bible_io

Added

• Version 1 of the Bible content contract, with an explicit schemaVersion, deterministic bookOrder, stable edition IDs, and lossless JSON-compatible annotations on Bible, book, chapter, verse, and metadata values
• Strict, path-aware decoded-content validation through BibleDataFormatError, plus an explicit permissive policy for intentionally skeletal data
• Background JSON processing where isolates are available, structured reading/processing/completion progress, and eager, lazy, or disabled search index policies
• Edition-aware BibleVerseKey values and JSON-restorable BibleLocation values for bookmarks, highlights, notes, and reading progress
• Result pagination metadata, snippet-relative highlight ranges, and explicit snippet bounds for UI search rendering
• Platform-neutral loading from Flutter-style asset bundles, UTF-8 bytes, decoded JSON maps, and JSON strings, while retaining asynchronous file loading on dart:io platforms
• Translation metadata, BibleSource and BibleCatalog helpers, stable BibleLocation values, and display-ready search hits with snippets and match ranges
• Full bible_io_references 1.1 integration, including typed non-throwing parsing, edition-aware book aliases, multilingual language preferences, rich passage expressions, cross-book ranges, localized formatting, and OSIS/USFM interoperability
• Support for short, full-name, OSIS, and USFM book identifiers in Bible JSON
• A conventional example/bible_io_example.dart package example

Changed

• Legacy data without bookOrder now falls back to canonical book order rather than JSON map insertion order; chapter and verse ordering follows declared numeric identifiers
• Search now applies canonical Unicode normalization by default, offers opt-in diacritic folding, and supports substrings in Chinese, Japanese, Thai, Lao, Khmer, and Myanmar text that does not delimit every word with spaces
• Search index construction is configurable for startup latency and memory, search/page collection stops without constructing discarded hit objects, and lazy indexes can be prewarmed asynchronously on isolate-capable targets
• Catalog source IDs are validated and indexed as unique edition identities; unknown metadata fields and nested source provenance survive round trips
• Book, chapter, verse, search-result, catalog-result, and passage-result collections are defensively copied and exposed as immutable values
• Chapter and verse access now uses declared numbers instead of list offsets, safely supporting sparse and out-of-order source JSON
• Chapter navigation follows the edition's actual declared chapter numbers and book order
• Search value types now live in a dedicated module, removing the previous Bible/extensions import cycle
• Fuzzy search and word statistics now use consistent Unicode tokenization and Unicode scalar-aware edit distances
• Internal imports and the public re-export now use the conventional bible_io_references.dart entry point
• The minimum Dart SDK was widened from 3.10.7 to 3.4.0 after removing unnecessary newer-language syntax and adding minimum-SDK CI coverage
• Direct and transitive dependencies were refreshed to their latest resolvable versions

Fixed

• Canonically equivalent NFC/NFD text now matches consistently, while match ranges continue to point into the original UTF-16 verse text
• Cropped snippets now expose their origin and snippet-relative ranges, and avoid splitting grapheme clusters
• All/any indexed search no longer misses query substrings in commonly unspaced writing systems
• Fuzzy search now applies the same unspaced-script substring behavior and consistently rejects negative result limits
• Malformed Bible JSON now reports stable typed errors and precise paths instead of leaking TypeError, cast, or integer parsing failures
• Translation identity, description, custom annotation fields, and explicit canon order are no longer lost during serialization
• Parsed cross-book ranges now honor an edition's explicit non-default book order, and conflicting loaded book aliases are rejected as ambiguous
• Unknown nested BibleSource fields stay nested instead of being flattened into metadata or lost on key collisions
• Public model lists and annotation trees can no longer be mutated through shared references
• Bible metadata now preserves nested source information across JSON round-trips
• Nested language catalog maps no longer create bogus source entries
• Fuzzy search now handles blank queries, non-positive result limits, invalid distance limits, and non-ASCII text correctly
• Search match ranges remain accurate for case-insensitive Unicode matching
• Duplicate books, chapters, and verses, invalid parent relationships, and non-positive declared numbers are rejected instead of corrupting indexes
• Result helpers no longer include captured stack traces in user-facing error strings

Documentation and tests

• Documented the version 1 content schema, migration path, edition-aware UI state, validation policies, search normalization, and index lifecycle
• Clarified the intentional AGPL-3.0-or-later code license and the independent copyright and licensing requirements of Bible translation content
• Added CI checks for formatting, analysis, tests, coverage, web compilation, and the pub.dev archive
• Expanded the README with platform-neutral loading, catalogs, metadata, multilingual parsing, rich passages, OSIS/USFM, locations, and search-hit examples
• Added regression coverage for model invariants, metadata/catalog round-tripping, Unicode fuzzy search, cross-book and rich-passage resolution, language detection, and machine identifiers
• Replaced the placeholder package test and reduced repeated loading of the large KJV fixture across test groups
• Deprecated the legacy lib/bible_example.dart entry point; it remains available for compatibility and will be removed in a future major release

Bug-fix note

• Public model collections were always intended to be immutable. The exposed mutation paths were fixed; derive changed values with constructors or copyWith() rather than mutating Bible.books, Book.chapters, or Chapter.verses.