Simplify octal handling

joefarebrother · joefarebrother · commit bc109521aa22 · 2022-05-04T15:41:34.000+01:00
diff --git a/java/ql/lib/semmle/code/java/regex/regex.qll b/java/ql/lib/semmle/code/java/regex/regex.qll
@@ -280,13 +280,14 @@ abstract class RegexString extends StringLiteral {
       or
       // octal value \0o, \0oo, or \0ooo. Max of 0377.
       this.getChar(start + 1) = "0" and
-      end in [start + 3 .. start + 5] and
-      forall(int i | i in [start + 1 .. end - 1] | this.isOctal(i)) and
-      (end = start + 5 implies this.getChar(start + 2) <= "3") and
-      not (
-        end < start + 5 and
-        this.isOctal(end) and
-        (end = start + 4 implies this.getChar(start + 2) <= "3")
+      this.isOctal(start + 2) and
+      (
+        if this.isOctal(start + 3)
+        then
+          if this.isOctal(start + 4) and this.getChar(start + 2) in ["0", "1", "2", "3"]
+          then end = start + 5
+          else end = start + 4
+        else end = start + 3
       )
       or
       // 16-bit hex value \uhhhh
diff --git a/java/ql/test/library-tests/regex/RegexParseTests.expected b/java/ql/test/library-tests/regex/RegexParseTests.expected
@@ -130,3 +130,10 @@ parseFailures
 | Test.java:18:18:18:18 | e | [RegExpConstant,RegExpNormalChar] |
 | Test.java:18:20:18:20 | f | [RegExpConstant,RegExpNormalChar] |
 | Test.java:18:22:18:22 | g | [RegExpConstant,RegExpNormalChar] |
+| Test.java:19:10:19:12 | \\01 | [RegExpConstant,RegExpEscape] |
+| Test.java:19:10:19:27 | \\018\\033\\0377\\0777 | [RegExpSequence] |
+| Test.java:19:13:19:13 | 8 | [RegExpConstant,RegExpNormalChar] |
+| Test.java:19:14:19:17 | \\033 | [RegExpConstant,RegExpEscape] |
+| Test.java:19:18:19:22 | \\0377 | [RegExpConstant,RegExpEscape] |
+| Test.java:19:23:19:26 | \\077 | [RegExpConstant,RegExpEscape] |
+| Test.java:19:27:19:27 | 7 | [RegExpConstant,RegExpNormalChar] |
diff --git a/java/ql/test/library-tests/regex/Test.java b/java/ql/test/library-tests/regex/Test.java
@@ -15,7 +15,8 @@ class Test {
         "(?>hi)(?<name>hell*?o*+)123\\k<name>",
         "a+b*c?d{2}e{3,4}f{,5}g{6,}h+?i*?j??k{7}?l{8,9}?m{,10}?n{11,}?o++p*+q?+r{12}+s{13,14}+t{,15}+u{16,}+",
         "(?i)(?=a)(?!b)(?<=c)(?<!d)",
-        "a|b|c(d|e)f|g"
+        "a|b|c(d|e)f|g",
+        "\\018\\033\\0377\\0777"
     };
 
     void test() {
