;
                break;
        }

        case '0':
        case '1':
        case '2':
        case '3':
        case '4':
        case '5':
        case '6':
        case '7': {
                /* octal encoding */
                int a, b, c;
                char32_t m;

                if (length != SIZE_MAX && length < 3)
                        return -EINVAL;

                a = unoctchar(p[0]);
                if (a < 0)
                        return -EINVAL;

                b = unoctchar(p[1]);
                if (b < 0)
                        return -EINVAL;

                c = unoctchar(p[2]);
                if (c < 0)
                        return -EINVAL;

                /* don't allow NUL bytes */
                if (a == 0 && b == 0 && c == 0 && !accept_nul)
                        return -EINVAL;

                /* Don't allow bytes above 255 */
                m = ((uint32_t) a << 6U) | ((uint32_t) b << 3U) | (uint32_t) c;
                if (m > 255)
                        return -EINVAL;

                *ret = m;
                *eight_bit = true;
                r = 3;
                break;
        }

        default:
                return -EINVAL;
        }

        return r;
}

ssize_t cunescape_length_with_prefix(const char *s, size_t length, const char *prefix, UnescapeFlags flags, char **ret) {
        _cleanup_free_ char *ans = NULL;
        char *t;
        const char *f;
        size_t pl;
        int r;

        assert(s);
        assert(ret);

        /* Undoes C style string escaping, and optionally prefixes it. */

        if (length == SIZE_MAX)
                length = strlen(s);

        pl = strlen_ptr(prefix);

        ans = new(char, pl+length+1);
        if (!ans)
                return -ENOMEM;

        if (prefix)
                memcpy(ans, prefix, pl);

        for (f = s, t = ans + pl; f < s + length; f++) {
                size_t remaining;
                bool eight_bit = false;
                char32_t u;

                remaining = s + length - f;
                assert(remaining > 0);

                if (*f != '\\') {
                        /* A literal, copy verbatim */
                        *(t++) = *f;
                        continue;
                }

                if (remaining == 1) {
                        if (flags & UNESCAPE_RELAX) {
                                /* A trailing backslash, copy verbatim */
                                *(t++) = *f;
                                continue;
                        }

                        return -EINVAL;
                }

                r = cunescape_one(f + 1, remaining - 1, &u, &eight_bit, flags & UNESCAPE_ACCEPT_NUL);
                if (r < 0) {
                        if (flags & UNESCAPE_RELAX) {
                                /* Invalid escape code, let's take it literal then */
                                *(t++) = '\\';
                                continue;
                        }

                        return r;
                }

                f += r;
                if (eight_bit)
                        /* One byte? Set directly as specified */
                        *(t++) = u;
                else
                        /* Otherwise encode as multi-byte UTF-8 */
                        t += utf8_encode_unichar(t, u);
        }

        *t = 0;

        assert(t >= ans); /* Let static analyzers know that the answer is non-negative. */
        *ret = TAKE_PTR(ans);
        return t - *ret;
}

char* xescape_full(const char *s, const char *bad, size_t console_width, XEscapeFlags flags) {
        char *ans, *t, *prev, *prev2;
        const char *f;

        assert(s);

        /* Escapes all chars in bad, in addition to \ and all special chars, in \xFF style escaping. May be
         * reversed with cunescape(). If XESCAPE_8_BIT is specified, characters >= 127 are let through
         * unchanged. This corresponds to non-ASCII printable characters in pre-unicode encodings.
         *
         * If console_width is reached, or XESCAPE_FORCE_ELLIPSIS is set, output is truncated and "..." is
         * appended. */

        if (console_width == 0)
                return strdup("");

        ans = new(char, MIN(strlen(s), console_width) * 4 + 1);
        if (!ans)
                return NULL;

        memset(ans, '_', MIN(strlen(s), console_width) * 4);
        ans[MIN(strlen(s), console_width) * 4] = 0;

        bool force_ellipsis = FLAGS_SET(flags, XESCAPE_FORCE_ELLIPSIS);

        for (f = s, t = prev = prev2 = ans; ; f++) {
                char *tmp_t = t;

                if (!*f) {
                        if (force_ellipsis)
                                break;

                        *t = 0;
                        return ans;
                }

                if ((unsigned char) *f < ' ' ||
                    (!FLAGS_SET(flags, XESCAPE_8_BIT) && (unsigned char) *f >= 127) ||
                    *f == '\\' || (bad && strchr(bad, *f))) {
                        if ((size_t) (t - ans) + 4 + 3 * force_ellipsis > console_width)
                                break;

