-/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/
-
/***
This file is part of systemd.
along with systemd; If not, see <http://www.gnu.org/licenses/>.
***/
+#include <errno.h>
+#include <stdlib.h>
+#include <string.h>
+
#include "alloc-util.h"
#include "escape.h"
#include "hexdecoct.h"
-#include "string-util.h"
+#include "macro.h"
#include "utf8.h"
-#include "util.h"
size_t cescape_char(char c, char *buf) {
char * buf_old = buf;
return buf - buf_old;
}
-char *cescape(const char *s) {
- char *r, *t;
+char *cescape_length(const char *s, size_t n) {
const char *f;
+ char *r, *t;
- assert(s);
+ assert(s || n == 0);
/* Does C style string escaping. May be reversed with
* cunescape(). */
- r = new(char, strlen(s)*4 + 1);
+ r = new(char, n*4 + 1);
if (!r)
return NULL;
- for (f = s, t = r; *f; f++)
+ 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, char *ret, uint32_t *ret_unicode) {
+char *cescape(const char *s) {
+ assert(s);
+
+ return cescape_length(s, strlen(s));
+}
+
+int cunescape_one(const char *p, size_t length, char32_t *ret, bool *eight_bit) {
int r = 1;
assert(p);
assert(*p);
assert(ret);
- /* Unescapes C style. Returns the unescaped character in ret,
- * unless we encountered a \u sequence in which case the full
- * unicode character is returned in ret_unicode, instead. */
+ /* 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_t) -1 && length < 1)
return -EINVAL;
if (a == 0 && b == 0)
return -EINVAL;
- *ret = (char) ((a << 4U) | b);
+ *ret = (a << 4U) | b;
+ *eight_bit = true;
r = 3;
break;
}
if (c == 0)
return -EINVAL;
- if (c < 128)
- *ret = c;
- else {
- if (!ret_unicode)
- return -EINVAL;
-
- *ret = 0;
- *ret_unicode = c;
- }
-
+ *ret = c;
r = 5;
break;
}
int a[8];
unsigned i;
- uint32_t c;
+ char32_t c;
if (length != (size_t) -1 && length < 9)
return -EINVAL;
if (!unichar_is_valid(c))
return -EINVAL;
- if (c < 128)
- *ret = c;
- else {
- if (!ret_unicode)
- return -EINVAL;
-
- *ret = 0;
- *ret_unicode = c;
- }
-
+ *ret = c;
r = 9;
break;
}
case '7': {
/* octal encoding */
int a, b, c;
- uint32_t m;
+ char32_t m;
if (length != (size_t) -1 && length < 3)
return -EINVAL;
return -EINVAL;
*ret = m;
+ *eight_bit = true;
r = 3;
break;
}
for (f = s, t = r + pl; f < s + length; f++) {
size_t remaining;
- uint32_t u;
- char c;
+ bool eight_bit = false;
+ char32_t u;
int k;
remaining = s + length - f;
assert(remaining > 0);
if (*f != '\\') {
- /* A literal literal, copy verbatim */
+ /* A literal, copy verbatim */
*(t++) = *f;
continue;
}
return -EINVAL;
}
- k = cunescape_one(f + 1, remaining - 1, &c, &u);
+ k = cunescape_one(f + 1, remaining - 1, &u, &eight_bit);
if (k < 0) {
if (flags & UNESCAPE_RELAX) {
/* Invalid escape code, let's take it literal then */
return k;
}
- if (c != 0)
- /* Non-Unicode? Let's encode this directly */
- *(t++) = c;
+ f += k;
+ if (eight_bit)
+ /* One byte? Set directly as specified */
+ *(t++) = u;
else
- /* Unicode? Then let's encode this in UTF-8 */
+ /* Otherwise encode as multi-byte UTF-8 */
t += utf8_encode_unichar(t, u);
-
- f += k;
}
*t = 0;
return r;
}
-static char *strcpy_backslash_escaped(char *t, const char *s, const char *bad) {
+#if 0 /// UNNEEDED by elogind
+char *octescape(const char *s, size_t len) {
+ char *r, *t;
+ const char *f;
+
+ /* Escapes all chars in bad, in addition to \ and " chars,
+ * in \nnn style escaping. */
+
+ r = new(char, len * 4 + 1);
+ if (!r)
+ return NULL;
+
+ for (f = s, t = r; f < s + len; f++) {
+
+ if (*f < ' ' || *f >= 127 || *f == '\\' || *f == '"') {
+ *(t++) = '\\';
+ *(t++) = '0' + (*f >> 6);
+ *(t++) = '0' + ((*f >> 3) & 8);
+ *(t++) = '0' + (*f & 8);
+ } else
+ *(t++) = *f;
+ }
+
+ *t = 0;
+
+ return r;
+
+}
+
+static char *strcpy_backslash_escaped(char *t, const char *s, const char *bad, bool escape_tab_nl) {
assert(bad);
for (; *s; s++) {
+ if (escape_tab_nl && IN_SET(*s, '\n', '\t')) {
+ *(t++) = '\\';
+ *(t++) = *s == '\n' ? 'n' : 't';
+ continue;
+ }
+
if (*s == '\\' || strchr(bad, *s))
*(t++) = '\\';
if (!r)
return NULL;
- t = strcpy_backslash_escaped(r, s, bad);
+ t = strcpy_backslash_escaped(r, s, bad, false);
*t = 0;
return r;
}
-char *shell_maybe_quote(const char *s) {
+char* shell_maybe_quote(const char *s, EscapeStyle style) {
const char *p;
char *r, *t;
assert(s);
- /* Encloses a string in double quotes if necessary to make it
- * OK as shell string. */
+ /* Encloses a string in quotes if necessary to make it OK as a shell
+ * string. Note that we treat benign UTF-8 characters as needing
+ * escaping too, but that should be OK. */
for (p = s; *p; p++)
if (*p <= ' ' ||
if (!*p)
return strdup(s);
- r = new(char, 1+strlen(s)*2+1+1);
+ r = new(char, (style == ESCAPE_POSIX) + 1 + strlen(s)*2 + 1 + 1);
if (!r)
return NULL;
t = r;
- *(t++) = '"';
+ if (style == ESCAPE_BACKSLASH)
+ *(t++) = '"';
+ else if (style == ESCAPE_POSIX) {
+ *(t++) = '$';
+ *(t++) = '\'';
+ } else
+ assert_not_reached("Bad EscapeStyle");
+
t = mempcpy(t, s, p - s);
- t = strcpy_backslash_escaped(t, p, SHELL_NEED_ESCAPE);
+ if (style == ESCAPE_BACKSLASH)
+ t = strcpy_backslash_escaped(t, p, SHELL_NEED_ESCAPE, false);
+ else
+ t = strcpy_backslash_escaped(t, p, SHELL_NEED_ESCAPE_POSIX, true);
- *(t++)= '"';
+ if (style == ESCAPE_BACKSLASH)
+ *(t++) = '"';
+ else
+ *(t++) = '\'';
*t = 0;
return r;
}
+#endif // 0