Aristidis Papaioannou

This commit implements a **minor documentation fix** by correcting the order of a comment line within the **Thrift object model documentation**. It specifically targets the `index.md` file in `thrift/doc/object-model/`, ensuring a comment appears correctly after the main heading. This **maintenance change** improves the readability and structural correctness of the **Thrift object model guide**, providing a more polished and accurate resource for developers. The scope of this change is limited to documentation formatting and has no impact on Thrift's functional behavior or generated code.

This commit **updates the Thrift documentation** by adding an HTML comment to `thrift/doc/object-model/index.md`. Specifically, it marks the **object model documentation** as a **generated file**, indicating its automated origin. This is a **maintenance** change that helps tools and processes correctly identify the nature of this documentation, ensuring it's handled appropriately and preventing manual edits from being inadvertently overwritten.

This commit **relaxes the default validation level** for the `unnecessary_allow_unsafe_non_sealed_key_type` option within the **Thrift compiler**. The compiler's **semantic analysis** will now issue a 'warn' instead of an 'error' when encountering potentially unsafe non-sealed key types, as updated in `thrift/compiler/compiler.cc` and `thrift/compiler/sema/sema_context.h`. This **maintenance change** addresses the previous 'error' being overly aggressive, particularly for patch generation workflows where full type knowledge might be incomplete. The adjustment means users will now receive **warnings instead of build-breaking errors** for this specific issue, improving compiler flexibility without compromising safety, and is reflected in updated test expectations.

This commit introduces a **new validation rule** within the **Thrift compiler's semantic analysis** to prevent enum value names from matching their enclosing type name, mirroring existing checks for field names. It enhances the `standard_validator` to implement this check and updates the `sema_context` with new `validation_level` parameters. Furthermore, **command-line options** are added to `compiler.cc` to configure the diagnostic level for both enum and field name validation, allowing these checks to eventually be promoted to errors. This **feature enhancement** improves the robustness of Thrift definitions by enforcing clearer naming conventions.

This commit introduces a **new validation flag** `no_package` to the **Thrift compiler**, specifically targeting `.thrift` files that completely lack a `package` directive. It **refactors** the existing `missing_package` check, renaming it to `empty_or_no_package`, to differentiate between a truly absent package and an explicitly empty `package;` directive. This **enhances the compiler's semantic analysis** by providing more granular control over package validation within the `validate_program_package` function. Users can now configure these distinct checks via new command-line options and internal `sema_params`, leading to more precise diagnostics for **Thrift IDL files**. The change includes updates to argument parsing and comprehensive test coverage for the new validation behaviors.

This commit performs a widespread **refactoring** across numerous **Thrift test fixtures** by introducing the `@thrift.AllowUnsafeNonSealedKeyType` annotation. This annotation is applied to various set and map fields within structs, unions, and typedefs, as well as parameters in service methods, across fixtures such as `complex-struct`, `mcpp2-compare`, and `fatal`. This **maintenance** effort explicitly permits the use of non-sealed types as keys in collections, likely to align with updated compiler rules or to enable specific test cases that require this behavior. The broad application across the test suite indicates a systemic adjustment to how certain key types are handled in Thrift collections.

This commit performs a **refactoring** by adding the `@thrift.AllowUnsafeNonSealedKeyType` annotation to numerous set and map fields and typedefs across various **Thrift IDL definitions**. This annotation explicitly permits the use of **unsealed types as keys** in these specific Thrift containers, which might otherwise be disallowed by stricter type checking. The changes span across C++, Python, and Rust test suites, as well as core Thrift utility definitions, ensuring compatibility for existing structures or enabling specific use cases where such key types are required. This broad application indicates a systemic adjustment to allow more flexible, albeit potentially less safe, key types in specific Thrift structures.

This commit **fixes a bug** in the **Thrift compiler's codemod tool** that incorrectly placed `thrift.AllowUnsafeNonSealedKeyType` annotations when function parameters shared a line with the function signature. A new helper function, `annotate_function_param()`, was introduced in `thrift/compiler/codemod/annotate_allow_unsafe_non_sealed_key_type.cc` to detect this specific case. It now ensures the annotation is correctly placed on its own indented line before the parameter, improving the **accuracy of code formatting**. This **bug fix** prevents potential style or compilation issues related to misaligned annotations in generated Thrift code, with new test cases added to validate the corrected placement logic.

This commit introduces a **new codemod tool** (`remove_duplicate_namespaces.cc`) designed to **clean up Thrift IDL files** by **removing duplicate `namespace` directives**. The tool intelligently identifies and retains only the first `namespace` declaration for each language, aligning with the deprecation of redundant directives. A crucial **safety mechanism** is implemented to skip files where duplicate `namespace` lines contain trailing content, preventing unintended modifications. This **maintenance** effort enhances the **Thrift compiler's codemod subsystem** and is thoroughly validated with new **unit tests** covering various scenarios.

This commit performs a widespread **refactoring** across numerous **Thrift compiler test fixtures** by explicitly adding the `@thrift.AllowUnsafeRequiredFieldQualifier` annotation. This annotation is applied to existing `required` fields within various structs and exceptions, including examples like `Example`, `Internship`, `Fiery`, and `MinPadding`, found in modules such as `complex-struct`, `exceptions`, and `types`. The change ensures that these test definitions remain compatible with evolving Thrift compiler rules regarding required field qualifiers. This **maintenance** effort prevents potential build failures or unexpected validation behavior within the **Thrift compiler's test suite** by aligning test definitions with current compiler expectations.

