Correct Regular Expressions Behavior Related to Annex B (#58320)

Co-authored-by: Ron Buckton <[email protected]>
microsoft · May 29, 2024 · 0cc62c3 · 0cc62c3
1 parent fa58c61
commit 0cc62c3
Show file tree

Hide file tree

Showing 20 changed files with 955 additions and 131 deletions.
diff --git a/src/compiler/diagnosticMessages.json b/src/compiler/diagnosticMessages.json
@@ -1789,14 +1789,22 @@
         "category": "Error",
         "code": 1533
     },
-    "This backreference is invalid because the containing regular expression contains no capturing groups.": {
+    "This backreference refers to a group that does not exist. There are no capturing groups in this regular expression.": {
         "category": "Error",
         "code": 1534
     },
     "This character cannot be escaped in a regular expression.": {
         "category": "Error",
         "code": 1535
     },
+    "Octal escape sequences and backreferences are not allowed in a character class. If this was intended as an escape sequence, use the syntax '{0}' instead.": {
+        "category": "Error",
+        "code": 1536
+    },
+    "Decimal escape sequences and backreferences are not allowed in a character class.": {
+        "category": "Error",
+        "code": 1537
+    },
 
     "The types of '{0}' are incompatible between these types.": {
         "category": "Error",

diff --git a/src/compiler/scanner.ts b/src/compiler/scanner.ts
diff --git a/tests/baselines/reference/regularExpressionAnnexB.errors.txt b/tests/baselines/reference/regularExpressionAnnexB.errors.txt
@@ -0,0 +1,269 @@
+regularExpressionAnnexB.ts(2,8): error TS1125: Hexadecimal digit expected.
+regularExpressionAnnexB.ts(2,22): error TS1125: Hexadecimal digit expected.
+regularExpressionAnnexB.ts(2,28): error TS1125: Hexadecimal digit expected.
+regularExpressionAnnexB.ts(3,9): error TS1125: Hexadecimal digit expected.
+regularExpressionAnnexB.ts(3,23): error TS1125: Hexadecimal digit expected.
+regularExpressionAnnexB.ts(3,29): error TS1125: Hexadecimal digit expected.
+regularExpressionAnnexB.ts(7,4): error TS1535: This character cannot be escaped in a regular expression.
+regularExpressionAnnexB.ts(7,8): error TS1125: Hexadecimal digit expected.
+regularExpressionAnnexB.ts(7,10): error TS1512: '\c' must be followed by an ASCII letter.
+regularExpressionAnnexB.ts(7,12): error TS1510: '\k' must be followed by a capturing group name enclosed in angle brackets.
+regularExpressionAnnexB.ts(7,14): error TS1535: This character cannot be escaped in a regular expression.
+regularExpressionAnnexB.ts(7,18): error TS1535: This character cannot be escaped in a regular expression.
+regularExpressionAnnexB.ts(7,22): error TS1125: Hexadecimal digit expected.
+regularExpressionAnnexB.ts(7,24): error TS1535: This character cannot be escaped in a regular expression.
+regularExpressionAnnexB.ts(7,28): error TS1125: Hexadecimal digit expected.
+regularExpressionAnnexB.ts(7,30): error TS1531: '\p' must be followed by a Unicode property value expression enclosed in braces.
+regularExpressionAnnexB.ts(8,5): error TS1535: This character cannot be escaped in a regular expression.
+regularExpressionAnnexB.ts(8,9): error TS1125: Hexadecimal digit expected.
+regularExpressionAnnexB.ts(8,11): error TS1512: '\c' must be followed by an ASCII letter.
+regularExpressionAnnexB.ts(8,13): error TS1535: This character cannot be escaped in a regular expression.
+regularExpressionAnnexB.ts(8,15): error TS1535: This character cannot be escaped in a regular expression.
+regularExpressionAnnexB.ts(8,19): error TS1535: This character cannot be escaped in a regular expression.
+regularExpressionAnnexB.ts(8,23): error TS1125: Hexadecimal digit expected.
+regularExpressionAnnexB.ts(8,25): error TS1535: This character cannot be escaped in a regular expression.
+regularExpressionAnnexB.ts(8,29): error TS1125: Hexadecimal digit expected.
