Python: Avoid multiple results for toString

Author: yoff

python/ql/src/semmle/python/regex.qll

@@ -121,12 +121,6 @@ deprecated string mode_from_mode_object(Value obj) {
 abstract class RegexString extends Expr {
   RegexString() { (this instanceof Bytes or this instanceof Unicode) }
 
-  override string toString() {
-    result = this.(Bytes).getText()
-    or
-    result = this.(Unicode).getText()
-  }
-
   /**
    * Helper predicate for `char_set_start(int start, int end)`.
    *

python/ql/test/library-tests/regex/SubstructureTests.ql

@@ -13,7 +13,7 @@ class CharacterSetTest extends InlineExpectationsTest {
     exists(Regex re, int start, int end |
       re.charSet(start, end) and
       location = re.getLocation() and
-      element = re.toString().substring(start, end) and
+      element = re.getText().substring(start, end) and
       value = start + ":" + end and
       tag = "charSet"
     )
@@ -31,7 +31,7 @@ class CharacterRangeTest extends InlineExpectationsTest {
     exists(Regex re, int start, int lower_end, int upper_start, int end |
       re.charRange(_, start, lower_end, upper_start, end) and
       location = re.getLocation() and
-      element = re.toString().substring(start, end) and
+      element = re.getText().substring(start, end) and
       value = start + ":" + lower_end + "-" + upper_start + ":" + end and
       tag = "charRange"
     )
@@ -49,7 +49,7 @@ class EscapeTest extends InlineExpectationsTest {
     exists(Regex re, int start, int end |
       re.escapedCharacter(start, end) and
       location = re.getLocation() and
-      element = re.toString().substring(start, end) and
+      element = re.getText().substring(start, end) and
       value = start + ":" + end and
       tag = "escapedCharacter"
     )
@@ -67,7 +67,7 @@ class GroupTest extends InlineExpectationsTest {
     exists(Regex re, int start, int end |
       re.group(start, end) and
       location = re.getLocation() and
-      element = re.toString().substring(start, end) and
+      element = re.getText().substring(start, end) and
       value = start + ":" + end and
       tag = "group"
     )

python/ql/test/query-tests/Expressions/Regex/BackspaceEscape.expected

@@ -1,4 +1,2 @@
 | test.py:17:12:17:22 | Str | Backspace escape in regular expression at offset 1. |
-| test.py:17:12:17:22 | [\\b\\t ] | Backspace escape in regular expression at offset 1. |
-| test.py:19:12:19:28 | E\\d+\\b[ \\b\\t] | Backspace escape in regular expression at offset 8. |
 | test.py:19:12:19:28 | Str | Backspace escape in regular expression at offset 8. |

python/ql/test/query-tests/Expressions/Regex/DuplicateCharacterInSet.expected

@@ -1,6 +1,3 @@
 | test.py:46:12:46:18 | Str | This regular expression includes duplicate character 'A' in a set of characters. |
-| test.py:46:12:46:18 | [AA] | This regular expression includes duplicate character 'A' in a set of characters. |
 | test.py:47:12:47:19 | Str | This regular expression includes duplicate character '0' in a set of characters. |
-| test.py:47:12:47:19 | [000] | This regular expression includes duplicate character '0' in a set of characters. |
 | test.py:48:12:48:21 | Str | This regular expression includes duplicate character '-' in a set of characters. |
-| test.py:48:12:48:21 | [-0-9-] | This regular expression includes duplicate character '-' in a set of characters. |

python/ql/test/query-tests/Expressions/Regex/MissingPartSpecialGroup.expected

@@ -1,4 +1,2 @@
-| test.py:22:12:22:29 | (P<name>[\\w]+) | Regular expression is missing '?' in named group. |
 | test.py:22:12:22:29 | Str | Regular expression is missing '?' in named group. |
-| test.py:23:12:23:33 | (_(P<name>[\\w]+)\|) | Regular expression is missing '?' in named group. |
 | test.py:23:12:23:33 | Str | Regular expression is missing '?' in named group. |

python/ql/test/query-tests/Expressions/Regex/UnmatchableCaret.expected

@@ -1,8 +1,4 @@
-| test.py:4:12:4:19 |  ^abc | This regular expression includes an unmatchable caret at offset 1. |
 | test.py:4:12:4:19 | Str | This regular expression includes an unmatchable caret at offset 1. |
-| test.py:5:12:5:23 | (?s) ^abc | This regular expression includes an unmatchable caret at offset 5. |
 | test.py:5:12:5:23 | Str | This regular expression includes an unmatchable caret at offset 5. |
 | test.py:6:12:6:21 | Str | This regular expression includes an unmatchable caret at offset 2. |
-| test.py:6:12:6:21 | \\[^123] | This regular expression includes an unmatchable caret at offset 2. |
-| test.py:79:12:79:27 | (?<=foo)^\\w+ | This regular expression includes an unmatchable caret at offset 8. |
 | test.py:79:12:79:27 | Str | This regular expression includes an unmatchable caret at offset 8. |

python/ql/test/query-tests/Expressions/Regex/UnmatchableDollar.expected

@@ -1,8 +1,4 @@
 | test.py:29:12:29:19 | Str | This regular expression includes an unmatchable dollar at offset 3. |
-| test.py:29:12:29:19 | abc$  | This regular expression includes an unmatchable dollar at offset 3. |
 | test.py:30:12:30:23 | Str | This regular expression includes an unmatchable dollar at offset 3. |
-| test.py:30:12:30:23 | abc$ (?s) | This regular expression includes an unmatchable dollar at offset 3. |
 | test.py:31:12:31:20 | Str | This regular expression includes an unmatchable dollar at offset 2. |
-| test.py:31:12:31:20 | \\[$]   | This regular expression includes an unmatchable dollar at offset 2. |
 | test.py:80:12:80:26 | Str | This regular expression includes an unmatchable dollar at offset 3. |
-| test.py:80:12:80:26 | \\w+$(?=foo) | This regular expression includes an unmatchable dollar at offset 3. |

python/ql/test/query-tests/Security/CWE-020/IncompleteHostnameRegExp.expected

@@ -1,2 +1 @@
-| hosttest.py:6:27:6:51 | (www\|beta).example.com/ | This regular expression has an unescaped '.' before 'example.com', so it might match more hosts than expected. |
 | hosttest.py:6:27:6:51 | Str | This regular expression has an unescaped '.' before 'example.com', so it might match more hosts than expected. |