refactor: fixes type annotations and makes the output of the time spe…

…nt on work checks more human-readable. (#200)

Refs: #200.

.pre-commit-config.yaml

@@ -2,7 +2,7 @@ default_language_version:
     python: python3.10
 repos:
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.3.4
+    rev: v0.6.4
     hooks:
       - id: ruff
         args: [ --fix ]

Makefile

@@ -1,4 +1,4 @@
-UTIL_VERSION            := 0.5.2
+UTIL_VERSION            := 0.5.3
 UTIL_NAME               := codeplag
 PWD                     := $(shell pwd)
 

docs/notebooks/utils.py

@@ -7,12 +7,13 @@
 import matplotlib.pyplot as plt
 import numpy as np
 import pandas as pd
+from decouple import Config, RepositoryEnv
+from scipy.optimize import curve_fit
+
 from codeplag.algorithms.featurebased import counter_metric, struct_compare
 from codeplag.algorithms.stringbased import gst
 from codeplag.algorithms.tokenbased import value_jakkar_coef
 from codeplag.pyplag.utils import get_ast_from_content, get_features_from_ast
-from decouple import Config, RepositoryEnv
-from scipy.optimize import curve_fit
 from webparsers.github_parser import GitHubParser
 
 
@@ -188,7 +189,7 @@ def plot_and_save_result(
 
 def get_time_algorithms(
     df: pd.DataFrame,
-    work,
+    work: pd.Series,
     iterations: int = 5,
     metric: Literal["fast", "gst", "structure"] = "fast",
 ) -> pd.DataFrame:

pyproject.toml

@@ -12,11 +12,19 @@ lint.select = [
     "I",    # isort
     "B",    # flake8-bugbear
     "C4",   # flake8-comprehensions
+    "ANN",  # flake8-annotations
     "SIM",  # flake8-simplify
     "ERA",  # eradicate
     "C90",  # mccabe
 ]
 lint.ignore = [
+    "ANN003",  # missing type annotations for `*kwargs`
+    "ANN101",
+    "ANN102",
+    "ANN201",
+    "ANN204",
+    "ANN206",
+    "ANN401",  # dynamically typed expressions (typing.Any)
     "D100",
     "D101",
     "D102",
@@ -25,13 +33,17 @@ lint.ignore = [
     "D105",
     "D107",
 ]
+lint.exclude = ["*.ipynb"]
 
 [tool.ruff.lint.pydocstyle]
 convention = "google"
 
 [tool.ruff.lint.mccabe]
 max-complexity = 12
 
+[tool.ruff.format]
+exclude = ["*.ipynb"]
+
 [tool.pyright]
 pythonVersion = "3.10"
 

src/codeplag/codeplagcli.py

@@ -6,8 +6,6 @@
 import builtins
 from pathlib import Path
 
-from webparsers.types import GitHubContentUrl
-
 from codeplag.consts import (
     DEFAULT_MODE,
     DEFAULT_REPORT_TYPE,
@@ -22,6 +20,7 @@
     WORKERS_CHOICE,
 )
 from codeplag.types import Settings
+from webparsers.types import GitHubContentUrl
 
 # FIXME: dirty hook for using logic without translations
 builtins.__dict__["_"] = builtins.__dict__.get("_", str)
@@ -51,7 +50,7 @@ def __call__(
 class DirPath(Path):
     """Raises `argparse.ArgumentTypeError` if the directory doesn't exist."""
 
-    def __new__(cls, *args, **kwargs):
+    def __new__(cls, *args: str, **kwargs):
         path = Path(*args, **kwargs).resolve()
         if not path.is_dir():
             raise argparse.ArgumentTypeError(
@@ -64,7 +63,7 @@ def __new__(cls, *args, **kwargs):
 class FilePath(Path):
     """Raises `argparse.ArgumentTypeError` if the file doesn't exist."""
 
