Fix parsing of alternations in character classes

joefarebrother · joefarebrother · commit 49374b877af3 · 2022-05-04T15:41:37.000+01:00
diff --git a/java/ql/lib/semmle/code/java/regex/regex.qll b/java/ql/lib/semmle/code/java/regex/regex.qll
@@ -807,6 +807,7 @@ abstract class RegexString extends StringLiteral {
   }
 
   private predicate topLevel(int start, int end) {
+    not this.inCharSet(start) and
     this.subalternation(start, end, _) and
     not this.isOptionDivider(end)
   }
diff --git a/java/ql/test/library-tests/regex/parser/RegexParseTests.expected b/java/ql/test/library-tests/regex/parser/RegexParseTests.expected
@@ -139,3 +139,5 @@ parseFailures
 | Test.java:19:26:19:30 | \\077 | [RegExpConstant,RegExpEscape] |
 | Test.java:19:31:19:31 | 7 | [RegExpConstant,RegExpNormalChar] |
 | Test.java:19:32:19:37 | \u1337 | [RegExpConstant,RegExpNormalChar] |
+| Test.java:20:10:20:12 | [\|] | [RegExpCharacterClass] |
+| Test.java:20:11:20:11 | \| | [RegExpConstant,RegExpNormalChar] |
diff --git a/java/ql/test/library-tests/regex/parser/Test.java b/java/ql/test/library-tests/regex/parser/Test.java
@@ -17,7 +17,7 @@ class Test {
         "(?i)(?=a)(?!b)(?<=c)(?<!d)",
         "a||b|c(d|e|)f|g",
         "\\018\\033\\0377\\0777\u1337",
-
+        "[|]"
     };
 
     void test() {

Original file line number	Diff line number	Diff line change
`@@ -807,6 +807,7 @@ abstract class RegexString extends StringLiteral {`
`807`	`807`	`}`
`808`	`808`
`809`	`809`	`private predicate topLevel(int start, int end) {`
	`810`	`+ not this.inCharSet(start) and`
`810`	`811`	`this.subalternation(start, end, _) and`
`811`	`812`	`not this.isOptionDivider(end)`
`812`	`813`	`}`