Cloner begining

Signed-off-by: Václav Valíček <valicek1994@gmail.com>

repo_cloner/lib/cloner.py

@@ -0,0 +1,97 @@
+from repo_cloner.lib.cloner_config import ClonerConfig
+from repo_cloner.lib.repo_dir_structure import RepoDirStructure
+from repo_cloner.lib.dir_not_found_error import DirNotFoundError
+from repo_cloner.lib.repo_tool import RepoTool
+from repo_cloner.lib.checksum import gen_repo_hashed_name
+from pathlib import Path
+from time import time
+import os
+import logging
+
+log = logging.getLogger("rc.cloner")
+
+
+class Cloner:
+	_dirs: RepoDirStructure = None
+	_config: ClonerConfig = None
+	_interval_file: str = "last-check-time"
+	_repo: RepoTool = None
+	_repo_url: str = ""
+
+	def __init__(self, dir_structure: RepoDirStructure):
+		self._dirs = dir_structure
+		self._config = self._dirs.config
+		if len(self._config.cloner_repo_url) == 0:
+			logging.critical(f"Undefined repo cloner URL in config!")
+			raise KeyError(f"cloner_repo_url not defined in config!")
+
+		# create cache dir, if missing
+		try:
+			assert self._dirs.cache_dir_exists
+		except DirNotFoundError:
+			log.info(f"Cache dir for project {self._config.cloner_project_name} not found -> creating")
+			Path(self._dirs.cache_dir).mkdir()
+			log.debug(f"Cache dir created")
+
+	def check_interval(self):
+		log.debug(f"Checking interval for {self._config.cloner_project_name}")
+		# get interval
+		interval = self._config.cloner_interval
+		# interval file?
+		interval_file: Path = Path(self._dirs.cache_dir).joinpath(self._interval_file)
+		log.debug(f"Interval file: {interval_file}")
+		file_stamp: int = 0
+		if interval_file.exists():
+			str_val = interval_file.read_text()
+			try:
+				file_stamp = int(str_val)
+			except ValueError:
+				log.warning(f"Interval file file is corrupted, keeping value as nothing happened")
+		# check time
+		if time() > file_stamp + interval * 60:
+			return True
+		return False
+
+	def open(self, url: str) -> bool:
+		log.debug(f"Opening repo with url: {url}")
+		repo_path = self._repo_path_by_url(url)
+		self._repo_url = url
+		self._repo = RepoTool(repo_path)
+		return self.__opened
+
+	@property
+	def __opened(self) -> bool:
+		if not self._repo:
+			return False
+		return self._repo.initialized
+
+	def _repo_path_by_url(self, url: str) -> str:
+		hashed_name: str = gen_repo_hashed_name(url)
+		log.debug(f"Repo hashed name for {url} is {hashed_name}")
+		return os.path.join(self._dirs.repos_dir, hashed_name)
+
+	@property
+	def __main_repo_path(self) -> str:
+		return self._repo_path_by_url(self._config.cloner_repo_url)
+
+	def sync(self) -> bool:
+		if not self.__opened:
+			self._repo = RepoTool(self.__main_repo_path)
+			if not self._repo.initialized:
+				return False
+		return self._repo.fetch()
+
+	def perform_check(self):
+		log.info(f"Started check for {self._config.cloner_project_name}, url: {self._config.cloner_repo_url}")
+		if self.check_interval():
+			self.sync()
+		log.info(f"Check finished")
+
+	def clone_from_url(self, url: str) -> bool:
+		path = self._repo_path_by_url(url)
+		self._repo_url = url
+		self._repo = RepoTool(path)
+		if self._repo.initialized:
+			log.critical(f"Repo path {path} is initialized... Refusing clone!")
+			return False
+		return self._repo.clone(url)

tests/lib/cloner_test_fixtures.py

@@ -17,11 +17,12 @@ def cloner_dir_struct(tmp_path: Path) -> Path:
 
 
 @pytest.fixture
-def cloner_dir_with_config(cloner_dir_struct: Path) -> Path:
+def cloner_dir_with_config(cloner_dir_struct: Path, support_data_path) -> Path:
+	repo_source = support_data_path.joinpath("test-repo-base").as_uri()
 	cfg_file = cloner_dir_struct.joinpath("config", "cloner.cfg")
 	cfg_file.touch()
-	cfg_file.write_text("# cloner.cfg"
+	cfg_file.write_text(f"# cloner.cfg"
-						"cloner_repo_url=https://git.sw3.cz/kamikaze/test-repo-base.git"
+						f"cloner_repo_url={repo_source}"
 						"cloner_project_name=test-repo"
 						)
 	return cloner_dir_struct

