| # Specify the Python packages available as dependencies to targets in Bazel. |
| # |
| # When modifying this list, always run |
| # the //third_party:python_requirements.update target: |
| # |
| # bazel run //third_party:python_requirements.update |
| # |
| # to compile (using pip-compile) this list of direct dependencies into a pinned |
| # requirements file---a complete list of direct and transitive dependencies, |
| # pinned by version and cryptographic hash. The pinned requirements file is |
| # used in @rules_python's pip_parse() in the WORKSPACE file to create the |
| # external repositories available as dependencies to py_binary() and |
| # py_library() targets. |
| # |
| # To upgrade dependencies to their latest version, run the update target with |
| # the option --upgrade: |
| # |
| # bazel run //third_party:python_requirements.update -- --upgrade |
| # |
| # Without the --upgrade option, the underlying pip-compile only adds or removes |
| # dependencies without upgrading them to the latest versions available in PyPI. |
| # |
| # Both this input file and the pinned requirements file should be committed to |
| # git. Avoid committing changes that break other developers by using an |
| # environment that meets the project's recommendations. Dependency resolution |
| # is sensitive to the Python environment (interpreter version, etc.) in which |
| # it is run. |
| |
| tensorflow-cpu |
| numpy |
| mako |
| pillow |
| yapf |
| protobuf |