Current File : //usr/share/doc/python-setuptools-0.9.8/docs/setuptools.txt
==================================================
Building and Distributing Packages with Setuptools
==================================================
``Setuptools`` is a collection of enhancements to the Python ``distutils``
(for Python 2.3.5 and up on most platforms; 64-bit platforms require a minimum
of Python 2.4) that allow you to more easily build and distribute Python
packages, especially ones that have dependencies on other packages.
Packages built and distributed using ``setuptools`` look to the user like
ordinary Python packages based on the ``distutils``. Your users don't need to
install or even know about setuptools in order to use them, and you don't
have to include the entire setuptools package in your distributions. By
including just a single `bootstrap module`_ (an 8K .py file), your package will
automatically download and install ``setuptools`` if the user is building your
package from source and doesn't have a suitable version already installed.
.. _bootstrap module: https://bitbucket.org/pypa/setuptools/downloads/ez_setup.py
Feature Highlights:
* Automatically find/download/install/upgrade dependencies at build time using
the `EasyInstall tool <http://peak.telecommunity.com/DevCenter/EasyInstall>`_,
which supports downloading via HTTP, FTP, Subversion, and SourceForge, and
automatically scans web pages linked from PyPI to find download links. (It's
the closest thing to CPAN currently available for Python.)
* Create `Python Eggs <http://peak.telecommunity.com/DevCenter/PythonEggs>`_ -
a single-file importable distribution format
* Include data files inside your package directories, where your code can
actually use them. (Python 2.4 distutils also supports this feature, but
setuptools provides the feature for Python 2.3 packages also, and supports
accessing data files in zipped packages too.)
* Automatically include all packages in your source tree, without listing them
individually in setup.py
* Automatically include all relevant files in your source distributions,
without needing to create a ``MANIFEST.in`` file, and without having to force
regeneration of the ``MANIFEST`` file when your source tree changes.
* Automatically generate wrapper scripts or Windows (console and GUI) .exe
files for any number of "main" functions in your project. (Note: this is not
a py2exe replacement; the .exe files rely on the local Python installation.)
* Transparent Pyrex support, so that your setup.py can list ``.pyx`` files and
still work even when the end-user doesn't have Pyrex installed (as long as
you include the Pyrex-generated C in your source distribution)
* Command aliases - create project-specific, per-user, or site-wide shortcut
names for commonly used commands and options
* PyPI upload support - upload your source distributions and eggs to PyPI
* Deploy your project in "development mode", such that it's available on
``sys.path``, yet can still be edited directly from its source checkout.
* Easily extend the distutils with new commands or ``setup()`` arguments, and
distribute/reuse your extensions for multiple projects, without copying code.
* Create extensible applications and frameworks that automatically discover
extensions, using simple "entry points" declared in a project's setup script.
In addition to the PyPI downloads, the development version of ``setuptools``
is available from the `Python SVN sandbox`_, and in-development versions of the
`0.6 branch`_ are available as well.
.. _0.6 branch: http://svn.python.org/projects/sandbox/branches/setuptools-0.6/#egg=setuptools-dev06
.. _Python SVN sandbox: http://svn.python.org/projects/sandbox/trunk/setuptools/#egg=setuptools-dev
.. contents:: **Table of Contents**
.. _distribute_setup.py: `bootstrap module`_
-----------------
Developer's Guide
-----------------
Installing ``setuptools``
=========================
Please follow the `EasyInstall Installation Instructions`_ to install the
current stable version of setuptools. In particular, be sure to read the
section on `Custom Installation Locations`_ if you are installing anywhere
other than Python's ``site-packages`` directory.
.. _EasyInstall Installation Instructions: http://peak.telecommunity.com/DevCenter/EasyInstall#installation-instructions
.. _Custom Installation Locations: http://peak.telecommunity.com/DevCenter/EasyInstall#custom-installation-locations
If you want the current in-development version of setuptools, you should first
install a stable version, and then run::
distribute_setup.py setuptools==dev
This will download and install the latest development (i.e. unstable) version
of setuptools from the Python Subversion sandbox.
Basic Use
=========
For basic use of setuptools, just import things from setuptools instead of
the distutils. Here's a minimal setup script using setuptools::
from setuptools import setup, find_packages
setup(
name = "HelloWorld",
version = "0.1",
packages = find_packages(),
)
As you can see, it doesn't take much to use setuptools in a project.
Just by doing the above, this project will be able to produce eggs, upload to
PyPI, and automatically include all packages in the directory where the
setup.py lives. See the `Command Reference`_ section below to see what
commands you can give to this setup script.
Of course, before you release your project to PyPI, you'll want to add a bit
more information to your setup script to help people find or learn about your
project. And maybe your project will have grown by then to include a few
dependencies, and perhaps some data files and scripts::
from setuptools import setup, find_packages
setup(
name = "HelloWorld",
version = "0.1",
packages = find_packages(),
scripts = ['say_hello.py'],
# Project uses reStructuredText, so ensure that the docutils get
# installed or upgraded on the target machine
install_requires = ['docutils>=0.3'],
package_data = {
# If any package contains *.txt or *.rst files, include them:
'': ['*.txt', '*.rst'],
# And include any *.msg files found in the 'hello' package, too:
'hello': ['*.msg'],
},
# metadata for upload to PyPI
author = "Me",
author_email = "me@example.com",
description = "This is an Example Package",
license = "PSF",
keywords = "hello world example examples",
url = "http://example.com/HelloWorld/", # project home page, if any
# could also include long_description, download_url, classifiers, etc.
)
In the sections that follow, we'll explain what most of these ``setup()``
arguments do (except for the metadata ones), and the various ways you might use
them in your own project(s).
Specifying Your Project's Version
---------------------------------
Setuptools can work well with most versioning schemes; there are, however, a
few special things to watch out for, in order to ensure that setuptools and
EasyInstall can always tell what version of your package is newer than another
version. Knowing these things will also help you correctly specify what
versions of other projects your project depends on.
A version consists of an alternating series of release numbers and pre-release
or post-release tags. A release number is a series of digits punctuated by
dots, such as ``2.4`` or ``0.5``. Each series of digits is treated
numerically, so releases ``2.1`` and ``2.1.0`` are different ways to spell the
same release number, denoting the first subrelease of release 2. But ``2.10``
is the *tenth* subrelease of release 2, and so is a different and newer release
from ``2.1`` or ``2.1.0``. Leading zeros within a series of digits are also
ignored, so ``2.01`` is the same as ``2.1``, and different from ``2.0.1``.
Following a release number, you can have either a pre-release or post-release
tag. Pre-release tags make a version be considered *older* than the version
they are appended to. So, revision ``2.4`` is *newer* than revision ``2.4c1``,
which in turn is newer than ``2.4b1`` or ``2.4a1``. Postrelease tags make
a version be considered *newer* than the version they are appended to. So,
revisions like ``2.4-1`` and ``2.4pl3`` are newer than ``2.4``, but are *older*
than ``2.4.1`` (which has a higher release number).
A pre-release tag is a series of letters that are alphabetically before
"final". Some examples of prerelease tags would include ``alpha``, ``beta``,
``a``, ``c``, ``dev``, and so on. You do not have to place a dot or dash
before the prerelease tag if it's immediately after a number, but it's okay to
do so if you prefer. Thus, ``2.4c1`` and ``2.4.c1`` and ``2.4-c1`` all
represent release candidate 1 of version ``2.4``, and are treated as identical
by setuptools.
In addition, there are three special prerelease tags that are treated as if
they were the letter ``c``: ``pre``, ``preview``, and ``rc``. So, version
``2.4rc1``, ``2.4pre1`` and ``2.4preview1`` are all the exact same version as
``2.4c1``, and are treated as identical by setuptools.
A post-release tag is either a series of letters that are alphabetically
greater than or equal to "final", or a dash (``-``). Post-release tags are
generally used to separate patch numbers, port numbers, build numbers, revision
numbers, or date stamps from the release number. For example, the version
``2.4-r1263`` might denote Subversion revision 1263 of a post-release patch of
version ``2.4``. Or you might use ``2.4-20051127`` to denote a date-stamped
post-release.
Notice that after each pre or post-release tag, you are free to place another
release number, followed again by more pre- or post-release tags. For example,
``0.6a9.dev-r41475`` could denote Subversion revision 41475 of the in-
development version of the ninth alpha of release 0.6. Notice that ``dev`` is
a pre-release tag, so this version is a *lower* version number than ``0.6a9``,
which would be the actual ninth alpha of release 0.6. But the ``-r41475`` is
a post-release tag, so this version is *newer* than ``0.6a9.dev``.
For the most part, setuptools' interpretation of version numbers is intuitive,
but here are a few tips that will keep you out of trouble in the corner cases:
* Don't stick adjoining pre-release tags together without a dot or number
between them. Version ``1.9adev`` is the ``adev`` prerelease of ``1.9``,
*not* a development pre-release of ``1.9a``. Use ``.dev`` instead, as in
``1.9a.dev``, or separate the prerelease tags with a number, as in
``1.9a0dev``. ``1.9a.dev``, ``1.9a0dev``, and even ``1.9.a.dev`` are
identical versions from setuptools' point of view, so you can use whatever
scheme you prefer.
* If you want to be certain that your chosen numbering scheme works the way
you think it will, you can use the ``pkg_resources.parse_version()`` function
to compare different version numbers::
>>> from pkg_resources import parse_version
>>> parse_version('1.9.a.dev') == parse_version('1.9a0dev')
True
>>> parse_version('2.1-rc2') < parse_version('2.1')
True
>>> parse_version('0.6a9dev-r41475') < parse_version('0.6a9')
True
Once you've decided on a version numbering scheme for your project, you can
have setuptools automatically tag your in-development releases with various
pre- or post-release tags. See the following sections for more details:
* `Tagging and "Daily Build" or "Snapshot" Releases`_
* `Managing "Continuous Releases" Using Subversion`_
* The `egg_info`_ command
New and Changed ``setup()`` Keywords
====================================
The following keyword arguments to ``setup()`` are added or changed by
``setuptools``. All of them are optional; you do not have to supply them
unless you need the associated ``setuptools`` feature.
``include_package_data``
If set to ``True``, this tells ``setuptools`` to automatically include any
data files it finds inside your package directories, that are either under
CVS or Subversion control, or which are specified by your ``MANIFEST.in``
file. For more information, see the section below on `Including Data
Files`_.
``exclude_package_data``
A dictionary mapping package names to lists of glob patterns that should
be *excluded* from your package directories. You can use this to trim back
any excess files included by ``include_package_data``. For a complete
description and examples, see the section below on `Including Data Files`_.
``package_data``
A dictionary mapping package names to lists of glob patterns. For a
complete description and examples, see the section below on `Including
Data Files`_. You do not need to use this option if you are using
``include_package_data``, unless you need to add e.g. files that are
generated by your setup script and build process. (And are therefore not
in source control or are files that you don't want to include in your
source distribution.)
``zip_safe``
A boolean (True or False) flag specifying whether the project can be
safely installed and run from a zip file. If this argument is not
supplied, the ``bdist_egg`` command will have to analyze all of your
project's contents for possible problems each time it buids an egg.
``install_requires``
A string or list of strings specifying what other distributions need to
be installed when this one is. See the section below on `Declaring
Dependencies`_ for details and examples of the format of this argument.
``entry_points``
A dictionary mapping entry point group names to strings or lists of strings
defining the entry points. Entry points are used to support dynamic
discovery of services or plugins provided by a project. See `Dynamic
Discovery of Services and Plugins`_ for details and examples of the format
of this argument. In addition, this keyword is used to support `Automatic
Script Creation`_.
``extras_require``
A dictionary mapping names of "extras" (optional features of your project)
to strings or lists of strings specifying what other distributions must be
installed to support those features. See the section below on `Declaring
Dependencies`_ for details and examples of the format of this argument.
``setup_requires``
A string or list of strings specifying what other distributions need to
be present in order for the *setup script* to run. ``setuptools`` will
attempt to obtain these (even going so far as to download them using
``EasyInstall``) before processing the rest of the setup script or commands.
This argument is needed if you are using distutils extensions as part of
your build process; for example, extensions that process setup() arguments
and turn them into EGG-INFO metadata files.
(Note: projects listed in ``setup_requires`` will NOT be automatically
installed on the system where the setup script is being run. They are
simply downloaded to the setup directory if they're not locally available
already. If you want them to be installed, as well as being available
when the setup script is run, you should add them to ``install_requires``
**and** ``setup_requires``.)
``dependency_links``
A list of strings naming URLs to be searched when satisfying dependencies.
These links will be used if needed to install packages specified by
``setup_requires`` or ``tests_require``. They will also be written into
the egg's metadata for use by tools like EasyInstall to use when installing
an ``.egg`` file.
``namespace_packages``
A list of strings naming the project's "namespace packages". A namespace
package is a package that may be split across multiple project
distributions. For example, Zope 3's ``zope`` package is a namespace
package, because subpackages like ``zope.interface`` and ``zope.publisher``
may be distributed separately. The egg runtime system can automatically
merge such subpackages into a single parent package at runtime, as long
as you declare them in each project that contains any subpackages of the
namespace package, and as long as the namespace package's ``__init__.py``
does not contain any code other than a namespace declaration. See the
section below on `Namespace Packages`_ for more information.
``test_suite``
A string naming a ``unittest.TestCase`` subclass (or a package or module
containing one or more of them, or a method of such a subclass), or naming
a function that can be called with no arguments and returns a
``unittest.TestSuite``. If the named suite is a module, and the module
has an ``additional_tests()`` function, it is called and the results are
added to the tests to be run. If the named suite is a package, any
submodules and subpackages are recursively added to the overall test suite.
Specifying this argument enables use of the `test`_ command to run the
specified test suite, e.g. via ``setup.py test``. See the section on the
`test`_ command below for more details.
``tests_require``
If your project's tests need one or more additional packages besides those
needed to install it, you can use this option to specify them. It should
be a string or list of strings specifying what other distributions need to
be present for the package's tests to run. When you run the ``test``
command, ``setuptools`` will attempt to obtain these (even going
so far as to download them using ``EasyInstall``). Note that these
required projects will *not* be installed on the system where the tests
are run, but only downloaded to the project's setup directory if they're
not already installed locally.
.. _test_loader:
``test_loader``
If you would like to use a different way of finding tests to run than what
setuptools normally uses, you can specify a module name and class name in
this argument. The named class must be instantiable with no arguments, and
its instances must support the ``loadTestsFromNames()`` method as defined
in the Python ``unittest`` module's ``TestLoader`` class. Setuptools will
pass only one test "name" in the `names` argument: the value supplied for
the ``test_suite`` argument. The loader you specify may interpret this
string in any way it likes, as there are no restrictions on what may be
contained in a ``test_suite`` string.
The module name and class name must be separated by a ``:``. The default
value of this argument is ``"setuptools.command.test:ScanningLoader"``. If
you want to use the default ``unittest`` behavior, you can specify
``"unittest:TestLoader"`` as your ``test_loader`` argument instead. This
will prevent automatic scanning of submodules and subpackages.