+regularExpressionAnnexB.ts(8,31): error TS1531: '\p' must be followed by a Unicode property value expression enclosed in braces.
+regularExpressionAnnexB.ts(9,4): error TS1531: '\P' must be followed by a Unicode property value expression enclosed in braces.
+regularExpressionAnnexB.ts(9,7): error TS1531: '\P' must be followed by a Unicode property value expression enclosed in braces.
+regularExpressionAnnexB.ts(9,9): error TS1516: A character class range must not be bounded by another character class.
+regularExpressionAnnexB.ts(23,4): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(23,8): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(24,4): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(24,9): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(25,4): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(25,10): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(26,4): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(26,5): error TS1506: Numbers out of order in quantifier.
+regularExpressionAnnexB.ts(26,10): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(29,4): error TS1508: Unexpected '{'. Did you mean to escape it with backslash?
+regularExpressionAnnexB.ts(30,4): error TS1508: Unexpected '{'. Did you mean to escape it with backslash?
+regularExpressionAnnexB.ts(31,4): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(31,5): error TS1505: Incomplete quantifier. Digit expected.
+regularExpressionAnnexB.ts(31,7): error TS1005: '}' expected.
+regularExpressionAnnexB.ts(31,8): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(32,4): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(32,6): error TS1005: '}' expected.
+regularExpressionAnnexB.ts(32,7): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(33,4): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(33,7): error TS1005: '}' expected.
+regularExpressionAnnexB.ts(33,8): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(34,4): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(34,8): error TS1005: '}' expected.
+regularExpressionAnnexB.ts(34,9): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(35,4): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(35,5): error TS1506: Numbers out of order in quantifier.
+regularExpressionAnnexB.ts(35,8): error TS1005: '}' expected.
+regularExpressionAnnexB.ts(35,9): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(36,4): error TS1508: Unexpected '{'. Did you mean to escape it with backslash?
+regularExpressionAnnexB.ts(36,5): error TS1508: Unexpected '}'. Did you mean to escape it with backslash?
+regularExpressionAnnexB.ts(37,4): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(37,5): error TS1505: Incomplete quantifier. Digit expected.
+regularExpressionAnnexB.ts(37,8): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(38,4): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(38,5): error TS1505: Incomplete quantifier. Digit expected.
+regularExpressionAnnexB.ts(38,9): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(39,4): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(39,8): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(40,4): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(40,9): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(41,4): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(41,10): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(42,4): error TS1507: There is nothing available for repetition.
+regularExpressionAnnexB.ts(42,5): error TS1506: Numbers out of order in quantifier.
+regularExpressionAnnexB.ts(42,10): error TS1507: There is nothing available for repetition.
+
+
+==== regularExpressionAnnexB.ts (74 errors) ====
+    const regexes: RegExp[] = [
+      /\q\u\i\c\k\_\f\o\x\-\j\u\m\p\s/,
+
+!!! error TS1125: Hexadecimal digit expected.
+
+!!! error TS1125: Hexadecimal digit expected.
+
+!!! error TS1125: Hexadecimal digit expected.
+      /[\q\u\i\c\k\_\f\o\x\-\j\u\m\p\s]/,
+
+!!! error TS1125: Hexadecimal digit expected.
+
+!!! error TS1125: Hexadecimal digit expected.
+
+!!! error TS1125: Hexadecimal digit expected.
+      /\P[\P\w-_]/,
+
+      // Compare to
+      /\q\u\i\c\k\_\f\o\x\-\j\u\m\p\s/u,
+       ~~
+!!! error TS1535: This character cannot be escaped in a regular expression.
+
+!!! error TS1125: Hexadecimal digit expected.
+             ~~
+!!! error TS1512: '\c' must be followed by an ASCII letter.
+               ~~
+!!! error TS1510: '\k' must be followed by a capturing group name enclosed in angle brackets.
+                 ~~
+!!! error TS1535: This character cannot be escaped in a regular expression.
+                     ~~
+!!! error TS1535: This character cannot be escaped in a regular expression.
+
+!!! error TS1125: Hexadecimal digit expected.
+                           ~~
+!!! error TS1535: This character cannot be escaped in a regular expression.
+
+!!! error TS1125: Hexadecimal digit expected.
+                                 ~~
+!!! error TS1531: '\p' must be followed by a Unicode property value expression enclosed in braces.
