More efficient `Chunk` and `NonEmptyChunk` implementations (#1060)

Misc fixes related to JSON schemas and syntax (#1049)

* Shrink too long error messages when validating enums

* Fix JSON schemas for some primitive values

* Memorize JSON schema values for derived codecs to avoid recalculations

* Fix regex patterns for "date-time" and "time" JSON schema types

* Test cleanup

* Fix runtime overhead for `toJson...` and `fromJson...` syntax

* Formatting

* Fix surrogate char filtering in tests

Fix TypeId hashCode/equals inconsistency for opaque type aliases (#1022)

Fix inconsistent TypeIds (#1012)

* Fix TypeId representation

* Fix another bug

* Fix

* Fix issue with tuples

* Fixes and tests for Tuple

Reduce bytecode generate by TypeId for common use cases (#1004)

* Add TypeId.nominal overload to reduce bytecode size

* Use predefined vals for common TypeDefKind cases

* Apply review comment

Filter common base types from TypeDefKind to reduce generated bytecode (

#1000)

* Filter common base types from TypeDefKind to reduce generated bytecode

Exclude Product, Equals, Serializable, Mirror.*, and Mirror.Sum from
the base types list in TypeDefKind. These standard library types are
automatically synthesized and generate substantial bytecode in macro
expansions, contributing to "Method Too Large" errors for complex ADTs.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* fmt

* Add tests + fix

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

Switch `Patch`, `DynamicPatch`, `JsonPatch`, `Differ`. `JsonDiffer`, …

…`LSC` to use `Chunk` instead of `Vector` (#991)

fix(schema): restore structural type derivation support for Scala 3 (#…

…962)

* fix(schema): restore structural type derivation support for Scala 3

Restores structural type schema derivation that was accidentally removed
in PR #800 (TypeId migration). Pure structural types like { def name: String }
can now derive schemas again.

Changes:
- Added isStructuralType() - detects refinement types
- Added getStructuralMembers() - extracts field names and types
- Added structuralFieldOffset() - calculates register offsets
- Added deriveSchemaForStructuralType() - entry point with JVM reflection check
- Added deriveSchemaForPureStructuralType() - full schema generation
- Added structural type branch in deriveSchema() before isNonAbstractScalaClass
- Enabled PureStructuralTypeSpec tests (4 tests)

Structural types require JVM reflection for deconstruction via getMethod/invoke.

* feat(schema): add Scala 2 structural type support

- Add RefinedType handling to Scala 2 TypeIdMacros for structural types
- Add structural type schema derivation to Scala 2 SchemaCompanionVersionSpecific
- Add 4 structural type tests mirroring Scala 3 PureStructuralTypeSpec

This restores structural type derivation for Scala 2, which was missing
after the TypeId migration in PR #800.

* Fix merging error

---------

Co-authored-by: Andriy Plokhotnyuk <plokhotnyuk@gmail.com>

Clean up Json.scala API for consistency and simplicity (#841)

* Clean up Json.scala API for consistency and simplicity

API changes:
- Rename apply(index) to get(index) for consistency with get(key)
- Rename encode → print, encodeToBytes → printBytes, encodeToChunk → printChunk
- Remove decode alias (use parse instead)
- Change Array.value, fields, elements return types from Vector to Chunk
- Delete shorthand constructors (str, number, bool, obj, arr) - use case classes directly
- Make jsonCodec implicit in Json companion
- Add toBigDecimalOption to Json.Number
- Consolidate MergeStrategy (Auto/Deep/Concat handled by default case)
- Make Json.toDynamicValue private (instance method still available)

Test updates:
- Update all test files to use new API patterns
- Replace Vector comparisons with Chunk for elements/fields

* refactor(json): Redesign MergeStrategy for consistency and simplicity

- Create JsonType.scala with sealed trait representing JSON value types
- Add Json#jsonType to all Json subtypes (Object, Array, String, Number, Boolean, Null)
- Refactor MergeStrategy to extend (DynamicOptic, Json, Json) => Json
- Add recurse(path, jsonType) method for recursion control decisions
- Remove redundant MergeStrategy.Deep (now covered by Auto)
- Reimplement MergeStrategy.Concat to recurse only into objects
- Replace merge implementation with symmetric mergeByKey/mergeByIndex approach

BREAKING CHANGE: Arrays now merge by index with Auto strategy (right wins on
shared indices, union of indices preserved). Use MergeStrategy.Concat for the
previous concatenation behavior.