The module and class you specify here may be contained in another package,
as long as you use the ``tests_require`` option to ensure that the package
containing the loader class is available when the ``test`` command is run.
``eager_resources``
A list of strings naming resources that should be extracted together, if
any of them is needed, or if any C extensions included in the project are
imported. This argument is only useful if the project will be installed as
a zipfile, and there is a need to have all of the listed resources be
extracted to the filesystem *as a unit*. Resources listed here
should be '/'-separated paths, relative to the source root, so to list a
resource ``foo.png`` in package ``bar.baz``, you would include the string
``bar/baz/foo.png`` in this argument.
If you only need to obtain resources one at a time, or you don't have any C
extensions that access other files in the project (such as data files or
shared libraries), you probably do NOT need this argument and shouldn't
mess with it. For more details on how this argument works, see the section
below on `Automatic Resource Extraction`_.
``use_2to3``
Convert the source code from Python 2 to Python 3 with 2to3 during the
build process. See :doc:`python3` for more details.
``convert_2to3_doctests``
List of doctest source files that need to be converted with 2to3.
See :doc:`python3` for more details.
``use_2to3_fixers``
A list of modules to search for additional fixers to be used during
the 2to3 conversion. See :doc:`python3` for more details.
Using ``find_packages()``
-------------------------
For simple projects, it's usually easy enough to manually add packages to
the ``packages`` argument of ``setup()``. However, for very large projects
(Twisted, PEAK, Zope, Chandler, etc.), it can be a big burden to keep the
package list updated. That's what ``setuptools.find_packages()`` is for.
``find_packages()`` takes a source directory, and a list of package names or
patterns to exclude. If omitted, the source directory defaults to the same
directory as the setup script. Some projects use a ``src`` or ``lib``
directory as the root of their source tree, and those projects would of course
use ``"src"`` or ``"lib"`` as the first argument to ``find_packages()``. (And
such projects also need something like ``package_dir = {'':'src'}`` in their
``setup()`` arguments, but that's just a normal distutils thing.)
Anyway, ``find_packages()`` walks the target directory, and finds Python
packages by looking for ``__init__.py`` files. It then filters the list of
packages using the exclusion patterns.
Exclusion patterns are package names, optionally including wildcards. For
example, ``find_packages(exclude=["*.tests"])`` will exclude all packages whose
last name part is ``tests``. Or, ``find_packages(exclude=["*.tests",
"*.tests.*"])`` will also exclude any subpackages of packages named ``tests``,
but it still won't exclude a top-level ``tests`` package or the children
thereof. In fact, if you really want no ``tests`` packages at all, you'll need
something like this::
find_packages(exclude=["*.tests", "*.tests.*", "tests.*", "tests"])
in order to cover all the bases. Really, the exclusion patterns are intended
to cover simpler use cases than this, like excluding a single, specified
package and its subpackages.
Regardless of the target directory or exclusions, the ``find_packages()``
function returns a list of package names suitable for use as the ``packages``
argument to ``setup()``, and so is usually the easiest way to set that
argument in your setup script. Especially since it frees you from having to
remember to modify your setup script whenever your project grows additional
top-level packages or subpackages.
Automatic Script Creation
=========================
Packaging and installing scripts can be a bit awkward with the distutils. For
one thing, there's no easy way to have a script's filename match local
conventions on both Windows and POSIX platforms. For another, you often have
to create a separate file just for the "main" script, when your actual "main"
is a function in a module somewhere. And even in Python 2.4, using the ``-m``
option only works for actual ``.py`` files that aren't installed in a package.
``setuptools`` fixes all of these problems by automatically generating scripts
for you with the correct extension, and on Windows it will even create an
``.exe`` file so that users don't have to change their ``PATHEXT`` settings.
The way to use this feature is to define "entry points" in your setup script
that indicate what function the generated script should import and run. For
example, to create two console scripts called ``foo`` and ``bar``, and a GUI
script called ``baz``, you might do something like this::
setup(
# other arguments here...
entry_points = {
'console_scripts': [
'foo = my_package.some_module:main_func',
'bar = other_module:some_func',
],
'gui_scripts': [
'baz = my_package_gui.start_func',
]
}
)
When this project is installed on non-Windows platforms (using "setup.py
install", "setup.py develop", or by using EasyInstall), a set of ``foo``,
``bar``, and ``baz`` scripts will be installed that import ``main_func`` and
``some_func`` from the specified modules. The functions you specify are called
with no arguments, and their return value is passed to ``sys.exit()``, so you
can return an errorlevel or message to print to stderr.
On Windows, a set of ``foo.exe``, ``bar.exe``, and ``baz.exe`` launchers are
created, alongside a set of ``foo.py``, ``bar.py``, and ``baz.pyw`` files. The
``.exe`` wrappers find and execute the right version of Python to run the
``.py`` or ``.pyw`` file.
You may define as many "console script" and "gui script" entry points as you
like, and each one can optionally specify "extras" that it depends on, that
will be added to ``sys.path`` when the script is run. For more information on
"extras", see the section below on `Declaring Extras`_. For more information
on "entry points" in general, see the section below on `Dynamic Discovery of
Services and Plugins`_.
"Eggsecutable" Scripts
----------------------
Occasionally, there are situations where it's desirable to make an ``.egg``
file directly executable. You can do this by including an entry point such
as the following::
setup(
# other arguments here...
entry_points = {
'setuptools.installation': [
'eggsecutable = my_package.some_module:main_func',
]
}
)
Any eggs built from the above setup script will include a short excecutable
prelude that imports and calls ``main_func()`` from ``my_package.some_module``.
The prelude can be run on Unix-like platforms (including Mac and Linux) by
invoking the egg with ``/bin/sh``, or by enabling execute permissions on the
``.egg`` file. For the executable prelude to run, the appropriate version of
Python must be available via the ``PATH`` environment variable, under its
"long" name. That is, if the egg is built for Python 2.3, there must be a
``python2.3`` executable present in a directory on ``PATH``.
This feature is primarily intended to support distribute_setup the installation of
setuptools itself on non-Windows platforms, but may also be useful for other
projects as well.
IMPORTANT NOTE: Eggs with an "eggsecutable" header cannot be renamed, or
invoked via symlinks. They *must* be invoked using their original filename, in
order to ensure that, once running, ``pkg_resources`` will know what project
and version is in use. The header script will check this and exit with an
error if the ``.egg`` file has been renamed or is invoked via a symlink that
changes its base name.
Declaring Dependencies
======================
``setuptools`` supports automatically installing dependencies when a package is
installed, and including information about dependencies in Python Eggs (so that
package management tools like EasyInstall can use the information).
``setuptools`` and ``pkg_resources`` use a common syntax for specifying a
project's required dependencies. This syntax consists of a project's PyPI
name, optionally followed by a comma-separated list of "extras" in square
brackets, optionally followed by a comma-separated list of version
specifiers. A version specifier is one of the operators ``<``, ``>``, ``<=``,
``>=``, ``==`` or ``!=``, followed by a version identifier. Tokens may be
separated by whitespace, but any whitespace or nonstandard characters within a
project name or version identifier must be replaced with ``-``.
Version specifiers for a given project are internally sorted into ascending
version order, and used to establish what ranges of versions are acceptable.
Adjacent redundant conditions are also consolidated (e.g. ``">1, >2"`` becomes
``">1"``, and ``"<2,<3"`` becomes ``"<3"``). ``"!="`` versions are excised from
the ranges they fall within. A project's version is then checked for
membership in the resulting ranges. (Note that providing conflicting conditions
for the same version (e.g. "<2,>=2" or "==2,!=2") is meaningless and may
therefore produce bizarre results.)
Here are some example requirement specifiers::
docutils >= 0.3
# comment lines and \ continuations are allowed in requirement strings
BazSpam ==1.1, ==1.2, ==1.3, ==1.4, ==1.5, \
==1.6, ==1.7 # and so are line-end comments
PEAK[FastCGI, reST]>=0.5a4
setuptools==0.5a7
The simplest way to include requirement specifiers is to use the
``install_requires`` argument to ``setup()``. It takes a string or list of
strings containing requirement specifiers. If you include more than one
requirement in a string, each requirement must begin on a new line.
This has three effects:
1. When your project is installed, either by using EasyInstall, ``setup.py
install``, or ``setup.py develop``, all of the dependencies not already
installed will be located (via PyPI), downloaded, built (if necessary),
and installed.
2. Any scripts in your project will be installed with wrappers that verify
the availability of the specified dependencies at runtime, and ensure that
the correct versions are added to ``sys.path`` (e.g. if multiple versions
have been installed).
3. Python Egg distributions will include a metadata file listing the
dependencies.
Note, by the way, that if you declare your dependencies in ``setup.py``, you do
*not* need to use the ``require()`` function in your scripts or modules, as
long as you either install the project or use ``setup.py develop`` to do
development work on it. (See `"Development Mode"`_ below for more details on
using ``setup.py develop``.)
Dependencies that aren't in PyPI
--------------------------------
If your project depends on packages that aren't registered in PyPI, you may
still be able to depend on them, as long as they are available for download
as:
- an egg, in the standard distutils ``sdist`` format,
- a single ``.py`` file, or
- a VCS repository (Subversion, Mercurial, or Git).
You just need to add some URLs to the ``dependency_links`` argument to
``setup()``.
The URLs must be either:
1. direct download URLs,
2. the URLs of web pages that contain direct download links, or
3. the repository's URL
In general, it's better to link to web pages, because it is usually less
complex to update a web page than to release a new version of your project.
You can also use a SourceForge ``showfiles.php`` link in the case where a
package you depend on is distributed via SourceForge.
If you depend on a package that's distributed as a single ``.py`` file, you
must include an ``"#egg=project-version"`` suffix to the URL, to give a project
name and version number. (Be sure to escape any dashes in the name or version
by replacing them with underscores.) EasyInstall will recognize this suffix
and automatically create a trivial ``setup.py`` to wrap the single ``.py`` file
as an egg.
In the case of a VCS checkout, you should also append ``#egg=project-version``
in order to identify for what package that checkout should be used. You can
append ``@REV`` to the URL's path (before the fragment) to specify a revision.
Additionally, you can also force the VCS being used by prepending the URL with
a certain prefix. Currently available are:
- ``svn+URL`` for Subversion,
- ``git+URL`` for Git, and
- ``hg+URL`` for Mercurial
A more complete example would be:
``vcs+proto://host/path@revision#egg=project-version``
Be careful with the version. It should match the one inside the project files.
If you want do disregard the version, you have to omit it both in the
``requires`` and in the URL's fragment.
This will do a checkout (or a clone, in Git and Mercurial parlance) to a
temporary folder and run ``setup.py bdist_egg``.
The ``dependency_links`` option takes the form of a list of URL strings. For
example, the below will cause EasyInstall to search the specified page for
eggs or source distributions, if the package's dependencies aren't already
installed::
setup(
...
dependency_links = [
"http://peak.telecommunity.com/snapshots/"
],
)
.. _Declaring Extras:
Declaring "Extras" (optional features with their own dependencies)
------------------------------------------------------------------
Sometimes a project has "recommended" dependencies, that are not required for
all uses of the project. For example, a project might offer optional PDF
output if ReportLab is installed, and reStructuredText support if docutils is
installed. These optional features are called "extras", and setuptools allows
you to define their requirements as well. In this way, other projects that
require these optional features can force the additional requirements to be
installed, by naming the desired extras in their ``install_requires``.
For example, let's say that Project A offers optional PDF and reST support::
setup(
name="Project-A",
...
extras_require = {
'PDF': ["ReportLab>=1.2", "RXP"],
'reST': ["docutils>=0.3"],
}
)
As you can see, the ``extras_require`` argument takes a dictionary mapping
names of "extra" features, to strings or lists of strings describing those
features' requirements. These requirements will *not* be automatically
installed unless another package depends on them (directly or indirectly) by
including the desired "extras" in square brackets after the associated project
name. (Or if the extras were listed in a requirement spec on the EasyInstall
command line.)
Extras can be used by a project's `entry points`_ to specify dynamic
dependencies. For example, if Project A includes a "rst2pdf" script, it might
declare it like this, so that the "PDF" requirements are only resolved if the
"rst2pdf" script is run::
setup(
name="Project-A",
...
entry_points = {
'console_scripts':
['rst2pdf = project_a.tools.pdfgen [PDF]'],
['rst2html = project_a.tools.htmlgen'],
# more script entry points ...
}
)
Projects can also use another project's extras when specifying dependencies.
For example, if project B needs "project A" with PDF support installed, it
might declare the dependency like this::
setup(
name="Project-B",
install_requires = ["Project-A[PDF]"],
...
)
This will cause ReportLab to be installed along with project A, if project B is
installed -- even if project A was already installed. In this way, a project
can encapsulate groups of optional "downstream dependencies" under a feature
name, so that packages that depend on it don't have to know what the downstream
dependencies are. If a later version of Project A builds in PDF support and
no longer needs ReportLab, or if it ends up needing other dependencies besides
ReportLab in order to provide PDF support, Project B's setup information does
not need to change, but the right packages will still be installed if needed.
Note, by the way, that if a project ends up not needing any other packages to
support a feature, it should keep an empty requirements list for that feature
in its ``extras_require`` argument, so that packages depending on that feature
don't break (due to an invalid feature name). For example, if Project A above
builds in PDF support and no longer needs ReportLab, it could change its
setup to this::
setup(
name="Project-A",
...
extras_require = {
'PDF': [],
'reST': ["docutils>=0.3"],
}
)
so that Package B doesn't have to remove the ``[PDF]`` from its requirement
specifier.
Including Data Files
====================
The distutils have traditionally allowed installation of "data files", which
are placed in a platform-specific location. However, the most common use case
for data files distributed with a package is for use *by* the package, usually
by including the data files in the package directory.
Setuptools offers three ways to specify data files to be included in your
packages. First, you can simply use the ``include_package_data`` keyword,
e.g.::
from setuptools import setup, find_packages
setup(
...
include_package_data = True
)
This tells setuptools to install any data files it finds in your packages.
The data files must be under CVS or Subversion control, or else they must be
specified via the distutils' ``MANIFEST.in`` file. (They can also be tracked
by another revision control system, using an appropriate plugin. See the
section below on `Adding Support for Other Revision Control Systems`_ for
information on how to write such plugins.)
