githubtakeout 0.1.16__tar.gz → 0.1.18__tar.gz

{githubtakeout-0.1.16/src/githubtakeout.egg-info → githubtakeout-0.1.18}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: githubtakeout
-Version: 0.1.16
+Version: 0.1.18
 Summary: Backup and archive Git Repos and Gists from GitHub
 Author: Corey Goldberg
 Maintainer: Corey Goldberg
@@ -101,24 +101,27 @@ the `--token` argument.
 ## CLI Options:
 
 ```
-usage: githubtakeout [-h] [--dir DIR] [--pattern PATTERN] [--format {tar,zip,none}]
-                     [--gists] [--history] [--keep] [--list] [--token]
+$ githubtakeout --help
+usage: githubtakeout [-h] [--dir DIR] [--pattern PATTERN] [--skip_pattern PATTERN]
+                     [--format {tar,zip,none}] [--gists] [--history]
+                     [--skip_forks] [--keep] [--list] [--token]
                      username
 
 positional arguments:
-  username                    github username
+  username                 github username
 
 options:
-  -h, --help                  show this help message and exit
-  --dir DIR                   output directory (default: .)
-  --pattern PATTERN           regex matching repo names to include
-  --skip_pattern PATTERN      regex matching repo names to skip
-  --format {tar,zip,none}     archive format (default: zip)
-  --gists                     include gists
-  --history                   include commit history and branches (.git directory)
-  --keep                      keep repos after archiving
-  --list                      list repos only
-  --token                     prompt for auth token
+  -h, --help               show this help message and exit
+  --dir DIR                output directory (default: .)
+  --pattern PATTERN        regex matching repo names to include
+  --skip_pattern PATTERN   regex matching repo names to skip
+  --format {tar,zip,none}  archive format (default: zip)
+  --gists                  include gists
+  --history                include commit history and branches (.git directory)
+  --skip_forks             skip repos that are forks
+  --keep                   keep repos after archiving
+  --list                   list repos only
+  --token                  prompt for auth token
 ```
 
 ## Usage Examples:

{githubtakeout-0.1.16 → githubtakeout-0.1.18}/README.md

@@ -74,24 +74,27 @@ the `--token` argument.
 ## CLI Options:
 
 ```
-usage: githubtakeout [-h] [--dir DIR] [--pattern PATTERN] [--format {tar,zip,none}]
-                     [--gists] [--history] [--keep] [--list] [--token]
+$ githubtakeout --help
+usage: githubtakeout [-h] [--dir DIR] [--pattern PATTERN] [--skip_pattern PATTERN]
+                     [--format {tar,zip,none}] [--gists] [--history]
+                     [--skip_forks] [--keep] [--list] [--token]
                      username
 
 positional arguments:
-  username                    github username
+  username                 github username
 
 options:
-  -h, --help                  show this help message and exit
-  --dir DIR                   output directory (default: .)
-  --pattern PATTERN           regex matching repo names to include
-  --skip_pattern PATTERN      regex matching repo names to skip
-  --format {tar,zip,none}     archive format (default: zip)
-  --gists                     include gists
-  --history                   include commit history and branches (.git directory)
-  --keep                      keep repos after archiving
-  --list                      list repos only
-  --token                     prompt for auth token
+  -h, --help               show this help message and exit
+  --dir DIR                output directory (default: .)
+  --pattern PATTERN        regex matching repo names to include
+  --skip_pattern PATTERN   regex matching repo names to skip
+  --format {tar,zip,none}  archive format (default: zip)
+  --gists                  include gists
+  --history                include commit history and branches (.git directory)
+  --skip_forks             skip repos that are forks
+  --keep                   keep repos after archiving
+  --list                   list repos only
+  --token                  prompt for auth token
 ```
 
 ## Usage Examples:

{githubtakeout-0.1.16 → githubtakeout-0.1.18}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "githubtakeout"
-version = "0.1.16"
+version = "0.1.18"
 description = "Backup and archive Git Repos and Gists from GitHub"
 license = "MIT"
 license-files = ["LICENSE"]
@@ -62,13 +62,13 @@ addopts = "-vv" # extra verbose
 tmp_path_retention_policy = "none"
 
 [tool.black]
-line-length = 120
+line-length = 88
 target-version = [
     "py312", "py313", "py314"
 ]
 
