lumi-reading-chapter-ingest

You can ingest a chapter by providing either raw chapter text or a PDF page range along with the chapter title and number. The system extracts characters, themes, and plot beats, then routes all mutations through the wiki engine to create wiki pages and register graph edges.

Narrative-to-graph extraction works by parsing ingested chapter text to identify character mentions, theme tags, and ordered plot beats. It then registers bidirectional edges like features/appears_in and tagged_with/appears_in, connecting these artifacts within the wiki graph.

Yes, you can extract text from a specific PDF page range for chapter ingestion by specifying the source PDF file along with the start and end pages. The extracted text is then converted into canonical chapter, character, theme, and plot-beat wiki pages.

The best way to ensure idempotent updates is to route all wiki mutations through the wiki engine using a consistent chapter slug. This guarantees that re-running the same chapter ingest is safe and produces byte-identical outputs without duplicating graph edges.

Yes, chapter ingestion supports automatic edge registration for characters and themes. It creates artifacts in the correct namespaces and registers bidirectional edges like features/appears_in and tagged_with/appears_in via the dedicated wiki engine scripts.

You need to use the wiki engine script for chapter ingest mutations to enforce idempotent, workspace-safe writes. This ensures all narrative-to-graph edge registrations and wiki page updates are handled correctly without corrupting the existing wiki graph.