If the data files are not under version control, or are not in a supported
version control system, or if you want finer-grained control over what files
are included (for example, if you have documentation files in your package
directories and want to exclude them from installation), then you can also use
the ``package_data`` keyword, e.g.::
from setuptools import setup, find_packages
setup(
...
package_data = {
# If any package contains *.txt or *.rst files, include them:
'': ['*.txt', '*.rst'],
# And include any *.msg files found in the 'hello' package, too:
'hello': ['*.msg'],
}
)
The ``package_data`` argument is a dictionary that maps from package names to
lists of glob patterns. The globs may include subdirectory names, if the data
files are contained in a subdirectory of the package. For example, if the
package tree looks like this::
setup.py
src/
mypkg/
__init__.py
mypkg.txt
data/
somefile.dat
otherdata.dat
The setuptools setup file might look like this::
from setuptools import setup, find_packages
setup(
...
packages = find_packages('src'), # include all packages under src
package_dir = {'':'src'}, # tell distutils packages are under src
package_data = {
# If any package contains *.txt files, include them:
'': ['*.txt'],
# And include any *.dat files found in the 'data' subdirectory
# of the 'mypkg' package, also:
'mypkg': ['data/*.dat'],
}
)
Notice that if you list patterns in ``package_data`` under the empty string,
these patterns are used to find files in every package, even ones that also
have their own patterns listed. Thus, in the above example, the ``mypkg.txt``
file gets included even though it's not listed in the patterns for ``mypkg``.
Also notice that if you use paths, you *must* use a forward slash (``/``) as
the path separator, even if you are on Windows. Setuptools automatically
converts slashes to appropriate platform-specific separators at build time.
(Note: although the ``package_data`` argument was previously only available in
``setuptools``, it was also added to the Python ``distutils`` package as of
Python 2.4; there is `some documentation for the feature`__ available on the
python.org website. If using the setuptools-specific ``include_package_data``
argument, files specified by ``package_data`` will *not* be automatically
added to the manifest unless they are tracked by a supported version control
system, or are listed in the MANIFEST.in file.)
__ http://docs.python.org/dist/node11.html
Sometimes, the ``include_package_data`` or ``package_data`` options alone
aren't sufficient to precisely define what files you want included. For
example, you may want to include package README files in your revision control
system and source distributions, but exclude them from being installed. So,
setuptools offers an ``exclude_package_data`` option as well, that allows you
to do things like this::
from setuptools import setup, find_packages
setup(
...
packages = find_packages('src'), # include all packages under src
package_dir = {'':'src'}, # tell distutils packages are under src
include_package_data = True, # include everything in source control
# ...but exclude README.txt from all packages
exclude_package_data = { '': ['README.txt'] },
)
The ``exclude_package_data`` option is a dictionary mapping package names to
lists of wildcard patterns, just like the ``package_data`` option. And, just
as with that option, a key of ``''`` will apply the given pattern(s) to all
packages. However, any files that match these patterns will be *excluded*
from installation, even if they were listed in ``package_data`` or were
included as a result of using ``include_package_data``.
In summary, the three options allow you to:
``include_package_data``
Accept all data files and directories matched by ``MANIFEST.in`` or found
in source control.
``package_data``
Specify additional patterns to match files and directories that may or may
not be matched by ``MANIFEST.in`` or found in source control.
``exclude_package_data``
Specify patterns for data files and directories that should *not* be
included when a package is installed, even if they would otherwise have
been included due to the use of the preceding options.
NOTE: Due to the way the distutils build process works, a data file that you
include in your project and then stop including may be "orphaned" in your
project's build directories, requiring you to run ``setup.py clean --all`` to
fully remove them. This may also be important for your users and contributors
if they track intermediate revisions of your project using Subversion; be sure
to let them know when you make changes that remove files from inclusion so they
can run ``setup.py clean --all``.
Accessing Data Files at Runtime
-------------------------------
Typically, existing programs manipulate a package's ``__file__`` attribute in
order to find the location of data files. However, this manipulation isn't
compatible with PEP 302-based import hooks, including importing from zip files
and Python Eggs. It is strongly recommended that, if you are using data files,
you should use the `Resource Management API`_ of ``pkg_resources`` to access
them. The ``pkg_resources`` module is distributed as part of setuptools, so if
you're using setuptools to distribute your package, there is no reason not to
use its resource management API. See also `Accessing Package Resources`_ for
a quick example of converting code that uses ``__file__`` to use
``pkg_resources`` instead.
.. _Resource Management API: http://peak.telecommunity.com/DevCenter/PythonEggs#resource-management
.. _Accessing Package Resources: http://peak.telecommunity.com/DevCenter/PythonEggs#accessing-package-resources
Non-Package Data Files
----------------------
The ``distutils`` normally install general "data files" to a platform-specific
location (e.g. ``/usr/share``). This feature intended to be used for things
like documentation, example configuration files, and the like. ``setuptools``
does not install these data files in a separate location, however. They are
bundled inside the egg file or directory, alongside the Python modules and
packages. The data files can also be accessed using the `Resource Management
API`_, by specifying a ``Requirement`` instead of a package name::
from pkg_resources import Requirement, resource_filename
filename = resource_filename(Requirement.parse("MyProject"),"sample.conf")
The above code will obtain the filename of the "sample.conf" file in the data
root of the "MyProject" distribution.
Note, by the way, that this encapsulation of data files means that you can't
actually install data files to some arbitrary location on a user's machine;
this is a feature, not a bug. You can always include a script in your
distribution that extracts and copies your the documentation or data files to
a user-specified location, at their discretion. If you put related data files
in a single directory, you can use ``resource_filename()`` with the directory
name to get a filesystem directory that then can be copied with the ``shutil``
module. (Even if your package is installed as a zipfile, calling
``resource_filename()`` on a directory will return an actual filesystem
directory, whose contents will be that entire subtree of your distribution.)
(Of course, if you're writing a new package, you can just as easily place your
data files or directories inside one of your packages, rather than using the
distutils' approach. However, if you're updating an existing application, it
may be simpler not to change the way it currently specifies these data files.)
Automatic Resource Extraction
-----------------------------
If you are using tools that expect your resources to be "real" files, or your
project includes non-extension native libraries or other files that your C
extensions expect to be able to access, you may need to list those files in
the ``eager_resources`` argument to ``setup()``, so that the files will be
extracted together, whenever a C extension in the project is imported.
This is especially important if your project includes shared libraries *other*
than distutils-built C extensions, and those shared libraries use file
extensions other than ``.dll``, ``.so``, or ``.dylib``, which are the
extensions that setuptools 0.6a8 and higher automatically detects as shared
libraries and adds to the ``native_libs.txt`` file for you. Any shared
libraries whose names do not end with one of those extensions should be listed
as ``eager_resources``, because they need to be present in the filesystem when
he C extensions that link to them are used.
The ``pkg_resources`` runtime for compressed packages will automatically
extract *all* C extensions and ``eager_resources`` at the same time, whenever
*any* C extension or eager resource is requested via the ``resource_filename()``
API. (C extensions are imported using ``resource_filename()`` internally.)
This ensures that C extensions will see all of the "real" files that they
expect to see.
Note also that you can list directory resource names in ``eager_resources`` as
well, in which case the directory's contents (including subdirectories) will be
extracted whenever any C extension or eager resource is requested.
Please note that if you're not sure whether you need to use this argument, you
don't! It's really intended to support projects with lots of non-Python
dependencies and as a last resort for crufty projects that can't otherwise
handle being compressed. If your package is pure Python, Python plus data
files, or Python plus C, you really don't need this. You've got to be using
either C or an external program that needs "real" files in your project before
there's any possibility of ``eager_resources`` being relevant to your project.
Extensible Applications and Frameworks
======================================
.. _Entry Points:
Dynamic Discovery of Services and Plugins
-----------------------------------------
``setuptools`` supports creating libraries that "plug in" to extensible
applications and frameworks, by letting you register "entry points" in your
project that can be imported by the application or framework.
For example, suppose that a blogging tool wants to support plugins
that provide translation for various file types to the blog's output format.
The framework might define an "entry point group" called ``blogtool.parsers``,
and then allow plugins to register entry points for the file extensions they
support.
This would allow people to create distributions that contain one or more
parsers for different file types, and then the blogging tool would be able to
find the parsers at runtime by looking up an entry point for the file
extension (or mime type, or however it wants to).
Note that if the blogging tool includes parsers for certain file formats, it
can register these as entry points in its own setup script, which means it
doesn't have to special-case its built-in formats. They can just be treated
the same as any other plugin's entry points would be.
If you're creating a project that plugs in to an existing application or
framework, you'll need to know what entry points or entry point groups are
defined by that application or framework. Then, you can register entry points
in your setup script. Here are a few examples of ways you might register an
``.rst`` file parser entry point in the ``blogtool.parsers`` entry point group,
for our hypothetical blogging tool::
setup(
# ...
entry_points = {'blogtool.parsers': '.rst = some_module:SomeClass'}
)
setup(
# ...
entry_points = {'blogtool.parsers': ['.rst = some_module:a_func']}
)
setup(
# ...
entry_points = """
[blogtool.parsers]
.rst = some.nested.module:SomeClass.some_classmethod [reST]
""",
extras_require = dict(reST = "Docutils>=0.3.5")
)
The ``entry_points`` argument to ``setup()`` accepts either a string with
``.ini``-style sections, or a dictionary mapping entry point group names to
either strings or lists of strings containing entry point specifiers. An
entry point specifier consists of a name and value, separated by an ``=``
sign. The value consists of a dotted module name, optionally followed by a
``:`` and a dotted identifier naming an object within the module. It can
also include a bracketed list of "extras" that are required for the entry
point to be used. When the invoking application or framework requests loading
of an entry point, any requirements implied by the associated extras will be
passed to ``pkg_resources.require()``, so that an appropriate error message
can be displayed if the needed package(s) are missing. (Of course, the
invoking app or framework can ignore such errors if it wants to make an entry
point optional if a requirement isn't installed.)
Defining Additional Metadata
----------------------------
Some extensible applications and frameworks may need to define their own kinds
of metadata to include in eggs, which they can then access using the
``pkg_resources`` metadata APIs. Ordinarily, this is done by having plugin
developers include additional files in their ``ProjectName.egg-info``
directory. However, since it can be tedious to create such files by hand, you
may want to create a distutils extension that will create the necessary files
from arguments to ``setup()``, in much the same way that ``setuptools`` does
for many of the ``setup()`` arguments it adds. See the section below on
`Creating distutils Extensions`_ for more details, especially the subsection on
`Adding new EGG-INFO Files`_.
"Development Mode"
==================
Under normal circumstances, the ``distutils`` assume that you are going to
build a distribution of your project, not use it in its "raw" or "unbuilt"
form. If you were to use the ``distutils`` that way, you would have to rebuild
and reinstall your project every time you made a change to it during
development.
Another problem that sometimes comes up with the ``distutils`` is that you may
need to do development on two related projects at the same time. You may need
to put both projects' packages in the same directory to run them, but need to
keep them separate for revision control purposes. How can you do this?
Setuptools allows you to deploy your projects for use in a common directory or
staging area, but without copying any files. Thus, you can edit each project's
code in its checkout directory, and only need to run build commands when you
change a project's C extensions or similarly compiled files. You can even
deploy a project into another project's checkout directory, if that's your
preferred way of working (as opposed to using a common independent staging area
or the site-packages directory).
To do this, use the ``setup.py develop`` command. It works very similarly to
``setup.py install`` or the EasyInstall tool, except that it doesn't actually
install anything. Instead, it creates a special ``.egg-link`` file in the
deployment directory, that links to your project's source code. And, if your
deployment directory is Python's ``site-packages`` directory, it will also
update the ``easy-install.pth`` file to include your project's source code,
thereby making it available on ``sys.path`` for all programs using that Python
installation.
If you have enabled the ``use_2to3`` flag, then of course the ``.egg-link``
will not link directly to your source code when run under Python 3, since
that source code would be made for Python 2 and not work under Python 3.
Instead the ``setup.py develop`` will build Python 3 code under the ``build``
directory, and link there. This means that after doing code changes you will
have to run ``setup.py build`` before these changes are picked up by your
Python 3 installation.
In addition, the ``develop`` command creates wrapper scripts in the target
script directory that will run your in-development scripts after ensuring that
all your ``install_requires`` packages are available on ``sys.path``.
You can deploy the same project to multiple staging areas, e.g. if you have
multiple projects on the same machine that are sharing the same project you're
doing development work.
When you're done with a given development task, you can remove the project
source from a staging area using ``setup.py develop --uninstall``, specifying
the desired staging area if it's not the default.
There are several options to control the precise behavior of the ``develop``
command; see the section on the `develop`_ command below for more details.
Note that you can also apply setuptools commands to non-setuptools projects,
using commands like this::
python -c "import setuptools; execfile('setup.py')" develop
That is, you can simply list the normal setup commands and options following
the quoted part.
Distributing a ``setuptools``-based project
===========================================
Using ``setuptools``... Without bundling it!
---------------------------------------------
Your users might not have ``setuptools`` installed on their machines, or even
if they do, it might not be the right version. Fixing this is easy; just
download `distribute_setup.py`_, and put it in the same directory as your ``setup.py``
script. (Be sure to add it to your revision control system, too.) Then add
these two lines to the very top of your setup script, before the script imports
anything from setuptools:
.. code-block:: python
import distribute_setup
distribute_setup.use_setuptools()
That's it. The ``distribute_setup`` module will automatically download a matching
version of ``setuptools`` from PyPI, if it isn't present on the target system.
Whenever you install an updated version of setuptools, you should also update
your projects' ``distribute_setup.py`` files, so that a matching version gets installed
on the target machine(s).
By the way, setuptools supports the new PyPI "upload" command, so you can use
``setup.py sdist upload`` or ``setup.py bdist_egg upload`` to upload your
source or egg distributions respectively. Your project's current version must
be registered with PyPI first, of course; you can use ``setup.py register`` to
do that. Or you can do it all in one step, e.g. ``setup.py register sdist
bdist_egg upload`` will register the package, build source and egg
distributions, and then upload them both to PyPI, where they'll be easily
found by other projects that depend on them.
(By the way, if you need to distribute a specific version of ``setuptools``,
you can specify the exact version and base download URL as parameters to the
``use_setuptools()`` function. See the function's docstring for details.)
What Your Users Should Know
---------------------------
In general, a setuptools-based project looks just like any distutils-based
project -- as long as your users have an internet connection and are installing
to ``site-packages``, that is. But for some users, these conditions don't
apply, and they may become frustrated if this is their first encounter with
a setuptools-based project. To keep these users happy, you should review the
following topics in your project's installation instructions, if they are
relevant to your project and your target audience isn't already familiar with
setuptools and ``easy_install``.
Network Access
If your project is using ``distribute_setup``, you should inform users of the
need to either have network access, or to preinstall the correct version of
setuptools using the `EasyInstall installation instructions`_. Those
instructions also have tips for dealing with firewalls as well as how to
manually download and install setuptools.
Custom Installation Locations
You should inform your users that if they are installing your project to
somewhere other than the main ``site-packages`` directory, they should
first install setuptools using the instructions for `Custom Installation
Locations`_, before installing your project.
Your Project's Dependencies
If your project depends on other projects that may need to be downloaded
from PyPI or elsewhere, you should list them in your installation
instructions, or tell users how to find out what they are. While most
users will not need this information, any users who don't have unrestricted
internet access may have to find, download, and install the other projects
manually. (Note, however, that they must still install those projects
using ``easy_install``, or your project will not know they are installed,
and your setup script will try to download them again.)
