[menny]

Overview

This PR implements the interactive reader engine in //core/reader, providing sentence segmentation, automated interlinear glossing, token inspection, and the session state machine for both curated lessons and custom/imported texts.

Components Implemented

1. Reader Domain Models (//core/reader:ReaderModels.kt):
   • ReaderMode: SENTENCE_FOCUS and PARAGRAPH.
   • ReadingSource: Curated(lesson), CustomText(customText), and RawText(title, content).
   • EnrichedSentence: tokens, interlinear glosses, translation scaffolding, and grammar callouts.
   • TokenInspection: lemma, POS, English gloss, cognate rule notes, false cognate trap cards, clitic surgery deconstructions, and vault status.
   • ReaderSessionState & SessionSummary.
2. Sentence Segmenter (SentenceSegmenter.kt):
   • Robust Spanish sentence boundary segmentation handling inverted question/exclamation marks (¿, ¡) and abbreviation boundary preservation (Dr., Sr., Dra., etc.).
3. Interlinear Aligner (InterlinearAligner.kt):
   • On-demand literal 1-to-1 word gloss generation for freeform texts derived from dictionary entries, clitic deconstructions, and false cognate disambiguations.
4. Token Inspector (TokenInspector.kt):
   • Deep inspection builder linking tokens to cognate rules, clitic surgery, and learner vault encounter logging.
5. Session Sentence Builder (SessionSentenceBuilder.kt):
   • Builds enriched sentences for both curated lessons and custom passages.
6. Reader Session Engine (ReaderSessionEngine.kt):
   • Pure Kotlin state machine coordinating sentence navigation (next, previous, jump), mode switching, interlinear and translation toggles, token selection/deselection, vault status updates, and session summary metrics.
7. Unit Tests:
   • SentenceSegmenterTest: Spanish sentence segmentation, inverted marks, and abbreviations.
   • InterlinearAlignerTest: 1-to-1 literal gloss alignment, false cognate prioritization, clitic compounds.
   • ReaderSessionEngineTest: Curated and custom text loading, navigation, toggles, token inspection, encounter logging, and vault actions.

Verification

• bazelisk run //:format and bazelisk run //:format.check passed.
• bazelisk build --config=lint //... (Detekt) passed with 0 issues.
• bazelisk test //... passed all 25 tests across the workspace.