nx_text_ingest.nx

nx_text_ingest.nx -- text-medium ingester targeting media-agnostic schema. Tier-1 ingester in the multi-media stack. Consumes a MediaItem of type TEXT or HTML from a StoryDb's MediaPool and emits StoryBeats + StoryScenes + StoryChapters into the same StoryDb. Supersedes nx_narrative_ingest.nx (kept in tree as deprecation pointer until concurrent-agent wiring catches up). The two differ in ONE thing: this one writes beats into the canonical nx_storybeat.nx schema so the rest of the director pipeline (nx_arc / nx_moment_resolve / nx_directors_note) consumes them uniformly with image / gif / video / audio beats. Scope: structural parse only. Marks paragraphs as NARRATION, scene breaks as TRANSITION, headings as HEADING. Semantic classification (DIALOGUE / ACTION / DESCRIPTION / INTERIORITY / GESTURE) is the next layer (nx_text_beat_classify.nx). Sources expected: * URL via nishi-browser fetch -> nx_media_fetch.nx * Manual paste via stdin -> nx_media_fetch.nx * Filesystem path -> nx_media_fetch.nx * Voice dictation -> nx_media_fetch.nx * Batch directory walk -> nx_ingest_runner.nx (Tier 5) genealogy_id: propp_morphology_1928 (narrative structural roots) + commonmark_2014 (heading / blank-line rules) Composes with: nx_storybeat.nx -- writes here nx_storydb.nx -- container nx_media_pool.nx -- reads source bytes via media_id nx_media_sniff.nx -- caller already classified nx_safety_envelope: intended_use: "Parse text-medium MediaItem into structured beats / scenes / chapters in the canonical media-agnostic StoryBeat schema." sil_target: SIL2 asil_target: QM