github-backup 0.64.0__tar.gz → 0.64.2__tar.gz

{github_backup-0.64.0 → github_backup-0.64.2}/CHANGES.rst

@@ -1,9 +1,21 @@
 Changelog
 =========
 
-0.64.0 (2026-07-07)
+0.64.2 (2026-07-18)
 -------------------
 ------------------------
+- Add tests. [Duncan Ogilvie]
+- Skip fetching gists that have not been updated. [Duncan Ogilvie]
+- Skip retrieving repositories when not necessary. [Duncan Ogilvie]
+
+
+0.64.1 (2026-07-17)
+-------------------
+- Fix exception when DMCA'd/empty gists were backed up. [Duncan Ogilvie]
+
+
+0.64.0 (2026-07-07)
+-------------------
 
 Fix
 ~~~

{github_backup-0.64.0/github_backup.egg-info → github_backup-0.64.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: github-backup
-Version: 0.64.0
+Version: 0.64.2
 Summary: backup a github user or organization
 Home-page: http://github.com/josegonzalez/python-github-backup
 Author: Jose Diaz-Gonzalez

github_backup-0.64.2/github_backup/__init__.py

@@ -0,0 +1 @@
+__version__ = "0.64.2"

{github_backup-0.64.0 → github_backup-0.64.2}/github_backup/github_backup.py

@@ -1789,7 +1789,39 @@ def check_git_lfs_install():
         )
 
 
+# Resources fetched from their own endpoints, not from the repository listing
+# (gists and account-level data). Every other include_* flag needs the listing.
+NON_REPOSITORY_RESOURCES = frozenset(
+    {
+        "include_gists",
+        "include_starred_gists",
+        "include_starred",
+        "include_watched",
+        "include_followers",
+        "include_following",
+    }
+)
+
+
+def repository_list_needed(args):
+    """The base repository listing is only needed when a per-repository
+    resource was requested; e.g. a gists-only backup can skip it entirely.
+    Flags not listed in NON_REPOSITORY_RESOURCES default to needing it, so a
+    newly added resource can never be silently skipped."""
+    if args.repository:
+        return True
+    return any(
+        value
+        for name, value in vars(args).items()
+        if name.startswith("include_") and name not in NON_REPOSITORY_RESOURCES
+    )
+
+
 def retrieve_repositories(args, authenticated_user):
+    if not repository_list_needed(args):
+        repos = []
+        return retrieve_additional_repositories(args, authenticated_user, repos)
+
     logger.info("Retrieving repositories")
     paginated = True
     if args.user == authenticated_user["login"]:
@@ -1825,6 +1857,12 @@ def retrieve_repositories(args, authenticated_user):
             logger.warning(f"Legal notice: {e.legal_url}")
         return []
 
+    return retrieve_additional_repositories(args, authenticated_user, repos)
+
+
+def retrieve_additional_repositories(args, authenticated_user, repos):
+    """Append starred repositories, gists and starred gists to the base
+    repository list when requested."""
     if args.all_starred:
         starred_template = "https://{0}/users/{1}/starred".format(
             get_github_api_host(args), args.user
@@ -2058,6 +2096,24 @@ def remove_legacy_last_update_if_migrated(
     )
 
 
+def gist_backup_is_current(repository, repo_cwd, repo_dir):
+    """Whether fetching a gist can be skipped because a clone exists and the
+    stored gist.json matches the listing's updated_at. Any edit to a gist —
+    including a git push — bumps updated_at, so this can never miss a content
+    change; at worst a comment bumps it and causes a harmless refetch."""