If you want to be especially friendly to users with limited network access,
you may wish to build eggs for your project and its dependencies, making
them all available for download from your site, or at least create a page
with links to all of the needed eggs. In this way, users with limited
network access can manually download all the eggs to a single directory,
then use the ``-f`` option of ``easy_install`` to specify the directory
to find eggs in. Users who have full network access can just use ``-f``
with the URL of your download page, and ``easy_install`` will find all the
needed eggs using your links directly. This is also useful when your
target audience isn't able to compile packages (e.g. most Windows users)
and your package or some of its dependencies include C code.
Subversion or CVS Users and Co-Developers
Users and co-developers who are tracking your in-development code using
CVS, Subversion, or some other revision control system should probably read
this manual's sections regarding such development. Alternately, you may
wish to create a quick-reference guide containing the tips from this manual
that apply to your particular situation. For example, if you recommend
that people use ``setup.py develop`` when tracking your in-development
code, you should let them know that this needs to be run after every update
or commit.
Similarly, if you remove modules or data files from your project, you
should remind them to run ``setup.py clean --all`` and delete any obsolete
``.pyc`` or ``.pyo``. (This tip applies to the distutils in general, not
just setuptools, but not everybody knows about them; be kind to your users
by spelling out your project's best practices rather than leaving them
guessing.)
Creating System Packages
Some users want to manage all Python packages using a single package
manager, and sometimes that package manager isn't ``easy_install``!
Setuptools currently supports ``bdist_rpm``, ``bdist_wininst``, and
``bdist_dumb`` formats for system packaging. If a user has a locally-
installed "bdist" packaging tool that internally uses the distutils
``install`` command, it should be able to work with ``setuptools``. Some
examples of "bdist" formats that this should work with include the
``bdist_nsi`` and ``bdist_msi`` formats for Windows.
However, packaging tools that build binary distributions by running
``setup.py install`` on the command line or as a subprocess will require
modification to work with setuptools. They should use the
``--single-version-externally-managed`` option to the ``install`` command,
combined with the standard ``--root`` or ``--record`` options.
See the `install command`_ documentation below for more details. The
``bdist_deb`` command is an example of a command that currently requires
this kind of patching to work with setuptools.
If you or your users have a problem building a usable system package for
your project, please report the problem via the mailing list so that
either the "bdist" tool in question or setuptools can be modified to
resolve the issue.
Managing Multiple Projects
--------------------------
If you're managing several projects that need to use ``distribute_setup``, and you
are using Subversion as your revision control system, you can use the
"svn:externals" property to share a single copy of ``distribute_setup`` between
projects, so that it will always be up-to-date whenever you check out or update
an individual project, without having to manually update each project to use
a new version.
However, because Subversion only supports using directories as externals, you
have to turn ``distribute_setup.py`` into ``distribute_setup/__init__.py`` in order
to do this, then create "externals" definitions that map the ``distribute_setup``
directory into each project. Also, if any of your projects use
``find_packages()`` on their setup directory, you will need to exclude the
resulting ``distribute_setup`` package, to keep it from being included in your
distributions, e.g.::
setup(
...
packages = find_packages(exclude=['distribute_setup']),
)
Of course, the ``distribute_setup`` package will still be included in your
packages' source distributions, as it needs to be.
For your convenience, you may use the following external definition, which will
track the latest version of setuptools::
ez_setup svn://svn.eby-sarna.com/svnroot/ez_setup
You can set this by executing this command in your project directory::
svn propedit svn:externals .
And then adding the line shown above to the file that comes up for editing.
Setting the ``zip_safe`` flag
-----------------------------
For maximum performance, Python packages are best installed as zip files.
Not all packages, however, are capable of running in compressed form, because
they may expect to be able to access either source code or data files as
normal operating system files. So, ``setuptools`` can install your project
as a zipfile or a directory, and its default choice is determined by the
project's ``zip_safe`` flag.
You can pass a True or False value for the ``zip_safe`` argument to the
``setup()`` function, or you can omit it. If you omit it, the ``bdist_egg``
command will analyze your project's contents to see if it can detect any
conditions that would prevent it from working in a zipfile. It will output
notices to the console about any such conditions that it finds.
Currently, this analysis is extremely conservative: it will consider the
project unsafe if it contains any C extensions or datafiles whatsoever. This
does *not* mean that the project can't or won't work as a zipfile! It just
means that the ``bdist_egg`` authors aren't yet comfortable asserting that
the project *will* work. If the project contains no C or data files, and does
no ``__file__`` or ``__path__`` introspection or source code manipulation, then
there is an extremely solid chance the project will work when installed as a
zipfile. (And if the project uses ``pkg_resources`` for all its data file
access, then C extensions and other data files shouldn't be a problem at all.
See the `Accessing Data Files at Runtime`_ section above for more information.)
However, if ``bdist_egg`` can't be *sure* that your package will work, but
you've checked over all the warnings it issued, and you are either satisfied it
*will* work (or if you want to try it for yourself), then you should set
``zip_safe`` to ``True`` in your ``setup()`` call. If it turns out that it
doesn't work, you can always change it to ``False``, which will force
``setuptools`` to install your project as a directory rather than as a zipfile.
Of course, the end-user can still override either decision, if they are using
EasyInstall to install your package. And, if you want to override for testing
purposes, you can just run ``setup.py easy_install --zip-ok .`` or ``setup.py
easy_install --always-unzip .`` in your project directory. to install the
package as a zipfile or directory, respectively.
In the future, as we gain more experience with different packages and become
more satisfied with the robustness of the ``pkg_resources`` runtime, the
"zip safety" analysis may become less conservative. However, we strongly
recommend that you determine for yourself whether your project functions
correctly when installed as a zipfile, correct any problems if you can, and
then make an explicit declaration of ``True`` or ``False`` for the ``zip_safe``
flag, so that it will not be necessary for ``bdist_egg`` or ``EasyInstall`` to
try to guess whether your project can work as a zipfile.
Namespace Packages
------------------
Sometimes, a large package is more useful if distributed as a collection of
smaller eggs. However, Python does not normally allow the contents of a
package to be retrieved from more than one location. "Namespace packages"
are a solution for this problem. When you declare a package to be a namespace
package, it means that the package has no meaningful contents in its
``__init__.py``, and that it is merely a container for modules and subpackages.
The ``pkg_resources`` runtime will then automatically ensure that the contents
of namespace packages that are spread over multiple eggs or directories are
combined into a single "virtual" package.
The ``namespace_packages`` argument to ``setup()`` lets you declare your
project's namespace packages, so that they will be included in your project's
metadata. The argument should list the namespace packages that the egg
participates in. For example, the ZopeInterface project might do this::
setup(
# ...
namespace_packages = ['zope']
)
because it contains a ``zope.interface`` package that lives in the ``zope``
namespace package. Similarly, a project for a standalone ``zope.publisher``
would also declare the ``zope`` namespace package. When these projects are
installed and used, Python will see them both as part of a "virtual" ``zope``
package, even though they will be installed in different locations.
Namespace packages don't have to be top-level packages. For example, Zope 3's
``zope.app`` package is a namespace package, and in the future PEAK's
``peak.util`` package will be too.
Note, by the way, that your project's source tree must include the namespace
packages' ``__init__.py`` files (and the ``__init__.py`` of any parent
packages), in a normal Python package layout. These ``__init__.py`` files
*must* contain the line::
__import__('pkg_resources').declare_namespace(__name__)
This code ensures that the namespace package machinery is operating and that
the current package is registered as a namespace package.
You must NOT include any other code and data in a namespace package's
``__init__.py``. Even though it may appear to work during development, or when
projects are installed as ``.egg`` files, it will not work when the projects
are installed using "system" packaging tools -- in such cases the
``__init__.py`` files will not be installed, let alone executed.
You must include the ``declare_namespace()`` line in the ``__init__.py`` of
*every* project that has contents for the namespace package in question, in
order to ensure that the namespace will be declared regardless of which
project's copy of ``__init__.py`` is loaded first. If the first loaded
``__init__.py`` doesn't declare it, it will never *be* declared, because no
other copies will ever be loaded!)
TRANSITIONAL NOTE
~~~~~~~~~~~~~~~~~
Setuptools 0.6a automatically calls ``declare_namespace()`` for you at runtime,
but the 0.7a versions will *not*. This is because the automatic declaration
feature has some negative side effects, such as needing to import all namespace
packages during the initialization of the ``pkg_resources`` runtime, and also
the need for ``pkg_resources`` to be explicitly imported before any namespace
packages work at all. Beginning with the 0.7a releases, you'll be responsible
for including your own declaration lines, and the automatic declaration feature
will be dropped to get rid of the negative side effects.
During the remainder of the 0.6 development cycle, therefore, setuptools will
warn you about missing ``declare_namespace()`` calls in your ``__init__.py``
files, and you should correct these as soon as possible before setuptools 0.7a1
is released. Namespace packages without declaration lines will not work
correctly once a user has upgraded to setuptools 0.7a1, so it's important that
you make this change now in order to avoid having your code break in the field.
Our apologies for the inconvenience, and thank you for your patience.
Tagging and "Daily Build" or "Snapshot" Releases
------------------------------------------------
When a set of related projects are under development, it may be important to
track finer-grained version increments than you would normally use for e.g.
"stable" releases. While stable releases might be measured in dotted numbers
with alpha/beta/etc. status codes, development versions of a project often
need to be tracked by revision or build number or even build date. This is
especially true when projects in development need to refer to one another, and
therefore may literally need an up-to-the-minute version of something!
To support these scenarios, ``setuptools`` allows you to "tag" your source and
egg distributions by adding one or more of the following to the project's
"official" version identifier:
* A manually-specified pre-release tag, such as "build" or "dev", or a
manually-specified post-release tag, such as a build or revision number
(``--tag-build=STRING, -bSTRING``)
* A "last-modified revision number" string generated automatically from
Subversion's metadata (assuming your project is being built from a Subversion
"working copy") (``--tag-svn-revision, -r``)
* An 8-character representation of the build date (``--tag-date, -d``), as
a postrelease tag
You can add these tags by adding ``egg_info`` and the desired options to
the command line ahead of the ``sdist`` or ``bdist`` commands that you want
to generate a daily build or snapshot for. See the section below on the
`egg_info`_ command for more details.
(Also, before you release your project, be sure to see the section above on
`Specifying Your Project's Version`_ for more information about how pre- and
post-release tags affect how setuptools and EasyInstall interpret version
numbers. This is important in order to make sure that dependency processing
tools will know which versions of your project are newer than others.)
Finally, if you are creating builds frequently, and either building them in a
downloadable location or are copying them to a distribution server, you should
probably also check out the `rotate`_ command, which lets you automatically
delete all but the N most-recently-modified distributions matching a glob
pattern. So, you can use a command line like::
setup.py egg_info -rbDEV bdist_egg rotate -m.egg -k3
to build an egg whose version info includes 'DEV-rNNNN' (where NNNN is the
most recent Subversion revision that affected the source tree), and then
delete any egg files from the distribution directory except for the three
that were built most recently.
If you have to manage automated builds for multiple packages, each with
different tagging and rotation policies, you may also want to check out the
`alias`_ command, which would let each package define an alias like ``daily``
that would perform the necessary tag, build, and rotate commands. Then, a
simpler script or cron job could just run ``setup.py daily`` in each project
directory. (And, you could also define sitewide or per-user default versions
of the ``daily`` alias, so that projects that didn't define their own would
use the appropriate defaults.)
Generating Source Distributions
-------------------------------
``setuptools`` enhances the distutils' default algorithm for source file
selection, so that all files managed by CVS or Subversion in your project tree
are included in any source distribution you build. This is a big improvement
over having to manually write a ``MANIFEST.in`` file and try to keep it in
sync with your project. So, if you are using CVS or Subversion, and your
source distributions only need to include files that you're tracking in
revision control, don't create a a ``MANIFEST.in`` file for your project.
(And, if you already have one, you might consider deleting it the next time
you would otherwise have to change it.)
(NOTE: other revision control systems besides CVS and Subversion can be
supported using plugins; see the section below on `Adding Support for Other
Revision Control Systems`_ for information on how to write such plugins.)
If you need to include automatically generated files, or files that are kept in
an unsupported revision control system, you'll need to create a ``MANIFEST.in``
file to specify any files that the default file location algorithm doesn't
catch. See the distutils documentation for more information on the format of
the ``MANIFEST.in`` file.
But, be sure to ignore any part of the distutils documentation that deals with
``MANIFEST`` or how it's generated from ``MANIFEST.in``; setuptools shields you
from these issues and doesn't work the same way in any case. Unlike the
distutils, setuptools regenerates the source distribution manifest file
every time you build a source distribution, and it builds it inside the
project's ``.egg-info`` directory, out of the way of your main project
directory. You therefore need not worry about whether it is up-to-date or not.
Indeed, because setuptools' approach to determining the contents of a source
distribution is so much simpler, its ``sdist`` command omits nearly all of
the options that the distutils' more complex ``sdist`` process requires. For
all practical purposes, you'll probably use only the ``--formats`` option, if
you use any option at all.
(By the way, if you're using some other revision control system, you might
consider creating and publishing a `revision control plugin for setuptools`_.)
.. _revision control plugin for setuptools: `Adding Support for Other Revision Control Systems`_
Making your package available for EasyInstall
---------------------------------------------
If you use the ``register`` command (``setup.py register``) to register your
package with PyPI, that's most of the battle right there. (See the
`docs for the register command`_ for more details.)
.. _docs for the register command: http://docs.python.org/dist/package-index.html
If you also use the `upload`_ command to upload actual distributions of your
package, that's even better, because EasyInstall will be able to find and
download them directly from your project's PyPI page.
However, there may be reasons why you don't want to upload distributions to
PyPI, and just want your existing distributions (or perhaps a Subversion
checkout) to be used instead.
So here's what you need to do before running the ``register`` command. There
are three ``setup()`` arguments that affect EasyInstall:
``url`` and ``download_url``
These become links on your project's PyPI page. EasyInstall will examine
them to see if they link to a package ("primary links"), or whether they are
HTML pages. If they're HTML pages, EasyInstall scans all HREF's on the
page for primary links
``long_description``
EasyInstall will check any URLs contained in this argument to see if they
are primary links.
A URL is considered a "primary link" if it is a link to a .tar.gz, .tgz, .zip,
.egg, .egg.zip, .tar.bz2, or .exe file, or if it has an ``#egg=project`` or
``#egg=project-version`` fragment identifier attached to it. EasyInstall
attempts to determine a project name and optional version number from the text
of a primary link *without* downloading it. When it has found all the primary
links, EasyInstall will select the best match based on requested version,
platform compatibility, and other criteria.
So, if your ``url`` or ``download_url`` point either directly to a downloadable
source distribution, or to HTML page(s) that have direct links to such, then
EasyInstall will be able to locate downloads automatically. If you want to
make Subversion checkouts available, then you should create links with either
``#egg=project`` or ``#egg=project-version`` added to the URL. You should
replace ``project`` and ``version`` with the values they would have in an egg
filename. (Be sure to actually generate an egg and then use the initial part
of the filename, rather than trying to guess what the escaped form of the
project name and version number will be.)
