Merge pull request #20 from tpiekarski/feature/type-hints

Merging feature/type-hints, resolves #17

Makefile

@@ -60,6 +60,10 @@ run:
 	$(info Testing if derl runs with $(TEST_DIRECTORY) (use args="" to pass arguments))
 	derl $(TEST_DIRECTORY) $(args)
 
+show-function-signatures:
+	$(info Showing all function (and method) signatures)
+	find src/ tests/ -name "*.py" | xargs grep -E "def.*\(.*\).*:"
+
 sonar:
 	$(info Running Sonar Scanner (only when automatic analysis is turned off))
 	sonar-scanner -Dsonar.login=${SONAR_KEY}

src/derl/checker.py

@@ -6,6 +6,7 @@
 
 import sys
 
+from argparse import Namespace
 from logging import getLogger
 from os import path
 from magic import from_file
@@ -18,32 +19,32 @@
 _logger = getLogger(__name__)
 
 
-def is_directory(value):
+def is_directory(value: str) -> bool:
     _logger.debug("Checking provided directory %s", value)
 
     return path.isdir(value)
 
 
-def is_retry_value(value):
+def is_retry_value(value: str) -> bool:
     return 0 < value <= 10
 
 
-def is_text_file(file):
+def is_text_file(file: str) -> bool:
     _logger.debug("Checking file %s", file)
     mimetype = from_file(str(file), mime=True)
 
     return file.is_file() and mimetype[:4] == "text"
 
 
-def is_timeout_value(value):
+def is_timeout_value(value: int) -> bool:
     return value > 0
 
 
-def is_url(value):
+def is_url(value: str) -> bool:
     return url(value)
 
 
-def check_arguments(args):
+def check_arguments(args: Namespace):
     if not is_timeout_value(args.timeout):
         _logger.error("Invalid timeout, timeout must be greater than 0")
         sys.exit(_INVALID_TIMEOUT)

src/derl/collector.py

@@ -6,12 +6,14 @@
 
 from logging import getLogger
 
+from derl.model.url import URL
+
 _logger = getLogger(__name__)
 
 _DEFAULT_ENCODING = "utf-8"
 
 
-def _extract_context(url, lines):
+def _extract_context(url: URL, lines: list) -> list:
     context = []
     line_index = url.line_number - 1
 
@@ -26,7 +28,7 @@ def _extract_context(url, lines):
     return context
 
 
-def collect_context(files):
+def collect_context(files: list) -> list:
     for current_file in files:
         for current_url in current_file.urls:
             with open(current_file.filename, "r", encoding=_DEFAULT_ENCODING) as open_file:

src/derl/dispatcher.py

@@ -19,7 +19,7 @@
 _DEFAULT_TIMEOUT = 10
 
 
-def request(files, retry=_DEFAULT_RETRY, timeout=_DEFAULT_TIMEOUT):
+def request(files: list, retry: int = _DEFAULT_RETRY, timeout: int = _DEFAULT_TIMEOUT) -> list:
     if len(files) == 0:
         _logger.debug("No matches for HTTP requests")
         return []

src/derl/filterer.py

@@ -9,7 +9,7 @@
 _logger = getLogger(__name__)
 
 
-def filter_not_matching(files):
+def filter_not_matching(files: list) -> list:
     filtered_files = []
 
     for current_file in files:

src/derl/main.py

@@ -27,7 +27,7 @@
 _tracker = get_tracker()
 
 
-def setup_logging(loglevel):
+def setup_logging(loglevel: int):
     basicConfig(
         datefmt="%Y-%m-%d %H:%M:%S",
         format="[%(asctime)s] %(levelname)s:%(name)s:%(message)s",
@@ -36,7 +36,7 @@ def setup_logging(loglevel):
     )
 
 
-def main(args):
+def main(args: list):
     args = parse_args(args)
     setup_logging(args.loglevel)
     check_arguments(args)

src/derl/model/file.py

@@ -11,23 +11,23 @@ class File:
     filename = None
     urls: []
 
-    def __init__(self, filename):
+    def __init__(self: "File", filename: str):
         self.filename = filename
         self.urls = []
 