tests/lib/test_cloner.py

@@ -0,0 +1,251 @@
+import git
+import pytest
+import repo_cloner.lib.dir_not_found_error
+from cloner_test_fixtures import *
+from repo_cloner.lib.cloner import Cloner
+from pathlib import Path
+import logging
+
+
+def mock_time() -> float:
+	return 1658702099.4258687
+
+
+class MockConfig:
+	def __init__(self):
+		self.cloner_repo_url = ""
+		self.cloner_project_name = "Mocked Project"
+		self.cloner_interval = 0
+
+
+class MockDirStruct:
+	raise_cache_exists = False
+
+	def __init__(self, tmp: Path):
+		self.config = MockConfig()
+		self.cache_dir = tmp.joinpath("cache")
+		self.config_dir = tmp.joinpath("config")
+		self.repos_dir = tmp.joinpath("repos")
+		self.raise_cache_exists = False
+
+	@property
+	def cache_dir_exists(self):
+		if self.raise_cache_exists:
+			raise repo_cloner.lib.dir_not_found_error.DirNotFoundError("mock_dir")
+		return True
+
+
+class MockRepoTool:
+	path = None
+	initialized = False
+
+	def __init__(self, path: str):
+		self.path = path
+		self.initialized = False
+
+
+def test_init_invalid_config(tmp_path, caplog):
+	caplog.set_level(logging.INFO)
+	mock = MockDirStruct(tmp_path)
+	with pytest.raises(KeyError) as e:
+		Cloner(mock)
+	assert "KeyError: 'cloner_repo_url not defined in config!'" == e.exconly()
+	assert "CRITICAL root:cloner.py:" in caplog.text
+	assert "Undefined repo cloner URL in config!\n" in caplog.text
+
+	# valid config
+	caplog.clear()
+	# set mocks
+	mock.config.cloner_repo_url = "git@mocked:/path"
+	cache_dir = tmp_path.joinpath("cache")
+	# validate dir does not exist
+	assert not cache_dir.exists()
+	# set mock to raise error
+	mock.raise_cache_exists = True
+	# create new cloner class
+	Cloner(mock)
+	# dir should exist now
+	assert cache_dir.exists()
+	assert "Cache dir for project Mocked Project not found -> creating" in caplog.text
+
+
+def test_check_interval(cloner_dir_struct: Path, monkeypatch):
+	mock = MockDirStruct(cloner_dir_struct)
+	mock.config.cloner_repo_url = "git@mocked:/path"
+	timestamp_file = cloner_dir_struct.joinpath("cache", "last-check-time")
+
+	with monkeypatch.context() as m:
+		m.setattr("repo_cloner.lib.cloner.time", mock_time)
+		cloner = Cloner(mock)
+		# no timestamp file exists -> timer has to run
+		assert cloner.check_interval()
+
+		# timestamp file in future -> no run
+		timestamp_file.touch()
+		timestamp_file.write_text(str(int(mock_time() + 1)))
+		assert not cloner.check_interval()
+
+		# timestamp in history, but run everytime is set >> RUN
+		timestamp_file.write_text(str(int(mock_time() - 1)))
+		assert cloner.check_interval()
+
+		# second ago, but with 5 minute interval
+		mock.config.cloner_interval = 5
+		assert not cloner.check_interval()
+
+		# 6 minutes ago, with 5 min interval -> run!
+		timestamp_file.write_text(str(int(mock_time() - 360)))
+		assert cloner.check_interval()
+
+		# corrupted file
+		timestamp_file.write_text("1234WTF")
+		assert cloner.check_interval()
+
+
+def test__opened(cloner_dir_with_config, monkeypatch):
+	ds = MockDirStruct(cloner_dir_with_config)
+	ds.config.cloner_repo_url = cloner_dir_with_config.as_uri()
+	c = Cloner(ds)
+	# no repo defined
+	assert not c._Cloner__opened
+	# with mocked repo
+	monkeypatch.setattr(c, "_repo", MockRepoTool(cloner_dir_with_config.as_posix()))
+	c._repo.initialized = True
+	assert c._Cloner__opened
+	c._repo.initialized = False
+	assert not c._Cloner__opened
+
+
+def test_repo_path_by_url(cloner_dir_struct: Path, path_repo_base: Path):
+	ds = MockDirStruct(cloner_dir_struct)
+	ds.config.cloner_repo_url = path_repo_base.as_uri()
+	c = Cloner(ds)
+	x = c._repo_path_by_url("git@server:namespace/repo.git")
+	assert x == ds.repos_dir.joinpath("namespace_repo_3375634822.git").as_posix()
+
+