Note that Subversion checkout links are of lower precedence than other kinds
of distributions, so EasyInstall will not select a Subversion checkout for
downloading unless it has a version included in the ``#egg=`` suffix, and
it's a higher version than EasyInstall has seen in any other links for your
project.
As a result, it's a common practice to use mark checkout URLs with a version of
"dev" (i.e., ``#egg=projectname-dev``), so that users can do something like
this::
easy_install --editable projectname==dev
in order to check out the in-development version of ``projectname``.
Managing "Continuous Releases" Using Subversion
-----------------------------------------------
If you expect your users to track in-development versions of your project via
Subversion, there are a few additional steps you should take to ensure that
things work smoothly with EasyInstall. First, you should add the following
to your project's ``setup.cfg`` file:
.. code-block:: ini
[egg_info]
tag_build = .dev
tag_svn_revision = 1
This will tell ``setuptools`` to generate package version numbers like
``1.0a1.dev-r1263``, which will be considered to be an *older* release than
``1.0a1``. Thus, when you actually release ``1.0a1``, the entire egg
infrastructure (including ``setuptools``, ``pkg_resources`` and EasyInstall)
will know that ``1.0a1`` supersedes any interim snapshots from Subversion, and
handle upgrades accordingly.
(Note: the project version number you specify in ``setup.py`` should always be
the *next* version of your software, not the last released version.
Alternately, you can leave out the ``tag_build=.dev``, and always use the
*last* release as a version number, so that your post-1.0 builds are labelled
``1.0-r1263``, indicating a post-1.0 patchlevel. Most projects so far,
however, seem to prefer to think of their project as being a future version
still under development, rather than a past version being patched. It is of
course possible for a single project to have both situations, using
post-release numbering on release branches, and pre-release numbering on the
trunk. But you don't have to make things this complex if you don't want to.)
Commonly, projects releasing code from Subversion will include a PyPI link to
their checkout URL (as described in the previous section) with an
``#egg=projectname-dev`` suffix. This allows users to request EasyInstall
to download ``projectname==dev`` in order to get the latest in-development
code. Note that if your project depends on such in-progress code, you may wish
to specify your ``install_requires`` (or other requirements) to include
``==dev``, e.g.:
.. code-block:: python
install_requires = ["OtherProject>=0.2a1.dev-r143,==dev"]
The above example says, "I really want at least this particular development
revision number, but feel free to follow and use an ``#egg=OtherProject-dev``
link if you find one". This avoids the need to have actual source or binary
distribution snapshots of in-development code available, just to be able to
depend on the latest and greatest a project has to offer.
A final note for Subversion development: if you are using SVN revision tags
as described in this section, it's a good idea to run ``setup.py develop``
after each Subversion checkin or update, because your project's version number
will be changing, and your script wrappers need to be updated accordingly.
Also, if the project's requirements have changed, the ``develop`` command will
take care of fetching the updated dependencies, building changed extensions,
etc. Be sure to also remind any of your users who check out your project
from Subversion that they need to run ``setup.py develop`` after every update
in order to keep their checkout completely in sync.
Making "Official" (Non-Snapshot) Releases
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
When you make an official release, creating source or binary distributions,
you will need to override the tag settings from ``setup.cfg``, so that you
don't end up registering versions like ``foobar-0.7a1.dev-r34832``. This is
easy to do if you are developing on the trunk and using tags or branches for
your releases - just make the change to ``setup.cfg`` after branching or
tagging the release, so the trunk will still produce development snapshots.
Alternately, if you are not branching for releases, you can override the
default version options on the command line, using something like::
python setup.py egg_info -RDb "" sdist bdist_egg register upload
The first part of this command (``egg_info -RDb ""``) will override the
configured tag information, before creating source and binary eggs, registering
the project with PyPI, and uploading the files. Thus, these commands will use
the plain version from your ``setup.py``, without adding the Subversion
revision number or build designation string.
Of course, if you will be doing this a lot, you may wish to create a personal
alias for this operation, e.g.::
python setup.py alias -u release egg_info -RDb ""
You can then use it like this::
python setup.py release sdist bdist_egg register upload
Or of course you can create more elaborate aliases that do all of the above.
See the sections below on the `egg_info`_ and `alias`_ commands for more ideas.
Distributing Extensions compiled with Pyrex
-------------------------------------------
``setuptools`` includes transparent support for building Pyrex extensions, as
long as you define your extensions using ``setuptools.Extension``, *not*
``distutils.Extension``. You must also not import anything from Pyrex in
your setup script.
If you follow these rules, you can safely list ``.pyx`` files as the source
of your ``Extension`` objects in the setup script. ``setuptools`` will detect
at build time whether Pyrex is installed or not. If it is, then ``setuptools``
will use it. If not, then ``setuptools`` will silently change the
``Extension`` objects to refer to the ``.c`` counterparts of the ``.pyx``
files, so that the normal distutils C compilation process will occur.
Of course, for this to work, your source distributions must include the C
code generated by Pyrex, as well as your original ``.pyx`` files. This means
that you will probably want to include current ``.c`` files in your revision
control system, rebuilding them whenever you check changes in for the ``.pyx``
source files. This will ensure that people tracking your project in CVS or
Subversion will be able to build it even if they don't have Pyrex installed,
and that your source releases will be similarly usable with or without Pyrex.
-----------------
Command Reference
-----------------
.. _alias:
``alias`` - Define shortcuts for commonly used commands
=======================================================
Sometimes, you need to use the same commands over and over, but you can't
necessarily set them as defaults. For example, if you produce both development
snapshot releases and "stable" releases of a project, you may want to put
the distributions in different places, or use different ``egg_info`` tagging
options, etc. In these cases, it doesn't make sense to set the options in
a distutils configuration file, because the values of the options changed based
on what you're trying to do.
Setuptools therefore allows you to define "aliases" - shortcut names for
an arbitrary string of commands and options, using ``setup.py alias aliasname
expansion``, where aliasname is the name of the new alias, and the remainder of
the command line supplies its expansion. For example, this command defines
a sitewide alias called "daily", that sets various ``egg_info`` tagging
options::
setup.py alias --global-config daily egg_info --tag-svn-revision \
--tag-build=development
Once the alias is defined, it can then be used with other setup commands,
e.g.::
setup.py daily bdist_egg # generate a daily-build .egg file
setup.py daily sdist # generate a daily-build source distro
setup.py daily sdist bdist_egg # generate both
The above commands are interpreted as if the word ``daily`` were replaced with
``egg_info --tag-svn-revision --tag-build=development``.
Note that setuptools will expand each alias *at most once* in a given command
line. This serves two purposes. First, if you accidentally create an alias
loop, it will have no effect; you'll instead get an error message about an
unknown command. Second, it allows you to define an alias for a command, that
uses that command. For example, this (project-local) alias::
setup.py alias bdist_egg bdist_egg rotate -k1 -m.egg
redefines the ``bdist_egg`` command so that it always runs the ``rotate``
command afterwards to delete all but the newest egg file. It doesn't loop
indefinitely on ``bdist_egg`` because the alias is only expanded once when
used.
You can remove a defined alias with the ``--remove`` (or ``-r``) option, e.g.::
setup.py alias --global-config --remove daily
would delete the "daily" alias we defined above.
Aliases can be defined on a project-specific, per-user, or sitewide basis. The
default is to define or remove a project-specific alias, but you can use any of
the `configuration file options`_ (listed under the `saveopts`_ command, below)
to determine which distutils configuration file an aliases will be added to
(or removed from).
Note that if you omit the "expansion" argument to the ``alias`` command,
you'll get output showing that alias' current definition (and what
configuration file it's defined in). If you omit the alias name as well,
you'll get a listing of all current aliases along with their configuration
file locations.
``bdist_egg`` - Create a Python Egg for the project
===================================================
This command generates a Python Egg (``.egg`` file) for the project. Python
Eggs are the preferred binary distribution format for EasyInstall, because they
are cross-platform (for "pure" packages), directly importable, and contain
project metadata including scripts and information about the project's
dependencies. They can be simply downloaded and added to ``sys.path``
directly, or they can be placed in a directory on ``sys.path`` and then
automatically discovered by the egg runtime system.
This command runs the `egg_info`_ command (if it hasn't already run) to update
the project's metadata (``.egg-info``) directory. If you have added any extra
metadata files to the ``.egg-info`` directory, those files will be included in
the new egg file's metadata directory, for use by the egg runtime system or by
any applications or frameworks that use that metadata.
You won't usually need to specify any special options for this command; just
use ``bdist_egg`` and you're done. But there are a few options that may
be occasionally useful:
``--dist-dir=DIR, -d DIR``
Set the directory where the ``.egg`` file will be placed. If you don't
supply this, then the ``--dist-dir`` setting of the ``bdist`` command
will be used, which is usually a directory named ``dist`` in the project
directory.
``--plat-name=PLATFORM, -p PLATFORM``
Set the platform name string that will be embedded in the egg's filename
(assuming the egg contains C extensions). This can be used to override
the distutils default platform name with something more meaningful. Keep
in mind, however, that the egg runtime system expects to see eggs with
distutils platform names, so it may ignore or reject eggs with non-standard
platform names. Similarly, the EasyInstall program may ignore them when
searching web pages for download links. However, if you are
cross-compiling or doing some other unusual things, you might find a use
for this option.
``--exclude-source-files``
Don't include any modules' ``.py`` files in the egg, just compiled Python,
C, and data files. (Note that this doesn't affect any ``.py`` files in the
EGG-INFO directory or its subdirectories, since for example there may be
scripts with a ``.py`` extension which must still be retained.) We don't
recommend that you use this option except for packages that are being
bundled for proprietary end-user applications, or for "embedded" scenarios
where space is at an absolute premium. On the other hand, if your package
is going to be installed and used in compressed form, you might as well
exclude the source because Python's ``traceback`` module doesn't currently
understand how to display zipped source code anyway, or how to deal with
files that are in a different place from where their code was compiled.
There are also some options you will probably never need, but which are there
because they were copied from similar ``bdist`` commands used as an example for
creating this one. They may be useful for testing and debugging, however,
which is why we kept them:
``--keep-temp, -k``
Keep the contents of the ``--bdist-dir`` tree around after creating the
``.egg`` file.
``--bdist-dir=DIR, -b DIR``
Set the temporary directory for creating the distribution. The entire
contents of this directory are zipped to create the ``.egg`` file, after
running various installation commands to copy the package's modules, data,
and extensions here.
``--skip-build``
Skip doing any "build" commands; just go straight to the
install-and-compress phases.
.. _develop:
``develop`` - Deploy the project source in "Development Mode"
=============================================================
This command allows you to deploy your project's source for use in one or more
"staging areas" where it will be available for importing. This deployment is
done in such a way that changes to the project source are immediately available
in the staging area(s), without needing to run a build or install step after
each change.
The ``develop`` command works by creating an ``.egg-link`` file (named for the
project) in the given staging area. If the staging area is Python's
``site-packages`` directory, it also updates an ``easy-install.pth`` file so
that the project is on ``sys.path`` by default for all programs run using that
Python installation.
The ``develop`` command also installs wrapper scripts in the staging area (or
a separate directory, as specified) that will ensure the project's dependencies
are available on ``sys.path`` before running the project's source scripts.
And, it ensures that any missing project dependencies are available in the
staging area, by downloading and installing them if necessary.
Last, but not least, the ``develop`` command invokes the ``build_ext -i``
command to ensure any C extensions in the project have been built and are
up-to-date, and the ``egg_info`` command to ensure the project's metadata is
updated (so that the runtime and wrappers know what the project's dependencies
are). If you make any changes to the project's setup script or C extensions,
you should rerun the ``develop`` command against all relevant staging areas to
keep the project's scripts, metadata and extensions up-to-date. Most other
kinds of changes to your project should not require any build operations or
rerunning ``develop``, but keep in mind that even minor changes to the setup
script (e.g. changing an entry point definition) require you to re-run the
``develop`` or ``test`` commands to keep the distribution updated.
Here are some of the options that the ``develop`` command accepts. Note that
they affect the project's dependencies as well as the project itself, so if you
have dependencies that need to be installed and you use ``--exclude-scripts``
(for example), the dependencies' scripts will not be installed either! For
this reason, you may want to use EasyInstall to install the project's
dependencies before using the ``develop`` command, if you need finer control
over the installation options for dependencies.
``--uninstall, -u``
Un-deploy the current project. You may use the ``--install-dir`` or ``-d``
option to designate the staging area. The created ``.egg-link`` file will
be removed, if present and it is still pointing to the project directory.
The project directory will be removed from ``easy-install.pth`` if the
staging area is Python's ``site-packages`` directory.
Note that this option currently does *not* uninstall script wrappers! You
must uninstall them yourself, or overwrite them by using EasyInstall to
activate a different version of the package. You can also avoid installing
script wrappers in the first place, if you use the ``--exclude-scripts``
(aka ``-x``) option when you run ``develop`` to deploy the project.
``--multi-version, -m``
"Multi-version" mode. Specifying this option prevents ``develop`` from
adding an ``easy-install.pth`` entry for the project(s) being deployed, and
if an entry for any version of a project already exists, the entry will be
removed upon successful deployment. In multi-version mode, no specific
version of the package is available for importing, unless you use
``pkg_resources.require()`` to put it on ``sys.path``, or you are running
a wrapper script generated by ``setuptools`` or EasyInstall. (In which
case the wrapper script calls ``require()`` for you.)
Note that if you install to a directory other than ``site-packages``,
this option is automatically in effect, because ``.pth`` files can only be
used in ``site-packages`` (at least in Python 2.3 and 2.4). So, if you use
the ``--install-dir`` or ``-d`` option (or they are set via configuration
file(s)) your project and its dependencies will be deployed in multi-
version mode.
``--install-dir=DIR, -d DIR``
Set the installation directory (staging area). If this option is not
directly specified on the command line or in a distutils configuration
file, the distutils default installation location is used. Normally, this
will be the ``site-packages`` directory, but if you are using distutils
configuration files, setting things like ``prefix`` or ``install_lib``,
then those settings are taken into account when computing the default
staging area.
``--script-dir=DIR, -s DIR``
Set the script installation directory. If you don't supply this option
(via the command line or a configuration file), but you *have* supplied
an ``--install-dir`` (via command line or config file), then this option
defaults to the same directory, so that the scripts will be able to find
their associated package installation. Otherwise, this setting defaults
to the location where the distutils would normally install scripts, taking
any distutils configuration file settings into account.
``--exclude-scripts, -x``
Don't deploy script wrappers. This is useful if you don't want to disturb
existing versions of the scripts in the staging area.
``--always-copy, -a``
Copy all needed distributions to the staging area, even if they
are already present in another directory on ``sys.path``. By default, if
a requirement can be met using a distribution that is already available in
a directory on ``sys.path``, it will not be copied to the staging area.
