mirror of
https://github.com/DeusData/codebase-memory-mcp.git
synced 2026-11-11 09:25:35 +08:00
Martin VogelandJames
30941983cb
Generalizes #224's Svelte/Vue script-block import extraction into a spec-driven embedded-language pattern that also covers HTML and Astro, so any future host grammar that keeps sub-language source as raw text gets coverage with a one-line declaration. The existing CBMLangSpec already exposes the per-language tree-sitter grammar via cbm_ts_language(); add a parallel embedded_imports field that lists which AST node types hold which sub-language. The new generic walker in extract_imports.c reads that list, locates each embedded content node in the host AST, re-parses the slice with the embedded grammar resolved through cbm_ts_language(), and runs the standard ES import extractor (walk_es_imports) over the inner AST. No grammar symbols are referenced from the extraction file. - internal/cbm/lang_specs.h: add CBMEmbeddedLangSpec + embedded_imports field on CBMLangSpec. - internal/cbm/lang_specs.c: declare vue/svelte/html/astro embed lists ({"script_element", "raw_text", CBM_LANG_JAVASCRIPT}); set the new field on those four spec entries; explicit NULL initializer on every other entry to satisfy -Wmissing-field-initializers. - internal/cbm/extract_imports.c: add embedded_collect_content_nodes + parse_embedded_imports; route SVELTE/VUE/HTML/ASTRO through the generic helper. - tests/test_extraction.c: cover Svelte (with and without <script>), Vue, and HTML. Full suite: 2824 passed, 0 failed. Closes #223. Closes #224. Co-authored-by: James <jmcmacnz@gmail.com>