Catch unsupported operation errors

regexploit/ast/sre.py

@@ -15,6 +15,10 @@
 SreOp = Tuple[SreConstant, SreOpData]
 
 
+class UnsupportedSreOpException(Exception):
+    pass
+
+
 class SreOpParser:
     def __init__(self):
         self._groups = {}
@@ -24,6 +28,8 @@ def parse_sre(self, pattern: str, flags: int = 0):
         return self.sequence_or_singleton(sre_parse.parse(pattern, flags))
 
     def parse_op(self, op: SreConstant, data: SreOpData):
+        if not hasattr(self, f"from_{op.name}"):
+            raise UnsupportedSreOpException(f"Unsupported SRE op: {op.name}")
         return getattr(self, f"from_{op.name}")(data)
 
     def sequence_or_singleton(self, ops: List[SreOp]):

regexploit/bin/regexploit-python-env

@@ -39,9 +39,13 @@ def main():
                 hooked_regex: regexploit.hook.CompiledRegex
                 for hooked_regex in regexploit.hook.get_and_clear_regexes():
                     output.next()
-                    parsed = SreOpParser().parse_sre(
-                        hooked_regex.pattern, hooked_regex.flags
-                    )
+                    try:
+                        parsed = SreOpParser().parse_sre(
+                            hooked_regex.pattern, hooked_regex.flags
+                        )
+                    except Exception as e:
+                        print(f"Error parsing regex {hooked_regex.pattern} from {p}: {e}")
+                        continue
                     for redos in find(parsed):
                         if redos.starriness > 2:
                             output.record(

regexploit/bin/regexploit_python_ast.py

@@ -6,7 +6,7 @@
 import traceback
 import warnings
 
-from regexploit.ast.sre import SreOpParser
+from regexploit.ast.sre import SreOpParser, UnsupportedSreOpException
 from regexploit.bin.files import file_generator
 from regexploit.languages.python_node_visitor import PythonNodeVisitor
 from regexploit.output.text import TextOutput
@@ -31,6 +31,9 @@ def handle_file(filename: str, output: TextOutput):
             parsed = SreOpParser().parse_sre(regex.pattern, regex.flags)
         except re.error:
             continue  # We will have many strings which aren't actually regexes
+        except UnsupportedSreOpException as e:
+            print(f"Error parsing regex {regex.pattern} from {filename}: {e}")
+            continue
         try:
             output.next()
             for redos in find(parsed):

tests/test_at.py

@@ -1,7 +1,7 @@
 import pytest
 
 from regexploit.ast.at import EndOfString
-from regexploit.ast.sre import SreOpParser
+from regexploit.ast.sre import SreOpParser, UnsupportedSreOpException
 
 
 def from_regex(pattern: str):
@@ -60,3 +60,16 @@ def test_real():
     dollar = EndOfString()
     dollar.set_character(from_regex(r"-\d+(\s*\s*\s*)").elements)
     assert dollar.character == from_regex(r"[\s]")
+
+
+@pytest.mark.parametrize(
+    "r",
+    [
+        r"a++b",
+        r"a?+b",
+        r"a*+b",
+    ],
+)
+def test_unsupported_op(r):
+    with pytest.raises(UnsupportedSreOpException):
+        from_regex(r)