``--egg-path=DIR``
Force the generated ``.egg-link`` file to use a specified relative path
to the source directory. This can be useful in circumstances where your
installation directory is being shared by code running under multiple
platforms (e.g. Mac and Windows) which have different absolute locations
for the code under development, but the same *relative* locations with
respect to the installation directory. If you use this option when
installing, you must supply the same relative path when uninstalling.
In addition to the above options, the ``develop`` command also accepts all of
the same options accepted by ``easy_install``. If you've configured any
``easy_install`` settings in your ``setup.cfg`` (or other distutils config
files), the ``develop`` command will use them as defaults, unless you override
them in a ``[develop]`` section or on the command line.
``easy_install`` - Find and install packages
============================================
This command runs the `EasyInstall tool
<http://peak.telecommunity.com/DevCenter/EasyInstall>`_ for you. It is exactly
equivalent to running the ``easy_install`` command. All command line arguments
following this command are consumed and not processed further by the distutils,
so this must be the last command listed on the command line. Please see
the EasyInstall documentation for the options reference and usage examples.
Normally, there is no reason to use this command via the command line, as you
can just use ``easy_install`` directly. It's only listed here so that you know
it's a distutils command, which means that you can:
* create command aliases that use it,
* create distutils extensions that invoke it as a subcommand, and
* configure options for it in your ``setup.cfg`` or other distutils config
files.
.. _egg_info:
``egg_info`` - Create egg metadata and set build tags
=====================================================
This command performs two operations: it updates a project's ``.egg-info``
metadata directory (used by the ``bdist_egg``, ``develop``, and ``test``
commands), and it allows you to temporarily change a project's version string,
to support "daily builds" or "snapshot" releases. It is run automatically by
the ``sdist``, ``bdist_egg``, ``develop``, ``register``, and ``test`` commands
in order to update the project's metadata, but you can also specify it
explicitly in order to temporarily change the project's version string while
executing other commands. (It also generates the``.egg-info/SOURCES.txt``
manifest file, which is used when you are building source distributions.)
In addition to writing the core egg metadata defined by ``setuptools`` and
required by ``pkg_resources``, this command can be extended to write other
metadata files as well, by defining entry points in the ``egg_info.writers``
group. See the section on `Adding new EGG-INFO Files`_ below for more details.
Note that using additional metadata writers may require you to include a
``setup_requires`` argument to ``setup()`` in order to ensure that the desired
writers are available on ``sys.path``.
Release Tagging Options
-----------------------
The following options can be used to modify the project's version string for
all remaining commands on the setup command line. The options are processed
in the order shown, so if you use more than one, the requested tags will be
added in the following order:
``--tag-build=NAME, -b NAME``
Append NAME to the project's version string. Due to the way setuptools
processes "pre-release" version suffixes beginning with the letters "a"
through "e" (like "alpha", "beta", and "candidate"), you will usually want
to use a tag like ".build" or ".dev", as this will cause the version number
to be considered *lower* than the project's default version. (If you
want to make the version number *higher* than the default version, you can
always leave off --tag-build and then use one or both of the following
options.)
If you have a default build tag set in your ``setup.cfg``, you can suppress
it on the command line using ``-b ""`` or ``--tag-build=""`` as an argument
to the ``egg_info`` command.
``--tag-svn-revision, -r``
If the current directory is a Subversion checkout (i.e. has a ``.svn``
subdirectory, this appends a string of the form "-rNNNN" to the project's
version string, where NNNN is the revision number of the most recent
modification to the current directory, as obtained from the ``svn info``
command.
If the current directory is not a Subversion checkout, the command will
look for a ``PKG-INFO`` file instead, and try to find the revision number
from that, by looking for a "-rNNNN" string at the end of the version
number. (This is so that building a package from a source distribution of
a Subversion snapshot will produce a binary with the correct version
number.)
If there is no ``PKG-INFO`` file, or the version number contained therein
does not end with ``-r`` and a number, then ``-r0`` is used.
``--no-svn-revision, -R``
Don't include the Subversion revision in the version number. This option
is included so you can override a default setting put in ``setup.cfg``.
``--tag-date, -d``
Add a date stamp of the form "-YYYYMMDD" (e.g. "-20050528") to the
project's version number.
``--no-date, -D``
Don't include a date stamp in the version number. This option is included
so you can override a default setting in ``setup.cfg``.
(Note: Because these options modify the version number used for source and
binary distributions of your project, you should first make sure that you know
how the resulting version numbers will be interpreted by automated tools
like EasyInstall. See the section above on `Specifying Your Project's
Version`_ for an explanation of pre- and post-release tags, as well as tips on
how to choose and verify a versioning scheme for your your project.)
For advanced uses, there is one other option that can be set, to change the
location of the project's ``.egg-info`` directory. Commands that need to find
the project's source directory or metadata should get it from this setting:
Other ``egg_info`` Options
--------------------------
``--egg-base=SOURCEDIR, -e SOURCEDIR``
Specify the directory that should contain the .egg-info directory. This
should normally be the root of your project's source tree (which is not
necessarily the same as your project directory; some projects use a ``src``
or ``lib`` subdirectory as the source root). You should not normally need
to specify this directory, as it is normally determined from the
``package_dir`` argument to the ``setup()`` function, if any. If there is
no ``package_dir`` set, this option defaults to the current directory.
``egg_info`` Examples
---------------------
Creating a dated "nightly build" snapshot egg::
python setup.py egg_info --tag-date --tag-build=DEV bdist_egg
Creating and uploading a release with no version tags, even if some default
tags are specified in ``setup.cfg``::
python setup.py egg_info -RDb "" sdist bdist_egg register upload
(Notice that ``egg_info`` must always appear on the command line *before* any
commands that you want the version changes to apply to.)
.. _install command:
``install`` - Run ``easy_install`` or old-style installation
============================================================
The setuptools ``install`` command is basically a shortcut to run the
``easy_install`` command on the current project. However, for convenience
in creating "system packages" of setuptools-based projects, you can also
use this option:
``--single-version-externally-managed``
This boolean option tells the ``install`` command to perform an "old style"
installation, with the addition of an ``.egg-info`` directory so that the
installed project will still have its metadata available and operate
normally. If you use this option, you *must* also specify the ``--root``
or ``--record`` options (or both), because otherwise you will have no way
to identify and remove the installed files.
This option is automatically in effect when ``install`` is invoked by another
distutils command, so that commands like ``bdist_wininst`` and ``bdist_rpm``
will create system packages of eggs. It is also automatically in effect if
you specify the ``--root`` option.
``install_egg_info`` - Install an ``.egg-info`` directory in ``site-packages``
==============================================================================
Setuptools runs this command as part of ``install`` operations that use the
``--single-version-externally-managed`` options. You should not invoke it
directly; it is documented here for completeness and so that distutils
extensions such as system package builders can make use of it. This command
has only one option:
``--install-dir=DIR, -d DIR``
The parent directory where the ``.egg-info`` directory will be placed.
Defaults to the same as the ``--install-dir`` option specified for the
``install_lib`` command, which is usually the system ``site-packages``
directory.
This command assumes that the ``egg_info`` command has been given valid options
via the command line or ``setup.cfg``, as it will invoke the ``egg_info``
command and use its options to locate the project's source ``.egg-info``
directory.
.. _rotate:
``rotate`` - Delete outdated distribution files
===============================================
As you develop new versions of your project, your distribution (``dist``)
directory will gradually fill up with older source and/or binary distribution
files. The ``rotate`` command lets you automatically clean these up, keeping
only the N most-recently modified files matching a given pattern.
``--match=PATTERNLIST, -m PATTERNLIST``
Comma-separated list of glob patterns to match. This option is *required*.
The project name and ``-*`` is prepended to the supplied patterns, in order
to match only distributions belonging to the current project (in case you
have a shared distribution directory for multiple projects). Typically,
you will use a glob pattern like ``.zip`` or ``.egg`` to match files of
the specified type. Note that each supplied pattern is treated as a
distinct group of files for purposes of selecting files to delete.
``--keep=COUNT, -k COUNT``
Number of matching distributions to keep. For each group of files
identified by a pattern specified with the ``--match`` option, delete all
but the COUNT most-recently-modified files in that group. This option is
*required*.
``--dist-dir=DIR, -d DIR``
Directory where the distributions are. This defaults to the value of the
``bdist`` command's ``--dist-dir`` option, which will usually be the
project's ``dist`` subdirectory.
**Example 1**: Delete all .tar.gz files from the distribution directory, except
for the 3 most recently modified ones::
setup.py rotate --match=.tar.gz --keep=3
**Example 2**: Delete all Python 2.3 or Python 2.4 eggs from the distribution
directory, except the most recently modified one for each Python version::
setup.py rotate --match=-py2.3*.egg,-py2.4*.egg --keep=1
.. _saveopts:
``saveopts`` - Save used options to a configuration file
========================================================
Finding and editing ``distutils`` configuration files can be a pain, especially
since you also have to translate the configuration options from command-line
form to the proper configuration file format. You can avoid these hassles by
using the ``saveopts`` command. Just add it to the command line to save the
options you used. For example, this command builds the project using
the ``mingw32`` C compiler, then saves the --compiler setting as the default
for future builds (even those run implicitly by the ``install`` command)::
setup.py build --compiler=mingw32 saveopts
The ``saveopts`` command saves all options for every commmand specified on the
command line to the project's local ``setup.cfg`` file, unless you use one of
the `configuration file options`_ to change where the options are saved. For
example, this command does the same as above, but saves the compiler setting
to the site-wide (global) distutils configuration::
setup.py build --compiler=mingw32 saveopts -g
Note that it doesn't matter where you place the ``saveopts`` command on the
command line; it will still save all the options specified for all commands.
For example, this is another valid way to spell the last example::
setup.py saveopts -g build --compiler=mingw32
Note, however, that all of the commands specified are always run, regardless of
where ``saveopts`` is placed on the command line.
Configuration File Options
--------------------------
Normally, settings such as options and aliases are saved to the project's
local ``setup.cfg`` file. But you can override this and save them to the
global or per-user configuration files, or to a manually-specified filename.
``--global-config, -g``
Save settings to the global ``distutils.cfg`` file inside the ``distutils``
package directory. You must have write access to that directory to use
this option. You also can't combine this option with ``-u`` or ``-f``.
``--user-config, -u``
Save settings to the current user's ``~/.pydistutils.cfg`` (POSIX) or
``$HOME/pydistutils.cfg`` (Windows) file. You can't combine this option
with ``-g`` or ``-f``.
``--filename=FILENAME, -f FILENAME``
Save settings to the specified configuration file to use. You can't
combine this option with ``-g`` or ``-u``. Note that if you specify a
non-standard filename, the ``distutils`` and ``setuptools`` will not
use the file's contents. This option is mainly included for use in
testing.
These options are used by other ``setuptools`` commands that modify
configuration files, such as the `alias`_ and `setopt`_ commands.
.. _setopt:
``setopt`` - Set a distutils or setuptools option in a config file
==================================================================
This command is mainly for use by scripts, but it can also be used as a quick
and dirty way to change a distutils configuration option without having to
remember what file the options are in and then open an editor.
**Example 1**. Set the default C compiler to ``mingw32`` (using long option
names)::
setup.py setopt --command=build --option=compiler --set-value=mingw32
**Example 2**. Remove any setting for the distutils default package
installation directory (short option names)::
setup.py setopt -c install -o install_lib -r
Options for the ``setopt`` command:
``--command=COMMAND, -c COMMAND``
Command to set the option for. This option is required.
``--option=OPTION, -o OPTION``
The name of the option to set. This option is required.
``--set-value=VALUE, -s VALUE``
The value to set the option to. Not needed if ``-r`` or ``--remove`` is
set.
``--remove, -r``
Remove (unset) the option, instead of setting it.
In addition to the above options, you may use any of the `configuration file
options`_ (listed under the `saveopts`_ command, above) to determine which
distutils configuration file the option will be added to (or removed from).
.. _test:
``test`` - Build package and run a unittest suite
=================================================
When doing test-driven development, or running automated builds that need
testing before they are deployed for downloading or use, it's often useful
to be able to run a project's unit tests without actually deploying the project
anywhere, even using the ``develop`` command. The ``test`` command runs a
project's unit tests without actually deploying it, by temporarily putting the
project's source on ``sys.path``, after first running ``build_ext -i`` and
``egg_info`` to ensure that any C extensions and project metadata are
up-to-date.
To use this command, your project's tests must be wrapped in a ``unittest``
test suite by either a function, a ``TestCase`` class or method, or a module
or package containing ``TestCase`` classes. If the named suite is a module,
and the module has an ``additional_tests()`` function, it is called and the
result (which must be a ``unittest.TestSuite``) is added to the tests to be
run. If the named suite is a package, any submodules and subpackages are
recursively added to the overall test suite. (Note: if your project specifies
a ``test_loader``, the rules for processing the chosen ``test_suite`` may
differ; see the `test_loader`_ documentation for more details.)
Note that many test systems including ``doctest`` support wrapping their
non-``unittest`` tests in ``TestSuite`` objects. So, if you are using a test
package that does not support this, we suggest you encourage its developers to
implement test suite support, as this is a convenient and standard way to
aggregate a collection of tests to be run under a common test harness.
By default, tests will be run in the "verbose" mode of the ``unittest``
package's text test runner, but you can get the "quiet" mode (just dots) if
you supply the ``-q`` or ``--quiet`` option, either as a global option to
the setup script (e.g. ``setup.py -q test``) or as an option for the ``test``
command itself (e.g. ``setup.py test -q``). There is one other option
available:
``--test-suite=NAME, -s NAME``
Specify the test suite (or module, class, or method) to be run
(e.g. ``some_module.test_suite``). The default for this option can be
set by giving a ``test_suite`` argument to the ``setup()`` function, e.g.::
setup(
# ...
test_suite = "my_package.tests.test_all"
)
If you did not set a ``test_suite`` in your ``setup()`` call, and do not
provide a ``--test-suite`` option, an error will occur.
.. _upload:
``upload`` - Upload source and/or egg distributions to PyPI
===========================================================
PyPI now supports uploading project files for redistribution; uploaded files
are easily found by EasyInstall, even if you don't have download links on your
project's home page.
Although Python 2.5 will support uploading all types of distributions to PyPI,
setuptools only supports source distributions and eggs. (This is partly
because PyPI's upload support is currently broken for various other file
types.) To upload files, you must include the ``upload`` command *after* the
``sdist`` or ``bdist_egg`` commands on the setup command line. For example::
setup.py bdist_egg upload # create an egg and upload it
setup.py sdist upload # create a source distro and upload it
setup.py sdist bdist_egg upload # create and upload both
Note that to upload files for a project, the corresponding version must already
be registered with PyPI, using the distutils ``register`` command. It's
usually a good idea to include the ``register`` command at the start of the
command line, so that any registration problems can be found and fixed before
building and uploading the distributions, e.g.::
setup.py register sdist bdist_egg upload
This will update PyPI's listing for your project's current version.
