Skip to content

Latest commit

 

History

History
198 lines (135 loc) · 6.74 KB

README.rst

File metadata and controls

198 lines (135 loc) · 6.74 KB
Raw

Python Source Extractor

This is the Python Source Extractor (PySE) for extraction and measurement of sources in radio astronomical images, or, more generally, images from aperture synthesis. This means that PySE has been tested thoroughly on images with correlated noise. PySE was primarily designed for the LOFAR Transients Key Project (TKP), i.e. targeting the detection of transient sources, which will be mostly unresolved.

Recent development - 2020 - 2023 - has been funded by grant 27019G06 from the Netherlands eScience Center for PADRE.

Current development is internally funded by The Netherlands eScience Center through the Software Sustainability Project "PySE".

The code in this repo was formerly part of the LOFAR Transients Pipeline (TraP) repo. Compatibility with TraP is maintained, i.e. the output of PySE is suitable for futher processing in TraP. Also, PySE is still included and regularly updated in Trap. A reason for starting this repo is to offer PySE as a standalone application.

Documentation

Currently only offered as part of the TraP documentation.

Installation

PySE has an entry on PyPI:

$ pip install radio-pyse

PySE is compatible with Python 3.10 or up.

License

PySE was released under the BSD-2 license. This is still the case for the master branch.

However, some of the fast branches of PySE make use of SEP. Please have a look at the SEP license info. This means that the license for PySE as a whole, for some of the fast branches, is LGPLv3. Only sourcefinder/image.py makes use of SEP, so the other Python modules have a BSD-2 license.

We will be including C source code for least squares fitting into one or more of the fast branches. This fitting code was written decades ago as part of the Groningen Image Processing System (GIPSY). GIPSY has an ASCL record and a LGPL2+ license.

Authors

The list of authors, sorted by the number of commits:

  • Hanno Spreeuw
  • John Swinbank
  • Gijs Molenaar
  • Tim Staley
  • Evert Rol
  • John Sanders
  • Bart Scheers
  • Mark Kuiack

Developer information

PySE uses hatch to manage the different environments during development. So make sure you have hatch installed globally. You could either use your system's package manager to install hatch, or use pipx to install as a regular user. Please ensure that you are using a version hatch>=1.10, otherwise you might encounter this bug.

In general, to run command in a specific environment managed by hatch, you can do this:

$ hatch run <environment>:<command> [--options]

You can see a summary of all the evironments managed by hatch like this:

$ hatch env show
             Standalone
┏━━━━━━━━━┳━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Name    ┃ Type    ┃ Dependencies ┃
┡━━━━━━━━━╇━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ default │ virtual │              │
├─────────┼─────────┼──────────────┤
│ test    │ virtual │ pytest       │
│         │         │ pytest-cov   │
├─────────┼─────────┼──────────────┤
│ lint    │ virtual │ black        │
│         │         │ flake8       │
│         │         │ mypy         │
│         │         │ ruff         │
└─────────┴─────────┴──────────────┘

Some common tasks using hatch are summarised below.

Package builds

hatch does package builds in an isolated environment. The package build setup also uses a dynamic hook (also known as build hooks) to generate the package version from Git repository release tags. So to do a local package build, you need to ensure all Git tags are present.

  1. Fetch all Git release tags locally.

    $ git fetch --tags

  2. You can now build a distribution (a wheel file and a source tarball) locally using:

    $ hatch build

    This creates the distribution files in the dist/ directory in the project root.

    $ tree dist/
dist/
├── radio_pyse-0.3.2.dev9+gfb04dc7.d20240729-py3-none-any.whl
└── radio_pyse-0.3.2.dev9+gfb04dc7.d20240729.tar.gz

  3. If you want to trigger only the build hooks (like generating the package version), you can do:

    $ hatch build --hooks-only

    This is necessary to refresh the version information if you update any of the build configuration in pyproject.toml, or if you are implementing something that depends on the version, e.g. making a new capability available only for a newer version.

Running the test suite

$ hatch run test:pytest [tests/test_iwanttorun.py] [-k match_string] [--options]
$ hatch run test:pytest --no-cov  # to disable coverage

Running formatters and static analysis tools

You can run supported linters/formatters (see the environment definition for lint) like this.

$ hatch run lint:mypy [--options]
$ hatch run lint:flake8 [--options]
$ hatch run lint:ruff check sourcefinder
$ hatch run lint:black --check sourcefinder

Note that on first run, mypy might need to install type-stubs. You can do that with:

$ hatch run lint:mypy --install-type --non-interactive

Running scripts that use PySE

Normally a regular user would install a released version from PyPI, but to use a development version you may run such scripts like this:

$ hatch run scripts/pyse [--options]

Since the development environment is the default, you don't need to specify the <envrironment>: prefix in the run command.