Cloner: sync - check fingerprint before costy discovery

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

repo_cloner/lib/cloner.py

@@ -1,9 +1,9 @@
 from repo_cloner.lib import gen_repo_hashed_name
 from repo_cloner.lib import DirNotFoundError
-from repo_cloner.lib import ClonerConfig, DiskStoredList, RepoDirStructure, RepoTool
+from repo_cloner.lib import ClonerConfig, DiskStoredList, RepoDirStructure, RepoTool, DetectedCommit, Detector
 
 from pathlib import Path
-from typing import Optional
+from typing import Optional, Callable
 from time import time
 import os
 import logging
@@ -119,11 +119,17 @@ class Cloner:
 		if not self._config.cloner_submodules:
 			return self._repo.fetch()
 
+		fingerprint = self._repo.repo_fingerprint
+
 		# recursive now
 		if not self._repo.fetch():
 			log.critical(f"Repo fetch failed for {self._config.cloner_project_name}")
 			return False
 
+		if fingerprint == self._repo.repo_fingerprint:
+			log.info(f"Repo fingerpring unchanged - submodule discovery skipped")
+			return True
+
 		log.debug(f"Loading submodules.cache")
 		submodules = DiskStoredList(os.path.join(self.__submodule_cache, "submodules.cache"))
 		log.debug(f"Loaded submodules.cache - {len(submodules)} items")
@@ -207,3 +213,7 @@ class Cloner:
 	@property
 	def detector_enabled(self) -> bool:
 		return os.path.exists(os.path.join(self._dirs.conf_dir, self.__detector_cfg))
+
+	def run_detector(self, callback: Callable[[DetectedCommit], None]):
+		detector = Detector(self.__main_repo_path, self._dirs.cache_dir, self._config.cloner_project_name)
+		detector.run(callback)

tests/lib/test_cloner.py

@@ -309,6 +309,7 @@ def test_submodules_sync_succeed(cloner_dir_with_config, caplog):
 		'initialized':      PropertyMock(side_effect = [True, False, True, True, True]),
 		'fetch':            MagicMock(return_value = True),
 		'clone':            MagicMock(return_value = True),
+		'repo_fingerprint': PropertyMock(side_effect = ["fp1", "fp2", "fp3"]),
 	}
 
 	with patch.multiple("repo_cloner.lib.cloner.RepoTool", **mocks):
@@ -320,6 +321,33 @@ def test_submodules_sync_succeed(cloner_dir_with_config, caplog):
 		assert repo_cloner.lib.cloner.RepoTool.clone.call_count == 1
 
 
+def test_submodules_sync_unchanged_main_repo(cloner_dir_with_config, caplog):
+	caplog.set_level(0)
+	cloner_dir_with_config.joinpath("cache", "submodules").mkdir(parents = True)
+	cloner_dir_with_config.joinpath("cache", "submodules", "submodules.cache") \
+		.write_text("https://git.hosting:previous/submodule.git")
+	ds = MockDirStruct(cloner_dir_with_config)
+	ds.config.cloner_repo_url = "https://git.hosting:namespace/whatever.git"
+	ds.config.cloner_submodules = True
+
+	# mock almost everything
+	mocks = {
+		'initialized':      PropertyMock(side_effect = [True, False, True, True, True]),
+		'fetch':            MagicMock(return_value = True),
+		'clone':            MagicMock(return_value = True),
+		# fingerprint is same for both queries, so nothing is fetched/synced afterwards
+		'repo_fingerprint': PropertyMock(return_value = "fp1"),
+	}
+
+	with patch.multiple("repo_cloner.lib.cloner.RepoTool", **mocks):
+		Cloner.check_submodules_repo = my_check_submodules_repo
+		cl = Cloner(ds)
+		assert cl.sync()
+
+		assert repo_cloner.lib.cloner.RepoTool.fetch.call_count == 1
+		assert repo_cloner.lib.cloner.RepoTool.clone.call_count == 0
+
+
 def test_submodules_sync_one_fail(cloner_dir_with_config, caplog):
 	cloner_dir_with_config.joinpath("cache", "submodules").mkdir(parents = True)
 	cloner_dir_with_config.joinpath("cache", "submodules", "submodules.cache") \
@@ -335,6 +363,7 @@ def test_submodules_sync_one_fail(cloner_dir_with_config, caplog):
 		'initialized':      PropertyMock(return_value = True),
 		'fetch':            MagicMock(side_effect = [True, True, True, False, True]),
 		'clone':            MagicMock(return_value = True),
+		'repo_fingerprint': PropertyMock(side_effect = ["fp1", "fp2", "fp3"]),
 	}
 
 	with patch.multiple("repo_cloner.lib.cloner.RepoTool", **mocks):
@@ -458,6 +487,7 @@ def test_clone_recursive(tmp_path, path_repo_base, caplog):
 		patch_clone_recursive.assert_called_with(
 			'https://repo', tmp_path.joinpath('cache', 'submodules').as_posix(), scan_depth = None)
 
+
 def test_detector_enabled(cloner_dir_with_config):
 	ds = MockDirStruct(cloner_dir_with_config)
 	ds.config.cloner_repo_url = "http://mock"
@@ -465,4 +495,3 @@ def test_detector_enabled(cloner_dir_with_config):
 	assert not cl.detector_enabled
 	Path(cloner_dir_with_config).joinpath("config", "detector.cfg").touch()
 	assert cl.detector_enabled
-