workspacemanager 0.2.10

This tool can manage a workspace by providing some useful functions (generation of the setup file and others, generate a virtual env according to the project name, install your own dependencies, generate dist of all dependencies...). See the readme for more informations.

In order to use this tool, we assume:

• You are familiar with pip / setuptools and python virtual envs

• You want to organize both dependencies between your own projects and with external libs using python virtual envs

Requirements

This lib works on a workspace which is a main folder within recursively projects and folders (which are sub-workspaces to gather projects). Here an example :

└── MyPythonWorkspace
    ├── Project1
    │   └── project1
    │        └── __init__.py
    ├── Basics
    │   ├── Project2
    │   │    └── project2
    │   │        └── __init__.py
    │   └── Project3
    │        └── project3
    │            └── __init__.py
    └── wm-conf.json

You need to create a blank wm-conf.json in the root folder to indicate which one is the main workspace.

To manage an eclipse workspace this way, you need to set MyPythonWorkspace as the workspace and add all projects as external projects, then manage them with eclipse working sets to have the same tree organization.

You also have to install Pew and pipsi in order to use wm-pew:

sudo pip install pipsi
sudo pipsi install pew

Installation

sudo pip install workspacemanager

Then cd to your workspace and execute touch wm-conf.json.

Generate the setup file

Usage (the default path is the current):

wm-setup [-a /path/to/the/project]

1. This function will create some files (LICENCE.txt, setup.py, requirements.txt…). Each file can be edited in the /path/to/the/lib/setup-templates. For instance, you can paste a different licence in the templates folder(default is the MIT licence)

2. Each file will be edited according to the username, email, date given when executing the command line

3. If there are no __init__.py files or an empty one, a version is added

You can add lines in the MANIFEST.in, choose topics in setup.py, add requirements…

Then you can subscribe to PyPi and upload your project using python setup.py sdist upload.

Or install your project on the current activated venv using python setup.py install

You can create wm-conf.json in your workspace (or edit it if it already exists) to set default values:

{
    "author_email": "[e-mail]",
    "author": "yourusername"
}

On Linux you need to install pandoc and pypandoc to handle markdown to reStructuredText convertion:

sudo apt-get install pandoc
pip install pypandoc

Generate a venv linked to the project using Pew

wm-pew [-a /path/to/the/project] [-p /path/to/python]

This command line is equivalent to pew new -a /path/to/the/project projectname-venv

If a python bin is given (e.g. /usr/bin/python3.6), the command is equivalent to pew new -p /path/to/python -a /path/to/the/project projectname-venv

To use the venv in eclipse PyDev, right click on your project, properties, interpreter, configure, new, add the venv path (e.g. /home/username/.virtualenvs/projectname-venv/bin/python2.7), ok, unselect all but the venv libs, ok, select the new venv, ok.

If this function doesn’t work, add the pew path to /bin using sudo ln -s ~/.local/bin/pew /bin

If you want python 3 as default python, you can add alias wm-pew="wm-pew -p /usr/bin/python3.5" in your ~/.bash_aliases.

Install internal workspace dependencies on the project venv

Each project has one or more packages with __init__.py files.

Create local-dependencies.txt at the root of the current project:

cd /path/to/the/project
touch local-dependencies.txt

Then add local dependencies, i.e. projects (in your workspace) that your project depends on:

echo "Utils" >> local-dependencies.txt
echo "MachineLearningTools" >> local-dependencies.txt

The current project and all projects in local-dependencies.txt must have setups files which work. And the current project must have an associated venv.

This command will install all dependencies (recursively) in the current project venv, so you can work on all project (e.g. through eclipse), install all update but work on a totally independent venv:

wm-deps [-a /path/to/the/project] [-r filename.txt]

If a local project (your own) is also on PyPi, you can write the PyPi project name after the local project name following by a “/” in local-dependencies.txt :

echo "Utils/hjutils" >> local-dependencies.txt

So, in this example, the hjutils project from PyPi you uploaded will not be installed using wm-req. Instead, you will need wm-deps to install the local project.

If you want to set a custom project name to upload on PyPi, for example hjutils instead of utils (because the project utils probably still exists), just edit the name param in the setup file.

Create a dist of your project and all dependencies

wm-dist

This function will package the project and all internal dependencies in wm-dist folder.

You can then edit the conf.json file and use rsync-all.sh to install all on a remote server.

To use this script, you will need jq to read json conf from bash script:

sudo apt-get install jq

You can add an authorized key in the host to avoid asking a password.

You can write your own script in the wm-dist folder to run the project on the remote server. This script won’t be erased by re-launching wm-dist.

You can set erase_wm-dist_templates to true in the wm-conf.json file if you want the dist templates to be erased at each re-use.

If pew is not found on the remote server, add the pew path to /bin using sudo ln -s ~/.local/bin/pew /bin

You can set multiple addresses in dist/conf.json by separating all addresses with spaces, e.g. "localhost 10.10.10.200 test.com"

Others

Use wm-workon to disp the pew workon command line according to the current project.

Use wm-freeze to see which lib is installed in the linked venv.

Use wm-req to install all requirements of requirements.txt in the linked venv.