Felix

[Python]: Modernize setup and drop Python 2 (#8955)

This commit **enhances error reporting** within the **Python IDL generator** by providing more specific diagnostic information. The `GeneratePython` function in `src/idl_gen_python.cpp` has been modified to return detailed error messages as `const char*` instead of a simple boolean indicator. Consequently, the `GenerateCode` function now captures and reports these specific error strings, replacing generic failure notifications. This **improves the clarity of error messages** for users, making it significantly easier to diagnose and resolve issues during Python code generation.

This commit addresses a **CI maintenance chore** by fixing a **typo** within the `.github/workflows/label.yml` GitHub Actions workflow. Specifically, it updates the version reference for the `actions/labeler` action from `'@6'` to the correct `'@v6'`. This ensures the **automatic labeling process** for pull requests and issues functions as intended, preventing potential CI failures or unexpected behavior due to an incorrectly specified action version.

This commit performs a significant **maintenance update** across various **GitHub Actions CI/CD workflows**, modernizing the project's build and release infrastructure. It updates numerous actions, including `actions/checkout`, `actions/stale`, `softprops/action-gh-release`, `microsoft/setup-msbuild`, `actions/setup-dotnet`, `actions/setup-java`, `gradle/actions/setup-gradle`, `jiro4989/setup-nim-action`, and `actions/labeler`, to their latest stable versions or to use tags instead of SHAs. This **refactoring** ensures the **build, release, documentation, and labeling pipelines** leverage the most current and secure tooling, improving reliability and potentially resolving issues with build images. The update also implicitly supports newer environments, such as Swift on Ubuntu 24.04, enhancing the overall robustness and future-proofing of the CI system.

This commit delivers a **bug fix** for the **Python code generation** module, specifically within `src/idl_gen_python.cpp`. It addresses an issue where the `GenerateCode` function was incorrectly generating `__init__.py` files due to an invalid starting index when parsing directory paths. The fix ensures that path separators are correctly identified from the beginning of the path, resolving errors that occurred with specific directory structures like "MyGame/Sample/". This change guarantees the proper generation and functionality of **Python modules** by ensuring `__init__.py` files are created in the correct locations.

This commit **refactors** the **testing infrastructure** to ensure **C++11 compatibility**. Specifically, it replaces instances of `std::make_unique` with the equivalent `std::unique_ptr` direct initialization for `Native::Matrix` objects within the `NativeTypeTest` function in `tests/test.cpp`. This **maintenance** effort addresses the fact that `std::make_unique` was introduced in C++14. The change is localized to the test suite, with no impact on runtime behavior or core functionality, only on how tests are compiled under C++11.

This commit delivers a **bug fix** addressing a **lifetime issue** within the **FlatBuffers gRPC integration**. Specifically, it **corrects the return value** of the `Reallocate` method in the `SliceAllocator` class, defined in `include/flatbuffers/grpc.h`. The previous implementation incorrectly managed pointer ownership, potentially leading to memory errors or misleading warnings. By ensuring `Reallocate` now points to the beginning of the new memory slice, this change enhances **memory safety and correctness** for gRPC-related allocations, preventing future issues.

This commit provides a **documentation fix** for the **C++ language examples**, specifically targeting the verifier functionality. It updates the `docs/source/languages/cpp.md` file to modify the example code for instantiating a `Verifier` object. The change involves using a named `Verifier` object instead of a temporary one, which improves the clarity and idiomatic correctness of the provided code snippet. This **maintenance** task ensures that the C++ documentation offers best practices for `Verifier` instantiation, benefiting users learning from the examples.

This commit **corrects an inaccuracy in the project documentation** by removing a stray 'required' keyword that was erroneously present. It specifically targets instances within the **documentation files** where a parameter or field was incorrectly marked as mandatory. This **documentation maintenance** improves the clarity and correctness of the user-facing guides, ensuring developers and users have accurate information regarding configuration or API usage.

