Metadata-Version: 2.1
Name: mirror-git-to-s3
Version: 0.0.12
Summary: Python functions and CLI to mirror git repositories available on HTTP(S) to S3
Project-URL: Homepage, https://github.com/uktrade/mirror-git-to-s3
Project-URL: Bug Tracker, https://github.com/uktrade/mirror-git-to-s3/issues
Author-email: Department for International Trade <sre@digital.trade.gov.uk>
License-File: LICENSE
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Classifier: Programming Language :: Python :: 3
Classifier: Topic :: Software Development :: Version Control :: Git
Requires-Python: >=3.8
Requires-Dist: boto3>=1.26.27
Requires-Dist: click>=8.1.3
Requires-Dist: httpx>=0.23.1
Provides-Extra: dev
Requires-Dist: pytest>=7.2.0; extra == 'dev'
Description-Content-Type: text/markdown

# mirror-git-to-s3 ![Build Status](https://github.com/uktrade/mirror-git-to-s3/actions/workflows/tests.yml/badge.svg?branch=main)

Python functions and CLI to mirror public git repositories available on HTTP(S) to S3. Essentially converts smart protocol git repositories to the so-called dumb protocol. Does not use temporary disk space, and uses streaming under the hood. This should allow the mirroring to be run on systems that don't have much disk or available memory, even on repositories with large objects.

This project mirrors objects stored on Large File Storage (LFS). Note however LFS objects are not accessible via the dumb protocol. To work around this, you can use [git-lfs-http-mirror](https://github.com/uktrade/git-lfs-http-mirror)  that fires up a temporary local LFS server during git commands.

At the time of writing repositories with many objects can be slow to mirror.


## Installation

```bash
pip install mirror-git-to-s3
```


## Usage

To mirror one or more repositories from Python, use the `mirror_repos` function, passing it an iterable of (source, target) mappings.

```python
from mirror_git_to_s3 import mirror_repos

mirror_repos((
    ('https://example.test/my-first-repo', 's3://my-bucket/my-first-repo'),
    ('https://example.test/my-second-repo', 's3://my-bucket/my-second-repo'),
))
```

Once a repository is mirrored to a bucket that doesn't need authentication to read, it can be cloned using standard git commands using the [virtual host or path of the S3 bucket](https://docs.aws.amazon.com/AmazonS3/latest/userguide/VirtualHosting.html).

```bash
git clone https://my-bucket.s3.eu-west-2.amazonaws.com/my-first-repo
````

Under the hood, boto3 is used to communicate with S3. The boto3 client is constructed automatically, but you can override the default by using the `get_s3_client` argument.

```python
import boto3
from mirror_git_to_s3 import mirror_repos

mirror_repos(mappings(), get_s3_client=lambda: boto3.client('s3'))
```

This can be used to mirror to S3-compatible storage.

```python
import boto3
from mirror_git_to_s3 import mirror_repos

mirror_repos(mappings(), get_s3_client=lambda: boto3.client('s3', endpoint_url='http://my-host.com/'))
```

To mirror repositories from the the command line pairs of `--source` `--target` options can be passed to `mirror-git-to-s3`.

```bash
mirror-git-to-s3 \
    --source 'https://example.test/my-first-repo' --target 's3://my-bucket/my-first-repo' \
    --source 'https://example.test/my-second-repo' --target 's3://my-bucket/my-second-repo'
```

At the time of writing, there is no known standard way of discovering a set of associated git repositories, hence to remain general, this project must be told the source and target addresses of each repository explicitly.