-    def __new__(cls, *args, **kwargs):
+    def __new__(cls, *args: str, **kwargs):
         path = Path(*args, **kwargs).resolve()
         if not path.is_file():
             raise argparse.ArgumentTypeError(

src/codeplag/cplag/utils.py

@@ -4,14 +4,14 @@
 from typing import Final
 
 from clang.cindex import Config, Cursor, Index, TranslationUnit
-from webparsers.types import WorkInfo
 
 from codeplag.consts import FILE_DOWNLOAD_PATH, GET_FRAZE, SUPPORTED_EXTENSIONS
 from codeplag.cplag.const import COMPILE_ARGS
 from codeplag.cplag.tree import get_features
 from codeplag.getfeatures import AbstractGetter, get_files_path_from_directory
 from codeplag.logger import log_err
 from codeplag.types import ASTFeatures
+from webparsers.types import WorkInfo
 
 # FIXME: Dirty hook for finding libclang so file
 LIBCLANG_SO_FILE_PATH: Final[Path] = Path("/usr/lib/llvm-14/lib/libclang-14.so.1")

src/codeplag/getfeatures.py

@@ -6,15 +6,14 @@
 from pathlib import Path
 from typing import Callable, Literal, ParamSpec, overload
 
-from webparsers.github_parser import GitHubParser
-from webparsers.types import WorkInfo
-
 from codeplag.consts import (
     ALL_EXTENSIONS,
     GET_FRAZE,
     UTIL_NAME,
 )
 from codeplag.types import ASTFeatures, Extension, Extensions
+from webparsers.github_parser import GitHubParser
+from webparsers.types import WorkInfo
 
 
 def get_files_path_from_directory(

src/codeplag/handlers/check.py

@@ -4,6 +4,7 @@
 import math
 import os
 from concurrent.futures import Future, ProcessPoolExecutor
+from datetime import timedelta
 from itertools import combinations
 from pathlib import Path
 from time import monotonic
@@ -13,7 +14,6 @@
 from decouple import Config, RepositoryEnv
 from numpy.typing import NDArray
 from requests import Session
-from webparsers.github_parser import GitHubParser
 
 from codeplag.algorithms.compare import compare_works
 from codeplag.config import read_settings_conf
@@ -38,6 +38,7 @@
     ProcessingWorksInfo,
     Threshold,
 )
+from webparsers.github_parser import GitHubParser
 
 
 class WorksComparator:
@@ -163,7 +164,7 @@ def check(
                 github_project_folders,
                 github_user,
             )
-        logger.debug("Time for all %.2fs.", monotonic() - begin_time)
+        logger.debug("Time for all %s.", timedelta(seconds=monotonic() - begin_time))
         logger.info("Ending searching for plagiarism ...")
         if isinstance(self.reporter, CSVReporter):
             self.reporter._write_df_to_fs()
@@ -175,7 +176,7 @@ def __many_to_many_check(
         features_from_gh_files: list[ASTFeatures],
         github_project_folders: list[str],
         github_user: str,
-    ):
+    ) -> None:
         works: list[ASTFeatures] = []
         works.extend(features_from_files)
         works.extend(self.features_getter.get_from_dirs(directories))
@@ -208,7 +209,7 @@ def __one_to_one_check(
         features_from_gh_files: list[ASTFeatures],
         github_project_folders: list[str],
         github_user: str,
-    ):
+    ) -> None:
         combined_elements = filter(
             bool,
             (
@@ -304,7 +305,7 @@ def _handle_compare_result(
     def _handle_completed_futures(
         self,
         processing: list[ProcessingWorksInfo],
-    ):
+    ) -> None:
         for proc_works_info in processing:
             metrics: CompareInfo = proc_works_info.compare_future.result()
             self._handle_compare_result(

src/codeplag/handlers/report.py

@@ -244,7 +244,7 @@ def _create_report(
     environment: jinja2.Environment,
     threshold: int = DEFAULT_THRESHOLD,
     language: Language = DEFAULT_LANGUAGE,
-):
+) -> None:
     template = environment.from_string(GENERAL_TEMPLATE_PATH.read_text())
     if save_path.is_dir():
         save_path = save_path / DEFAULT_GENERAL_REPORT_NAME

src/codeplag/logger.py

@@ -81,7 +81,7 @@ def set_handlers(
     logger.addHandler(get_stderr_handler())
 