                        *(t++) = '\\';
                        *(t++) = 'x';
                        *(t++) = hexchar(*f >> 4);
                        *(t++) = hexchar(*f);
                } else {
                        if ((size_t) (t - ans) + 1 + 3 * force_ellipsis > console_width)
                                break;

                        *(t++) = *f;
                }

                /* We might need to go back two cycles to fit three dots, so remember two positions */
                prev2 = prev;
                prev = tmp_t;
        }

        /* We can just write where we want, since chars are one-byte */
        size_t c = MIN(console_width, 3u); /* If the console is too narrow, write fewer dots */
        size_t off;
        if (console_width - c >= (size_t) (t - ans))
                off = (size_t) (t - ans);
        else if (console_width - c >= (size_t) (prev - ans))
                off = (size_t) (prev - ans);
        else if (console_width - c >= (size_t) (prev2 - ans))
                off = (size_t) (prev2 - ans);
        else
                off = console_width - c;
        assert(off <= (size_t) (t - ans));

        memcpy(ans + off, "...", c);
        ans[off + c] = '\0';
        return ans;
}

char* escape_non_printable_full(const char *str, size_t console_width, XEscapeFlags flags) {
        if (FLAGS_SET(flags, XESCAPE_8_BIT))
                return xescape_full(str, /* bad= */ NULL, console_width, flags);
        else
                return utf8_escape_non_printable_full(str,
                                                      console_width,
                                                      FLAGS_SET(flags, XESCAPE_FORCE_ELLIPSIS));
}

char* octescape(const char *s, size_t len) {
        char *buf, *t;

        /* Escapes \ and " chars, in \nnn style escaping. */

        assert(s || len == 0);

        if (len == SIZE_MAX)
                len = strlen(s);

        if (len > (SIZE_MAX - 1) / 4)
                return NULL;

        t = buf = new(char, len * 4 + 1);
        if (!buf)
                return NULL;

        for (size_t i = 0; i < len; i++) {
                uint8_t u = (uint8_t) s[i];

                if (u < ' ' || u >= 127 || IN_SET(u, '\\', '"')) {
                        *(t++) = '\\';
                        *(t++) = '0' + (u >> 6);
                        *(t++) = '0' + ((u >> 3) & 7);
                        *(t++) = '0' + (u & 7);
                } else
                        *(t++) = u;
        }

        *t = 0;
        return buf;
}

char* decescape(const char *s, size_t len, const char *bad) {
        char *buf, *t;

        /* Escapes all chars in bad, in addition to \ and " chars, in \nnn decimal style escaping. */

        assert(s || len == 0);

        if (len == SIZE_MAX)
                len = strlen(s);

        if (len > (SIZE_MAX - 1) / 4)
                return NULL;

        t = buf = new(char, len * 4 + 1);
        if (!buf)
                return NULL;

        for (size_t i = 0; i < len; i++) {
                uint8_t u = (uint8_t) s[i];

                if (u < ' ' || u >= 127 || IN_SET(u, '\\', '"') || strchr(bad, u)) {
                        *(t++) = '\\';
                        *(t++) = '0' + (u / 100);
                        *(t++) = '0' + ((u / 10) % 10);
                        *(t++) = '0' + (u % 10);
                } else
                        *(t++) = u;
        }

        *t = 0;
        return buf;
}

static char* strcpy_backslash_escaped(char *t, const char *s, const char *bad) {
        assert(bad);
        assert(t);
        assert(s);

        while (*s) {
                int l = utf8_encoded_valid_unichar(s, SIZE_MAX);

                if (char_is_cc(*s) || l < 0)
                        t += cescape_char(*(s++), t);
                else if (l == 1) {
                        if (*s == '\\' || strchr(bad, *s))
                                *(t++) = '\\';
                        *(t++) = *(s++);
                } else {
                        t = mempcpy(t, s, l);
                        s += l;
                }
        }

        return t;
}

char* shell_escape(const char *s, const char *bad) {
        char *buf, *t;

        buf = new(char, strlen(s)*4+1);
        if (!buf)
                return NULL;

        t = strcpy_backslash_escaped(buf, s, bad);
        *t = 0;

        return buf;
}

char* shell_maybe_quote(const char *s, ShellEscapeFlags flags) {
        const char *p;
        char *buf, *t;

        assert(s);