-    def __str__(self):
+    def __str__(self: "File") -> str:
         output = ""
 
         for current_url in self.urls:
             output += "{}:{}\n".format(self.filename, current_url)
 
         return output
 
-    def __repr__(self):
+    def __repr__(self: "File") -> str:
         return self.__str__()
 
-    def append(self, url, line_number):
+    def append(self: "File", url: str, line_number: int):
         self.urls.append(URL(url, line_number))
 
-    def contains_urls(self):
+    def contains_urls(self: "File") -> bool:
         return len(self.urls) > 0

src/derl/model/stats.py

@@ -13,7 +13,7 @@ class Stats():
     tokens = 0
     urls = 0
 
-    def __str__(self):
+    def __str__(self: "Stats") -> str:
         output = "Processed Directories/Files/Lines/Tokens/URLs: {0:d}/{1:d}/{2:d}/{3:d}/{4:d}".format(
             self.directories, self.files, self.lines, self.tokens, self.urls
         )
@@ -23,23 +23,23 @@ def __str__(self):
 
         return output
 
-    def __repr__(self):
+    def __repr__(self: "Stats") -> str:
         return self.__str__()
 
-    def inc_directories(self):
+    def inc_directories(self: "Stats"):
         self.directories += 1
 
-    def inc_files(self):
+    def inc_files(self: "Stats"):
         self.files += 1
 
-    def inc_lines(self):
+    def inc_lines(self: "Stats"):
         self.lines += 1
 
-    def inc_tokens(self):
+    def inc_tokens(self: "Stats"):
         self.tokens += 1
 
-    def inc_urls(self):
+    def inc_urls(self: "Stats"):
         self.urls += 1
 
-    def inc_requests(self):
+    def inc_requests(self: "Stats"):
         self.requests += 1

src/derl/model/url.py

@@ -7,7 +7,7 @@
 _CONTEXT_OUTPUT_MAP = {0: "", 1: "  {}\n", 2: "  {}\n  {}\n", 3: "  {}\n  {}\n  {}\n"}
 
 
-def _normalize_context(context):
+def _normalize_context(context: list) -> list:
     normalized_context = []
 
     for current_line in context:
@@ -22,11 +22,11 @@ class URL:
     line_number = None
     context = []
 
-    def __init__(self, location, line_number):
+    def __init__(self: "URL", location: str, line_number: int):
         self.location = location
         self.line_number = line_number
 
-    def __str__(self):
+    def __str__(self: "URL") -> str:
         if self.status_code is None:
             output = "{}, {}".format(self.line_number, self.location)
         else:
@@ -38,11 +38,11 @@ def __str__(self):
 
         return output
 
-    def __repr__(self):
+    def __repr__(self: "URL") -> str:
         return self.__str__()
 
-    def is_context_present(self):
-        return len(self.context)
+    def is_context_present(self: "URL") -> bool:
+        return len(self.context) > 0
 
-    def set_context(self, context):
+    def set_context(self: "URL", context: list):
         self.context = _normalize_context(context)

src/derl/outputer.py

@@ -11,7 +11,7 @@
 _tracker = get_tracker()
 
 
-def output(files, stats=False):
+def output(files: list, stats: bool = False):
     if len(files) == 0:
         _logger.debug("No matched files for output")
         return

src/derl/parser.py

@@ -4,14 +4,14 @@
 # Copyright 2020 Thomas Piekarski <[email protected]>
 #
 
-from argparse import ArgumentParser, HelpFormatter
+from argparse import ArgumentParser, HelpFormatter, Namespace
 from logging import DEBUG, INFO
 
 from derl import __version__
 from derl.dispatcher import _DEFAULT_RETRY, _DEFAULT_TIMEOUT
 
 
-def parse_args(args):
+def parse_args(args: list) -> Namespace:
     parser = ArgumentParser(
         prog="derl",
         formatter_class=lambda prog: HelpFormatter(prog, max_help_position=35, width=90),

src/derl/processor.py

@@ -7,6 +7,7 @@
 from logging import getLogger
 from pathlib import Path
 from re import compile as rcompile, IGNORECASE
+from typing import TextIO
 
 from derl.checker import is_text_file, is_url
 from derl.model.file import File
@@ -20,7 +21,7 @@
 _tracker = get_tracker()
 