 
-def log_err(*msgs) -> None:
+def log_err(*msgs: str) -> None:
     for msg in msgs:
         codeplag_logger.error(msg)
 

src/codeplag/pyplag/utils.py

@@ -2,8 +2,6 @@
 import logging
 from pathlib import Path
 
-from webparsers.types import WorkInfo
-
 from codeplag.consts import GET_FRAZE, SUPPORTED_EXTENSIONS
 from codeplag.display import red_bold
 from codeplag.getfeatures import (
@@ -15,6 +13,7 @@
 from codeplag.logger import log_err
 from codeplag.pyplag.astwalkers import ASTWalker
 from codeplag.types import ASTFeatures
+from webparsers.types import WorkInfo
 
 
 def get_ast_from_content(code: str, path: Path | str) -> ast.Module | None:

src/codeplag/types.py

@@ -93,12 +93,12 @@ def __post_init__(self) -> None:
         else:
             self.modify_date = ""
 
-    def __eq__(self, other) -> bool:
+    def __eq__(self, other: "ASTFeatures") -> bool:
         if not isinstance(other, self.__class__):
             raise NotImplementedError
         return str(self.filepath) == str(other.filepath)
 
-    def __lt__(self, other) -> bool:
+    def __lt__(self, other: "ASTFeatures") -> bool:
         if not isinstance(other, self.__class__):
             raise NotImplementedError
         return str(self.filepath) < str(other.filepath)

test/auto/functional/test_check.py

@@ -5,11 +5,12 @@
 import re
 
 import pytest
-from codeplag.consts import CONFIG_PATH, UTIL_NAME, UTIL_VERSION
-from codeplag.types import WorksReport
 from const import REPORTS_FOLDER
 from utils import modify_settings, run_check, run_util
 
+from codeplag.consts import CONFIG_PATH, UTIL_NAME, UTIL_VERSION
+from codeplag.types import WorksReport
+
 CWD = os.getcwd()
 CPP_FILES = [
     "test/unit/codeplag/cplag/data/sample1.cpp",

test/auto/functional/test_report.py

@@ -1,11 +1,12 @@
 from pathlib import Path
 
 import pytest
-from codeplag.consts import DEFAULT_GENERAL_REPORT_NAME, DEFAULT_SOURCES_REPORT_NAME
-from codeplag.types import ReportType
 from const import REPORTS_FOLDER
 from utils import create_report, modify_settings, run_check
 
+from codeplag.consts import DEFAULT_GENERAL_REPORT_NAME, DEFAULT_SOURCES_REPORT_NAME
+from codeplag.types import ReportType
+
 
 @pytest.fixture(scope="function", autouse=True)
 def setup(create_reports_folder: None):

test/auto/functional/test_settings.py

@@ -4,9 +4,10 @@
 from typing import Literal
 
 import pytest
+from utils import modify_settings
+
 from codeplag.consts import CONFIG_PATH, UTIL_NAME
 from codeplag.types import Language, LogLevel, ReportsExtension, Threshold
-from utils import modify_settings
 
 
 @pytest.fixture(scope="module", autouse=True)

test/conftest.py

@@ -1,9 +1,10 @@
 import logging
 
 import pytest
-from codeplag.consts import UTIL_NAME
 from pytest_mock import MockerFixture
 
+from codeplag.consts import UTIL_NAME
+
 
 @pytest.fixture
 def dummy_logger(mocker: MockerFixture):

test/unit/codeplag/algorithms/test_compare.py

@@ -1,4 +1,5 @@
 import pytest
+
 from codeplag.algorithms.compare import compare_works, fast_compare
 from codeplag.types import ASTFeatures, CompareInfo
 

test/unit/codeplag/algorithms/test_featurebased.py

@@ -2,6 +2,7 @@
 import unittest
 
 import numpy as np
+
 from codeplag.algorithms.featurebased import (
     add_not_counted,
     counter_metric,

test/unit/codeplag/conftest.py

@@ -3,6 +3,7 @@
 from pathlib import Path
 
 import pytest
+
 from codeplag.algorithms.compare import compare_works
 from codeplag.pyplag.utils import get_ast_from_filename, get_features_from_ast
 from codeplag.types import ASTFeatures, CompareInfo