        /* Encloses a string in quotes if necessary to make it OK as a shell string. */

        if (FLAGS_SET(flags, SHELL_ESCAPE_EMPTY) && isempty(s))
                return strdup("\"\""); /* We don't use $'' here in the POSIX mode. "" is fine too. */

        for (p = s; *p; ) {
                int l = utf8_encoded_valid_unichar(p, SIZE_MAX);

                if (char_is_cc(*p) || l < 0 ||
                    strchr(WHITESPACE SHELL_NEED_QUOTES, *p))
                        break;

                p += l;
        }

        if (!*p)
                return strdup(s);

        buf = new(char, FLAGS_SET(flags, SHELL_ESCAPE_POSIX) + 1 + strlen(s)*4 + 1 + 1);
        if (!buf)
                return NULL;

        t = buf;
        if (FLAGS_SET(flags, SHELL_ESCAPE_POSIX)) {
                *(t++) = '$';
                *(t++) = '\'';
        } else
                *(t++) = '"';

        t = mempcpy(t, s, p - s);

        t = strcpy_backslash_escaped(t, p,
                                     FLAGS_SET(flags, SHELL_ESCAPE_POSIX) ? SHELL_NEED_ESCAPE_POSIX : SHELL_NEED_ESCAPE);

        if (FLAGS_SET(flags, SHELL_ESCAPE_POSIX))
                *(t++) = '\'';
        else
                *(t++) = '"';
        *t = 0;

        return str_realloc(buf);
}

char* quote_command_line(char **argv, ShellEscapeFlags flags) {
        _cleanup_free_ char *result = NULL;

        assert(argv);

        STRV_FOREACH(a, argv) {
                _cleanup_free_ char *t = NULL;

                t = shell_maybe_quote(*a, flags);
                if (!t)
                        return NULL;

                if (!strextend_with_separator(&result, " ", t))
                        return NULL;
        }