This commit **fixes a bug** in the **Python code generation** for FlatBuffers, specifically addressing the `__eq__` method for **NumPy vector types**. It ensures that equality comparisons for fields backed by NumPy arrays correctly use `np.array_equal`, resolving issues where direct array comparison led to incorrect results. The `src/idl_gen_python.cpp` generator is updated to produce the correct comparison logic, and **new test cases** in `py_test.py` validate this behavior. This change significantly improves the **robustness and correctness** of FlatBuffers Python bindings when working with numerical data represented as NumPy arrays.

This commit **updates the documentation** on the **start page** (`docs/source/index.md`) to accurately reflect **FlatBuffers' schema evolution capabilities**. It specifically clarifies the project's robust handling of **backwards and forwards compatibility**, which is a core feature for data structure evolution. This **documentation fix** ensures that users reading the main introductory content receive precise information regarding how FlatBuffers manages schema changes over time. The update improves the overall clarity and correctness of critical project information.

This commit introduces a **fix** to the **Python stub generator** within `src/idl_gen_python.cpp`. It specifically modifies the `PythonStubGenerator::Generate` function to prevent the `flatbuffers` module from being included redundantly in generated `.pyi` files. This **maintenance** improvement ensures that **Python type hint files** are correctly formed, avoiding duplicate import statements. The change enhances the quality and correctness of generated Python code, providing a cleaner and more robust experience for developers using FlatBuffers with Python.

This commit provides a **bug fix** for the **Python stub generator** (`idl_gen_python.cpp`) to correctly define enum members in generated `.pyi` files. Previously, using type annotations with the `:` syntax for enum members led to them being interpreted as non-member attributes by type checkers, violating Python's typing specification for `Enum`s. The generator now uses `cast` for enum member definitions, ensuring that **generated Python enum stubs** accurately represent enum members. This change resolves type checking issues and improves the correctness of generated Python code for `Enum` types.

This commit **fixes a bug** in the **Python stub generator** that caused `mypy` errors due to self-imports within generated `.pyi` files. It introduces `Export` methods and an `exports` vector to the `python::Imports` class, declared in `include/codegen/python.h` and implemented in `include/codegen/python.cc`. The core `idl_gen_python.cpp` module is updated to leverage this new functionality, preventing generated stub files from attempting to import types they themselves define. This ensures correct type hint generation, resolving "Name already defined" errors and improving the robustness of generated Python code.

This commit performs a **documentation fix** by correcting a typo in a code comment within the `src/binary_annotator.cpp` file. Specifically, it improves the clarity of a comment describing the mapping between fields and their relative offsets in a VTable within the **binary annotator** component. This is a minor **maintenance** change that enhances code readability and understanding for developers, with no functional impact on the system's behavior.

This commit provides a **bug fix** and **maintenance update** to the **Python code generator**, ensuring more accurate type hints for generated Python code. It modifies `src/idl_gen_python.cpp` to emit `Optional[str]` and `Optional[List]` for fields that can be `None`, rather than just `str` or `List`. This change resolves `mypy` type errors that previously occurred when `None` was assigned to such fields, significantly improving the **type correctness** and static analysis capabilities of generated FlatBuffers Python code. Consequently, Python developers will experience fewer type-related issues and better IDE support when working with FlatBuffers objects.

This commit primarily **fixes a bug** in the `scripts/generate_code.py` script by correcting a typo in the `PYTHON_OPTS` list. Specifically, it changes the option from `--python-decode-obj-api-string` to the correct `--python-decode-obj-api-strings`, which ensures the **`flatc` compiler correctly processes this flag** during Python code generation. Concurrently, it **updates the documentation** in `docs/source/flatc.md` to formally introduce and describe the new `--python-decode-obj-api-strings` flag. This **maintenance and documentation update** ensures that the **Python object API string decoding feature** is both correctly enabled by the build system and properly documented for users.

This commit provides a **bug fix** for **Python gRPC code generation**, specifically addressing issues with generated filenames and directory structures. It modifies `grpc/src/compiler/python_generator.cc` to replace dots with underscores in gRPC filename suffixes, preventing conflicts with Python's namespace handling, and restores the generation of gRPC files into namespace-based directories, fixing a recent regression. Additionally, the `flatc` documentation in `docs/source/flatc.md` is updated to reflect `--gen-onefile` support for Java and Kotlin. These changes significantly improve the **usability and correctness of Python gRPC output**.