I'd like pip to install a dependency that I have on GitHub when the user issues the command to install the original software, also from source on GitHub. Neither of these packages are on PyPi (and never will be).
The user issues the command:
pip -e git+https://github.com/Lewisham/cvsanaly@develop#egg=cvsanaly
This repo has a requirements.txt
file, with another dependency on GitHub:
-e git+https://github.com/Lewisham/repositoryhandler#egg=repositoryhandler
What I'd like is a single command that a user can issue开发者_开发知识库 to install the original package, have pip find the requirements file, then install the dependency too.
This answer helped me solve the same problem you're talking about.
There doesn't seem to be an easy way for setup.py to use the requirements file directly to define its dependencies, but the same information can be put into the setup.py itself.
I have this requirements.txt:
PIL
-e git://github.com/gabrielgrant/django-ckeditor.git#egg=django-ckeditor
But when installing that requirements.txt's containing package, the requirements are ignored by pip.
This setup.py seems to coerce pip into installing the dependencies (including my github version of django-ckeditor):
from setuptools import setup
setup(
name='django-articles',
...,
install_requires=[
'PIL',
'django-ckeditor>=0.9.3',
],
dependency_links = [
'http://github.com/gabrielgrant/django-ckeditor/tarball/master#egg=django-ckeditor-0.9.3',
]
)
Edit:
This answer also contains some useful information.
Specifying the version as part of the "#egg=..." is required to identify which version of the package is available at the link. Note, however, that if you always want to depend on your latest version, you can set the version to dev
in install_requires, dependency_links and the other package's setup.py
Edit: using dev
as the version isn't a good idea, as per comments below.
Here's a small script I used to generate install_requires
and dependency_links
from a requirements file.
import os
import re
def which(program):
"""
Detect whether or not a program is installed.
Thanks to http://stackoverflow.com/a/377028/70191
"""
def is_exe(fpath):
return os.path.exists(fpath) and os.access(fpath, os.X_OK)
fpath, _ = os.path.split(program)
if fpath:
if is_exe(program):
return program
else:
for path in os.environ['PATH'].split(os.pathsep):
exe_file = os.path.join(path, program)
if is_exe(exe_file):
return exe_file
return None
EDITABLE_REQUIREMENT = re.compile(r'^-e (?P<link>(?P<vcs>git|svn|hg|bzr).+#egg=(?P<package>.+)-(?P<version>\d(?:\.\d)*))$')
install_requires = []
dependency_links = []
for requirement in (l.strip() for l in open('requirements')):
match = EDITABLE_REQUIREMENT.match(requirement)
if match:
assert which(match.group('vcs')) is not None, \
"VCS '%(vcs)s' must be installed in order to install %(link)s" % match.groupdict()
install_requires.append("%(package)s==%(version)s" % match.groupdict())
dependency_links.append(match.group('link'))
else:
install_requires.append(requirement)
does this answer your question?
setup(name='application-xpto',
version='1.0',
author='me,me,me',
author_email='xpto@mail.com',
packages=find_packages(),
include_package_data=True,
description='web app',
install_requires=open('app/requirements.txt').readlines(),
)
精彩评论