Installing pysto with pip

1. Simply run from your local environment or whole system

   pip install pysto

Installing pysto from the cloned github repository

1. Clone the pysto repository

   git clone https://github.com/rcasero/pysto.git

2. Activate the local environment of the other project, or you can also work without a local environment. For example, to activate a local conda environment called “myproject”

   source activate myproject

3. Install pysto from the pysto root directory (note that if you are installing it system-wide, you may need to run the command as root)

   cd pysto
   pip install .

Uninstalling pysto

1. Uninstall the package

   pip uninstall pysto

Install pysto project for development

1. Clone the pysto repository

   git clone https://github.com/rcasero/pysto.git

2. Run install_dependencies.sh to install development tools, create local environments for python 2.7 and 3.6, and install python dependencies. pysto depends on SimpleITK, and there are two options:

   1. Install the official SimpleITK package

      cd pysto
      ./install_dependencies.sh

   2. Build and install SimpleElastix, which is an extension of SimpleITK

      cd pysto
      ./install_dependencies.sh SimpleElastix

Developing source code for pysto

1. Activate one of the pysto local environments

   source activate pysto_3.6

2. If you are making changes to the code, you want your python environment to import the code you are working with in ~/Software/pysto, not the package installed in your local conda environment. Thus, add the project’s source directory to PYTHONPATH

   export PYTHONPATH=~/Software/pysto:$PYTHONPATH

3. Launch the development IDE, e.g.

   spyder&

4. In your code, import the pysto modules/functions in the usual way, e.g.

   import pysto.imgproc as pymg
   [...]
   imf = pymg.imfuse(im1, im2)

5. While developing, you can run all tests (both for python 2.7 and 3.6) from the command line with

   make test

6. You need to have a local file ~/.pypirc (replace <the password> by the password). This will be used by twine to release packages to PyPI

   [distutils]
   index-servers =
     pypi
     pypitest
   
   [pypi]
   username=rcasero
   password=<the password>
   
   [pypitest]
   repository = https://test.pypi.org/legacy/
   username=rcasero
   password=<the password>

7. Protect the file so that it can be read only by you

   chmod 600 ~/.pypirc

Uninstalling pysto

1. Uninstall the package

   pip uninstall pysto

Releasing a new version of pysto to PyPI

We provide a Makefile to simplify testing and releasing.

1. Run tests to make sure nothing obvious got broken

   make test

2. Commit and push all the code that should go in the release to github.

3. Update setup.py with release version, any new dependencies, the new download URL, changes to the description…

   from setuptools import setup, find_packages
   
   setup(
       name='pysto',
       version='1.0.0',
       download_url = 'https://github.com/rcasero/pysto/archive/1.0.0.tar.gz',
       packages=find_packages(),
       python_requires='>=3.6',
       install_requires=['matplotlib>=2.0','numpy>=1.13','opencv-python>=3.3.0'],
       description='Miscellanea image processing functions',
       url='https://github.com/rcasero/pysto',
       author='Ramón Casero',
       author_email='rcasero@gmail.com',
       license='GPL v3',
   )

4. Update ChangeLog.md with the main changes to this release. For example,

   ## v1.0.0
   ### Added
   
   - imgproc.matchHist(): "Modify image intensities to match the
     histogram of a reference image" by
     [rcasero](https://github.com/rcasero)
   - imgproc.imfuse(): "Composite of two images" by
     [rcasero](https://github.com/rcasero)
   - testdata/*.png: Stereo cloud images with ROI masks (left_mask.png,
     left.png, right_mask.png, right.png) by
     [rcasero](https://github.com/rcasero)

5. Tag the release in github, create the package/wheel and upload to the test PyPI server

   make test-package

6. You should be able to see your package in

   https://test.pypi.org/project/pysto/

7. If everything goes well, upload to PyPI Live

   make package

8. You should be able to see your package in

   https://pypi.org/project/pysto/