Skip to content

feat(rag): add pure Python chunking and RRF foundation - #17

Closed
adihex wants to merge 1 commit into
mainfrom
split/rag-core
Closed

adihex wants to merge 1 commit into
mainfrom
split/rag-core

Conversation

@adihex

@adihex adihex commented Aug 28, 2026

Copy link
Copy Markdown
Owner

Summary

  • add a corpus-free, standard-library Python core under apps/rag-pipeline
  • provide deterministic sentence-aware overlapping text chunking
  • provide generic weighted reciprocal-rank fusion with stable tie handling
  • cover boundaries, validation, weighting, duplicate handling, and ordering with focused unit tests

This is an independent foundation slice replacing the pure reusable portion of #10. It has no dependency on #10 or any other split/slice. It intentionally excludes Chroma/Gemini clients, FastAPI, NotebookLM integration, downloaded sources/data, notebooks, generated wiki content, duplicated agent configuration, and root hook/tooling changes. This PR does not modify or close #10.

Verification

  • python3 -m unittest discover -s tests -v (from apps/rag-pipeline): 10 passed
  • mise exec -- pnpm --filter @agentx/rag-pipeline test: 10 passed
  • python3 -m pip wheel --no-deps --wheel-dir <temp> .: built agentx_rag_core-0.1.0-py3-none-any.whl
  • mise exec -- pnpm build: passed (19 workspace projects)
  • git diff --check: passed

Existing root-check failures

  • mise exec -- pnpm test: 360 passed, 1 skipped, 1 failed. The pre-existing packages/core/tests/injection.test.ts custom ADP handler assertion also fails when run alone; this PR does not touch that code.
  • mise exec -- pnpm check: pre-existing formatting findings in 23 files outside this PR. The new package.json passes the formatter; the scoped lint phase reports no lintable files.

The initial red TDD run failed with ModuleNotFoundError: No module named 'rag' before implementation; the final Python suite is green.

Amp-Thread-ID: https://ampcode.com/threads/T-01a047e7-d402-72a8-ad22-eb5d3d3df6b1
Co-authored-by: Aditya Balakrishnan <adibalak1925@gmail.com>
@adihex adihex closed this Aug 28, 2026
@adihex adihex reopened this Aug 28, 2026

adihex commented Aug 28, 2026

Copy link
Copy Markdown
Owner Author

Landed on main as part of the consolidated, rebased stack ending at e9bd4bd. Verified this PR has no patches absent from origin/main, so closing the now-redundant stacked PR.

@adihex adihex closed this Aug 28, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants