Support UESCAPE in PostgreSQL unicode strings & identifiers

nene · Feb 11, 2024 · 6a8712a · 6a8712a
1 parent f715a15
commit 6a8712a
Show file tree

Hide file tree

Showing 4 changed files with 63 additions and 2 deletions.
diff --git a/src/cst/Expr.ts b/src/cst/Expr.ts
@@ -144,6 +144,7 @@ type KeywordOperator =
   | [Keyword<"MEMBER">, Keyword<"OF">] // MySQL
   | [Keyword<"SOUNDS">, Keyword<"LIKE">] // MySQL
   | Keyword<"ESCAPE"> // SQLite, MySQL, PostgreSQL
+  | Keyword<"UESCAPE"> // PostgreSQL
   // Timezone
   | [Keyword<"AT">, Keyword<"TIME">, Keyword<"ZONE">]; // PostgreSQL
 

diff --git a/src/parser.pegjs b/src/parser.pegjs
@@ -6962,7 +6962,15 @@ quoted_ident
   / (&sqlite / &mysql) ident:backticks_quoted_ident_qq { return ident; }
   / &bigquery ident:(bigquery_quoted_member_expr / backticks_quoted_ident_bs) { return ident; }
   / (&sqlite / &postgres) str:string_literal_double_quoted_qq { return loc(createIdentifier(str.text, str.value)); }
-  / &postgres str:string_literal_unicode_double_quoted_qq { return loc(createIdentifier(str.text, str.value)); }
+  / &postgres ident:postgres_unicode_ident { return ident; }
+
+postgres_unicode_ident
+  = head:ident_unicode_double_quoted_qq tail:(__ UESCAPE __ string_literal_single_quoted_qq)|0..1| {
+    return createBinaryExprChain(head, tail);
+  }
+
+ident_unicode_double_quoted_qq
+  = str:string_literal_unicode_double_quoted_qq { return loc(createIdentifier(str.text, str.value)); }
 
 backticks_quoted_ident_qq
   = "`" chars:([^`] / escaped_backtick_qq)+ "`" { return loc(createIdentifier(text(), chars.join(""))); }
@@ -7178,7 +7186,7 @@ string_literal_plain
       string_literal_single_quoted_qq
     / string_literal_dollar_quoted
     / string_literal_e_single_quoted_bs
-    / string_literal_unicode_single_quoted_qq) { return s; }
+    / postgres_unicode_string) { return s; }
 
 mysql_string_literal_chain
   = head:mysql_string_literal_plain tail:(__ mysql_string_literal_plain)* {
@@ -7200,6 +7208,11 @@ charset_introducer
 charset_name
   = ident_name_basic { return text(); }
 
+postgres_unicode_string
+  = head:string_literal_unicode_single_quoted_qq tail:(__ UESCAPE __ string_literal_single_quoted_qq)|0..1| {
+    return createBinaryExprChain(head, tail);
+  }
+
 string_literal_single_quoted_qq_bs // with repeated quote or backslash for escaping
   = "'" chars:([^'\\] / escaped_single_quote_qq / backslash_escape)* "'" {
     return loc({
@@ -8235,6 +8248,7 @@ TRUE                = kw:"TRUE"i                !ident_part { return loc(createK
 TRUNCATE            = kw:"TRUNCATE"i            !ident_part { return loc(createKeyword(kw)); }
 TUESDAY             = kw:"TUESDAY"i             !ident_part { return loc(createKeyword(kw)); }
 TYPE                = kw:"TYPE"i                !ident_part { return loc(createKeyword(kw)); }
+UESCAPE             = kw:"UESCAPE"i             !ident_part { return loc(createKeyword(kw)); }
 UNBOUNDED           = kw:"UNBOUNDED"i           !ident_part { return loc(createKeyword(kw)); }
 UNDEFINED           = kw:"UNDEFINED"i           !ident_part { return loc(createKeyword(kw)); }
 UNION               = kw:"UNION"i               !ident_part { return loc(createKeyword(kw)); }

diff --git a/test/identifier.test.ts b/test/identifier.test.ts
@@ -216,5 +216,28 @@ describe("identifier", () => {
     it(`supports ""-escaping in unicode identifer`, () => {
       testExpr(`U&"my "" why"`);
     });
+
+    it("supports custom unicode escape character on unicode identifier", () => {
+      expect(parseExpr(`U&"!0441!043B!043E!043D" UESCAPE '!'`)).toMatchInlineSnapshot(`
+        {
+          "left": {
+            "name": "!0441!043B!043E!043D",
+            "text": "U&"!0441!043B!043E!043D"",
+            "type": "identifier",
+          },
+          "operator": {
+            "name": "UESCAPE",
+            "text": "UESCAPE",
+            "type": "keyword",
+          },
+          "right": {
+            "text": "'!'",
+            "type": "string_literal",
+            "value": "!",
+          },
+          "type": "binary_expr",
+        }
+      `);
+    });
   });
 });
diff --git a/test/literal/string.test.ts b/test/literal/string.test.ts
@@ -340,5 +340,28 @@ describe("string literal", () => {
         }
       `);
     });
+
+    it("supports custom unicode escape character on unicode strings", () => {
+      expect(parseExpr(`U&'!0441!043B!043E!043D' UESCAPE '!'`)).toMatchInlineSnapshot(`
+        {
+          "left": {
+            "text": "U&'!0441!043B!043E!043D'",
+            "type": "string_literal",
+            "value": "!0441!043B!043E!043D",
+          },
+          "operator": {
+            "name": "UESCAPE",
+            "text": "UESCAPE",
+            "type": "keyword",
+          },
+          "right": {
+            "text": "'!'",
+            "type": "string_literal",
+            "value": "!",
+          },
+          "type": "binary_expr",
+        }
+      `);
+    });
   });
 });