Structuring and packaging your python project
Download as pptx, pdf2 likes384 views
This document discusses best practices for structuring Python projects. It recommends keeping the root directory clean and organized with files like README, LICENSE, requirements.txt, tests, and documentation. Python projects should be logically broken up into modules and packages to separate concerns and avoid issues like circular dependencies. Modules are individual .py files while packages are folders containing an __init__.py file. The document also covers importing, sharing code as installable packages, and using setup.py to define packages for distribution.
More Related Content
Similar to Structuring and packaging your python project (20)
More from Eyal Trabelsi (6)
Recently uploaded (20)
Structuring and packaging your python project
- 2. In practical terms, “structure” means: making clean code whose logic and dependencies are clear as well as how the files and folders are organized in the filesystem. What does structure means
- 3. Keep the root clean: - General files like README, LICENSE, etc - requirements.txt - tests / (or test.py) - doc/ - project_name/ (or project_name.py) - setup.py - Manage.py (Django) - Other tooling like .gitignore, .circleci/ etc… Principle for any repository
- 4. Things to avoid: - Circular dependencies - Hidden coupling - Spaghetti - Ravioli Generic considerations: - Separate concerns with decorators - Use context manager when setup and teardown is needed Principle for any repository
- 6. - Modules are Just .py files. - __name__ == ‘__main__’ to be usable as scripts - Modules are compiled into .pyc files - Module names standards - Each module is its own isolated world (namespace). - Module execution Modules - basics
- 7. - Search path - from module import * - from module import x Modules - Imports
- 9. - Packages are folders that contain __init__ file - What should __init__ file used for - Executable submodule using __main__.py file - Each package should be consumable from outside and consumed by its “public” interface. - Divide and conquer, divide huge project to small packages (Import) Packages- basics
- 11. - Shareable/installable compiled package - PyPI is public python package repository - .pypirc for PyPI settings *. (Distribution) Packages- basics - Create distribution files
- 12. - Packaging and distribution manager (Distribution) Packages- Setup.py - Installation developer mode by pip install –e
- 15. • Writing the Setup Script • Structuring large python projects • Packaging Python Projects • Comprehensive Modules and Packages talk • Installing Packages Resources
Editor's Notes
- #4: A pip requirements file should be placed at the root of the repository. It should specify the dependencies required to contribute to the project: testing, building, and generating documentation. If your project has no development dependencies, or you prefer development environment setup via setup.py, this file may be unnecessary.
- #5: Matplotlib for example for state Spaghetti long functions Ravioli too short with 100 of classes
- #7: Module name - keep module names short, lowercase, and be sure to avoid using special symbols like the dot (.) - or question mark (?). Module execution - when a module is imported all statement execute until end of file reached the content of module namespace are all of the global names that still defined at the end of the execution process
- #8: Search path – if a file is not on the path it wont import sys.path sometimes you might hack it by sys.path.append From module import * – take all symbols from a module and place them into local scope (__all__) From module import x – allows parts of the module to be used without having to type the module prefix lifts selected symbols out of a module after import it ! And make it available locally
- #10: Init file can be used as façade to combine multiple modules a simple init file is good one (no initialization or monkey patching even not logs) We can do it without __init__.py but than weird namespace black magic occur namespace packaegs Main file explicitly marks the entry point
- #12: You might require wheel twhine and tox to interable with pypi
- #13: Source distribution vs built distribution - Hold Meta information of your package Hold executable to complete your packing tasks Uploading it to pypi or artifactory Setup.py also has: -classifier where your package can be used for search - Install_requires for requirements - find_p[ackages to find packages - Entry points console scripts
- #16: Decorator example cache