bijux-canon-ingest 0.3.8

Deterministic document ingestion, chunking, and retrieval preparation for the bijux-canon package family from Bijux.

bijux-canon-ingest

bijux-canon-ingest is the package that turns raw documents into deterministic ingest artifacts and retrieval-ready structures. It is where cleaning, chunking, package-local retrieval assembly, and ingest-facing boundaries live.

This package should help a maintainer answer practical questions such as:

• how does a source document become stable ingest output
• where do retrieval-oriented assembly steps belong
• which code is pure transformation logic and which code is adapter work

What This Package Takes And Produces

• takes: raw documents, ingest configuration, and package-local retrieval assembly inputs
• produces: cleaned records, chunks, retrieval-ready artifacts, and ingest boundary failures when source data breaks contract
• guarantees: deterministic transforms stay separable from adapters and the root import remains safe for dependency-light consumers
• does not do: own runtime replay policy, expose every internal helper as a stable promise, or define vector execution semantics for the rest of canon

Public API Routing

Use the package root for stable, dependency-light ingestion primitives. Reach into submodules only when you need a specific boundary:

• bijux_canon_ingest for stable transforms, result helpers, and shared ingest primitives
• bijux_canon_ingest.application for workflow orchestration
• bijux_canon_ingest.interfaces for CLI and HTTP edges
• bijux_canon_ingest.config for builder-style package configuration

Legacy continuity

• compatibility package: bijux-rag
• legacy import root: bijux_rag
• legacy command: bijux-rag
• canonical migration guide: Migration guidance
• retired repository target: https://github.com/bijux/bijux-rag (see Repository consolidation notes)

What this package owns

• document cleaning, normalization, and chunking
• ingest-local retrieval and indexing assembly
• package-local CLI and HTTP boundaries
• ingest-specific adapters, safeguards, and observability helpers

What this package does not own

• standalone vector execution semantics
• runtime-wide governance, persistence, or replay authority
• repository tooling and release automation

Source map

• src/bijux_canon_ingest/processing for deterministic document transforms
• src/bijux_canon_ingest/retrieval for retrieval-oriented models and assembly
• src/bijux_canon_ingest/application for package workflows
• src/bijux_canon_ingest/infra and src/bijux_canon_ingest/integrations for adapters
• src/bijux_canon_ingest/interfaces for CLI and HTTP edges
• tests for behavior, layout, and corpus-backed checks

Read this next

• Package guide
• Package overview
• Ownership boundary
• Architecture overview
• Operator workflows
• Compatibility packages
• Changelog

Primary entrypoint

• console script: bijux-canon-ingest

Release Readiness

• release line prepared for publish: 0.3.8
• release date: 2026-06-28
• package changelog: CHANGELOG.md