ApkPy 1.3.0 — Data Core

Version 1.3.0 turns ApkPy's low-level SQLite support into a declarative data layer suitable for real local application state.

notes = db.model(
    "notes",
    fields={
        "id": db.integer(primary_key=True, auto_increment=True),
        "title": db.text(required=True, max_length=120),
        "favorite": db.boolean(default=False),
        "metadata": db.json(optional=True),
        "updated_at": db.datetime(default=db.now()),
    },
    indexes=[
        db.index("idx_notes_favorite_updated", ["favorite", "updated_at"]),
    ],
)

schema = db.schema("notes_app", version=1, models=[notes])

What changed

• typed integer, real, text, boolean, datetime, json and blob fields;
• defaults, nullability, primary keys, uniqueness, numeric bounds, text length, choices and composite indexes;
• asynchronous insert, batch insert, get, find, update, delete and count;
• parameterized filters, ordering, limit and offset pagination;
• atomic multi-model transactions;
• explicit consecutive migrations and stored schema hashes;
• private backup and restore around destructive migration paths;
• generated SQLiteOpenHelper, one ordered executor and per-model Java repositories, without Room or a Python runtime.

Knowledge Vault

The 1.3.0 example is an English two-screen notes app. Library performs indexed search, favorite filtering and paged reads. Data Lab creates, updates, deletes and archives records in an atomic batch. Initial records are seeded through a transaction and the schema declares its v1-to-v2 path.

It also demonstrates the important UI states: loading, no matching data, an explicit database error, retry and the end of pagination.

Create and reload

def created(note_id):
    status.set_value("Saved note #" + str(note_id))
    reload_library()

notes.insert(
    {
        "title": title_input.get_value(),
        "content": content_input.get_value(),
        "favorite": favorite_input.get_value(),
        "priority": priority_input.get_value(),
    },
    on_result=created,
    on_error=database_failed,
)

Seed several records atomically

def seed_library(tx):
    tx.insert_many(notes, initial_notes)
    return tx.count(notes)

db.transaction(
    run=seed_library,
    on_result=lambda total: status.set_value(str(total) + " notes ready"),
    on_error=database_failed,
)

Migration safety

The data layer refuses to guess how an old database should change. A schema change at the same version is an error; a missing migration step is an error; a downgrade is refused. Destructive paths require an explicit flag and create a recoverable private backup first.

migration_1_2 = db.migration(
    1,
    2,
    operations=[
        db.add_column(notes, "favorite", db.boolean(default=False)),
        db.create_index(
            notes,
            "idx_notes_favorite_updated",
            ["favorite", "updated_at"],
        ),
    ],
)

Native output and validation

The release candidate has been checked against the existing feed, rich-content and Gradle-generation suites. Its generated Knowledge Vault project compiles as a real Android debug APK. The generated source uses SQLiteStatement, a single data executor, main-thread callbacks, repository-specific queries and sequential migration code.

Validation	Result
typed declarations and constraints	passed
CRUD, JSON, dates, ordering and pagination	passed
batch and transaction rollback	passed
v1→v2 data preservation	passed
failed destructive migration recovery	passed
generated Android Gradle compilation	passed
projects without typed models	no Data Core output

The Android benchmark is documented separately because it measures a small in-memory UI app, not SQLite query throughput.

Read the complete Data Core guide for the API contract, examples, migration rules and current limits.

Not in this release

Relations, observable queries, automatic offline-first synchronization, conflict resolution and persistent background jobs remain outside 1.3.0. The low-level SQL API remains compatible.