        return str_realloc(TAKE_PTR(result));
}
Commit	Line	Data
db9ecf05	1	/* SPDX-License-Identifier: LGPL-2.1-or-later */
4f5dd394	2
11c3a366 TA	3	#include <stdlib.h>
	4	#include <string.h>
	5
b5efdb8a	6	#include "alloc-util.h"
e4e73a63 LP	7	#include "escape.h"
e4e73a63 LP	8	#include "hexdecoct.h"
0c15577a	9	#include "string-util.h"
eeb91d29	10	#include "strv.h"
4f5dd394	11	#include "utf8.h"
4f5dd394	12
b778252b ZJS	13	int cescape_char(char c, char *buf) {
b778252b ZJS	14	char *buf_old = buf;
4f5dd394	15
76a35973 LP	16	/* Needs space for 4 characters in the buffer */
76a35973 LP	17
4f5dd394 LP	18	switch (c) {
	19
	20	case '\a':
	21	*(buf++) = '\\';
	22	*(buf++) = 'a';
	23	break;
	24	case '\b':
	25	*(buf++) = '\\';
	26	*(buf++) = 'b';
	27	break;
	28	case '\f':
	29	*(buf++) = '\\';
	30	*(buf++) = 'f';
	31	break;
	32	case '\n':
	33	*(buf++) = '\\';
	34	*(buf++) = 'n';
	35	break;
	36	case '\r':
	37	*(buf++) = '\\';
	38	*(buf++) = 'r';
	39	break;
	40	case '\t':
	41	*(buf++) = '\\';
	42	*(buf++) = 't';
	43	break;
	44	case '\v':
	45	*(buf++) = '\\';
	46	*(buf++) = 'v';
	47	break;
	48	case '\\':
	49	*(buf++) = '\\';
	50	*(buf++) = '\\';
	51	break;
	52	case '"':
	53	*(buf++) = '\\';
	54	*(buf++) = '"';
	55	break;
	56	case '\'':
	57	*(buf++) = '\\';
	58	*(buf++) = '\'';
	59	break;
	60
	61	default:
	62	/* For special chars we prefer octal over
	63	* hexadecimal encoding, simply because glib's
	64	* g_strescape() does the same */
	65	if ((c < ' ') \|\| (c >= 127)) {
	66	*(buf++) = '\\';
	67	*(buf++) = octchar((unsigned char) c >> 6);
	68	*(buf++) = octchar((unsigned char) c >> 3);
	69	*(buf++) = octchar((unsigned char) c);
	70	} else
	71	*(buf++) = c;
	72	break;
	73	}
	74
	75	return buf - buf_old;
	76	}
	77
31be0e9e	78	char* cescape_length(const char *s, size_t n) {
4f5dd394	79	const char *f;
a5ef3638	80	char r, t;
4f5dd394	81
7de7c7b6 MY	82	/* Does C style string escaping. May be reversed with cunescape(). */
7de7c7b6 MY	83
a5ef3638	84	assert(s \|\| n == 0);
4f5dd394	85
7de7c7b6 MY	86	if (n == SIZE_MAX)
	87	n = strlen(s);
	88
	89	if (n > (SIZE_MAX - 1) / 4)
	90	return NULL;
4f5dd394	91
a5ef3638	92	r = new(char, n*4 + 1);
4f5dd394 LP	93	if (!r)
	94	return NULL;
	95
a5ef3638	96	for (f = s, t = r; f < s + n; f++)
4f5dd394 LP	97	t += cescape_char(*f, t);
	98
	99	*t = 0;
	100
	101	return r;
	102	}
	103
0e72e469	104	int cunescape_one(const char p, size_t length, char32_t ret, bool *eight_bit, bool accept_nul) {
4f5dd394 LP	105	int r = 1;
	106
	107	assert(p);
4f5dd394 LP	108	assert(ret);
4f5dd394 LP	109
3565e095 ZJS	110	/* Unescapes C style. Returns the unescaped character in ret.
	111	* Sets *eight_bit to true if the escaped sequence either fits in
	112	* one byte in UTF-8 or is a non-unicode literal byte and should
	113	* instead be copied directly.
	114	*/
4f5dd394	115
f5fbe71d	116	if (length != SIZE_MAX && length < 1)
4f5dd394 LP	117	return -EINVAL;
	118
	119	switch (p[0]) {
	120
	121	case 'a':
	122	*ret = '\a';
	123	break;
	124	case 'b':
	125	*ret = '\b';
	126	break;
	127	case 'f':
	128	*ret = '\f';
	129	break;
	130	case 'n':
	131	*ret = '\n';
	132	break;
	133	case 'r':
	134	*ret = '\r';
	135	break;
	136	case 't':
	137	*ret = '\t';
	138	break;
	139	case 'v':
	140	*ret = '\v';
	141	break;
	142	case '\\':
	143	*ret = '\\';
	144	break;
	145	case '"':
	146	*ret = '"';
	147	break;
	148	case '\'':