+    if not os.path.exists(repo_dir):
+        return False
+    updated_at = repository.get("updated_at")
+    if not updated_at:
+        return False
+    try:
+        with codecs.open(os.path.join(repo_cwd, "gist.json"), encoding="utf-8") as f:
+            stored = json.load(f)
+    except (OSError, ValueError):
+        return False
+    return stored.get("updated_at") == updated_at
+
+
 def backup_repositories(args, output_directory, repositories):
     logger.info("Backing up repositories")
     repos_template = "https://{0}/repos".format(get_github_api_host(args))
@@ -2065,6 +2121,7 @@ def backup_repositories(args, output_directory, repositories):
         args, output_directory
     )
     incremental_resource_work_attempted = False
+    skipped_unchanged_gists = 0
 
     for repository in repositories:
         if repository.get("is_gist"):
@@ -2098,18 +2155,30 @@ def backup_repositories(args, output_directory, repositories):
                 if not repository.get("is_gist")
                 else repository.get("id")
             )
-            fetch_repository(
-                repo_name,
-                repo_url,
-                repo_dir,
-                skip_existing=args.skip_existing,
-                bare_clone=args.bare_clone,
-                lfs_clone=args.lfs_clone,
-                no_prune=args.no_prune,
-            )
+            if repository.get("is_gist") and gist_backup_is_current(
+                repository, repo_cwd, repo_dir
+            ):
+                logger.debug(
+                    "Skipping gist {0} (unchanged since last backup)".format(repo_name)
+                )
+                skipped_unchanged_gists += 1
+            else:
+                fetch_repository(
+                    repo_name,
+                    repo_url,
+                    repo_dir,
+                    skip_existing=args.skip_existing,
+                    bare_clone=args.bare_clone,
+                    lfs_clone=args.lfs_clone,
+                    no_prune=args.no_prune,
+                )
 
             if repository.get("is_gist"):
-                # dump gist information to a file as well
+                # dump gist information to a file as well, even when the fetch
+                # was skipped, so listing metadata (fork counts etc.) stays
+                # fresh; the clone may also have been skipped entirely (e.g.
+                # DMCA-blocked or empty gist), so make sure the directory exists
+                mkdir_p(repo_cwd)
                 output_file = "{0}/gist.json".format(repo_cwd)
                 with codecs.open(output_file, "w", encoding="utf-8") as f:
                     json_dump(repository, f)
@@ -2178,6 +2247,9 @@ def backup_repositories(args, output_directory, repositories):
             logger.info(f"Skipping remaining resources for {repository['full_name']}")
             continue
 
+    if skipped_unchanged_gists:
+        logger.info("Skipped {0} unchanged gists".format(skipped_unchanged_gists))
+
     if incremental_resource_work_attempted:
         remove_legacy_last_update_if_migrated(
             args, output_directory, legacy_last_update_path

{github_backup-0.64.0 → github_backup-0.64.2/github_backup.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: github-backup
-Version: 0.64.0
+Version: 0.64.2
 Summary: backup a github user or organization
 Home-page: http://github.com/josegonzalez/python-github-backup
 Author: Jose Diaz-Gonzalez

{github_backup-0.64.0 → github_backup-0.64.2}/github_backup.egg-info/SOURCES.txt

@@ -23,6 +23,7 @@ tests/test_auth.py
 tests/test_case_sensitivity.py
 tests/test_disabled_issues_pulls.py
 tests/test_discussions.py
+tests/test_gists.py
 tests/test_http_451.py
 tests/test_incremental_per_repository.py
 tests/test_json_dump_if_changed.py

github_backup-0.64.2/tests/test_gists.py

@@ -0,0 +1,227 @@
+"""Tests for gist backup behavior: skipping the repository listing for
+gists-only backups, skipping fetches for unchanged gists, and writing
+gist.json even when the clone was skipped (e.g. DMCA-blocked gists)."""
+
+import json
+import os
+
+import pytest
+from unittest.mock import patch
+
+from github_backup.github_backup import (
+    backup_repositories,
+    gist_backup_is_current,
+    repository_list_needed,
+    retrieve_repositories,
+)
+
+
+class TestRepositoryListNeeded:
+    """The base /user/repos listing should only be fetched when a
+    per-repository resource was requested."""