Note, by the way, that the metadata in your ``setup()`` call determines what
will be listed in PyPI for your package. Try to fill out as much of it as
possible, as it will save you a lot of trouble manually adding and updating
your PyPI listings. Just put it in ``setup.py`` and use the ``register``
comamnd to keep PyPI up to date.
The ``upload`` command has a few options worth noting:
``--sign, -s``
Sign each uploaded file using GPG (GNU Privacy Guard). The ``gpg`` program
must be available for execution on the system ``PATH``.
``--identity=NAME, -i NAME``
Specify the identity or key name for GPG to use when signing. The value of
this option will be passed through the ``--local-user`` option of the
``gpg`` program.
``--show-response``
Display the full response text from server; this is useful for debugging
PyPI problems.
``--repository=URL, -r URL``
The URL of the repository to upload to. Defaults to
https://pypi.python.org/pypi (i.e., the main PyPI installation).
.. _upload_docs:
``upload_docs`` - Upload package documentation to PyPI
======================================================
PyPI now supports uploading project documentation to the dedicated URL
https://pythonhosted.org/<project>/.
The ``upload_docs`` command will create the necessary zip file out of a
documentation directory and will post to the repository.
Note that to upload the documentation of a project, the corresponding version
must already be registered with PyPI, using the distutils ``register``
command -- just like the ``upload`` command.
Assuming there is an ``Example`` project with documentation in the
subdirectory ``docs``, e.g.::
Example/
|-- example.py
|-- setup.cfg
|-- setup.py
|-- docs
| |-- build
| | `-- html
| | | |-- index.html
| | | `-- tips_tricks.html
| |-- conf.py
| |-- index.txt
| `-- tips_tricks.txt
You can simply pass the documentation directory path to the ``upload_docs``
command::
python setup.py upload_docs --upload-dir=docs/build/html
If no ``--upload-dir`` is given, ``upload_docs`` will attempt to run the
``build_sphinx`` command to generate uploadable documentation.
For the command to become available, `Sphinx <http://sphinx.pocoo.org/>`_
must be installed in the same environment as distribute.
As with other ``setuptools``-based commands, you can define useful
defaults in the ``setup.cfg`` of your Python project, e.g.:
.. code-block:: ini
[upload_docs]
upload-dir = docs/build/html
The ``upload_docs`` command has the following options:
``--upload-dir``
The directory to be uploaded to the repository.
``--show-response``
Display the full response text from server; this is useful for debugging
PyPI problems.
``--repository=URL, -r URL``
The URL of the repository to upload to. Defaults to
https://pypi.python.org/pypi (i.e., the main PyPI installation).
--------------------------------
Extending and Reusing Distribute
--------------------------------
Creating ``distutils`` Extensions
=================================
It can be hard to add new commands or setup arguments to the distutils. But
the ``setuptools`` package makes it a bit easier, by allowing you to distribute
a distutils extension as a separate project, and then have projects that need
the extension just refer to it in their ``setup_requires`` argument.
With ``setuptools``, your distutils extension projects can hook in new
commands and ``setup()`` arguments just by defining "entry points". These
are mappings from command or argument names to a specification of where to
import a handler from. (See the section on `Dynamic Discovery of Services and
Plugins`_ above for some more background on entry points.)
Adding Commands
---------------
You can add new ``setup`` commands by defining entry points in the
``distutils.commands`` group. For example, if you wanted to add a ``foo``
command, you might add something like this to your distutils extension
project's setup script::
setup(
# ...
entry_points = {
"distutils.commands": [
"foo = mypackage.some_module:foo",
],
},
)
(Assuming, of course, that the ``foo`` class in ``mypackage.some_module`` is
a ``setuptools.Command`` subclass.)
Once a project containing such entry points has been activated on ``sys.path``,
(e.g. by running "install" or "develop" with a site-packages installation
directory) the command(s) will be available to any ``setuptools``-based setup
scripts. It is not necessary to use the ``--command-packages`` option or
to monkeypatch the ``distutils.command`` package to install your commands;
``setuptools`` automatically adds a wrapper to the distutils to search for
entry points in the active distributions on ``sys.path``. In fact, this is
how setuptools' own commands are installed: the setuptools project's setup
script defines entry points for them!
Adding ``setup()`` Arguments
----------------------------
Sometimes, your commands may need additional arguments to the ``setup()``
call. You can enable this by defining entry points in the
``distutils.setup_keywords`` group. For example, if you wanted a ``setup()``
argument called ``bar_baz``, you might add something like this to your
distutils extension project's setup script::
setup(
# ...
entry_points = {
"distutils.commands": [
"foo = mypackage.some_module:foo",
],
"distutils.setup_keywords": [
"bar_baz = mypackage.some_module:validate_bar_baz",
],
},
)
The idea here is that the entry point defines a function that will be called
to validate the ``setup()`` argument, if it's supplied. The ``Distribution``
object will have the initial value of the attribute set to ``None``, and the
validation function will only be called if the ``setup()`` call sets it to
a non-None value. Here's an example validation function::
def assert_bool(dist, attr, value):
"""Verify that value is True, False, 0, or 1"""
if bool(value) != value:
raise DistutilsSetupError(
"%r must be a boolean value (got %r)" % (attr,value)
)
Your function should accept three arguments: the ``Distribution`` object,
the attribute name, and the attribute value. It should raise a
``DistutilsSetupError`` (from the ``distutils.errors`` module) if the argument
is invalid. Remember, your function will only be called with non-None values,
and the default value of arguments defined this way is always None. So, your
commands should always be prepared for the possibility that the attribute will
be ``None`` when they access it later.
If more than one active distribution defines an entry point for the same
``setup()`` argument, *all* of them will be called. This allows multiple
distutils extensions to define a common argument, as long as they agree on
what values of that argument are valid.
Also note that as with commands, it is not necessary to subclass or monkeypatch
the distutils ``Distribution`` class in order to add your arguments; it is
sufficient to define the entry points in your extension, as long as any setup
script using your extension lists your project in its ``setup_requires``
argument.
Adding new EGG-INFO Files
-------------------------
Some extensible applications or frameworks may want to allow third parties to
develop plugins with application or framework-specific metadata included in
the plugins' EGG-INFO directory, for easy access via the ``pkg_resources``
metadata API. The easiest way to allow this is to create a distutils extension
to be used from the plugin projects' setup scripts (via ``setup_requires``)
that defines a new setup keyword, and then uses that data to write an EGG-INFO
file when the ``egg_info`` command is run.
The ``egg_info`` command looks for extension points in an ``egg_info.writers``
group, and calls them to write the files. Here's a simple example of a
distutils extension defining a setup argument ``foo_bar``, which is a list of
lines that will be written to ``foo_bar.txt`` in the EGG-INFO directory of any
project that uses the argument::
setup(
# ...
entry_points = {
"distutils.setup_keywords": [
"foo_bar = setuptools.dist:assert_string_list",
],
"egg_info.writers": [
"foo_bar.txt = setuptools.command.egg_info:write_arg",
],
},
)
This simple example makes use of two utility functions defined by setuptools
for its own use: a routine to validate that a setup keyword is a sequence of
strings, and another one that looks up a setup argument and writes it to
a file. Here's what the writer utility looks like::
def write_arg(cmd, basename, filename):
argname = os.path.splitext(basename)[0]
value = getattr(cmd.distribution, argname, None)
if value is not None:
value = '\n'.join(value)+'\n'
cmd.write_or_delete_file(argname, filename, value)
As you can see, ``egg_info.writers`` entry points must be a function taking
three arguments: a ``egg_info`` command instance, the basename of the file to
write (e.g. ``foo_bar.txt``), and the actual full filename that should be
written to.
In general, writer functions should honor the command object's ``dry_run``
setting when writing files, and use the ``distutils.log`` object to do any
console output. The easiest way to conform to this requirement is to use
the ``cmd`` object's ``write_file()``, ``delete_file()``, and
``write_or_delete_file()`` methods exclusively for your file operations. See
those methods' docstrings for more details.
Adding Support for Other Revision Control Systems
-------------------------------------------------
If you would like to create a plugin for ``setuptools`` to find files in other
source control systems besides CVS and Subversion, you can do so by adding an
entry point to the ``setuptools.file_finders`` group. The entry point should
be a function accepting a single directory name, and should yield
all the filenames within that directory (and any subdirectories thereof) that
are under revision control.
For example, if you were going to create a plugin for a revision control system
called "foobar", you would write a function something like this:
.. code-block:: python
def find_files_for_foobar(dirname):
# loop to yield paths that start with `dirname`
And you would register it in a setup script using something like this::
entry_points = {
"setuptools.file_finders": [
"foobar = my_foobar_module:find_files_for_foobar"
]
}
Then, anyone who wants to use your plugin can simply install it, and their
local setuptools installation will be able to find the necessary files.
It is not necessary to distribute source control plugins with projects that
simply use the other source control system, or to specify the plugins in
``setup_requires``. When you create a source distribution with the ``sdist``
command, setuptools automatically records what files were found in the
``SOURCES.txt`` file. That way, recipients of source distributions don't need
to have revision control at all. However, if someone is working on a package
by checking out with that system, they will need the same plugin(s) that the
original author is using.
A few important points for writing revision control file finders:
* Your finder function MUST return relative paths, created by appending to the
passed-in directory name. Absolute paths are NOT allowed, nor are relative
paths that reference a parent directory of the passed-in directory.
* Your finder function MUST accept an empty string as the directory name,
meaning the current directory. You MUST NOT convert this to a dot; just
yield relative paths. So, yielding a subdirectory named ``some/dir`` under
the current directory should NOT be rendered as ``./some/dir`` or
``/somewhere/some/dir``, but *always* as simply ``some/dir``
* Your finder function SHOULD NOT raise any errors, and SHOULD deal gracefully
with the absence of needed programs (i.e., ones belonging to the revision
control system itself. It *may*, however, use ``distutils.log.warn()`` to
inform the user of the missing program(s).
Subclassing ``Command``
-----------------------
Sorry, this section isn't written yet, and neither is a lot of what's below
this point, except for the change log. You might want to `subscribe to changes
in this page <setuptools?action=subscribe>`_ to see when new documentation is
added or updated.
XXX
Reusing ``setuptools`` Code
===========================
``distribute_setup``
--------------------
XXX
``setuptools.archive_util``
---------------------------
XXX
``setuptools.sandbox``
----------------------
XXX
``setuptools.package_index``
----------------------------
XXX
History
=======
0.6c9
* Fixed a missing files problem when using Windows source distributions on
non-Windows platforms, due to distutils not handling manifest file line
endings correctly.
* Updated Pyrex support to work with Pyrex 0.9.6 and higher.
* Minor changes for Jython compatibility, including skipping tests that can't
work on Jython.
* Fixed not installing eggs in ``install_requires`` if they were also used for
``setup_requires`` or ``tests_require``.
* Fixed not fetching eggs in ``install_requires`` when running tests.
* Allow ``ez_setup.use_setuptools()`` to upgrade existing setuptools
installations when called from a standalone ``setup.py``.
* Added a warning if a namespace package is declared, but its parent package
is not also declared as a namespace.
* Support Subversion 1.5
* Removed use of deprecated ``md5`` module if ``hashlib`` is available
* Fixed ``bdist_wininst upload`` trying to upload the ``.exe`` twice
* Fixed ``bdist_egg`` putting a ``native_libs.txt`` in the source package's
``.egg-info``, when it should only be in the built egg's ``EGG-INFO``.
* Ensure that _full_name is set on all shared libs before extensions are
checked for shared lib usage. (Fixes a bug in the experimental shared
library build support.)
* Fix to allow unpacked eggs containing native libraries to fail more
gracefully under Google App Engine (with an ``ImportError`` loading the
C-based module, instead of getting a ``NameError``).
0.6c7
* Fixed ``distutils.filelist.findall()`` crashing on broken symlinks, and
``egg_info`` command failing on new, uncommitted SVN directories.
* Fix import problems with nested namespace packages installed via
``--root`` or ``--single-version-externally-managed``, due to the
parent package not having the child package as an attribute.
0.6c6
* Added ``--egg-path`` option to ``develop`` command, allowing you to force
``.egg-link`` files to use relative paths (allowing them to be shared across
platforms on a networked drive).
* Fix not building binary RPMs correctly.
* Fix "eggsecutables" (such as setuptools' own egg) only being runnable with
bash-compatible shells.
* Fix ``#!`` parsing problems in Windows ``.exe`` script wrappers, when there
was whitespace inside a quoted argument or at the end of the ``#!`` line
(a regression introduced in 0.6c4).
* Fix ``test`` command possibly failing if an older version of the project
being tested was installed on ``sys.path`` ahead of the test source
directory.
* Fix ``find_packages()`` treating ``ez_setup`` and directories with ``.`` in
their names as packages.
0.6c5
* Fix uploaded ``bdist_rpm`` packages being described as ``bdist_egg``
packages under Python versions less than 2.5.
* Fix uploaded ``bdist_wininst`` packages being described as suitable for
"any" version by Python 2.5, even if a ``--target-version`` was specified.
0.6c4
* Overhauled Windows script wrapping to support ``bdist_wininst`` better.
Scripts installed with ``bdist_wininst`` will always use ``#!python.exe`` or
``#!pythonw.exe`` as the executable name (even when built on non-Windows
platforms!), and the wrappers will look for the executable in the script's
parent directory (which should find the right version of Python).
* Fix ``upload`` command not uploading files built by ``bdist_rpm`` or
``bdist_wininst`` under Python 2.3 and 2.4.
* Add support for "eggsecutable" headers: a ``#!/bin/sh`` script that is
prepended to an ``.egg`` file to allow it to be run as a script on Unix-ish
platforms. (This is mainly so that setuptools itself can have a single-file
installer on Unix, without doing multiple downloads, dealing with firewalls,
etc.)
* Fix problem with empty revision numbers in Subversion 1.4 ``entries`` files
* Use cross-platform relative paths in ``easy-install.pth`` when doing
``develop`` and the source directory is a subdirectory of the installation
target directory.
* Fix a problem installing eggs with a system packaging tool if the project
contained an implicit namespace package; for example if the ``setup()``
listed a namespace package ``foo.bar`` without explicitly listing ``foo``
as a namespace package.
0.6c3
* Fixed breakages caused by Subversion 1.4's new "working copy" format
0.6c2
* The ``ez_setup`` module displays the conflicting version of setuptools (and
its installation location) when a script requests a version that's not
available.
* Running ``setup.py develop`` on a setuptools-using project will now install
setuptools if needed, instead of only downloading the egg.
0.6c1
* Fixed ``AttributeError`` when trying to download a ``setup_requires``
dependency when a distribution lacks a ``dependency_links`` setting.
* Made ``zip-safe`` and ``not-zip-safe`` flag files contain a single byte, so
as to play better with packaging tools that complain about zero-length
files.
* Made ``setup.py develop`` respect the ``--no-deps`` option, which it
previously was ignoring.
