modify unescape_byte

datafusion/sql/src/utils.rs

@@ -273,11 +273,11 @@ pub(crate) fn unescape(s: &str) -> Option<String> {
                 'n' => '\n',
                 'r' => '\r',
                 't' => '\t',
-                '\'' | '\"' | '\\' | '/' => c,
                 'u' => return_if_none!(unescape_unicode_16(&mut queue)),
                 'U' => return_if_none!(unescape_unicode_32(&mut queue)),
                 'x' => return_if_none!(unescape_byte(&mut queue)),
-                maybe_octal => return_if_none!(unescape_octal(maybe_octal, &mut queue)),
+                c if c.is_digit(8) => return_if_none!(unescape_octal(c, &mut queue)),
+                c => c,
             },
             None => return None,
         };
@@ -289,20 +289,37 @@ pub(crate) fn unescape(s: &str) -> Option<String> {
 
 // Hexadecimal byte value. \xh, \xhh (h = 0–9, A–F)
 fn unescape_byte(queue: &mut VecDeque<char>) -> Option<char> {
-    unescape_hexadecimal::<2>(queue)
+    let mut s = String::new();
+    let mut try_get_next = || match queue.front() {
+        Some(c) if c.is_ascii_hexdigit() => queue.pop_front(),
+        _ => None,
+    };
+
+    for _ in 0..2 {
+        match try_get_next() {
+            Some(c) => s.push(c),
+            None => break,
+        }
+    }
+
+    if s.is_empty() {
+        return Some('x');
+    }
+
+    to_char::<16>(&s)
 }
 
 // 16-bit hexadecimal Unicode character value. \uxxxx (x = 0–9, A–F)
 fn unescape_unicode_16(queue: &mut VecDeque<char>) -> Option<char> {
-    unescape_hexadecimal::<4>(queue)
+    unescape_unicode::<4>(queue)
 }
 
 // 32-bit hexadecimal Unicode character value. \Uxxxxxxxx (x = 0–9, A–F)
 fn unescape_unicode_32(queue: &mut VecDeque<char>) -> Option<char> {
-    unescape_hexadecimal::<8>(queue)
+    unescape_unicode::<8>(queue)
 }
 
-fn unescape_hexadecimal<const NUM: usize>(queue: &mut VecDeque<char>) -> Option<char> {
+fn unescape_unicode<const NUM: usize>(queue: &mut VecDeque<char>) -> Option<char> {
     let mut s = String::new();
     for _ in 0..NUM {
         s.push(return_if_none!(queue.pop_front()));
@@ -312,12 +329,8 @@ fn unescape_hexadecimal<const NUM: usize>(queue: &mut VecDeque<char>) -> Option<
 
 // Octal byte value. \o, \oo, \ooo (o = 0–7)
 fn unescape_octal(c: char, queue: &mut VecDeque<char>) -> Option<char> {
-    if !c.is_digit(8) {
-        return None;
-    }
-
     let mut s = String::new();
-    let mut try_get_next = || match queue.get(0) {
+    let mut try_get_next = || match queue.front() {
         Some(c) if c.is_digit(8) => queue.pop_front(),
         _ => None,
     };
@@ -345,7 +358,7 @@ fn unescape_octal(c: char, queue: &mut VecDeque<char>) -> Option<char> {
 
 #[inline]
 fn to_char<const RADIX: u32>(s: &str) -> Option<char> {
-    match u32::from_str_radix(&s, RADIX) {
+    match u32::from_str_radix(s, RADIX) {
         Err(_) => None,
         Ok(n) => char::from_u32(n),
     }
@@ -368,18 +381,23 @@ mod tests {
         check_unescape(r"\/", Some("/"));
         check_unescape(r"/", Some("/"));
         check_unescape(r"\\", Some("\\"));
-        check_unescape(r"\x", None);
 
         // 16 and 32-bit hexadecimal Unicode character value
         check_unescape(r"\u4c91", Some("\u{4c91}"));
         check_unescape(r"\u4c916", Some("\u{4c91}6"));
         check_unescape(r"\U0010FFFF", Some("\u{10FFFF}"));
+        check_unescape(r"\u", None);
+        check_unescape(r"\U", None);
         check_unescape(r"\U1010FFFF", None);
 
         // hexadecimal byte value
         check_unescape(r"\xCAD", Some("\u{00ca}D"));
         check_unescape(r"\xA9", Some("\u{00a9}"));
         check_unescape(r"\x4B", Some("\u{004b}"));
+        check_unescape(r"\x4", Some("\u{0004}"));
+        check_unescape(r"\x4L", Some("\u{0004}L"));
+        check_unescape(r"\x", Some("x"));
+        check_unescape(r"\xP", Some("xP"));
 
         // octal byte value
         check_unescape(r"\1", Some("\u{0001}"));
@@ -389,6 +407,8 @@ mod tests {
         check_unescape(r"\4", Some("\u{0004}"));
         check_unescape(r"\45", Some("\u{0025}"));
         check_unescape(r"\450", Some("\u{0025}0"));
-        check_unescape(r"\9", None);
+
+        // others
+        check_unescape(r"\9", Some("9"));
     }
 }