+
+    def test_gists_only_does_not_need_listing(self, create_args):
+        args = create_args(include_gists=True, include_starred_gists=True)
+        assert not repository_list_needed(args)
+
+    def test_account_level_flags_do_not_need_listing(self, create_args):
+        args = create_args(
+            include_starred=True,
+            include_watched=True,
+            include_followers=True,
+            include_following=True,
+        )
+        assert not repository_list_needed(args)
+
+    def test_all_starred_does_not_need_listing(self, create_args):
+        # starred repos come from /users/{user}/starred, not /user/repos
+        args = create_args(all_starred=True)
+        assert not repository_list_needed(args)
+
+    @pytest.mark.parametrize(
+        "flag",
+        [
+            "include_repository",
+            "include_everything",
+            "include_wiki",
+            "include_issues",
+            "include_pulls",
+            "include_releases",
+            "include_attachments",
+        ],
+    )
+    def test_repository_scoped_flags_need_listing(self, create_args, flag):
+        args = create_args(**{flag: True})
+        assert repository_list_needed(args)
+
+    def test_single_repository_needs_listing(self, create_args):
+        args = create_args(repository="some-repo")
+        assert repository_list_needed(args)
+
+    def test_unknown_include_flag_defaults_to_needing_listing(self, create_args):
+        # Fail-safe: a future include_* flag that nobody registered in
+        # NON_REPOSITORY_RESOURCES must not silently skip the listing.
+        args = create_args(include_some_future_resource=True)
+        assert repository_list_needed(args)
+
+
+class TestRetrieveRepositoriesSkipsListing:
+    """retrieve_repositories should not hit the repository endpoints for a
+    gists-only backup, but must still fetch gists and starred gists."""
+
+    @patch("github_backup.github_backup.retrieve_data")
+    def test_gists_only_skips_repo_listing(self, mock_retrieve, create_args):
+        args = create_args(include_gists=True, include_starred_gists=True)
+        mock_retrieve.return_value = [{"id": "abc123"}]
+
+        repos = retrieve_repositories(args, {"login": "testuser"})
+
+        requested = [call.args[1] for call in mock_retrieve.call_args_list]
+        assert not any("/repos" in url for url in requested), requested
+        assert any(url.endswith("/users/testuser/gists") for url in requested)
+        assert any(url.endswith("/gists/starred") for url in requested)
+        assert all(r["is_gist"] for r in repos)
+
+    @patch("github_backup.github_backup.retrieve_data")
+    def test_repository_backup_still_fetches_listing(self, mock_retrieve, create_args):
+        args = create_args(include_repository=True)
+        mock_retrieve.return_value = []
+
+        retrieve_repositories(args, {"login": "testuser"})
+
+        requested = [call.args[1] for call in mock_retrieve.call_args_list]
+        assert any(url.endswith("/user/repos") for url in requested), requested
+
+
+class TestGistBackupIsCurrent:
+    """A gist fetch is skipped only when a clone exists and the stored
+    gist.json updated_at matches the listing."""
+
+    GIST = {"id": "abc123", "is_gist": True, "updated_at": "2026-07-17T10:00:00Z"}
+
+    def _write_backup(self, repo_cwd, updated_at="2026-07-17T10:00:00Z", clone=True):
+        if clone:
+            os.makedirs(os.path.join(repo_cwd, "repository"))
+        else:
+            os.makedirs(repo_cwd)
+        with open(os.path.join(repo_cwd, "gist.json"), "w") as f:
+            json.dump({"id": "abc123", "updated_at": updated_at}, f)
+
+    def test_current_when_unchanged(self, tmp_path):
+        repo_cwd = str(tmp_path / "gists" / "abc123")
+        self._write_backup(repo_cwd)
+        assert gist_backup_is_current(
+            self.GIST, repo_cwd, os.path.join(repo_cwd, "repository")
+        )
+
+    def test_not_current_when_updated(self, tmp_path):
+        repo_cwd = str(tmp_path / "gists" / "abc123")
+        self._write_backup(repo_cwd, updated_at="2020-01-01T00:00:00Z")
+        assert not gist_backup_is_current(
+            self.GIST, repo_cwd, os.path.join(repo_cwd, "repository")
+        )
+
+    def test_not_current_without_clone(self, tmp_path):
+        # gist.json alone is not enough; e.g. the clone was skipped because
+        # the gist was inaccessible, so keep retrying the fetch
+        repo_cwd = str(tmp_path / "gists" / "abc123")
+        self._write_backup(repo_cwd, clone=False)
+        assert not gist_backup_is_current(
+            self.GIST, repo_cwd, os.path.join(repo_cwd, "repository")
+        )
+
+    def test_not_current_with_corrupt_gist_json(self, tmp_path):
+        repo_cwd = str(tmp_path / "gists" / "abc123")
+        os.makedirs(os.path.join(repo_cwd, "repository"))
+        with open(os.path.join(repo_cwd, "gist.json"), "w") as f:
+            f.write("{not json")
+        assert not gist_backup_is_current(
+            self.GIST, repo_cwd, os.path.join(repo_cwd, "repository")
+        )
+
+    def test_not_current_without_updated_at(self, tmp_path):
+        repo_cwd = str(tmp_path / "gists" / "abc123")
+        self._write_backup(repo_cwd)
+        assert not gist_backup_is_current(
+            {"id": "abc123", "is_gist": True},
+            repo_cwd,
+            os.path.join(repo_cwd, "repository"),
+        )
+
+
+class TestGistBackup:
+    """End-to-end behavior of backup_repositories for gists."""
