speedyk-005 / chunklet-py / 22282498046

91%

Build Type

push

github

Committed by web-flow

Commit Message

# v2.2.0: The Unification Edition (#12)

* feat(code_chunker): upgrade metadata parsing and expand language patterns

- Upgrade METADATA pattern to use recursive sub-routines (?1), enabling
  proper parsing of nested parentheses in decorators and attributes
- Add support for PHP 8 attributes (#[Attribute]) and VB.NET metadata (<...>)
- Expand line comment patterns to include Forth (\|)
- Move Erlang (%%) and Nim (##) documentation markers to Style 2
  prefixed comments for better chunking continuity
- Broaden function and namespace detection to include ColdFusion
  components and 'export' modifiers
- Improve OPENER/CLOSURE reliability by supporting Pascal-style
  case-insensitive BEGIN/END and stricter line-end anchoring
- Add open curly bracket in the function pattern to support more languages

* refactor(code_chunker): consolidate block state management and clean up extraction logic

- Refactored _code_structure_extractor.py to centralize structural state management and resolve logic bugs in block detection.
- SRP: Moved block_indent_level updates from _handle_annotated_line to _handle_block_start to ensure metadata handling remains stateless.
- Logic Cleanup: Refined boolean guards to correctly distinguish between nested closures and sibling methods, ensuring proper snippet flushing.
- State Machine: Centralized the indentation 'anchor' logic to prevent improper block merging while maintaining parent-child relations.
- Additional improvements to decorator handling, function detection, and overall chunking reliability.

* feat(code_chunker): add multi-line string assignment pattern to prevent improper chunking

Introduced MULTI_LINE_STRING_ASSIGN to patterns.py to ensure large string blocks are treated as single units.

Extractor safety: Prevents the code chunker from splitting or flushing inside multi-line string declarations.

Robustness: Ensures that content within triple-quotes (or equivalent multi-line syntax) remains intact during structure ex... (continued)

Coverage Stats

285 of 336 new or added lines in 24 files covered. (84.82%)

5 existing lines in 2 files now uncovered.

1406 of 1616 relevant lines covered (87.0%)

4.35 hits per line