 [tool.ruff]
-line-length = 120
+line-length = 88
 respect-gitignore = true
 target-version = "py312"
 
@@ -84,10 +84,10 @@ fixable = ["ALL"]
 
 [tool.ruff.format]
 docstring-code-format = true
-docstring-code-line-length = 120
+docstring-code-line-length = 88
 
 [tool.refurb]
-python_version = "3.9"
+python_version = "3.12"
 enable_all = true
 disable = [
     "FURB104", "FURB107", "FURB147", "FURB173", "FURB183", "FURB184"

{githubtakeout-0.1.16 → githubtakeout-0.1.18/src/githubtakeout.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: githubtakeout
-Version: 0.1.16
+Version: 0.1.18
 Summary: Backup and archive Git Repos and Gists from GitHub
 Author: Corey Goldberg
 Maintainer: Corey Goldberg
@@ -101,24 +101,27 @@ the `--token` argument.
 ## CLI Options:
 
 ```
-usage: githubtakeout [-h] [--dir DIR] [--pattern PATTERN] [--format {tar,zip,none}]
-                     [--gists] [--history] [--keep] [--list] [--token]
+$ githubtakeout --help
+usage: githubtakeout [-h] [--dir DIR] [--pattern PATTERN] [--skip_pattern PATTERN]
+                     [--format {tar,zip,none}] [--gists] [--history]
+                     [--skip_forks] [--keep] [--list] [--token]
                      username
 
 positional arguments:
-  username                    github username
+  username                 github username
 
 options:
-  -h, --help                  show this help message and exit
-  --dir DIR                   output directory (default: .)
-  --pattern PATTERN           regex matching repo names to include
-  --skip_pattern PATTERN      regex matching repo names to skip
-  --format {tar,zip,none}     archive format (default: zip)
-  --gists                     include gists
-  --history                   include commit history and branches (.git directory)
-  --keep                      keep repos after archiving
-  --list                      list repos only
-  --token                     prompt for auth token
+  -h, --help               show this help message and exit
+  --dir DIR                output directory (default: .)
+  --pattern PATTERN        regex matching repo names to include
+  --skip_pattern PATTERN   regex matching repo names to skip
+  --format {tar,zip,none}  archive format (default: zip)
+  --gists                  include gists
+  --history                include commit history and branches (.git directory)
+  --skip_forks             skip repos that are forks
+  --keep                   keep repos after archiving
+  --list                   list repos only
+  --token                  prompt for auth token
 ```
 
 ## Usage Examples:

{githubtakeout-0.1.16 → githubtakeout-0.1.18}/src/githubtakeout.py

@@ -51,27 +51,28 @@ def add_creds(url, username, token):
     return new_url
 
 
-def archive(local_repo_dir, archive_format="zip", is_gist=False):
+def archive(local_repo_dir, archive_format="zip", archive_basename=None):
     if archive_format not in ARCHIVE_FORMATS:
         raise ValueError(f"{archive_format} is not a valid archive format")
     if archive_format == "none":
         return None
-    base_name = os.path.basename(local_repo_dir)
+    basename = os.path.basename(local_repo_dir)
     extension = "tar.gz" if archive_format == "tar" else archive_format
-    archive_name = f"{base_name}.{extension}"
-    if is_gist:
-        archive_name = f"gist-{archive_name}"
+    if archive_basename is None:
+        archive_name = f"{basename}.{extension}"
+    else:
+        archive_name = f"{archive_basename}.{extension}"
     parent_dir = os.path.dirname(local_repo_dir)
     archive_path = os.path.join(parent_dir, archive_name)
     logger.info(f"creating archive: {archive_path}")
     if archive_format == "tar":
         with tarfile.open(archive_path, "w:gz") as tar_archive:
-            tar_archive.add(local_repo_dir, arcname=base_name)
+            tar_archive.add(local_repo_dir, arcname=basename)
     elif archive_format == "zip":
         with zipfile.ZipFile(archive_path, "w", zipfile.ZIP_DEFLATED) as zip_archive:
             repo_path = Path(local_repo_dir)
             for entry in repo_path.rglob("*"):
-                path = os.path.join(base_name, entry.relative_to(repo_path))
+                path = os.path.join(basename, entry.relative_to(repo_path))
                 zip_archive.write(entry, arcname=path)
     return archive_path
 