+
+    def _gist(self, **extra):
+        gist = {
+            "id": "abc123",
+            "is_gist": True,
+            "updated_at": "2026-07-17T10:00:00Z",
+            "git_pull_url": "https://gist.github.com/abc123.git",
+        }
+        gist.update(extra)
+        return gist
+
+    @patch("github_backup.github_backup.fetch_repository")
+    def test_new_gist_is_fetched_and_metadata_written(
+        self, mock_fetch, create_args, tmp_path
+    ):
+        args = create_args(include_gists=True)
+
+        backup_repositories(args, str(tmp_path), [self._gist()])
+
+        mock_fetch.assert_called_once()
+        gist_json = tmp_path / "gists" / "abc123" / "gist.json"
+        assert json.loads(gist_json.read_text())["id"] == "abc123"
+
+    @patch("github_backup.github_backup.fetch_repository")
+    def test_unchanged_gist_skips_fetch_but_refreshes_metadata(
+        self, mock_fetch, create_args, tmp_path
+    ):
+        args = create_args(include_gists=True)
+        repo_cwd = tmp_path / "gists" / "abc123"
+        (repo_cwd / "repository").mkdir(parents=True)
+        (repo_cwd / "gist.json").write_text(
+            json.dumps({"id": "abc123", "updated_at": "2026-07-17T10:00:00Z"})
+        )
+
+        backup_repositories(args, str(tmp_path), [self._gist(forks=5)])
+
+        assert not mock_fetch.called, "unchanged gist should not be fetched"
+        # gist.json is rewritten from the listing so metadata that changes
+        # without bumping updated_at (fork counts etc.) stays fresh
+        assert json.loads((repo_cwd / "gist.json").read_text())["forks"] == 5
+
+    @patch("github_backup.github_backup.fetch_repository")
+    def test_updated_gist_is_fetched(self, mock_fetch, create_args, tmp_path):
+        args = create_args(include_gists=True)
+        repo_cwd = tmp_path / "gists" / "abc123"
+        (repo_cwd / "repository").mkdir(parents=True)
+        (repo_cwd / "gist.json").write_text(
+            json.dumps({"id": "abc123", "updated_at": "2020-01-01T00:00:00Z"})
+        )
+
+        backup_repositories(args, str(tmp_path), [self._gist()])
+
+        mock_fetch.assert_called_once()
+
+    @patch("github_backup.github_backup.fetch_repository")
+    def test_inaccessible_gist_does_not_crash_backup(
+        self, mock_fetch, create_args, tmp_path
+    ):
+        """Regression test: fetch_repository skips inaccessible gists (e.g.
+        DMCA-blocked) without creating the clone directory; writing gist.json
+        must not crash and the next gist must still be processed."""
+        args = create_args(include_gists=True)
+        # mock fetch never creates the directory, like a skipped clone
+        blocked = self._gist(id="deadbeef")
+        second = self._gist()
+
+        backup_repositories(args, str(tmp_path), [blocked, second])
+
+        assert (tmp_path / "gists" / "deadbeef" / "gist.json").is_file()
+        assert (tmp_path / "gists" / "abc123" / "gist.json").is_file()
+
+
+if __name__ == "__main__":
+    pytest.main([__file__, "-v"])

github_backup-0.64.0/github_backup/__init__.py

@@ -1 +0,0 @@
-__version__ = "0.64.0"