	149	*ret = '\'';
	150	break;
	151
	152	case 's':
	153	/* This is an extension of the XDG syntax files */
	154	*ret = ' ';
	155	break;
	156
	157	case 'x': {
	158	/* hexadecimal encoding */
	159	int a, b;
	160
f5fbe71d	161	if (length != SIZE_MAX && length < 3)
4f5dd394 LP	162	return -EINVAL;
	163
	164	a = unhexchar(p[1]);
	165	if (a < 0)
	166	return -EINVAL;
	167
	168	b = unhexchar(p[2]);
	169	if (b < 0)
	170	return -EINVAL;
	171
	172	/* Don't allow NUL bytes */
0e72e469	173	if (a == 0 && b == 0 && !accept_nul)
4f5dd394 LP	174	return -EINVAL;
4f5dd394 LP	175
3565e095 ZJS	176	*ret = (a << 4U) \| b;
3565e095 ZJS	177	*eight_bit = true;
4f5dd394 LP	178	r = 3;
	179	break;
	180	}
	181
	182	case 'u': {
da890466	183	/* C++11 style 16-bit unicode */
4f5dd394 LP	184
4f5dd394 LP	185	int a[4];
da6053d0	186	size_t i;
4f5dd394 LP	187	uint32_t c;
4f5dd394 LP	188
f5fbe71d	189	if (length != SIZE_MAX && length < 5)
4f5dd394 LP	190	return -EINVAL;
	191
	192	for (i = 0; i < 4; i++) {
	193	a[i] = unhexchar(p[1 + i]);
	194	if (a[i] < 0)
	195	return a[i];
	196	}
	197
	198	c = ((uint32_t) a[0] << 12U) \| ((uint32_t) a[1] << 8U) \| ((uint32_t) a[2] << 4U) \| (uint32_t) a[3];
	199
	200	/* Don't allow 0 chars */
0e72e469	201	if (c == 0 && !accept_nul)
4f5dd394 LP	202	return -EINVAL;
4f5dd394 LP	203
3565e095	204	*ret = c;
4f5dd394 LP	205	r = 5;
	206	break;
	207	}
	208
	209	case 'U': {
da890466	210	/* C++11 style 32-bit unicode */
4f5dd394 LP	211
4f5dd394 LP	212	int a[8];
da6053d0	213	size_t i;
c932fb71	214	char32_t c;
4f5dd394	215
f5fbe71d	216	if (length != SIZE_MAX && length < 9)
4f5dd394 LP	217	return -EINVAL;
	218
	219	for (i = 0; i < 8; i++) {
	220	a[i] = unhexchar(p[1 + i]);
	221	if (a[i] < 0)
	222	return a[i];
	223	}
	224
dcd12626 LP	225	c = ((uint32_t) a[0] << 28U) \| ((uint32_t) a[1] << 24U) \| ((uint32_t) a[2] << 20U) \| ((uint32_t) a[3] << 16U) \|
dcd12626 LP	226	((uint32_t) a[4] << 12U) \| ((uint32_t) a[5] << 8U) \| ((uint32_t) a[6] << 4U) \| (uint32_t) a[7];
4f5dd394 LP	227
4f5dd394 LP	228	/* Don't allow 0 chars */
0e72e469	229	if (c == 0 && !accept_nul)
4f5dd394 LP	230	return -EINVAL;
	231
	232	/* Don't allow invalid code points */
	233	if (!unichar_is_valid(c))
	234	return -EINVAL;
	235
3565e095	236	*ret = c;
4f5dd394 LP	237	r = 9;
	238	break;
	239	}
	240
	241	case '0':
	242	case '1':
	243	case '2':
	244	case '3':
	245	case '4':
	246	case '5':
	247	case '6':
	248	case '7': {
	249	/* octal encoding */
	250	int a, b, c;
c932fb71	251	char32_t m;
4f5dd394	252
f5fbe71d	253	if (length != SIZE_MAX && length < 3)
4f5dd394 LP	254	return -EINVAL;
	255
	256	a = unoctchar(p[0]);
	257	if (a < 0)
	258	return -EINVAL;
	259
	260	b = unoctchar(p[1]);
	261	if (b < 0)
	262	return -EINVAL;
	263
	264	c = unoctchar(p[2]);
	265	if (c < 0)
	266	return -EINVAL;
	267
	268	/* don't allow NUL bytes */
0e72e469	269	if (a == 0 && b == 0 && c == 0 && !accept_nul)
4f5dd394 LP	270	return -EINVAL;
	271
	272	/* Don't allow bytes above 255 */
dcd12626	273	m = ((uint32_t) a << 6U) \| ((uint32_t) b << 3U) \| (uint32_t) c;
4f5dd394 LP	274	if (m > 255)
	275	return -EINVAL;
	276
	277	*ret = m;
3565e095	278	*eight_bit = true;
4f5dd394 LP	279	r = 3;
	280	break;
	281	}
	282
	283	default:
	284	return -EINVAL;
	285	}
	286
	287	return r;
	288	}
	289
e437538f	290	ssize_t cunescape_length_with_prefix(const char s, size_t length, const char prefix, UnescapeFlags flags, char **ret) {
