How do their APIs compare?

Comparing "APIs" across these two projects is a bit apples-to-oranges since they serve very different purposes — one is a mobile database SDK, the other is a YouTube client app — but here's how they compare based on what's in the retrieved data:

realm/realm-java: a public library API

Realm-java exposes a deliberate, documented public API meant for third-party app developers to build on. This is evident from:

• Formal Javadoc conventions enforced via contribution guidelines — every public class/method needs Javadoc with specific rules for @param, @throws, @deprecated, @see, verb-phrase descriptions, etc. [CONTRIBUTING.md:L153-L196]
• Annotation-driven schema API — things like @Index for marking searchable fields [realm-annotations/src/main/java/io/realm/annotations/Index.java:L1-L35] and RealmNamingPolicy for controlling how Java field names map to internal Realm file names (with policies like CAMEL_CASE, PASCAL_CASE, LOWER_CASE_WITH_UNDERSCORES) [realm-annotations/src/main/java/io/realm/annotations/RealmNamingPolicy.java:L81-L120, realm/realm-annotations-processor/src/main/java/io/realm/processor/Utils.kt:L556-L595].
• Kotlin extension functions layered on top of the core Java API to give idiomatic Kotlin call sites, e.g. RealmQuery<T>.oneOf(...) which wraps the underlying in(...) query method [realm/kotlin-extensions/src/main/kotlin/io/realm/kotlin/RealmQueryExtensions.kt:L1-L40].
• An annotation processor (TypeMirrors.kt, Utils.kt) that generates proxy classes and interfaces from model classes at compile time, which is how Realm turns plain Java/Kotlin objects into persisted, queryable database objects [realm/realm-annotations-processor/src/main/java/io/realm/processor/TypeMirrors.kt:L1-L33].

This is a classic SDK-style API surface: stable, versioned, documented, annotation-based, with compile-time code generation and query builders (RealmQuery) as the primary interaction points for consumers.

yuliskov/smarttube: internal app-only interfaces

Smarttube's "APIs" in the retrieved data are internal application interfaces, not public library contracts meant for external consumption:

• SearchTagsProvider — a small internal callback interface for search-tag lookups, with a ResultsCallback for async results [common/src/main/java/com/liskovsoft/smartyoutubetv2/common/app/models/search/SearchTagsProvider.java:L7-L13], implemented by MediaServiceSearchTagProvider which wraps an RxJava Observable-based content service [common/src/main/java/com/liskovsoft/smartyoutubetv2/common/app/models/search/MediaServiceSearchTagProvider.java:L13-L41].
• Video model class with custom equals/hashCode logic tailored to UI adapter behavior (e.g. avoiding duplicate list items) rather than general-purpose correctness [common/src/main/java/com/liskovsoft/smartyoutubetv2/common/app/models/data/Video.java:L236-L275], plus mutation-style sync(...) methods that merge in fresh metadata from YouTube responses [common/src/main/java/com/liskovsoft/smartyoutubetv2/common/app/models/data/Video.java:L716-L755].
• Error-handling logic embedded directly in a playback controller (ErrorFixerController) that pattern-matches on raw HTTP-style error strings ("Response code: 429", "Exception in CronetUrlRequest", etc.) to decide runtime remediation like disabling subtitles or switching data sources [common/src/main/java/com/liskovsoft/smartyoutubetv2/common/app/models/playback/controllers/ErrorFixerController.java:L142-L181].
• The app also communicates with external third-party services (SponsorBlock, DeArrow, Return YouTube Dislike) as documented in its privacy policy, but these are consumed, not exposed, by the app [PRIVACY.md:L21-L26].

Key differences

| Aspect | real