+      /[\q\u\i\c\k\_\f\o\x\-\j\u\m\p\s]/u,
+        ~~
+!!! error TS1535: This character cannot be escaped in a regular expression.
+
+!!! error TS1125: Hexadecimal digit expected.
+              ~~
+!!! error TS1512: '\c' must be followed by an ASCII letter.
+                ~~
+!!! error TS1535: This character cannot be escaped in a regular expression.
+                  ~~
+!!! error TS1535: This character cannot be escaped in a regular expression.
+                      ~~
+!!! error TS1535: This character cannot be escaped in a regular expression.
+
+!!! error TS1125: Hexadecimal digit expected.
+                            ~~
+!!! error TS1535: This character cannot be escaped in a regular expression.
+
+!!! error TS1125: Hexadecimal digit expected.
+                                  ~~
+!!! error TS1531: '\p' must be followed by a Unicode property value expression enclosed in braces.
+      /\P[\P\w-_]/u,
+       ~~
+!!! error TS1531: '\P' must be followed by a Unicode property value expression enclosed in braces.
+          ~~
+!!! error TS1531: '\P' must be followed by a Unicode property value expression enclosed in braces.
+            ~~
+!!! error TS1516: A character class range must not be bounded by another character class.
+    ];
+
+    const regexesWithBraces: RegExp[] = [
+      /{??/,
+      /{,??/,
+      /{,1??/,
+      /{1??/,
+      /{1,??/,
+      /{1,2??/,
+      /{2,1??/,
+      /{}??/,
+      /{,}??/,
+      /{,1}??/,
+      /{1}??/,
+       ~~~~
+!!! error TS1507: There is nothing available for repetition.
+           ~
+!!! error TS1507: There is nothing available for repetition.
+      /{1,}??/,
+       ~~~~~
+!!! error TS1507: There is nothing available for repetition.
+            ~
+!!! error TS1507: There is nothing available for repetition.
+      /{1,2}??/,
+       ~~~~~~
+!!! error TS1507: There is nothing available for repetition.
+             ~
+!!! error TS1507: There is nothing available for repetition.
+      /{2,1}??/,
+       ~~~~~~
+!!! error TS1507: There is nothing available for repetition.
+        ~~~
+!!! error TS1506: Numbers out of order in quantifier.
+             ~
+!!! error TS1507: There is nothing available for repetition.
+
+      // Compare to
+      /{??/u,
+       ~
+!!! error TS1508: Unexpected '{'. Did you mean to escape it with backslash?
+      /{,??/u,
+       ~
+!!! error TS1508: Unexpected '{'. Did you mean to escape it with backslash?
+      /{,1??/u,
+       ~~~~
+!!! error TS1507: There is nothing available for repetition.
+
+!!! error TS1505: Incomplete quantifier. Digit expected.
+
+!!! error TS1005: '}' expected.
+           ~
+!!! error TS1507: There is nothing available for repetition.
+      /{1??/u,
+       ~~~
+!!! error TS1507: There is nothing available for repetition.
+
+!!! error TS1005: '}' expected.
+          ~
+!!! error TS1507: There is nothing available for repetition.
+      /{1,??/u,
+       ~~~~
+!!! error TS1507: There is nothing available for repetition.
+
+!!! error TS1005: '}' expected.
+           ~
+!!! error TS1507: There is nothing available for repetition.
+      /{1,2??/u,
+       ~~~~~
+!!! error TS1507: There is nothing available for repetition.
+
+!!! error TS1005: '}' expected.
+            ~
+!!! error TS1507: There is nothing available for repetition.
+      /{2,1??/u,
+       ~~~~~
+!!! error TS1507: There is nothing available for repetition.
+        ~~~
+!!! error TS1506: Numbers out of order in quantifier.
+
+!!! error TS1005: '}' expected.
+            ~
+!!! error TS1507: There is nothing available for repetition.
+      /{}??/u,
+       ~
+!!! error TS1508: Unexpected '{'. Did you mean to escape it with backslash?
+        ~
+!!! error TS1508: Unexpected '}'. Did you mean to escape it with backslash?
+      /{,}??/u,
+       ~~~~
+!!! error TS1507: There is nothing available for repetition.
