From b02635c220b1fd8ee684848914b869deec0a2b4c Mon Sep 17 00:00:00 2001
From: Runji Wang <wangrunji0408@163.com>
Date: Wed, 25 Oct 2023 12:30:29 +0800
Subject: [PATCH] fix parsing array literal

Signed-off-by: Runji Wang <wangrunji0408@163.com>
---
 src/expr/impl/src/scalar/cast.rs | 49 ++++++++++++++++++++------------
 1 file changed, 31 insertions(+), 18 deletions(-)

diff --git a/src/expr/impl/src/scalar/cast.rs b/src/expr/impl/src/scalar/cast.rs
index 707c943520066..02ac5a6f4be9a 100644
--- a/src/expr/impl/src/scalar/cast.rs
+++ b/src/expr/impl/src/scalar/cast.rs
@@ -177,6 +177,7 @@ pub fn str_to_bytea(elem: &str) -> Result<Box<[u8]>> {
 
 // TODO(nanderstabel): optimize for multidimensional List. Depth can be given as a parameter to this
 // function.
+// FIXME(runji): handle escape characters in double quoted strings.
 /// Takes a string input in the form of a comma-separated list enclosed in braces, and returns a
 /// vector of strings containing the list items.
 ///
@@ -184,8 +185,8 @@ pub fn str_to_bytea(elem: &str) -> Result<Box<[u8]>> {
 /// - "{1, 2, 3}" => ["1", "2", "3"]
 /// - "{1, {2, 3}}" => ["1", "{2, 3}"]
 /// - "{"1,2"}" => ["1,2"]
-/// - "{null, "null"}" => [None, "null"]
-fn unnest(input: &str) -> Result<Vec<Option<&str>>> {
+/// - "{null, NuLL, "null"}" => [None, None, "null"]
+fn parse_array_literal(input: &str) -> Result<Vec<Option<&str>>> {
     let trimmed = input.trim();
     if !trimmed.starts_with('{') || !trimmed.ends_with('}') {
         return Err(ExprError::Parse("Input must be braced".into()));
@@ -215,7 +216,8 @@ fn unnest(input: &str) -> Result<Vec<Option<&str>>> {
             },
             ',' if depth == 0 => {
                 items.push(match trimmed[start..i].trim() {
-                    "null" => None,
+                    "" => return Err(ExprError::Parse("Unexpected \"}\" character.".into())),
+                    s if s.eq_ignore_ascii_case("null") => None,
                     s if s.starts_with('"') && s.ends_with('"') => Some(&s[1..s.len() - 1]),
                     s => Some(s),
                 });
@@ -228,8 +230,8 @@ fn unnest(input: &str) -> Result<Vec<Option<&str>>> {
         return Err(eof());
     }
     match trimmed[start..].trim() {
-        "" => {}
-        "null" => items.push(None),
+        "" => return Err(ExprError::Parse("Unexpected \"}\" character.".into())),
+        s if s.eq_ignore_ascii_case("null") => items.push(None),
         s if s.starts_with('"') && s.ends_with('"') => items.push(Some(&s[1..s.len() - 1])),
         s => items.push(Some(s)),
     }
@@ -245,7 +247,7 @@ fn str_to_list(input: &str, ctx: &Context) -> Result<ListValue> {
     )
     .unwrap();
     let mut values = vec![];
-    for item in unnest(input)? {
+    for item in parse_array_literal(input)? {
         match item {
             Some(item) => {
                 let v = cast
@@ -365,35 +367,46 @@ mod tests {
     }
 
     #[test]
-    fn test_unnest() {
-        assert_eq!(unnest("{ }").unwrap(), vec![]);
+    fn test_parse_array_literal() {
+        assert_eq!(parse_array_literal("{ }").unwrap(), vec![]);
+        assert_eq!(parse_array_literal("{1 2}").unwrap(), vec![Some("1 2")]);
         assert_eq!(
-            unnest("{1, 2, 3}").unwrap(),
+            parse_array_literal("{1, 2, 3}").unwrap(),
             vec![Some("1"), Some("2"), Some("3")]
         );
-        assert_eq!(unnest(r#"{"1,2"}"#).unwrap(), vec![Some("1,2")]);
-        assert_eq!(unnest(r#"{1, ""}"#).unwrap(), vec![Some("1"), Some("")]);
-        assert_eq!(unnest(r#"{"\""}"#).unwrap(), vec![Some("1"), Some("\"")]);
         assert_eq!(
-            unnest(r#"{"null", null}"#).unwrap(),
-            vec![Some("null"), None]
+            parse_array_literal(r#"{"1,2"}"#).unwrap(),
+            vec![Some("1,2")]
         );
         assert_eq!(
-            unnest("{{1, 2, 3}, {4, 5, 6}}").unwrap(),
+            parse_array_literal(r#"{1, ""}"#).unwrap(),
+            vec![Some("1"), Some("")]
+        );
+        // FIXME: handle escape characters in double quoted strings.
+        // assert_eq!(parse_array_literal(r#"{"\""}"#).unwrap(), vec![Some("\"")]);
+        assert_eq!(
+            parse_array_literal(r#"{"null", "NULL", null, NuLL}"#).unwrap(),
+            vec![Some("null"), Some("NULL"), None, None]
+        );
+        assert_eq!(
+            parse_array_literal("{{1, 2, 3}, {4, 5, 6}}").unwrap(),
             vec![Some("{1, 2, 3}"), Some("{4, 5, 6}")]
         );
         assert_eq!(
-            unnest("{{{1, 2, 3}}, {{4, 5, 6}}}").unwrap(),
+            parse_array_literal("{{{1, 2, 3}}, {{4, 5, 6}}}").unwrap(),
             vec![Some("{{1, 2, 3}}"), Some("{{4, 5, 6}}")]
         );
         assert_eq!(
-            unnest("{{{1, 2, 3}, {4, 5, 6}}}").unwrap(),
+            parse_array_literal("{{{1, 2, 3}, {4, 5, 6}}}").unwrap(),
             vec![Some("{{1, 2, 3}, {4, 5, 6}}")]
         );
         assert_eq!(
-            unnest("{{{aa, bb, cc}, {dd, ee, ff}}}").unwrap(),
+            parse_array_literal("{{{aa, bb, cc}, {dd, ee, ff}}}").unwrap(),
             vec![Some("{{aa, bb, cc}, {dd, ee, ff}}")]
         );
+        // Unexpected "," character.
+        assert!(parse_array_literal("{1,}").is_err());
+        assert!(parse_array_literal("{1,,3}").is_err());
     }
 
     #[test]