word breaking test case

Author: milseman

Tests/RegexTests/MatchTests.swift

@@ -1683,6 +1683,38 @@ extension RegexTests {
     firstMatchTest(#"\b÷\b"#, input: "3 ÷ 3 = 1", match: "÷")
   }
 
+  func testLevel2WordBoundaries_2() throws {
+    let re = try! Regex(#"\bA\b"#)
+
+    // OK: Space after colon works (note "NO ENTRY" emoji has U+FE0F after it)
+    try print(re.firstMatch(in: "⛔️: X ") == nil) // => true
+    try print(re.firstMatch(in: "日\u{FE0F}: X ") == nil) // => true
+    try print(re.firstMatch(in: "👨‍👨‍👧‍👦\u{FE0F}: X ") == nil) // => true
+
+    // OK: No U+FE0F to left of colon, no space after colon
+    try print(re.firstMatch(in: "Z:X ") == nil) // => true
+
+    // OK: Alphabetic followed by U+FE0F to left of colon
+    try print(re.firstMatch(in: "Z\u{FE0F}:X ") == nil) // => true
+    try print(re.firstMatch(in: "è\u{FE0F}:X ") == nil) // => true
+
+    // OK: Non-alphabetic without U+FE0F
+    try print(re.firstMatch(in: "日:X ") == nil) // => true
+    try print(re.firstMatch(in: "👨‍👨‍👧‍👦:X ") == nil) // => true
+
+    print("all the rest hang...")
+
+    // HANGS: Non-alphabetic followed by U+FE0F to left of colon (note "NO ENTRY" emoji has U+FE0F)
+    try print(re.firstMatch(in: "日\u{FE0F}:X ") == nil) // => HANGS!
+    try print(re.firstMatch(in: "👨‍👨‍👧‍👦\u{FE0F}:X ") == nil) // => HANGS!
+    try print(re.firstMatch(in: "⛔️:X ") == nil) // => HANGS!
+
+    // HANGS: Other MidLetters as well, not just colon (note "NO ENTRY" emoji has U+FE0F)
+    try print(re.firstMatch(in: "⛔️·X ") == nil) // => HANGS!
+    try print(re.firstMatch(in: "⛔️：X ") == nil) // => HANGS!
+
+  }
+
   func testMatchGroups() {
     // Must have new stdlib for character class ranges and word boundaries.
     guard ensureNewStdlib() else { return }
@@ -2717,4 +2749,14 @@ extension RegexTests {
       XCTAssertNotNil(str.wholeMatch(of: possessiveRegex))
     }
   }
+
+  func testFoo() throws {
+    let re = try Regex(#"^[\u{0000}-\u{024F}]+$"#)
+
+    XCTAssertNotNil("aaa".wholeMatch(of: re))
+    XCTAssertNotNil("aa\u{301}a".wholeMatch(of: re))
+    XCTAssertNil("aa\u{301}\u{302}a".wholeMatch(of: re))
+  }
 }
+
+