fix: do not error on empty json files

Author: sdn4z

src/twyn/dependency_parser/parsers/lock_parser.py

@@ -1,15 +1,21 @@
 import tomlkit
+import tomlkit.exceptions
 
 from twyn.dependency_parser.parsers.abstract_parser import AbstractParser
 from twyn.dependency_parser.parsers.constants import POETRY_LOCK, UV_LOCK
+from twyn.dependency_parser.parsers.exceptions import InvalidFileFormatError
 
 
 class TomlLockParser(AbstractParser):
     """Parser for TOML-based lock files."""
 
     def parse(self) -> set[str]:
         """Parse dependencies names and map them to a set."""
-        data = tomlkit.parse(self.file_handler.read())
+        try:
+            data = tomlkit.parse(self.file_handler.read())
+        except tomlkit.exceptions.ParseError as e:
+            raise InvalidFileFormatError("Invalid YAML format.") from e
+
         packages = data.get("package", [])
         return {pkg["name"] for pkg in packages if isinstance(pkg, dict) and "name" in pkg}
 

src/twyn/dependency_parser/parsers/package_lock_json.py

@@ -3,6 +3,7 @@
 
 from twyn.dependency_parser.parsers.abstract_parser import AbstractParser
 from twyn.dependency_parser.parsers.constants import PACKAGE_LOCK_JSON
+from twyn.dependency_parser.parsers.exceptions import InvalidFileFormatError
 
 
 class PackageLockJsonParser(AbstractParser):
@@ -14,7 +15,11 @@ def parse(self) -> set[str]:
 
         It supports v1, v2 and v3.
         """
-        data = json.loads(self.file_handler.read())
+        try:
+            data = json.loads(self.file_handler.read())
+        except json.JSONDecodeError as e:
+            raise InvalidFileFormatError("Invalid JSON format.") from e
+
         result: set[str] = set()
 
         # Handle v1 & v2
@@ -34,7 +39,7 @@ def parse(self) -> set[str]:
 
         return result
 
-    def _collect_deps(self, dep_tree: dict[str, Any], collected: set[str]):
+    def _collect_deps(self, dep_tree: dict[str, Any], collected: set[str]) -> None:
         """Recursively collect dependencies from dependency tree."""
         for name, info in dep_tree.items():
             collected.add(name)

src/twyn/dependency_parser/parsers/yarn_lock_parser.py

@@ -27,7 +27,7 @@ def parse(self) -> set[str]:
 
                 if "__metadata:" in line:
                     return self._parse_v2(fp)
-        raise InvalidFileFormatError
+        raise InvalidFileFormatError("Unkown file format.")
 
     def _parse_v1(self, fp: TextIO) -> set[str]:
         """Parse a yarn.lock file (v1) and return all the dependencies in it."""

src/twyn/file_handler/exceptions.py

@@ -11,3 +11,9 @@ class PathNotFoundError(TwynError):
     """Exception raised when a specified file path does not exist in the filesystem."""
 
     message = "Specified dependencies file path does not exist"
+
+
+class EmptyFileError(TwynError):
+    """Exception raised when the read file is empty."""
+
+    message = "Given file is empty."

src/twyn/file_handler/file_handler.py

@@ -5,7 +5,7 @@
 from pathlib import Path
 from typing import TextIO
 
-from twyn.file_handler.exceptions import PathIsNotFileError, PathNotFoundError
+from twyn.file_handler.exceptions import EmptyFileError, PathIsNotFileError, PathNotFoundError
 
 logger = logging.getLogger("twyn")
 
@@ -42,7 +42,7 @@ def exists(self) -> bool:
         """Check if file exists and is a valid file."""
         try:
             self._raise_for_file_exists()
-        except (PathNotFoundError, PathIsNotFileError):
+        except (PathNotFoundError, PathIsNotFileError, EmptyFileError):
             return False
         return True
 
@@ -54,6 +54,9 @@ def _raise_for_file_exists(self) -> None:
         if not self.file_path.is_file():
             raise PathIsNotFileError
 
+        if self.file_path.stat().st_size == 0:
+            raise EmptyFileError
+
     def write(self, data: str) -> None:
         """Write data to file."""
         self.file_path.write_text(data)

src/twyn/main.py

@@ -16,6 +16,8 @@
 )
 from twyn.dependency_parser.dependency_selector import DependencySelector
 from twyn.dependency_parser.parsers.abstract_parser import AbstractParser
+from twyn.dependency_parser.parsers.exceptions import InvalidFileFormatError
+from twyn.file_handler.exceptions import EmptyFileError
 from twyn.file_handler.file_handler import FileHandler
 from twyn.similarity.algorithm import EditDistance, SimilarityThreshold
 from twyn.trusted_packages.cache_handler import CacheHandler
@@ -185,8 +187,23 @@ def _analyze_packages_from_source(
         )
         results: list[TyposquatCheckResultFromSource] = []
         for parser in parsers:
+            try:
+                parsed_content = parser.parse()
+            except (InvalidFileFormatError, EmptyFileError) as e:
+                logger.warning("Could not parse %s. %s", parser.file_path, e)
+                continue
+
+            if not parsed_content:
+                logger.warning("No packages found in file %s. Skipping...", parser.file_path)
+                continue
+
             analyzed_dependencies = _analyze_dependencies(
-                top_package_reference, trusted_packages, parser.parse(), allowlist, show_progress_bar, parser.file_path
+                top_package_reference,
+                trusted_packages,
+                parsed_content,
+                allowlist,
+                show_progress_bar,
+                parser.file_path,
             )
 
             if analyzed_dependencies:

tests/config/test_config_handler.py

@@ -216,21 +216,17 @@ def test_allowlist_remove_non_existent_package_error(self, mock_toml: Mock, mock
         assert not mock_write_toml.called
 