@@ -90,7 +91,7 @@ def clone(repo_url, local_repo_dir, include_history):
             repo = git.Repo.clone_from(
                 url=repo_url,
                 to_path=local_repo_dir,
-                multi_options=["--single-branch", "--depth=1"],
+                multi_options=["--depth=1"],
                 progress=GitProgress(),
             )
     except git.GitCommandError as e:
@@ -123,7 +124,9 @@ def pull(local_repo_dir):
             pass
 
 
-def get_and_archive_repo(repo_url, local_repo_dir, archive_format, include_history, keep, is_gist=False):
+def get_and_archive_repo(
+    repo_url, local_repo_dir, archive_format, include_history, keep, description=None
+):
     def remove_readonly(func, path, _):
         # This is necessary so rmtree() doesn't fail if there are any readonly
         # dirs/files when trying to delete. This seems to happen after cloning on
@@ -161,7 +164,15 @@ def get_and_archive_repo(repo_url, local_repo_dir, archive_format, include_histo
             shutil.rmtree(git_dir, onexc=remove_readonly)
         except FileNotFoundError:
             pass
-    archive_path = archive(local_repo_dir, archive_format, is_gist)
+    if description:
+        # clean unsafe chars and truncate description to create a useable file name
+        clean_name = re.sub(r"[/\\?%*:|\"<>\x7F\x00-\x1F]", "-", description)[:255]
+        archive_basename = f"gist - {clean_name}"
+    else:
+        archive_basename = None
+    archive_path = archive(
+        local_repo_dir, archive_format, archive_basename=archive_basename
+    )
     if archive_path:
         size = convert_size(os.path.getsize(archive_path))
         logger.info(f"archive size: {size}")
@@ -206,10 +217,19 @@ def get_repos(username, token, include_gists):
     if include_gists:
         gists = user.get_gists()
     else:
-        gists = None
+        gists = []
     return repos, gists
 
 
+def filter_repos(repos, pattern, skip_pattern, skip_forks):
+    repos = (repo for repo in repos if re.match(pattern, repo.name))
+    if skip_pattern:
+        repos = (repo for repo in repos if not re.match(skip_pattern, repo.name))
+    if skip_forks:
+        repos = (repo for repo in repos if not repo.fork)
+    return repos
+
+
 def get_token(prompt_for_token):
     if prompt_for_token:
         token = getpass.getpass("Token:")
@@ -230,49 +250,52 @@ def run(
     archive_format,
     include_gists,
     include_history,
+    skip_forks,
     keep,
     list_only,
     prompt_for_token,
 ):
     working_dir = os.path.join(base_dir, "backups")
     token = get_token(prompt_for_token)
-    repos, gists = get_repos(username, token, include_gists)
-    num_repos = len(
-        [
-            repo
-            for repo in repos
-            if re.match(pattern, repo.name) and not (re.match(skip_pattern, repo.name) if skip_pattern else False)
-        ]
-    )
+
+    all_repos, gists = get_repos(username, token, include_gists)
+    repos = filter_repos(all_repos, pattern, skip_pattern, skip_forks)
+    num_repos = sum(1 for repo in repos)
+    num_gists = sum(1 for gist in gists)
+
+    # we have to fetch the repos again because we exhaust the generator
+    # when counting repos above
+    all_repos, gists = get_repos(username, token, include_gists)
+    repos = filter_repos(all_repos, pattern, skip_pattern, skip_forks)
     if not list_only:
         logger.info(f"creating archives in: {working_dir}\n")
     logger.info(f"found {num_repos} repos for user '{username}':\n")
     for repo in repos:
-        if re.match(pattern, repo.name) and not (re.match(skip_pattern, repo.name) if skip_pattern else False):
-            local_repo_dir = os.path.join(working_dir, repo.name)
-            url = add_creds(repo.clone_url, username, token)
-            if list_only:
-                logger.info(f"{username}/{repo.name}")
-            else:
-                get_and_archive_repo(url, local_repo_dir, archive_format, include_history, keep)
-    if gists is not None:
-        num_gists = len(
-            [
-                gist
-                for gist in gists
-                if re.match(pattern, gist.id) and not (re.match(skip_pattern, gist.id) if skip_pattern else False)
-            ]
-        )
+        local_repo_dir = os.path.join(working_dir, repo.name)
+        url = add_creds(repo.clone_url, username, token)
+        if list_only:
+            logger.info(f"{username}/{repo.name}")
+        else:
+            get_and_archive_repo(
+                url, local_repo_dir, archive_format, include_history, keep
+            )
+    if include_gists:
         logger.info("")
         logger.info(f"found {num_gists} gists for user '{username}':\n")
         for gist in gists:
-            if re.match(pattern, gist.id) and not (re.match(skip_pattern, gist.id) if skip_pattern else False):
-                local_repo_dir = os.path.join(working_dir, gist.id)
-                url = add_creds(gist.git_pull_url, username, token)
-                if list_only:
-                    logger.info(f"{username}/{gist.id}\n  - {gist.description}")
-                else:
-                    get_and_archive_repo(url, local_repo_dir, archive_format, include_history, keep, is_gist=True)
+            local_repo_dir = os.path.join(working_dir, gist.id)
+            url = add_creds(gist.git_pull_url, username, token)
+            if list_only:
+                logger.info(f"{username}/{gist.id}\n  - {gist.description}")
+            else:
+                get_and_archive_repo(
+                    url,
+                    local_repo_dir,
+                    archive_format,
+                    include_history,
+                    keep,
+                    description=gist.description,
+                )
 