 
-def process_file(file):
+def process_file(file: TextIO) -> list:
     _logger.debug("Spliting current file %s into lines...", file.name)
     _tracker.stats.inc_files()
 
@@ -44,7 +45,7 @@ def process_file(file):
     return urls
 
 
-def process_line(file, line, urls):
+def process_line(file: TextIO, line: tuple, urls: list) -> list:
     _logger.debug("Splitting current line into tokens...")
     _tracker.stats.inc_lines()
 
@@ -66,7 +67,7 @@ def process_line(file, line, urls):
     return urls
 
 
-def process_token(file, token, line_number):
+def process_token(file: TextIO, token: str, line_number: int) -> URL:
     _tracker.stats.inc_tokens()
 
     match = _pattern.match(token)
@@ -80,7 +81,7 @@ def process_token(file, token, line_number):
     return url
 
 
-def process_directory(directory, files):
+def process_directory(directory: str, files: list) -> list:
     _logger.info("Starting to process directory '%s'", directory)
     _tracker.stats.inc_directories()
 
@@ -89,7 +90,7 @@ def process_directory(directory, files):
         for current in path.iterdir():
             if current.is_dir():
                 _logger.debug("'%s' is a directory, descending...", current.name)
-                files = process_directory(current, files)
+                files = process_directory(str(current), files)
             elif is_text_file(current):
                 _logger.debug("Appending file '%s'", current.name)
                 files.append(File(current))

src/derl/searcher.py

@@ -13,7 +13,7 @@
 _DEFAULT_ENCODING = "utf-8"
 
 
-def search_urls(files):
+def search_urls(files: list) -> list:
     _logger.info("Starting to search for URLs in %i files...", len(files))
 
     if len(files) == 0:

src/derl/tracker.py

@@ -12,7 +12,7 @@
 class Singleton(type):
     _instances = {}
 
-    def __call__(cls, *args, **kwargs):
+    def __call__(cls: "Singleton", *args: tuple, **kwargs: dict) -> "Tracker":
         if cls not in cls._instances:
             cls._instances[cls] = super(Singleton, cls).__call__(*args, **kwargs)
 
@@ -25,29 +25,29 @@ class Tracker(metaclass=Singleton):
     stats = Stats()
     test = False
 
-    def start(self):
+    def start(self: "Tracker"):
         if self.start_time is None:
             self.start_time = perf_counter()
 
-    def stop(self):
+    def stop(self: "Tracker"):
         if self.stop_time is None:
             self.stop_time = perf_counter()
 
-    def calc_time(self):
+    def calc_time(self: "Tracker") -> float:
         if self.test:
             return -1
 
         return round(self.stop_time - self.start_time)
 
-    def reset(self):
+    def reset(self: "Tracker"):
         self.start_time = 0
         self.stop_time = 0
         self.stats = Stats()
 
-    def set_test(self):
+    def set_test(self: "Tracker"):
         self.test = True
 
-    def __str__(self):
+    def __str__(self: "Tracker") -> str:
         output = ""
 
         if self.start_time is not None and self.stop_time is not None:
@@ -57,9 +57,9 @@ def __str__(self):
 
         return output
 
-    def __repr__(self):
+    def __repr__(self: "Tracker") -> str:
         return self.__str__()
 
 
-def get_tracker():
+def get_tracker() -> "Tracker":
     return Tracker()

tests/test_checker.py

@@ -12,10 +12,10 @@
 
 class CheckerTest(TestCase):
 
-    def test_is_directory(self):
+    def test_is_directory(self: "CheckerTest"):
         self.assertTrue(is_directory("tests/test-directory"))
         self.assertFalse(is_directory("tests/not-existent-directory"))
 
-    def test_is_text_file(self):
+    def test_is_text_file(self: "CheckerTest"):
         self.assertTrue(is_text_file(Path("tests/test-files/plain-text")))
         self.assertFalse(is_text_file(Path("tests/test-files/binary-data")))