| Author: | Christopher Arndt |
|---|---|
| Version: | 0.1a |
| Date: | 2008-04-09 |
| Description: | A simple, lightweight Python Package Index (aka Cheeseshop) clone. |
EggBasket is a web application which provides a service similar and compatible to the Python Package Index (aka Cheeseshop). It allows you to maintain your own local repository of Python packages required by your installations.
It is implemented using the TurboGears web framework, Genshi and SQLAlchemy.
Warning
This is still alpha-stage software. All the basic operations necessary to support a setuptools-based infrastructure are there, but some convenience features are missing and the software has not been tested extensively. Use at your own risk!
Can be used by setuptools/easy_install as the package index and repository.
Supports distutils upload protocol.
Has simple role-based permission system to restrict package uploads (but see TODO).
Requires only SQLite database (included with Python 2.5).
Is able to read and display meta data from the following distribution package formats (source and binary):
.egg, .tar, .tar.bz2, .tar.gz, .tgz, .zip
Any other file format can be configured to be listed under the distribution files for a package (by default this includes .exe and .rpm and .tar.Z files in addition to the filetypes listed above).
Can be run without any configuration by just starting it from within a directory containing package directories (see "Usage").
This application is a re-implementation (almost no shared code) of the haufe.eggserver Grok application with some improvements.
To install EggBasket from the Cheeseshop use easy_install:
[sudo] easy_install EggBasket
This requires the setuptools package to be installed. If you have not done so already, download the ez_setup.py script and run it to install setuptools.
Your packages should all reside under a common root directory, with a sub-directory for each package with the same base name as the distribution. The sub-directories should each contain the egg files and source archives for all available versions of the package. The package diretories will be created by the application when using the upload command (see below).
To start the application server, open a terminal, change to the directory which contains the packages and then run:
eggbasket-server [<config file>]
You can also set the location of the package root directory in the configuration with the eggbasket.package_root setting and start the server anywheer you want.
If no configuration file is specified on the command line, the default configuration file included in the egg will be used. The default configuration file can also be found in the source distribution and be adapted for your environment.
The server either needs write permissions to directory where it is started, or you need to change the path to the database and the access log in the configuration so they can be written by the server.
To stop the server just hit Control-C in the terminal or kill the process.
You can look at the package index with your web browser by opening the URL http://localhost:3442/. The default port 3442 can be changed by setting the server.socket_port option in the configuration file.
You can instruct easy_install to search & download packages from your package repository by specifying the URL to your server with the -i option:
easy_install -i http://localhost:8080/ PACKAGE_NAME
You can upload a package to your repository with the distutils upload command, for example:
python setup.py bdist_egg upload -r http://localhost:3442/upload
This command will ask for your username and password on the server. You can store these and the repository URL in your .pypirc file. See the distutils documentation for more information.
By default, uploading is restricted to users in a group that has the upload permission. You can create an appropriate user, group and permission in the database by giving the --init option to the eggbasket-server command:
eggbasket-server --init [<config file>]
This will create a user with user_name/password "admin", who belongs to the group "maintainers", which has the "upload" permission. The database to use will be read from the config file.
See the TurboGears documentation on Identity for background information.
Of course you can always just copy package distribution files manually in the filesystem to your repository or upload them to the appropriate place with scp etc. The application will find and list new files without the need to "register" them as is necessary with the original PyPI.