 
 def main():
@@ -305,16 +328,30 @@ def main():
         default="zip",
         help="archive format (default: %(default)s)",
     )
-    parser.add_argument("--gists", action="store_true", default=False, help="include gists")
+    parser.add_argument(
+        "--gists", action="store_true", default=False, help="include gists"
+    )
     parser.add_argument(
         "--history",
         action="store_true",
         default=False,
         help="include commit history and branches (.git directory)",
     )
-    parser.add_argument("--keep", action="store_true", default=False, help="keep repos after archiving")
-    parser.add_argument("--list", action="store_true", default=False, help="list repos only")
-    parser.add_argument("--token", action="store_true", default=False, help="prompt for auth token")
+    parser.add_argument(
+        "--skip_forks",
+        action="store_true",
+        default=False,
+        help="skip repos that are forks",
+    )
+    parser.add_argument(
+        "--keep", action="store_true", default=False, help="keep repos after archiving"
+    )
+    parser.add_argument(
+        "--list", action="store_true", default=False, help="list repos only"
+    )
+    parser.add_argument(
+        "--token", action="store_true", default=False, help="prompt for auth token"
+    )
     args = parser.parse_args()
     try:
         run(
@@ -325,6 +362,7 @@ def main():
             archive_format=args.format,
             include_gists=args.gists,
             include_history=args.history,
+            skip_forks=args.skip_forks,
             keep=args.keep,
             list_only=args.list,
             prompt_for_token=args.token,

{githubtakeout-0.1.16 → githubtakeout-0.1.18}/src/progress.py

@@ -20,7 +20,9 @@ class GitProgress(git.RemoteProgress):
         "RESOLVING",
         "WRITING",
     ]
-    OP_CODE_MAP = {getattr(git.RemoteProgress, _op_code): _op_code for _op_code in OP_CODES}
+    OP_CODE_MAP = {
+        getattr(git.RemoteProgress, _op_code): _op_code for _op_code in OP_CODES
+    }
 
     def __init__(self):
         super().__init__()

{githubtakeout-0.1.16 → githubtakeout-0.1.18}/tests/test_githubtakeout.py

@@ -15,10 +15,15 @@ def require_token():
     load_dotenv()
     required = "GITHUB_TOKEN"
     if not os.getenv(required):
-        pytest.skip(reason=f"{required} environment variable or .env file is required to avoid rate limiting")
+        pytest.skip(
+            reason=f"{required} environment variable or .env file is "
+            "required to avoid rate limiting"
+        )
 
 
-@pytest.mark.skipif(os.getenv("GITHUB_TOKEN"), reason="GITHUB_TOKEN required to avoid rate limiting")
+@pytest.mark.skipif(
+    os.getenv("GITHUB_TOKEN"), reason="GITHUB_TOKEN required to avoid rate limiting"
+)
 def test_run_list_1_match(tmp_path, caplog):
     caplog.set_level("INFO")
     repo = "githubtakeout"
@@ -30,13 +35,14 @@ def test_run_list_1_match(tmp_path, caplog):
         archive_format="none",
         include_gists=False,
         include_history=False,
+        skip_forks=False,
         keep=False,
         list_only=True,
         prompt_for_token=False,
     )
     assert "creating archives" not in caplog.text
     assert f"found 1 repos for user '{USER}'" in caplog.text
-    assert "gists" not in caplog.text
+    assert not re.search(r"found \d* gists", caplog.text)
     assert f"{USER}/{repo}" in caplog.text
     assert not Path(tmp_path / "backups" / repo).exists()
 
