Rubmary Rojas

This commit **enhances** the **Glean reporting system** by populating crucial file information within **XRef targets** for **imported symbols**. Previously, the system only stored symbol names, causing functions like `get_additional_definitions` to miss the associated file context. Now, the `import_names` structure stores both the resolved definition's file and the intermediate module's file, enabling **Glean** to accurately **disambiguate symbols** with duplicate names across different library versions or directory branches. This **fix** significantly improves the precision of cross-reference data by ensuring all relevant file context is available for symbol resolution.

This commit **fixes a gap in the Pyrefly Glean indexer** by adding support for generating cross-references (xrefs) for exception type expressions within `except` handlers. Previously, the indexer would overlook these specific expressions in `try-except` blocks, leading to incomplete code navigation data. The **Pyrefly conversion logic** in `pyrefly/lib/report/glean/convert.rs` is updated to visit and index these previously missed elements. This **enhancement** improves the completeness and accuracy of the generated Glean data, benefiting downstream code analysis and navigation tools. A corresponding update to the `try_except.json` snapshot ensures the new cross-references are correctly captured.

This commit **fixes a bug** in the **`pyrefly` Glean reporter** that caused incorrect `class_declaration_facts` to be generated for **base classes**. Previously, the system would include all definition aliases when processing base classes, leading to facts being created for non-existent classes. The change modifies the definition finding logic within `pyrefly/lib/report/glean/convert.rs` to ensure that only the **original, resolved definition** is used for base classes, skipping intermediate and local aliases. This prevents the erroneous creation of facts for aliases and significantly improves the accuracy of the **Glean knowledge base** regarding class hierarchies.

This commit introduces a **refinement** to the **cross-reference (xrefs) generation logic** within the **`pyrefly`** project's `glean/convert.rs` module. Previously, xrefs were generated for all string literals, including those in comments, leading to potentially noisy data. Now, the system **conditionally generates xrefs** specifically for string literals found within **type annotations** and the `__all__` symbol, significantly improving the precision of the generated facts. This **feature enhancement** ensures that xrefs are only created for semantically relevant string literals, thereby **reducing noise** and enhancing the utility of the code analysis data, while also updating assignment information handling.

This commit **refactors** the **`pyrefly/lib/report/glean/convert.rs`** module to **consolidate the traversal of type annotation expressions**, such as parameter annotations, return types, and variable annotations. Previously, these expressions were visited twice for cross-reference generation and `TypeInfo` embedding; this change merges those visits into a single, more efficient pass. This **performance improvement** streamlines the generation of `xrefs` and type facts, and also enables future work by allowing the system to distinguish annotation visits from other expression visits, preventing incorrect reference additions in comments. The core logic around `make_xrefs` and `generate_expr_facts_and_xrefs` is updated to reflect this unified approach.

This commit **enhances the `pyrefly` static analysis tool** by **improving module import resolution** within its **Glean report conversion logic**. It specifically modifies how `from` imports and star imports (`from module import *`) are processed, ensuring that **file location information is correctly propagated** when computing the imported module. This **new capability** addresses a long-standing TODO in `pyrefly/lib/report/glean/convert.rs`, leading to **more accurate definition tracking** for imported symbols. The change significantly improves the overall precision of `pyrefly`'s output for downstream tools that rely on accurate code navigation and understanding.

This commit **fixes a bug** in the **`pyrefly` `glean` reporter** by **expanding the range calculation for variable definitions** within `for` loop statements. Previously, the span for these declarations was too narrow, causing inconsistencies with the old indexer and leading to gaps in the **dead code graph** used for static analysis. The change in `pyrefly/lib/report/glean/convert.rs` now ensures that the variable definition's span correctly encompasses the full loop header. This provides **more accurate and consistent data** for downstream analysis tools, improving the reliability of features like dead code detection by ensuring proper variable scope representation within the **`glean` indexer**.

This commit **fixes a bug** in the **`pyrefly`** library that caused missing cross-references (xrefs) for **relative Python imports**. Previously, the `find_definition` function failed to resolve these imports, leading to incomplete xref data. The **`pyrefly/lib/report/glean/convert.rs`** module has been refactored to correctly handle relative import resolution, leveraging `find_definition_for_imported_module` and ensuring xrefs are generated for both the module name and the "dots range". This **improves the accuracy and completeness of Python code navigation and analysis** by ensuring all relative imports are properly linked.

This commit provides a **bug fix** for incorrect **variable declaration and cross-reference generation** within the **`pyrefly` Glean fact conversion logic**. Previously, declarations were erroneously created for assignment targets that were merely referencing existing values, such as dictionary keys or instance attributes, instead of storing new ones. The fix in `pyrefly/lib/report/glean/convert.rs` ensures declarations are only generated when the assignment context is "Load" (indicating a new value is being stored), and explicitly visits target nodes to correctly record all necessary cross-references. This improves the semantic accuracy of generated Glean facts, as demonstrated by updated snapshots for class-related code.

This commit **fixes a workflow failure** within the **`pyrefly/lib/report/glean`** module by **updating the schema ID** to an existent one. Previously, the workflow was failing because the referenced schema did not exist, likely due to earlier issues with the config sync process. This **maintenance** change updates the `SCHEMA_ID` constant in `pyrefly/lib/report/glean/schema/builtin.rs` and propagates the new `schema_id` value across all relevant JSON snapshot files in `pyrefly/lib/report/glean/snapshots/`. This ensures the `glean` reporting mechanism can operate correctly by referencing a valid and existing schema.