Closes #841

* Unify Json type operations with path-dependent types

Add type members to JsonType for compile-time type safety:
- type Type <: Json: maps to corresponding Json subtype
- type Unwrap: maps to underlying Scala value type

Make JsonType extend (Json => Boolean) so it can be used directly
as a predicate for filtering operations.

Add unified methods to Json trait:
- is(jsonType): Boolean - type testing
- as(jsonType): Option[jsonType.Type] - type narrowing
- unwrap(jsonType): Option[jsonType.Unwrap] - value extraction

Refactor query methods:
- query(Json => Boolean) - filter by value predicate
- queryPath(DynamicOptic => Boolean) - filter by path predicate
- queryBoth((DynamicOptic, Json) => Boolean) - filter by both

Remove redundant type-specific methods:
- isObject, isArray, isString, isNumber, isBoolean, isNull
- asObject, asArray, asString, asNumber, asBoolean, asNull
- stringValue, numberValue, booleanValue

Update JsonSelection to use JsonType predicates for filtering.
Update all tests and documentation.

Amp-Thread-ID: https://ampcode.com/threads/T-019bf62a-58e0-7598-8bdc-4e2a1084811d
Co-authored-by: Amp <amp@ampcode.com>

* Add comprehensive tests for unified Json type operations

Add tests for:
- JsonType.apply as predicate function (Json => Boolean)
- query(Json => Boolean) with JsonType and custom predicates
- queryPath(DynamicOptic => Boolean) for path-based filtering
- queryBoth((DynamicOptic, Json) => Boolean) combining predicates
- Complete unwrap coverage for all JsonType cases

Ensures full coverage of new path-dependent type methods.

Amp-Thread-ID: https://ampcode.com/threads/T-019bf62a-58e0-7598-8bdc-4e2a1084811d
Co-authored-by: Amp <amp@ampcode.com>

* refactor(json): unify selection/query API with select, prune, retain

* Add partition/partitionPath/partitionBoth for API consistency

Align partition with the prune/retain/query trio pattern:
- partition(p: Json => Boolean) - value-only predicate
- partitionPath(p: DynamicOptic => Boolean) - path-only predicate
- partitionBoth(p: (DynamicOptic, Json) => Boolean) - both path and value

This makes partition consistent with prune, retain, and query which all
follow the foo/fooPath/fooBoth naming convention.

* Add fluent API methods to JsonSelection

Implement convenience methods on JsonSelection that delegate to underlying
Json methods, enabling fluent chaining without manual map(_.method) calls.

Unary methods (infallible):
- Normalization: sortKeys, dropNulls, dropEmpty, normalize
- Path ops: modify, set, delete, insert
- Transformation: transformUp, transformDown, transformKeys
- Pruning: prune, prunePath, pruneBoth
- Retention: retain, retainPath, retainBoth
- Projection: project

Binary method:
- merge(that, strategy): cartesian product semantics (N × M results)

Fallible mutation methods:
- modifyOrFail, setOrFail, deleteOrFail, insertOrFail

Type-directed extraction:
- as(jsonType), asAll(jsonType): narrow to specific JSON types
- unwrap(jsonType), unwrapAll(jsonType): extract underlying Scala values

Add comprehensive tests in JsonSelectionSpec covering all new methods.

---------

Co-authored-by: Amp <amp@ampcode.com>

Fix #768 by removing package-only visibility for OpticsHolder (#769)