@@ -52,13 +58,14 @@ def test_run_list_0_match(tmp_path, caplog):
         archive_format="none",
         include_gists=False,
         include_history=False,
+        skip_forks=False,
         keep=False,
         list_only=True,
         prompt_for_token=False,
     )
     assert "creating archives" not in caplog.text
-    assert f"found 0 repos for user '{USER}'" in caplog.text
-    assert "gists" not in caplog.text
+    assert "found 0 repos" in caplog.text
+    assert not re.search(r"found \d* gists", caplog.text)
     assert not Path(tmp_path / "backups" / repo).exists()
 
 
@@ -73,13 +80,14 @@ def test_run_list_0_match_with_gists(tmp_path, caplog):
         archive_format="none",
         include_gists=True,
         include_history=False,
+        skip_forks=False,
         keep=False,
         list_only=True,
         prompt_for_token=False,
     )
     assert "creating archives" not in caplog.text
-    assert f"found 0 repos for user '{USER}'" in caplog.text
-    assert f"found 0 gists for user '{USER}'" in caplog.text
+    assert "found 0 repos" in caplog.text
+    assert re.search(r"found \d* gists", caplog.text)
     assert not Path(tmp_path / "backups" / repo).exists()
 
 
@@ -89,16 +97,17 @@ def test_run_list_skip_all(tmp_path, caplog):
         username=USER,
         base_dir=tmp_path,
         pattern=".*",
-        skip_pattern=".",
+        skip_pattern=".*",
         archive_format="none",
         include_gists=False,
         include_history=False,
+        skip_forks=False,
         keep=False,
         list_only=True,
         prompt_for_token=False,
     )
     assert "creating archives" not in caplog.text
-    assert f"found 0 repos for user '{USER}'" in caplog.text
+    assert "found 0 repos" in caplog.text
 
 
 @pytest.mark.parametrize("archive_format", ["zip", "tar"])
@@ -114,6 +123,7 @@ def test_run_archive_1_match(archive_format, tmp_path, caplog):
         archive_format=archive_format,
         include_gists=False,
         include_history=False,
+        skip_forks=False,
         keep=False,
         list_only=False,
         prompt_for_token=False,
@@ -146,6 +156,7 @@ def test_run_no_archive_1_match_with_history(tmp_path, caplog):
         archive_format="none",
         include_gists=False,
         include_history=True,
+        skip_forks=False,
         keep=False,
         list_only=False,
         prompt_for_token=False,
@@ -176,6 +187,7 @@ def test_run_no_archive_1_match_without_history(tmp_path, caplog):
         archive_format="none",
         include_gists=False,
         include_history=False,
+        skip_forks=False,
         keep=False,  # this doesn't matter when archive_format is "none"
         list_only=False,
         prompt_for_token=False,
@@ -206,6 +218,7 @@ def test_run_archive_1_match_with_history_pull_with_keep(tmp_path, caplog):
         archive_format="none",
         include_gists=False,
         include_history=True,
+        skip_forks=False,
         keep=True,
         list_only=False,
         prompt_for_token=False,
@@ -229,6 +242,7 @@ def test_run_archive_1_match_with_history_pull_with_keep(tmp_path, caplog):
         archive_format="zip",
         include_gists=False,
         include_history=True,
+        skip_forks=False,
         keep=True,
         list_only=False,
         prompt_for_token=False,
@@ -236,7 +250,9 @@ def test_run_archive_1_match_with_history_pull_with_keep(tmp_path, caplog):
     assert f"found 1 repos for user '{USER}'" in caplog.text
     assert "gists" not in caplog.text
     assert not re.search(f"cloning repo: {USER}/{repo}.git to: .*backups", caplog.text)
-    assert re.search(f"pulling changes from repo: {USER}/{repo}.git to: .*backups", caplog.text)
+    assert re.search(
+        f"pulling changes from repo: {USER}/{repo}.git to: .*backups", caplog.text
+    )
     assert "creating archive:" in caplog.text
     assert "archive size:" in caplog.text
     assert "deleting repo" not in caplog.text