nx_reader_xray.nx

nx_reader_xray.nx -- SOVEREIGN, ZERO-DEPENDENCY, book-INTRINSIC X-Ray. Operator law: attack the BEHIND "aids ecosystem" axis of nx_reader_exceed_gate with a REAL capability, not a sovereignty restatement. Kindle's X-Ray is PROVISIONED-ONLY (exists only where Amazon pre-built the data, so most books show "X-Ray: Not available"). This organ DERIVES an X-Ray from the book's OWN text -- works on ANY book, offline, deterministic, auditable. Algorithm (deterministic, no random, no external data): - per chapter, skip the leading duplicate-title lines exactly as nx_reader_render does (index PROSE not headings) - tokenize into maximal A-Za-z runs; track for each token: capitalized? sentence-initial? space-joined-to-prev? - a NAME = a maximal run of consecutive capitalized tokens joined ONLY by spaces (so "White Rabbit" is one entity, but "Alice. Rabbit" is two). A multi-token run always qualifies (proper-noun phrase). A single capitalized token qualifies only if it is MID-sentence (not sentence-initial) AND not a stopword -- this is what separates "thought Alice" (entity) from "But when" / "There was" (sentence openers, not entities). - count occurrences per distinct name + remember the first chapter it appears in (for jump-to-first-mention) - keep names with count >= MINOCC; sort by count desc, then name asc (fully deterministic) Emits knowledge/staging/media/reader/<slug>/xray.json + prints a raw scorecard. Reads OUR own book.json + chap<N>.txt (emitted by nx_epub_book). expect_exit: 0 license_tier: ORIGINAL