+
+!!! error TS1505: Incomplete quantifier. Digit expected.
+           ~
+!!! error TS1507: There is nothing available for repetition.
+      /{,1}??/u,
+       ~~~~~
+!!! error TS1507: There is nothing available for repetition.
+
+!!! error TS1505: Incomplete quantifier. Digit expected.
+            ~
+!!! error TS1507: There is nothing available for repetition.
+      /{1}??/u,
+       ~~~~
+!!! error TS1507: There is nothing available for repetition.
+           ~
+!!! error TS1507: There is nothing available for repetition.
+      /{1,}??/u,
+       ~~~~~
+!!! error TS1507: There is nothing available for repetition.
+            ~
+!!! error TS1507: There is nothing available for repetition.
+      /{1,2}??/u,
+       ~~~~~~
+!!! error TS1507: There is nothing available for repetition.
+             ~
+!!! error TS1507: There is nothing available for repetition.
+      /{2,1}??/u,
+       ~~~~~~
+!!! error TS1507: There is nothing available for repetition.
+        ~~~
+!!! error TS1506: Numbers out of order in quantifier.
+             ~
+!!! error TS1507: There is nothing available for repetition.
+    ];
+
diff --git a/tests/baselines/reference/regularExpressionAnnexB.js b/tests/baselines/reference/regularExpressionAnnexB.js
@@ -0,0 +1,89 @@
+//// [tests/cases/compiler/regularExpressionAnnexB.ts] ////
+
+//// [regularExpressionAnnexB.ts]
+const regexes: RegExp[] = [
+  /\q\u\i\c\k\_\f\o\x\-\j\u\m\p\s/,
+  /[\q\u\i\c\k\_\f\o\x\-\j\u\m\p\s]/,
+  /\P[\P\w-_]/,
+
+  // Compare to
+  /\q\u\i\c\k\_\f\o\x\-\j\u\m\p\s/u,
+  /[\q\u\i\c\k\_\f\o\x\-\j\u\m\p\s]/u,
+  /\P[\P\w-_]/u,
+];
+
+const regexesWithBraces: RegExp[] = [
+  /{??/,
+  /{,??/,
+  /{,1??/,
+  /{1??/,
+  /{1,??/,
+  /{1,2??/,
+  /{2,1??/,
+  /{}??/,
+  /{,}??/,
+  /{,1}??/,
+  /{1}??/,
+  /{1,}??/,
+  /{1,2}??/,
+  /{2,1}??/,
+
+  // Compare to
+  /{??/u,
+  /{,??/u,
+  /{,1??/u,
+  /{1??/u,
+  /{1,??/u,
+  /{1,2??/u,
+  /{2,1??/u,
+  /{}??/u,
+  /{,}??/u,
+  /{,1}??/u,
+  /{1}??/u,
+  /{1,}??/u,
+  /{1,2}??/u,
+  /{2,1}??/u,
+];
+
+
+//// [regularExpressionAnnexB.js]
+const regexes = [
+    /\q\u\i\c\k\_\f\o\x\-\j\u\m\p\s/,
+    /[\q\u\i\c\k\_\f\o\x\-\j\u\m\p\s]/,
+    /\P[\P\w-_]/,
+    // Compare to
+    /\q\u\i\c\k\_\f\o\x\-\j\u\m\p\s/u,
+    /[\q\u\i\c\k\_\f\o\x\-\j\u\m\p\s]/u,
+    /\P[\P\w-_]/u,
+];
+const regexesWithBraces = [
+    /{??/,
+    /{,??/,
+    /{,1??/,
+    /{1??/,
+    /{1,??/,
+    /{1,2??/,
+    /{2,1??/,
+    /{}??/,
+    /{,}??/,
+    /{,1}??/,
+    /{1}??/,
+    /{1,}??/,
+    /{1,2}??/,
+    /{2,1}??/,
+    // Compare to
+    /{??/u,
+    /{,??/u,
+    /{,1??/u,
+    /{1??/u,
+    /{1,??/u,
+    /{1,2??/u,
+    /{2,1??/u,
+    /{}??/u,
+    /{,}??/u,
+    /{,1}??/u,
+    /{1}??/u,
+    /{1,}??/u,
+    /{1,2}??/u,
+    /{2,1}??/u,
+];