ddedf7ca ZJS	291	_cleanup_free_ char *ans = NULL;
ddedf7ca ZJS	292	char *t;
4f5dd394 LP	293	const char *f;
4f5dd394 LP	294	size_t pl;
ddedf7ca	295	int r;
4f5dd394 LP	296
	297	assert(s);
	298	assert(ret);
	299
	300	/* Undoes C style string escaping, and optionally prefixes it. */
	301
0c15577a DDM	302	if (length == SIZE_MAX)
	303	length = strlen(s);
	304
7bf7ce28	305	pl = strlen_ptr(prefix);
4f5dd394	306
ddedf7ca ZJS	307	ans = new(char, pl+length+1);
ddedf7ca ZJS	308	if (!ans)
4f5dd394 LP	309	return -ENOMEM;
	310
	311	if (prefix)
ddedf7ca	312	memcpy(ans, prefix, pl);
4f5dd394	313
ddedf7ca	314	for (f = s, t = ans + pl; f < s + length; f++) {
4f5dd394	315	size_t remaining;
3565e095	316	bool eight_bit = false;
c932fb71	317	char32_t u;
4f5dd394 LP	318
	319	remaining = s + length - f;
	320	assert(remaining > 0);
	321
	322	if (*f != '\\') {
629ff674	323	/* A literal, copy verbatim */
4f5dd394 LP	324	(t++) = f;
	325	continue;
	326	}
	327
	328	if (remaining == 1) {
	329	if (flags & UNESCAPE_RELAX) {
	330	/* A trailing backslash, copy verbatim */
	331	(t++) = f;
	332	continue;
	333	}
	334
4f5dd394 LP	335	return -EINVAL;
	336	}
	337
ddedf7ca ZJS	338	r = cunescape_one(f + 1, remaining - 1, &u, &eight_bit, flags & UNESCAPE_ACCEPT_NUL);
ddedf7ca ZJS	339	if (r < 0) {
4f5dd394 LP	340	if (flags & UNESCAPE_RELAX) {
	341	/* Invalid escape code, let's take it literal then */
	342	*(t++) = '\\';
	343	continue;
	344	}
	345
ddedf7ca	346	return r;
4f5dd394 LP	347	}
4f5dd394 LP	348
ddedf7ca	349	f += r;
3565e095 ZJS	350	if (eight_bit)
	351	/* One byte? Set directly as specified */
	352	*(t++) = u;
4f5dd394	353	else
3565e095	354	/* Otherwise encode as multi-byte UTF-8 */
4f5dd394	355	t += utf8_encode_unichar(t, u);
4f5dd394 LP	356	}
	357
	358	*t = 0;
	359
1421705d	360	assert(t >= ans); /* Let static analyzers know that the answer is non-negative. */
ddedf7ca ZJS	361	*ret = TAKE_PTR(ans);
ddedf7ca ZJS	362	return t - *ret;
4f5dd394 LP	363	}
4f5dd394 LP	364
b19f2116	365	char* xescape_full(const char s, const char bad, size_t console_width, XEscapeFlags flags) {
70d55819	366	char ans, t, prev, prev2;
4f5dd394 LP	367	const char *f;
4f5dd394 LP	368
ea844c49 LP	369	assert(s);
ea844c49 LP	370
70d55819	371	/* Escapes all chars in bad, in addition to \ and all special chars, in \xFF style escaping. May be
b19f2116 ZJS	372	* reversed with cunescape(). If XESCAPE_8_BIT is specified, characters >= 127 are let through
b19f2116 ZJS	373	* unchanged. This corresponds to non-ASCII printable characters in pre-unicode encodings.
70d55819	374	*
fc96e5c0 ZJS	375	* If console_width is reached, or XESCAPE_FORCE_ELLIPSIS is set, output is truncated and "..." is
fc96e5c0 ZJS	376	* appended. */
4f5dd394	377
70d55819 ZJS	378	if (console_width == 0)
	379	return strdup("");
	380
	381	ans = new(char, MIN(strlen(s), console_width) * 4 + 1);
	382	if (!ans)
4f5dd394 LP	383	return NULL;
4f5dd394 LP	384
70d55819 ZJS	385	memset(ans, '_', MIN(strlen(s), console_width) * 4);
	386	ans[MIN(strlen(s), console_width) * 4] = 0;
	387
fc96e5c0 ZJS	388	bool force_ellipsis = FLAGS_SET(flags, XESCAPE_FORCE_ELLIPSIS);
fc96e5c0 ZJS	389
70d55819 ZJS	390	for (f = s, t = prev = prev2 = ans; ; f++) {
	391	char *tmp_t = t;
	392
	393	if (!*f) {
fc96e5c0 ZJS	394	if (force_ellipsis)
	395	break;
	396
70d55819 ZJS	397	*t = 0;
	398	return ans;
