Fix corrupt git repo bug (#25)

.pre-commit-config.yaml

@@ -7,7 +7,7 @@ repos:
         files: "^(harvest|test)"
         stages: [commit]
 -   repo: https://gitlab.com/pycqa/flake8
-    rev: 3.8.3
+    rev: 3.8.4
     hooks:
     -   id: flake8
         args: [

CHANGES.md

@@ -1,11 +1,16 @@
-# 1.1.1 (2020-09-03)
+# [1.2.0](https://github.com/ComplianceAsCode/auditree-harvest/releases/tag/v1.2.0)
+
+- [FIXED] Invalid Git repo error is now handled by a refresh of the local Git repo.
+- [ADDED] Local path to the Git repo was added as a Collator property.
+
+# [1.1.1](https://github.com/ComplianceAsCode/auditree-harvest/releases/tag/v1.1.1)
 
 - [FIXED] Credentials are no longer expected when running in local repo mode.
 
-# 1.1.0 (2020-09-02)
+# [1.1.0](https://github.com/ComplianceAsCode/auditree-harvest/releases/tag/v1.1.0)
 
 - [ADDED] "local" `repo` positional argument option is now valid bypassing remote validation.
 
-# 1.0.0 (2020-07-27)
+# [1.0.0](https://github.com/ComplianceAsCode/auditree-harvest/releases/tag/v1.0.0)
 
 - [ADDED] Made the Auditree Harvest tool public.

harvest/__init__.py

@@ -14,4 +14,4 @@
 # limitations under the License.
 """The Auditree file collating and reporting tool."""
 
-__version__ = '1.1.1'
+__version__ = '1.2.0'

harvest/collator.py

@@ -14,8 +14,10 @@
 # limitations under the License.
 """Harvest file collator."""
 import os
+import shutil
 import tempfile
 from datetime import datetime, timedelta
+from pathlib import PurePath
 from urllib.parse import urlparse
 
 import git
@@ -38,6 +40,14 @@ def __init__(self, repo_url, creds, branch, repo_path=None, validate=True):
         self.git_repo = None
         self.validate = validate
 
+    @property
+    def local_path(self):
+        """Provide the local OS path to the Git repo."""
+        if self.repo_path:
+            return self.repo_path
+        tmpdir = PurePath(tempfile.gettempdir())
+        return str(tmpdir.joinpath('harvest', self.org, self.repo))
+
     def read(self, filepath, from_dt, until_dt):
         """
         Retrieve commits from the repository based on a date range.
@@ -93,43 +103,43 @@ def write(self, filepath, commits):
                 f.write(commit.tree[filepath].data_stream.read().decode())
 
     def checkout(self):
-        """Establish/Refresh the local git repository."""
+        """Establish/Refresh the local Git repository."""
         if self.repo_path and not self.git_repo:
             self.git_repo = git.Repo(self.repo_path)
         if self.git_repo:
             if self.validate and not self._valid_repo():
                 raise ValueError(f'{self.org}/{self.repo} repository mismatch')
             return
-        local_path = os.path.join(
-            tempfile.gettempdir(), 'harvest', self.org, self.repo
-        )
-        if os.path.isdir(os.path.join(local_path, '.git')):
-            self.git_repo = git.Repo(local_path)
-            self.git_repo.remote().fetch()
-            self.git_repo.remote().pull()
-        else:
-            token = None
-            if 'github.com' in self.hostname:
-                token = self.creds['github'].token
-            elif 'github' in self.hostname:
-                token = self.creds['github_enterprise'].token
-            elif 'bitbucket' in self.hostname:
-                token = self.creds['bitbucket'].token
-            elif 'gitlab' in self.hostname:
-                token = self.creds['gitlab'].token
-            url_path = f'{self.hostname}/{self.org}/{self.repo}.git'
+        if os.path.isdir(os.path.join(self.local_path, '.git')):
             try:
-                self.git_repo = git.Repo.clone_from(
-                    f'{self.scheme}://{token}@{url_path}',
-                    local_path,
-                    branch=self.branch
-                )
-            except git.exc.GitCommandError as e:
-                raise git.exc.GitCommandError(
-                    [c.replace(token, f'{"":*<10}') for c in e.command],
-                    e.status,
-                    e.stderr.strip('\n')
-                ) from None
+                self.git_repo = git.Repo(self.local_path)
+                self.git_repo.remote().fetch()
+                self.git_repo.remote().pull()
+                return
+            except git.exc.InvalidGitRepositoryError:
+                shutil.rmtree(self.local_path)
+        token = None
+        if 'github.com' in self.hostname:
+            token = self.creds['github'].token
+        elif 'github' in self.hostname:
+            token = self.creds['github_enterprise'].token
+        elif 'bitbucket' in self.hostname:
+            token = self.creds['bitbucket'].token
+        elif 'gitlab' in self.hostname:
+            token = self.creds['gitlab'].token
+        url_path = f'{self.hostname}/{self.org}/{self.repo}.git'
+        try:
+            self.git_repo = git.Repo.clone_from(
+                f'{self.scheme}://{token}@{url_path}',
+                self.local_path,
+                branch=self.branch
+            )
+        except git.exc.GitCommandError as e:
+            raise git.exc.GitCommandError(
+                [c.replace(token, f'{"":*<10}') for c in e.command],
+                e.status,
+                e.stderr.strip('\n')
+            ) from None
 
     def _valid_repo(self):
         remote_url = self.git_repo.remotes.origin.url

test/test_collator.py

@@ -17,6 +17,7 @@
 import tempfile
 import unittest
 from datetime import datetime
+from pathlib import PurePath
 from unittest.mock import MagicMock, call, create_autospec, mock_open, patch
 
 from git import Commit, Remote, Repo
@@ -57,6 +58,10 @@ def test_constructor_default(self):
         self.assertEqual(collator.branch, 'master')
         self.assertIsNone(collator.repo_path)
         self.assertIsNone(collator.git_repo)
+        tmpdir = PurePath(tempfile.gettempdir())
+        self.assertEqual(
+            collator.local_path, str(tmpdir.joinpath('harvest', 'foo', 'bar'))
+        )
 
     def test_constructor_with_repo_path(self):
         """Ensures collate object is constructed as expected."""
@@ -69,6 +74,7 @@ def test_constructor_with_repo_path(self):
         self.assertEqual(collator.branch, 'master')
         self.assertEqual(collator.repo_path, 'my/repo/path')
         self.assertIsNone(collator.git_repo)
+        self.assertEqual(collator.local_path, 'my/repo/path')
 
     def test_read_date_comparison(self):
         """Ensures read returns commits when date logic triggers completion."""