refactor: rewrite requirements.txt parser and change the way we read files (#311)

Author: sdn4z

pyproject.toml

@@ -12,7 +12,6 @@ maintainers = [
 requires-python = "<4,>=3.9"
 dependencies = [
     "requests<3.0.0,>=2.32.4",
-    "dparse<1.0.0,>=0.6.4",
     "click<9.0.0,>=8.1.8",
     "rich<15.0.0,>=14.0.0",
     "rapidfuzz<4.0.0,>=2.13.7",

src/twyn/base/utils.py

@@ -4,7 +4,7 @@
 from pathlib import Path
 from typing import Any, Optional, Union
 
-from tomlkit import TOMLDocument, dumps, parse, table
+from tomlkit import TOMLDocument, dumps, load, table
 
 from twyn.base.constants import (
     DEFAULT_PROJECT_TOML_FILE,
@@ -22,7 +22,7 @@
     TOMLError,
 )
 from twyn.file_handler.exceptions import PathNotFoundError
-from twyn.file_handler.file_handler import BaseFileHandler
+from twyn.file_handler.file_handler import FileHandler
 
 logger = logging.getLogger("twyn")
 
@@ -54,7 +54,7 @@ class ReadTwynConfiguration:
 class ConfigHandler:
     """Manage reading and writing configurations for Twyn."""
 
-    def __init__(self, file_handler: Optional[BaseFileHandler] = None) -> None:
+    def __init__(self, file_handler: Optional[FileHandler] = None) -> None:
         self.file_handler = file_handler
 
     def resolve_config(
@@ -160,7 +160,8 @@ def _read_toml(self) -> TOMLDocument:
         if not self.file_handler:
             raise ConfigFileNotConfiguredError("Config file not configured. Cannot perform read operation.")
         try:
-            return parse(self.file_handler.read())
+            with self.file_handler.open("rb") as fp:
+                return load(fp)
         except PathNotFoundError:
             if self.file_handler.is_handler_of_file(DEFAULT_PROJECT_TOML_FILE):
                 return TOMLDocument()

src/twyn/config/config_handler.py

@@ -1,10 +1,8 @@
 from dataclasses import dataclass
 from pathlib import Path
 
-from dparse import filetypes
-
 from twyn.dependency_managers.exceptions import NoMatchingDependencyManagerError
-from twyn.dependency_parser.parsers.constants import PACKAGE_LOCK_JSON, UV_LOCK
+from twyn.dependency_parser.parsers.constants import PACKAGE_LOCK_JSON, POETRY_LOCK, REQUIREMENTS_TXT, UV_LOCK
 from twyn.trusted_packages.references import AbstractPackageReference, TopNpmReference, TopPyPiReference
 
 
@@ -32,7 +30,7 @@ def matches_language_name(cls, name: str) -> bool:
 class PypiDependencyManager(BaseDependencyManager):
     name = "pypi"
     trusted_packages_source = TopPyPiReference
-    dependency_files = {UV_LOCK, filetypes.poetry_lock, filetypes.requirements_txt}
+    dependency_files = {UV_LOCK, POETRY_LOCK, REQUIREMENTS_TXT}
 
 
 @dataclass

src/twyn/dependency_managers/dependency_manager.py

@@ -22,9 +22,6 @@ def __init__(self, file_path: str) -> None:
     def __str__(self) -> str:
         return self.__class__.__name__
 
-    def _read(self) -> str:
-        return self.file_handler.read()
-
     def file_exists(self) -> bool:
         return self.file_handler.exists()
 

src/twyn/dependency_parser/lock_parser.py

@@ -1,2 +1,4 @@
 UV_LOCK = "uv.lock"
 PACKAGE_LOCK_JSON = "package-lock.json"
+POETRY_LOCK = "poetry.lock"
+REQUIREMENTS_TXT = "requirements.txt"

src/twyn/dependency_parser/package_lock_json.py

@@ -1,9 +1,7 @@
 import sys
 
-from dparse import filetypes
-
 from twyn.dependency_parser.parsers.abstract_parser import AbstractParser
-from twyn.dependency_parser.parsers.constants import UV_LOCK
+from twyn.dependency_parser.parsers.constants import POETRY_LOCK, UV_LOCK
 
 if sys.version_info >= (3, 11):
     import tomllib
@@ -16,14 +14,14 @@ class TomlLockParser(AbstractParser):
 
     def parse(self) -> set[str]:
         """Parse dependencies names and map them to a set."""
-        data = tomllib.loads(self._read())
+        data = tomllib.loads(self.file_handler.read())
         return {dependency["name"] for dependency in data["package"]}
 
 
 class PoetryLockParser(TomlLockParser):
     """Parser for poetry.lock files."""
 
-    def __init__(self, file_path: str = filetypes.poetry_lock) -> None:
+    def __init__(self, file_path: str = POETRY_LOCK) -> None:
         super().__init__(file_path)
 
 

src/twyn/dependency_parser/parsers/abstract_parser.py

@@ -14,7 +14,7 @@ def parse(self) -> set[str]:
 
         It supports v1, v2 and v3.
         """
-        data = json.loads(self._read())
+        data = json.loads(self.file_handler.read())
         result: set[str] = set()
 
         # Handle v1 & v2