+def test__main_repo_path(cloner_dir_struct: Path, path_repo_base: Path):
+	ds = MockDirStruct(cloner_dir_struct)
+	ds.config.cloner_repo_url = path_repo_base.as_uri()
+	c = Cloner(ds)
+	x = c._Cloner__main_repo_path
+	assert x == ds.repos_dir.joinpath("_support_data_test-repo-base_402961715.git").as_posix()
+
+
+def test_sync(cloner_dir_struct, tmp_path, monkeypatch):
+	called: bool = False
+
+	def fetch():
+		nonlocal called
+		called = True
+		return True
+
+	class PatchCloner(Cloner):
+		open = False
+
+		@property
+		def __opened(self) -> bool:
+			return self.open
+
+	ds = MockDirStruct(cloner_dir_struct)
+	ds.config.cloner_repo_url = "file://wut!"
+	repo = tmp_path.joinpath("repo.git")
+	c = PatchCloner(ds)
+	# invalid repo, closed
+	c.open = False
+	assert not c.sync()
+	assert not called
+	# mocked valid repo, opened
+	c._repo = MockRepoTool(repo.as_posix())
+	c._repo.initialized = True
+	c._repo.fetch = fetch
+	assert c.sync()
+	assert called
+
+	called = False
+	# mocked repo, invalid, closed
+	c._repo.initialized = False
+	assert not called
+	assert not c.sync()
+
+
+def test_perform_check(cloner_dir_with_config, monkeypatch, caplog):
+	call_counter: int = 0
+
+	def ret_true():
+		return True
+
+	def ret_false():
+		return False
+
+	def call_counter_inc():
+		nonlocal call_counter
+		call_counter += 1
+
+	# set logger
+	caplog.set_level(logging.INFO)
+
+	mock = MockDirStruct(cloner_dir_with_config)
+	mock.config.cloner_repo_url = "git://repo"
+	cloner = Cloner(mock)
+	# timer did not pass
+	monkeypatch.setattr(cloner, "check_interval", ret_false)
+	monkeypatch.setattr(cloner, "sync", call_counter_inc)
+	assert call_counter == 0
+	cloner.perform_check()
+	assert call_counter == 0
+	# timer passed
+	monkeypatch.setattr(cloner, "check_interval", ret_true)
+	cloner.perform_check()
+	assert call_counter == 1
+	# right logs
+	assert 2 == sum(
+		r.message == "Started check for Mocked Project, url: git://repo"
+		and r.levelname == "INFO"
+		for r in caplog.records
+	)
+	assert 2 == sum(
+		r.message == "Check finished"
+		and r.levelname == "INFO"
+		for r in caplog.records
+	)
+
+
+def test_open_uninitialized(cloner_dir_with_config, path_repo_base):
+	url = path_repo_base.as_uri()
+	mock = MockDirStruct(cloner_dir_with_config)
+	mock.config.cloner_repo_url = path_repo_base.as_uri()
+	c = Cloner(mock)
+	assert not c.open(mock.config.cloner_repo_url)
+	assert c._repo_url == path_repo_base.as_uri()
+
+
+def test_open_initialized(cloner_dir_with_config, path_repo_base, caplog):
+	mock = MockDirStruct(cloner_dir_with_config)
+	path = cloner_dir_with_config.joinpath("repos", "_support_data_test-repo-base_402961715.git").as_posix()
+	mock.config.cloner_repo_url = path_repo_base.as_uri()
+	r = git.Repo().clone_from(path_repo_base.as_uri(), path, bare = True)
+	commit = r.head.commit.hexsha
+	assert r.bare
+	c = Cloner(mock)
+	assert c.open(path_repo_base.as_uri())
+	assert c._repo._repo.head.commit.hexsha == commit
+
+
+def test_clone_from_url(tmp_path, path_repo_base):
+	mock = MockDirStruct(tmp_path)
+	mock.config.cloner_repo_url = "invalid"
+	c = Cloner(mock)
+	assert c.clone_from_url(path_repo_base.as_uri())
+	assert "e0c7e2a72579e24657c05e875201011d2b48bf94" == c._repo._repo.head.commit.hexsha
+
+
+def test_clone_from_url_initialized(tmp_path, path_repo_base, caplog):
+	mock = MockDirStruct(tmp_path)
+	path = tmp_path.joinpath("repos", "_support_data_test-repo-base_402961715.git").as_posix()
+	mock.config.cloner_repo_url = path_repo_base.as_uri()
+	r = git.Repo().init(path)
+	c = Cloner(mock)
+	assert not c.clone_from_url(path_repo_base.as_uri())
+	assert caplog.records[0].levelname == "CRITICAL"
+	assert caplog.records[0].message == f"Repo path {path} is initialized... Refusing clone!"