Merge pull request #16911 from keszybz/userdb-json-invalid-utf8

Deal properly with invalid utf-8 in userdb entries

src/basic/escape.c

@@ -75,7 +75,7 @@ int cescape_char(char c, char *buf) {
         return buf - buf_old;
 }
 
-char *cescape_length(const char *s, size_t n) {
+char* cescape_length(const char *s, size_t n) {
         const char *f;
         char *r, *t;
 
@@ -96,7 +96,7 @@ char *cescape_length(const char *s, size_t n) {
         return r;
 }
 
-char *cescape(const char *s) {
+char* cescape(const char *s) {
         assert(s);
 
         return cescape_length(s, strlen(s));
@@ -360,7 +360,7 @@ int cunescape_length_with_prefix(const char *s, size_t length, const char *prefi
         return t - r;
 }
 
-char *xescape_full(const char *s, const char *bad, size_t console_width, bool eight_bits) {
+char* xescape_full(const char *s, const char *bad, size_t console_width, bool eight_bits) {
         char *ans, *t, *prev, *prev2;
         const char *f;
 
@@ -427,14 +427,14 @@ char *xescape_full(const char *s, const char *bad, size_t console_width, bool ei
         return ans;
 }
 
-char *escape_non_printable_full(const char *str, size_t console_width, bool eight_bit) {
+char* escape_non_printable_full(const char *str, size_t console_width, bool eight_bit) {
         if (eight_bit)
                 return xescape_full(str, "", console_width, true);
         else
                 return utf8_escape_non_printable_full(str, console_width);
 }
 
-char *octescape(const char *s, size_t len) {
+char* octescape(const char *s, size_t len) {
         char *r, *t;
         const char *f;
 
@@ -462,7 +462,7 @@ char *octescape(const char *s, size_t len) {
 
 }
 
-static char *strcpy_backslash_escaped(char *t, const char *s, const char *bad, bool escape_tab_nl) {
+static char* strcpy_backslash_escaped(char *t, const char *s, const char *bad, bool escape_tab_nl) {
         assert(bad);
 
         for (; *s; s++) {
@@ -481,7 +481,7 @@ static char *strcpy_backslash_escaped(char *t, const char *s, const char *bad, b
         return t;
 }
 
-char *shell_escape(const char *s, const char *bad) {
+char* shell_escape(const char *s, const char *bad) {
         char *r, *t;
 
         r = new(char, strlen(s)*2+1);

src/basic/escape.h

@@ -43,8 +43,8 @@ typedef enum EscapeStyle {
                                         * syntax (a string enclosed in $'') instead of plain quotes. */
 } EscapeStyle;
 
-char *cescape(const char *s);
-char *cescape_length(const char *s, size_t n);
+char* cescape(const char *s);
+char* cescape_length(const char *s, size_t n);
 int cescape_char(char c, char *buf);
 
 int cunescape_length_with_prefix(const char *s, size_t length, const char *prefix, UnescapeFlags flags, char **ret);
@@ -56,12 +56,12 @@ static inline int cunescape(const char *s, UnescapeFlags flags, char **ret) {
 }
 int cunescape_one(const char *p, size_t length, char32_t *ret, bool *eight_bit, bool accept_nul);
 
-char *xescape_full(const char *s, const char *bad, size_t console_width, bool eight_bits);
-static inline char *xescape(const char *s, const char *bad) {
+char* xescape_full(const char *s, const char *bad, size_t console_width, bool eight_bits);
+static inline char* xescape(const char *s, const char *bad) {
         return xescape_full(s, bad, SIZE_MAX, false);
 }
-char *octescape(const char *s, size_t len);
-char *escape_non_printable_full(const char *str, size_t console_width, bool eight_bit);
+char* octescape(const char *s, size_t len);
+char* escape_non_printable_full(const char *str, size_t console_width, bool eight_bit);
 
-char *shell_escape(const char *s, const char *bad);
+char* shell_escape(const char *s, const char *bad);
 char* shell_maybe_quote(const char *s, EscapeStyle style);

src/basic/utf8.c

@@ -150,18 +150,22 @@ bool utf8_is_printable_newline(const char* str, size_t length, bool allow_newlin
         return true;
 }
 
-char *utf8_is_valid(const char *str) {
-        const char *p;
+char *utf8_is_valid_n(const char *str, size_t len_bytes) {
+        /* Check if the string is composed of valid utf8 characters. If length len_bytes is given, stop after
+         * len_bytes. Otherwise, stop at NUL. */
 
         assert(str);
 
-        p = str;
-        while (*p) {
+        for (const char *p = str; len_bytes != (size_t) -1 ? (size_t) (p - str) < len_bytes : *p != '\0'; ) {
                 int len;
 
-                len = utf8_encoded_valid_unichar(p, (size_t) -1);
-                if (len < 0)
-                        return NULL;
+                if (_unlikely_(*p == '\0') && len_bytes != (size_t) -1)
+                        return NULL; /* embedded NUL */
+
+                len = utf8_encoded_valid_unichar(p,
+                                                 len_bytes != (size_t) -1 ? len_bytes - (p - str) : (size_t) -1);
+                if (_unlikely_(len < 0))
+                        return NULL; /* invalid character */
 
                 p += len;
         }

src/basic/utf8.h

@@ -14,7 +14,10 @@
 
 bool unichar_is_valid(char32_t c);
 
-char *utf8_is_valid(const char *s) _pure_;
+char *utf8_is_valid_n(const char *str, size_t len_bytes) _pure_;
+static inline char *utf8_is_valid(const char *s) {
+        return utf8_is_valid_n(s, (size_t) -1);
+}
 char *ascii_is_valid(const char *s) _pure_;
 char *ascii_is_valid_n(const char *str, size_t len);