     @pytest.mark.parametrize("valid_selector", ["first-letter", "nearby-letter", "all"])
-    def test_valid_selector_methods_accepted(self, valid_selector: str, tmp_path: Path) -> None:
+    def test_valid_selector_methods_accepted(self, valid_selector: str, pyproject_toml_file: Path) -> None:
         """Test that all valid selector methods are accepted."""
-        pyproject_toml = tmp_path / "pyproject.toml"
-        pyproject_toml.write_text("")
-        config = ConfigHandler(FileHandler(str(pyproject_toml)))
+        config = ConfigHandler(FileHandler(str(pyproject_toml_file)))
 
         # Should not raise any exception
         resolved_config = config.resolve_config(selector_method=valid_selector)
         assert resolved_config.selector_method == valid_selector
 
-    def test_invalid_selector_method_rejected(self, tmp_path: Path) -> None:
+    def test_invalid_selector_method_rejected(self, pyproject_toml_file: Path) -> None:
         """Test that invalid selector methods are rejected with appropriate error."""
-        pyproject_toml = tmp_path / "pyproject.toml"
-        pyproject_toml.write_text("")
-        config = ConfigHandler(FileHandler(str(pyproject_toml)))
+        config = ConfigHandler(FileHandler(str(pyproject_toml_file)))
 
         with pytest.raises(InvalidSelectorMethodError) as exc_info:
             config.resolve_config(selector_method="random-selector")

tests/dependency_parser/test_dependency_parser.py

@@ -11,7 +11,7 @@
 )
 from twyn.dependency_parser.parsers.abstract_parser import AbstractParser
 from twyn.dependency_parser.parsers.yarn_lock_parser import YarnLockParser
-from twyn.file_handler.exceptions import PathIsNotFileError, PathNotFoundError
+from twyn.file_handler.exceptions import EmptyFileError, PathIsNotFileError, PathNotFoundError
 
 
 class TestAbstractParser:
@@ -22,26 +22,50 @@ def parse(self) -> set[str]:
             self._read()
             return set()
 
+    @patch("pathlib.Path.stat")
     @patch("pathlib.Path.exists")
     @patch("pathlib.Path.is_file")
-    def test_file_exists(self, mock_exists: Mock, mock_is_file: Mock) -> None:
+    def test_file_exists(self, mock_is_file: Mock, mock_exists: Mock, mock_stat: Mock) -> None:
         mock_exists.return_value = True
         mock_is_file.return_value = True
+
+        # Mock stat to return a mock object with st_size > 0 (non-empty file)
+        mock_stat_result = Mock()
+        mock_stat_result.st_size = 100
+        mock_stat.return_value = mock_stat_result
+
         parser = self.TemporaryParser("fake_path.txt")
         assert parser.file_exists() is True
 
+    @patch("pathlib.Path.stat")
     @patch("pathlib.Path.exists")
     @patch("pathlib.Path.is_file")
     @pytest.mark.parametrize(
-        ("file_exists", "is_file", "exception"),
-        [(False, False, PathNotFoundError), (True, False, PathIsNotFileError)],
+        ("file_exists", "is_file", "file_size", "exception"),
+        [
+            (False, False, 100, PathNotFoundError),
+            (True, False, 100, PathIsNotFileError),
+            (True, True, 0, EmptyFileError),
+        ],
     )
     def test_raise_for_valid_file(
-        self, mock_is_file: Mock, mock_exists: Mock, file_exists: Mock, is_file, exception: Mock
+        self,
+        mock_is_file: Mock,
+        mock_exists: Mock,
+        mock_stat: Mock,
+        file_exists: bool,
+        is_file: bool,
+        file_size: int,
+        exception: type,
     ) -> None:
         mock_exists.return_value = file_exists
         mock_is_file.return_value = is_file
 
+        # Mock stat to return a mock object with st_size attribute
+        mock_stat_result = Mock()
+        mock_stat_result.st_size = file_size
+        mock_stat.return_value = mock_stat_result
+
         parser = self.TemporaryParser("fake_path.txt")
         assert parser.file_exists() is False
 

tests/main/test_main.py

@@ -655,3 +655,23 @@ def test_auto_detect_dependency_file_parser_scans_subdirectories(
 
         assert len(error.results) == 1
         assert error.get_results_from_source(str(req_file)) is not None
+
+    @patch("twyn.trusted_packages.TopPyPiReference.get_packages")
+    def test_check_dependencies_continues_execution_with_empty_file_mixed_with_content_file(
+        self, mock_get_packages: Mock, tmp_path: Path, uv_lock_file_with_typo: Path
+    ) -> None:
+        """Test that execution continues when one file is empty and another contains dependencies."""
+        mock_get_packages.return_value = {"requests"}
+
+        empty_file = tmp_path / "requirements.txt"
+        with create_tmp_file(empty_file, ""):
+            error = check_dependencies(
+                dependency_files={str(empty_file), str(uv_lock_file_with_typo)},
+                use_cache=False,
+            )
+
+        assert len(error.results) == 1
+        assert error.results[0].source == str(uv_lock_file_with_typo)
+        assert error.results[0].errors == [TyposquatCheckResultEntry(dependency="reqests", similars=["requests"])]
+
+        assert error.get_results_from_source(str(empty_file)) is None