	399	}
	400
b19f2116 ZJS	401	if ((unsigned char) *f < ' ' \|\|
b19f2116 ZJS	402	(!FLAGS_SET(flags, XESCAPE_8_BIT) && (unsigned char) *f >= 127) \|\|
ea844c49	403	f == '\\' \|\| (bad && strchr(bad, f))) {
fc96e5c0	404	if ((size_t) (t - ans) + 4 + 3 * force_ellipsis > console_width)
70d55819	405	break;
4f5dd394	406
4f5dd394 LP	407	*(t++) = '\\';
	408	*(t++) = 'x';
	409	(t++) = hexchar(f >> 4);
	410	(t++) = hexchar(f);
70d55819	411	} else {
fc96e5c0	412	if ((size_t) (t - ans) + 1 + 3 * force_ellipsis > console_width)
70d55819 ZJS	413	break;
70d55819 ZJS	414
4f5dd394	415	(t++) = f;
70d55819	416	}
4f5dd394	417
70d55819 ZJS	418	/* We might need to go back two cycles to fit three dots, so remember two positions */
	419	prev2 = prev;
	420	prev = tmp_t;
	421	}
4f5dd394	422
70d55819 ZJS	423	/* We can just write where we want, since chars are one-byte */
	424	size_t c = MIN(console_width, 3u); /* If the console is too narrow, write fewer dots */
	425	size_t off;
	426	if (console_width - c >= (size_t) (t - ans))
	427	off = (size_t) (t - ans);
	428	else if (console_width - c >= (size_t) (prev - ans))
	429	off = (size_t) (prev - ans);
	430	else if (console_width - c >= (size_t) (prev2 - ans))
	431	off = (size_t) (prev2 - ans);
	432	else
	433	off = console_width - c;
	434	assert(off <= (size_t) (t - ans));
	435
	436	memcpy(ans + off, "...", c);
	437	ans[off + c] = '\0';
	438	return ans;
4f5dd394 LP	439	}
4f5dd394 LP	440
b19f2116 ZJS	441	char* escape_non_printable_full(const char *str, size_t console_width, XEscapeFlags flags) {
b19f2116 ZJS	442	if (FLAGS_SET(flags, XESCAPE_8_BIT))
ea844c49	443	return xescape_full(str, /* bad= */ NULL, console_width, flags);
e3b4efd2	444	else
fc96e5c0 ZJS	445	return utf8_escape_non_printable_full(str,
	446	console_width,
	447	FLAGS_SET(flags, XESCAPE_FORCE_ELLIPSIS));
e3b4efd2 ZJS	448	}
e3b4efd2 ZJS	449
31be0e9e	450	char* octescape(const char *s, size_t len) {
76519cec	451	char buf, t;
95052df3	452
c7a097d7	453	/* Escapes \ and " chars, in \nnn style escaping. */
95052df3	454
76519cec YW	455	assert(s \|\| len == 0);
76519cec YW	456
c6342e35 LP	457	if (len == SIZE_MAX)
	458	len = strlen(s);
	459
60cf4059	460	if (len > (SIZE_MAX - 1) / 4)
c6342e35 LP	461	return NULL;
c6342e35 LP	462
76519cec YW	463	t = buf = new(char, len * 4 + 1);
76519cec YW	464	if (!buf)
95052df3 ZJS	465	return NULL;
95052df3 ZJS	466
76519cec YW	467	for (size_t i = 0; i < len; i++) {
76519cec YW	468	uint8_t u = (uint8_t) s[i];
95052df3	469
76519cec	470	if (u < ' ' \|\| u >= 127 \|\| IN_SET(u, '\\', '"')) {
95052df3	471	*(t++) = '\\';
76519cec YW	472	*(t++) = '0' + (u >> 6);
	473	*(t++) = '0' + ((u >> 3) & 7);
	474	*(t++) = '0' + (u & 7);
95052df3	475	} else
76519cec	476	*(t++) = u;
95052df3 ZJS	477	}
	478
	479	*t = 0;
76519cec	480	return buf;
95052df3 ZJS	481	}
95052df3 ZJS	482
6bdbfb7e	483	char* decescape(const char s, size_t len, const char bad) {
b699f5f2 RP	484	char buf, t;
	485
	486	/* Escapes all chars in bad, in addition to \ and " chars, in \nnn decimal style escaping. */
	487
	488	assert(s \|\| len == 0);
	489
7de7c7b6 MY	490	if (len == SIZE_MAX)
	491	len = strlen(s);
	492
	493	if (len > (SIZE_MAX - 1) / 4)
	494	return NULL;
	495
b699f5f2 RP	496	t = buf = new(char, len * 4 + 1);
	497	if (!buf)
	498	return NULL;
	499
	500	for (size_t i = 0; i < len; i++) {
	501	uint8_t u = (uint8_t) s[i];
	502
