Swift: Fix parse failures due to mode flags.

geoffw0 · geoffw0 · commit f50345659e78 · 2023-07-10T19:14:42.000+01:00
diff --git a/swift/ql/lib/codeql/swift/regex/internal/ParseRegex.qll b/swift/ql/lib/codeql/swift/regex/internal/ParseRegex.qll
@@ -708,6 +708,8 @@ abstract class RegExp extends Expr {
     this.commentGroupStart(start, end)
     or
     this.simpleGroupStart(start, end)
+    or
+    this.flagGroupStart(start, end, _)
   }
 
   /** Matches the start of a non-capturing group, e.g. `(?:` */
diff --git a/swift/ql/test/library-tests/regex/parse.expected b/swift/ql/test/library-tests/regex/parse.expected
@@ -1618,7 +1618,12 @@ redos_variants.swift:
 
 #  142| [RegExpConstant, RegExpNormalChar] !
 
-#  146| [RegExpConstant, RegExpNormalChar] s
+#  146| [RegExpZeroWidthMatch] (?s)
+
+#  146| [RegExpSequence] (?s)(.|\n)*!
+#-----| 0 -> [RegExpZeroWidthMatch] (?s)
+#-----| 1 -> [RegExpStar] (.|\n)*
+#-----| 2 -> [RegExpConstant, RegExpNormalChar] !
 
 #  146| [RegExpGroup] (.|\n)
 #-----| 0 -> [RegExpAlt] .|\n
@@ -6484,15 +6489,30 @@ regex.swift:
 
 #  205| [RegExpNamedCharacterProperty] [:aaaaa:]
 
-#  209| [RegExpConstant, RegExpNormalChar] i
+#  209| [RegExpZeroWidthMatch] (?i)
+
+#  209| [RegExpSequence] (?i)abc
+#-----| 0 -> [RegExpZeroWidthMatch] (?i)
+#-----| 1 -> [RegExpConstant, RegExpNormalChar] abc
 
 #  209| [RegExpConstant, RegExpNormalChar] abc
 
-#  210| [RegExpConstant, RegExpNormalChar] s
+#  210| [RegExpZeroWidthMatch] (?s)
+
+#  210| [RegExpSequence] (?s)abc
+#-----| 0 -> [RegExpZeroWidthMatch] (?s)
+#-----| 1 -> [RegExpConstant, RegExpNormalChar] abc
 
 #  210| [RegExpConstant, RegExpNormalChar] abc
 
-#  211| [RegExpConstant, RegExpNormalChar] is
+#  211| [RegExpGroup] (?is)
+#-----| 0 -> [RegExpConstant, RegExpNormalChar] s
+
+#  211| [RegExpSequence] (?is)abc
+#-----| 0 -> [RegExpGroup] (?is)
+#-----| 1 -> [RegExpConstant, RegExpNormalChar] abc
+
+#  211| [RegExpConstant, RegExpNormalChar] s
 
 #  211| [RegExpConstant, RegExpNormalChar] abc
 
diff --git a/swift/ql/test/library-tests/regex/redos_variants.swift b/swift/ql/test/library-tests/regex/redos_variants.swift
@@ -143,7 +143,7 @@ func myRegexpVariantsTests(myUrl: URL) throws {
 
     // BAD
     // attack string: "\n".repeat(100) + "."
-    _ = try Regex(#"(?s)(.|\n)*!"#).firstMatch(in: tainted) // $ hasParseFailure modes=DOTALL MISSING: redos-vulnerable=
+    _ = try Regex(#"(?s)(.|\n)*!"#).firstMatch(in: tainted) // $ modes=DOTALL redos-vulnerable=
 
     // GOOD
     _ = try Regex(#"([\w.]+)*"#).firstMatch(in: tainted)
diff --git a/swift/ql/test/library-tests/regex/regex.swift b/swift/ql/test/library-tests/regex/regex.swift
@@ -206,9 +206,9 @@ func myRegexpMethodsTests(b: Bool, str_unknown: String) throws {
 
 	// --- parse modes ---
 
-    _ = try Regex("(?i)abc").firstMatch(in: input) // $ input=input modes=IGNORECASE regex=(?i)abc SPURIOUS: hasParseFailure
-    _ = try Regex("(?s)abc").firstMatch(in: input) // $ input=input modes=DOTALL regex=(?s)abc SPURIOUS: $hasParseFailure
-    _ = try Regex("(?is)abc").firstMatch(in: input) // $ input=input regex=(?is)abc MISSING: modes="DOTALL | IGNORECASE" SPURIOUS: modes=IGNORECASE $hasParseFailure
+    _ = try Regex("(?i)abc").firstMatch(in: input) // $ input=input modes=IGNORECASE regex=(?i)abc
+    _ = try Regex("(?s)abc").firstMatch(in: input) // $ input=input modes=DOTALL regex=(?s)abc
+    _ = try Regex("(?is)abc").firstMatch(in: input) // $ input=input regex=(?is)abc MISSING: modes="DOTALL | IGNORECASE" SPURIOUS: modes=IGNORECASE
 
     _ = try Regex("abc").dotMatchesNewlines(true).firstMatch(in: input) // $ input=input regex=abc MISSING: modes=DOTALL
     _ = try Regex("abc").dotMatchesNewlines(false).firstMatch(in: input) // $ input=input regex=abc

Original file line number	Diff line number	Diff line change
`@@ -708,6 +708,8 @@ abstract class RegExp extends Expr {`
`708`	`708`	`this.commentGroupStart(start, end)`
`709`	`709`	`or`
`710`	`710`	`this.simpleGroupStart(start, end)`
	`711`	`+ or`
	`712`	`+ this.flagGroupStart(start, end, _)`
`711`	`713`	`}`
`712`	`714`
`713`	`715`	/** Matches the start of a non-capturing group, e.g. `(?:` */