* Support ``extra_path`` option to ``setup()`` when ``install`` is run in
backward-compatibility mode.
* Source distributions now always include a ``setup.cfg`` file that explicitly
sets ``egg_info`` options such that they produce an identical version number
to the source distribution's version number. (Previously, the default
version number could be different due to the use of ``--tag-date``, or if
the version was overridden on the command line that built the source
distribution.)
0.6b4
* Fix ``register`` not obeying name/version set by ``egg_info`` command, if
``egg_info`` wasn't explicitly run first on the same command line.
* Added ``--no-date`` and ``--no-svn-revision`` options to ``egg_info``
command, to allow suppressing tags configured in ``setup.cfg``.
* Fixed redundant warnings about missing ``README`` file(s); it should now
appear only if you are actually a source distribution.
0.6b3
* Fix ``bdist_egg`` not including files in subdirectories of ``.egg-info``.
* Allow ``.py`` files found by the ``include_package_data`` option to be
automatically included. Remove duplicate data file matches if both
``include_package_data`` and ``package_data`` are used to refer to the same
files.
0.6b1
* Strip ``module`` from the end of compiled extension modules when computing
the name of a ``.py`` loader/wrapper. (Python's import machinery ignores
this suffix when searching for an extension module.)
0.6a11
* Added ``test_loader`` keyword to support custom test loaders
* Added ``setuptools.file_finders`` entry point group to allow implementing
revision control plugins.
* Added ``--identity`` option to ``upload`` command.
* Added ``dependency_links`` to allow specifying URLs for ``--find-links``.
* Enhanced test loader to scan packages as well as modules, and call
``additional_tests()`` if present to get non-unittest tests.
* Support namespace packages in conjunction with system packagers, by omitting
the installation of any ``__init__.py`` files for namespace packages, and
adding a special ``.pth`` file to create a working package in
``sys.modules``.
* Made ``--single-version-externally-managed`` automatic when ``--root`` is
used, so that most system packagers won't require special support for
setuptools.
* Fixed ``setup_requires``, ``tests_require``, etc. not using ``setup.cfg`` or
other configuration files for their option defaults when installing, and
also made the install use ``--multi-version`` mode so that the project
directory doesn't need to support .pth files.
* ``MANIFEST.in`` is now forcibly closed when any errors occur while reading
it. Previously, the file could be left open and the actual error would be
masked by problems trying to remove the open file on Windows systems.
0.6a10
* Fixed the ``develop`` command ignoring ``--find-links``.
0.6a9
* The ``sdist`` command no longer uses the traditional ``MANIFEST`` file to
create source distributions. ``MANIFEST.in`` is still read and processed,
as are the standard defaults and pruning. But the manifest is built inside
the project's ``.egg-info`` directory as ``SOURCES.txt``, and it is rebuilt
every time the ``egg_info`` command is run.
* Added the ``include_package_data`` keyword to ``setup()``, allowing you to
automatically include any package data listed in revision control or
``MANIFEST.in``
* Added the ``exclude_package_data`` keyword to ``setup()``, allowing you to
trim back files included via the ``package_data`` and
``include_package_data`` options.
* Fixed ``--tag-svn-revision`` not working when run from a source
distribution.
* Added warning for namespace packages with missing ``declare_namespace()``
* Added ``tests_require`` keyword to ``setup()``, so that e.g. packages
requiring ``nose`` to run unit tests can make this dependency optional
unless the ``test`` command is run.
* Made all commands that use ``easy_install`` respect its configuration
options, as this was causing some problems with ``setup.py install``.
* Added an ``unpack_directory()`` driver to ``setuptools.archive_util``, so
that you can process a directory tree through a processing filter as if it
were a zipfile or tarfile.
* Added an internal ``install_egg_info`` command to use as part of old-style
``install`` operations, that installs an ``.egg-info`` directory with the
package.
* Added a ``--single-version-externally-managed`` option to the ``install``
command so that you can more easily wrap a "flat" egg in a system package.
* Enhanced ``bdist_rpm`` so that it installs single-version eggs that
don't rely on a ``.pth`` file. The ``--no-egg`` option has been removed,
since all RPMs are now built in a more backwards-compatible format.
* Support full roundtrip translation of eggs to and from ``bdist_wininst``
format. Running ``bdist_wininst`` on a setuptools-based package wraps the
egg in an .exe that will safely install it as an egg (i.e., with metadata
and entry-point wrapper scripts), and ``easy_install`` can turn the .exe
back into an ``.egg`` file or directory and install it as such.
0.6a8
* Fixed some problems building extensions when Pyrex was installed, especially
with Python 2.4 and/or packages using SWIG.
* Made ``develop`` command accept all the same options as ``easy_install``,
and use the ``easy_install`` command's configuration settings as defaults.
* Made ``egg_info --tag-svn-revision`` fall back to extracting the revision
number from ``PKG-INFO`` in case it is being run on a source distribution of
a snapshot taken from a Subversion-based project.
* Automatically detect ``.dll``, ``.so`` and ``.dylib`` files that are being
installed as data, adding them to ``native_libs.txt`` automatically.
* Fixed some problems with fresh checkouts of projects that don't include
``.egg-info/PKG-INFO`` under revision control and put the project's source
code directly in the project directory. If such a package had any
requirements that get processed before the ``egg_info`` command can be run,
the setup scripts would fail with a "Missing 'Version:' header and/or
PKG-INFO file" error, because the egg runtime interpreted the unbuilt
metadata in a directory on ``sys.path`` (i.e. the current directory) as
being a corrupted egg. Setuptools now monkeypatches the distribution
metadata cache to pretend that the egg has valid version information, until
it has a chance to make it actually be so (via the ``egg_info`` command).
0.6a5
* Fixed missing gui/cli .exe files in distribution. Fixed bugs in tests.
0.6a3
* Added ``gui_scripts`` entry point group to allow installing GUI scripts
on Windows and other platforms. (The special handling is only for Windows;
other platforms are treated the same as for ``console_scripts``.)
0.6a2
* Added ``console_scripts`` entry point group to allow installing scripts
without the need to create separate script files. On Windows, console
scripts get an ``.exe`` wrapper so you can just type their name. On other
platforms, the scripts are written without a file extension.
0.6a1
* Added support for building "old-style" RPMs that don't install an egg for
the target package, using a ``--no-egg`` option.
* The ``build_ext`` command now works better when using the ``--inplace``
option and multiple Python versions. It now makes sure that all extensions
match the current Python version, even if newer copies were built for a
different Python version.
* The ``upload`` command no longer attaches an extra ``.zip`` when uploading
eggs, as PyPI now supports egg uploads without trickery.
* The ``ez_setup`` script/module now displays a warning before downloading
the setuptools egg, and attempts to check the downloaded egg against an
internal MD5 checksum table.
* Fixed the ``--tag-svn-revision`` option of ``egg_info`` not finding the
latest revision number; it was using the revision number of the directory
containing ``setup.py``, not the highest revision number in the project.
* Added ``eager_resources`` setup argument
* The ``sdist`` command now recognizes Subversion "deleted file" entries and
does not include them in source distributions.
* ``setuptools`` now embeds itself more thoroughly into the distutils, so that
other distutils extensions (e.g. py2exe, py2app) will subclass setuptools'
versions of things, rather than the native distutils ones.
* Added ``entry_points`` and ``setup_requires`` arguments to ``setup()``;
``setup_requires`` allows you to automatically find and download packages
that are needed in order to *build* your project (as opposed to running it).
* ``setuptools`` now finds its commands, ``setup()`` argument validators, and
metadata writers using entry points, so that they can be extended by
third-party packages. See `Creating distutils Extensions`_ above for more
details.
* The vestigial ``depends`` command has been removed. It was never finished
or documented, and never would have worked without EasyInstall - which it
pre-dated and was never compatible with.
0.5a12
* The zip-safety scanner now checks for modules that might be used with
``python -m``, and marks them as unsafe for zipping, since Python 2.4 can't
handle ``-m`` on zipped modules.
0.5a11
* Fix breakage of the "develop" command that was caused by the addition of
``--always-unzip`` to the ``easy_install`` command.
0.5a9
* Include ``svn:externals`` directories in source distributions as well as
normal subversion-controlled files and directories.
* Added ``exclude=patternlist`` option to ``setuptools.find_packages()``
* Changed --tag-svn-revision to include an "r" in front of the revision number
for better readability.
* Added ability to build eggs without including source files (except for any
scripts, of course), using the ``--exclude-source-files`` option to
``bdist_egg``.
* ``setup.py install`` now automatically detects when an "unmanaged" package
or module is going to be on ``sys.path`` ahead of a package being installed,
thereby preventing the newer version from being imported. If this occurs,
a warning message is output to ``sys.stderr``, but installation proceeds
anyway. The warning message informs the user what files or directories
need deleting, and advises them they can also use EasyInstall (with the
``--delete-conflicting`` option) to do it automatically.
* The ``egg_info`` command now adds a ``top_level.txt`` file to the metadata
directory that lists all top-level modules and packages in the distribution.
This is used by the ``easy_install`` command to find possibly-conflicting
"unmanaged" packages when installing the distribution.
* Added ``zip_safe`` and ``namespace_packages`` arguments to ``setup()``.
Added package analysis to determine zip-safety if the ``zip_safe`` flag
is not given, and advise the author regarding what code might need changing.
* Fixed the swapped ``-d`` and ``-b`` options of ``bdist_egg``.
0.5a8
* The "egg_info" command now always sets the distribution metadata to "safe"
forms of the distribution name and version, so that distribution files will
be generated with parseable names (i.e., ones that don't include '-' in the
name or version). Also, this means that if you use the various ``--tag``
options of "egg_info", any distributions generated will use the tags in the
version, not just egg distributions.
* Added support for defining command aliases in distutils configuration files,
under the "[aliases]" section. To prevent recursion and to allow aliases to
call the command of the same name, a given alias can be expanded only once
per command-line invocation. You can define new aliases with the "alias"
command, either for the local, global, or per-user configuration.
* Added "rotate" command to delete old distribution files, given a set of
patterns to match and the number of files to keep. (Keeps the most
recently-modified distribution files matching each pattern.)
* Added "saveopts" command that saves all command-line options for the current
invocation to the local, global, or per-user configuration file. Useful for
setting defaults without having to hand-edit a configuration file.
* Added a "setopt" command that sets a single option in a specified distutils
configuration file.
0.5a7
* Added "upload" support for egg and source distributions, including a bug
fix for "upload" and a temporary workaround for lack of .egg support in
PyPI.
0.5a6
* Beefed up the "sdist" command so that if you don't have a MANIFEST.in, it
will include all files under revision control (CVS or Subversion) in the
current directory, and it will regenerate the list every time you create a
source distribution, not just when you tell it to. This should make the
default "do what you mean" more often than the distutils' default behavior
did, while still retaining the old behavior in the presence of MANIFEST.in.
* Fixed the "develop" command always updating .pth files, even if you
specified ``-n`` or ``--dry-run``.
* Slightly changed the format of the generated version when you use
``--tag-build`` on the "egg_info" command, so that you can make tagged
revisions compare *lower* than the version specified in setup.py (e.g. by
using ``--tag-build=dev``).
0.5a5
* Added ``develop`` command to ``setuptools``-based packages. This command
installs an ``.egg-link`` pointing to the package's source directory, and
script wrappers that ``execfile()`` the source versions of the package's
scripts. This lets you put your development checkout(s) on sys.path without
having to actually install them. (To uninstall the link, use
use ``setup.py develop --uninstall``.)
* Added ``egg_info`` command to ``setuptools``-based packages. This command
just creates or updates the "projectname.egg-info" directory, without
building an egg. (It's used by the ``bdist_egg``, ``test``, and ``develop``
commands.)
* Enhanced the ``test`` command so that it doesn't install the package, but
instead builds any C extensions in-place, updates the ``.egg-info``
metadata, adds the source directory to ``sys.path``, and runs the tests
directly on the source. This avoids an "unmanaged" installation of the
package to ``site-packages`` or elsewhere.
* Made ``easy_install`` a standard ``setuptools`` command, moving it from
the ``easy_install`` module to ``setuptools.command.easy_install``. Note
that if you were importing or extending it, you must now change your imports
accordingly. ``easy_install.py`` is still installed as a script, but not as
a module.
0.5a4
* Setup scripts using setuptools can now list their dependencies directly in
the setup.py file, without having to manually create a ``depends.txt`` file.
The ``install_requires`` and ``extras_require`` arguments to ``setup()``
are used to create a dependencies file automatically. If you are manually
creating ``depends.txt`` right now, please switch to using these setup
arguments as soon as practical, because ``depends.txt`` support will be
removed in the 0.6 release cycle. For documentation on the new arguments,
see the ``setuptools.dist.Distribution`` class.
* Setup scripts using setuptools now always install using ``easy_install``
internally, for ease of uninstallation and upgrading.
0.5a1
* Added support for "self-installation" bootstrapping. Packages can now
include ``ez_setup.py`` in their source distribution, and add the following
to their ``setup.py``, in order to automatically bootstrap installation of
setuptools as part of their setup process::
from ez_setup import use_setuptools
use_setuptools()
from setuptools import setup
# etc...
0.4a2
* Added ``ez_setup.py`` installer/bootstrap script to make initial setuptools
installation easier, and to allow distributions using setuptools to avoid
having to include setuptools in their source distribution.
* All downloads are now managed by the ``PackageIndex`` class (which is now
subclassable and replaceable), so that embedders can more easily override
download logic, give download progress reports, etc. The class has also
been moved to the new ``setuptools.package_index`` module.
* The ``Installer`` class no longer handles downloading, manages a temporary
directory, or tracks the ``zip_ok`` option. Downloading is now handled
by ``PackageIndex``, and ``Installer`` has become an ``easy_install``
command class based on ``setuptools.Command``.
* There is a new ``setuptools.sandbox.run_setup()`` API to invoke a setup
script in a directory sandbox, and a new ``setuptools.archive_util`` module
with an ``unpack_archive()`` API. These were split out of EasyInstall to
allow reuse by other tools and applications.
* ``setuptools.Command`` now supports reinitializing commands using keyword
arguments to set/reset options. Also, ``Command`` subclasses can now set
their ``command_consumes_arguments`` attribute to ``True`` in order to
receive an ``args`` option containing the rest of the command line.
0.3a2
* Added new options to ``bdist_egg`` to allow tagging the egg's version number
with a subversion revision number, the current date, or an explicit tag
value. Run ``setup.py bdist_egg --help`` to get more information.
* Misc. bug fixes
0.3a1
* Initial release.
Mailing List and Bug Tracker
============================
Please use the `distutils-sig mailing list`_ for questions and discussion about
setuptools, and the `setuptools bug tracker`_ ONLY for issues you have
confirmed via the list are actual bugs, and which you have reduced to a minimal
set of steps to reproduce.
.. _distutils-sig mailing list: http://mail.python.org/pipermail/distutils-sig/
.. _setuptools bug tracker: http://bugs.python.org/setuptools/
Mini Shell