	503	if (u < ' ' \|\| u >= 127 \|\| IN_SET(u, '\\', '"') \|\| strchr(bad, u)) {
	504	*(t++) = '\\';
	505	*(t++) = '0' + (u / 100);
	506	*(t++) = '0' + ((u / 10) % 10);
	507	*(t++) = '0' + (u % 10);
	508	} else
	509	*(t++) = u;
	510	}
	511
	512	*t = 0;
	513	return buf;
	514	}
	515
566d06ae	516	static char* strcpy_backslash_escaped(char t, const char s, const char *bad) {
4f5dd394	517	assert(bad);
0b82a6fa	518	assert(t);
0b82a6fa	519	assert(s);
4f5dd394	520
00f57157	521	while (*s) {
	522	int l = utf8_encoded_valid_unichar(s, SIZE_MAX);
	523
	524	if (char_is_cc(*s) \|\| l < 0)
	525	t += cescape_char(*(s++), t);
	526	else if (l == 1) {
0089ab08 ZJS	527	if (s == '\\' \|\| strchr(bad, s))
0089ab08 ZJS	528	*(t++) = '\\';
00f57157	529	(t++) = (s++);
	530	} else {
	531	t = mempcpy(t, s, l);
	532	s += l;
804ee07c	533	}
00f57157	534	}
804ee07c	535
4f5dd394 LP	536	return t;
	537	}
	538
31be0e9e	539	char* shell_escape(const char s, const char bad) {
0089ab08	540	char buf, t;
4f5dd394	541
0089ab08 ZJS	542	buf = new(char, strlen(s)*4+1);
0089ab08 ZJS	543	if (!buf)
4f5dd394 LP	544	return NULL;
4f5dd394 LP	545
0089ab08	546	t = strcpy_backslash_escaped(buf, s, bad);
4f5dd394 LP	547	*t = 0;
4f5dd394 LP	548
0089ab08	549	return buf;
4f5dd394 LP	550	}
4f5dd394 LP	551
9e53c10a	552	char* shell_maybe_quote(const char *s, ShellEscapeFlags flags) {
4f5dd394	553	const char *p;
0089ab08	554	char buf, t;
4f5dd394 LP	555
	556	assert(s);
	557
0089ab08	558	/* Encloses a string in quotes if necessary to make it OK as a shell string. */
4f5dd394	559
1129cd8a ZJS	560	if (FLAGS_SET(flags, SHELL_ESCAPE_EMPTY) && isempty(s))
	561	return strdup("\"\""); /* We don't use $'' here in the POSIX mode. "" is fine too. */
	562
00f57157	563	for (p = s; *p; ) {
	564	int l = utf8_encoded_valid_unichar(p, SIZE_MAX);
	565
	566	if (char_is_cc(*p) \|\| l < 0 \|\|
0089ab08	567	strchr(WHITESPACE SHELL_NEED_QUOTES, *p))
4f5dd394 LP	568	break;
4f5dd394 LP	569
00f57157	570	p += l;
	571	}
	572
4f5dd394 LP	573	if (!*p)
	574	return strdup(s);
	575
0089ab08 ZJS	576	buf = new(char, FLAGS_SET(flags, SHELL_ESCAPE_POSIX) + 1 + strlen(s)*4 + 1 + 1);
0089ab08 ZJS	577	if (!buf)
4f5dd394 LP	578	return NULL;
4f5dd394 LP	579
0089ab08	580	t = buf;
9e53c10a	581	if (FLAGS_SET(flags, SHELL_ESCAPE_POSIX)) {
804ee07c ZJS	582	*(t++) = '$';
804ee07c ZJS	583	*(t++) = '\'';
9e53c10a ZJS	584	} else
9e53c10a ZJS	585	*(t++) = '"';
804ee07c	586
4f5dd394 LP	587	t = mempcpy(t, s, p - s);
4f5dd394 LP	588
9e53c10a	589	t = strcpy_backslash_escaped(t, p,
566d06ae	590	FLAGS_SET(flags, SHELL_ESCAPE_POSIX) ? SHELL_NEED_ESCAPE_POSIX : SHELL_NEED_ESCAPE);
4f5dd394	591
9e53c10a	592	if (FLAGS_SET(flags, SHELL_ESCAPE_POSIX))
804ee07c	593	*(t++) = '\'';
9e53c10a ZJS	594	else
9e53c10a ZJS	595	*(t++) = '"';
4f5dd394 LP	596	*t = 0;
4f5dd394 LP	597
0089ab08	598	return str_realloc(buf);
4f5dd394	599	}
eeb91d29	600
4ef15008	601	char* quote_command_line(char **argv, ShellEscapeFlags flags) {
eeb91d29 ZJS	602	_cleanup_free_ char *result = NULL;
	603
	604	assert(argv);
	605
eeb91d29 ZJS	606	STRV_FOREACH(a, argv) {
	607	_cleanup_free_ char *t = NULL;
	608
4ef15008	609	t = shell_maybe_quote(*a, flags);
eeb91d29 ZJS	610	if (!t)
	611	return NULL;
	612
	613	if (!strextend_with_separator(&result, " ", t))
	614	return NULL;
	615	}
	616
7d0cede0	617	return str_realloc(TAKE_PTR(result));
eeb91d29	618	}