This commit **regenerates the Python Rust schema** within the `pyrefly` project, specifically updating the `SCHEMA_ID` constant in `pyrefly/lib/report/glean/schema/builtin.rs`. As a **maintenance chore**, it also propagates this updated schema identifier to several generated JSON snapshot files in `pyrefly/lib/report/glean/snapshots/`. This ensures that the **Glean reporting subsystem** in `pyrefly` remains synchronized with the latest schema definition, maintaining data consistency and compatibility for generated reports.

This commit introduces a **new capability** within the **`pyrefly` Glean reporting module** by adding the `XRefsByFile` predicate. This predicate significantly enhances cross-reference data by including the **target definition** of a reference, rather than just its name. The `pyrefly/lib/report/glean/convert.rs` logic has been updated to generate this richer information, which is reflected in the updated `Glean` output snapshots. This change provides more precise and semantically complete cross-reference data for downstream tools consuming `pyrefly`'s `Glean` facts.

This commit **enhances the `pyrefly` indexer** by **adding logic** to correctly generate cross-references for **transitive imports** within Python `from` statements. Previously, the indexer failed to create xrefs for symbols that were re-exported from another module, leading to incomplete code navigation. This **bug fix** and **feature improvement** in `pyrefly/lib/report/glean/convert.rs` ensures that the original definition of such symbols is properly linked, significantly improving the accuracy and completeness of the generated index for Python codebases.

This commit introduces a **unified preference mechanism** for **definition-finding functions** within the **`pyrefly`** tool's **`glean` report conversion** module. It adds the `find_preference_glean` function and integrates it into several existing definition lookup methods, including `find_definition_for_name_use` and `find_definition`. This **enhancement** ensures that when searching for a definition, the system will now consistently **prefer `.py` implementation files over `.pyi` stub files**. This change improves the accuracy and consistency of code navigation and analysis by prioritizing concrete implementations.

This commit implements a **bug fix** within the **Pyreflly tool's `check` command** by modifying its **hashing mechanism**. Previously, the system used the module name to generate hashes, which led to **hash collisions** when multiple files (like `.py` and `.pyi` files) shared the same module identifier. To resolve this, the hashing now utilizes the **full file path** instead of the module name, ensuring each file has a unique hash and preventing unintended overwrites or incorrect processing. This change improves the **robustness and accuracy** of file identification within `pyrefly`.

This commit performs a **refactoring** to **remove duplicated code** within the `pyrefly/lib/report/glean/convert.rs` module. It introduces a new helper function, `find_definition_for_literal_symbol`, which centralizes the logic for finding definition locations for literal symbols. This change specifically reduces redundancy for string, boolean, and `None` literals, which were previously handled in a duplicated manner across functions like `get_xrefs_for_str_lit` and `get_xrefs`. The work improves the maintainability and clarity of the **Glean reporting module's** cross-reference generation by consolidating common logic.

This commit **improves cross-reference generation** for Python `import from` statements within the **Glean fact generation system**. Previously, an import like `from a.b import x` would only generate a cross-reference for `b`; this change now correctly adds references for **all module path components**, such as `a` and `b`. This **enhancement** leverages the Language Server Protocol (LSP) via `find_definition_for_imported_module` to accurately locate module definitions, ensuring more comprehensive code intelligence. The core logic is implemented in `pyrefly/lib/report/glean/convert.rs`, with a supporting visibility change in `pyrefly/lib/state/lsp.rs`.

This commit **improves cross-references (xrefs) for string literals** within the **`pyrefly` Glean report generation** by adopting a more robust definition-finding mechanism. The `pyrefly/lib/report/glean/convert.rs` module now utilizes `pyrefly`'s internal `find_definition_for_name_use` function to accurately locate the definition of string literals, including their originating file, replacing a less precise, manually constructed name approach. This **feature improvement** and **refactoring** enhances the semantic accuracy of code analysis. Consequently, users will observe **more precise and complete cross-reference data** in their Glean reports, with updated snapshot tests reflecting refined span and xref entries, such as the exclusion of quotes from reported ranges.

This commit **improves the accuracy of attribute definition lookups** within **PyreFly's LSP server**. It **refactors** the `pyrefly/lib/report/glean/convert.rs` module to leverage the `find_definition_for_base_type` function for attribute cross-referencing, replacing an older lookup mechanism. This change provides more precise results by correctly identifying the file where an attribute was defined. This required adjusting the visibility of `find_definition_for_base_type` in `pyrefly/lib/state/lsp.rs` and is reflected in updated test snapshots, leading to more reliable static analysis and LSP features.

This commit **enhances the Pyrefly indexer's Glean conversion logic** to **improve the accuracy and completeness of import cross-references**. It **fixes a long-standing issue** where the indexer failed to correctly link cross-references with their corresponding imports, particularly for **transitive imports** and when imports were erroneously identified as definitions. By refining functions like `record_name_for_import` and `add_xref` in `pyrefly/lib/report/glean/convert.rs`, this change ensures that **dead code analysis tools** receive more reliable data. This is a **critical maintenance update** that significantly improves the quality of code graph data.