This commit performs a **refactoring** within the **Thrift compiler's Abstract Syntax Tree (AST)** by renaming the `t_program` method `namespace_nodes` to `all_namespace_nodes`. This change improves clarity and consistency for internal compiler development by providing a more descriptive name for retrieving all namespace nodes. The update affects the `t_program.h` definition, its associated `add_namespace` logic, and propagates to usages in the `standard_validator` for semantic analysis and `thrift_ast_debug_tree_utils` for debugging output. This internal API improvement has no impact on generated code or user-facing functionality.

This commit **enhances the Thrift compiler's Abstract Syntax Tree (AST) generation** by ensuring that all `t_namespace` nodes are created and tracked within `t_program`, even when namespace declarations are duplicated. Previously, duplicate declarations might have been silently discarded from the AST, but now they are explicitly represented in a new `namespace_nodes_` list. This **feature improvement** provides a more complete and accurate AST representation of the source code, particularly for tools that analyze or process the AST. It also includes updates to the debug tree utilities to reflect these newly tracked nodes and a new helper function `addChildForNamespaceNode` in `thrift_ast_debug_tree_utils.cc`.

This commit introduces a **new validation flag**, `duplicate_namespace`, to the **Thrift compiler**, which reports errors or warnings when a single `.thrift` file declares the same language in multiple `namespace` directives. This **enhancement to semantic analysis** involves **extending the compiler's command-line argument parsing** in `compiler.cc` and **implementing the core validation logic** in `standard_validator.cc` via the `validate_duplicate_namespaces` function. The new capability helps maintain consistency and prevent potential issues arising from redundant namespace declarations within Thrift definitions.

This commit **enhances** the **Thrift Object Model** by introducing several key definitions and capabilities, marking it as version 1.2.0. It formally defines **Thrift identifiers** for enum and field names, providing clearer naming conventions within the model. Additionally, the model now explicitly **allows `NaN` and signed zero datums** for `Float{N}`-kind records, expanding the range of valid floating-point values. A formal definition for **sealed types** has also been added, further solidifying the object model's specification. These updates primarily affect the **Thrift Object Model's documentation** and its underlying specification, ensuring consistent interpretation and implementation across Thrift-based systems.

This commit **enhances the Thrift compiler's parser** by enabling the tracking of **source code ranges for `namespace` directives**. The `parse_namespace()` function in the **parser** now captures this range and passes it through the `on_namespace()` action to the **AST builder**. This **new capability** ensures that the **Abstract Syntax Tree (AST)** for Thrift definitions retains precise source location information for `namespace` statements. This is a significant improvement for downstream tools that rely on accurate source-level diagnostics or refactoring.

This commit introduces a new **`t_namespace` AST node** within the **Thrift compiler's Abstract Syntax Tree (AST)** to encapsulate namespace values along with their source location information. It **refactors** the `t_program` class to store pointers to these `t_namespace` objects instead of raw strings, and updates numerous downstream components like code generators and semantic analysis tools to access namespace strings via the new `->ns()` method. This change is a **no-functional-change refactor** that prepares the compiler for future enhancements, specifically enabling the **tracking of source ranges for namespace directives**. The update ensures that all parts of the compiler correctly handle the new structured namespace representation without altering current behavior.

This commit **fixes a parsing error** within the **`thrift/compiler/codemod/file_manager`** by preventing `namespace` directives from being incorrectly identified when they appear inside comments. It introduces a new helper function, `find_not_in_comment()`, which intelligently skips matches found within line (`//`, `#`) and block (`/* ... */`) comments. This ensures that core functions like `get_first_namespace_offset()` and `remove_namespace()` accurately locate actual namespace declarations. The **bug fix** improves the robustness of codemod tools by preventing them from misinterpreting commented-out code, leading to more reliable code transformations. A new test case in `annotate_allow_legacy_missing_uris_test.py` validates this correction.

Fix test fixtures

This commit **refactors** the **Thrift compiler's semantic validation** for **sealed types** by extracting the core logic into a new, reusable function `validate_sealed_type` within `standard_validator.cc`. A field-specific wrapper, `validate_sealed_field`, is also introduced to streamline validation. This **refactoring** improves code reuse, enabling the same validation logic to be applied to other Abstract Syntax Tree (AST) nodes like typedefs. Consequently, the `compiler_test.cc` is updated to align expected error messages with the new diagnostic output.

This commit introduces a **new codemod**, `annotate_allow_unsafe_non_sealed_key_type`, within the **Thrift compiler's codemod module**. This **new capability** automatically annotates Thrift **fields, typedefs, and function parameters** with `thrift.AllowUnsafeNonSealedKeyType` when they utilize `map` or `set` types with non-sealed key or element types. The primary purpose is to **exempt existing Thrift definitions** from new sealed-type validation being introduced in the compiler, providing a **migration path** for legacy code. This helps prevent immediate breaking changes for existing codebases that rely on non-sealed types in map keys or set elements. Unit tests are also added to ensure the codemod's correct behavior.