sqlite - Chirp

data.drivers.sqlite

module

SQLite driver helpers for chirp.data.

For file-backed databases the "pool" is a small bounded set ofAsyncConnectioninstances. WAL mode (enabled per connection) lets SQLite serve many concurrent readers alongside a single…

SQLite driver helpers for chirp.data.

For file-backed databases the "pool" is a small bounded set ofAsyncConnection instances. WAL mode (enabled per connection) lets SQLite serve many concurrent readers alongside a single writer, so readers acquire any pooled connection without the app-wide write lock; write serialization is handled at the Databaselayer.

In-memory databases (sqlite:///:memory: / sqlite://) are special. A plain:memory:connection is private to whichever connection opened it, so a multi-connection pool would hand each task its own empty database. Shared-cache mode (file::memory:?cache=shared) lets connections see one logical database but uses coarse table-level locks that raiseSQLITE_LOCKED(not retried by busy_timeout) under concurrent reader/writer access. So in-memory databases use a single shared connection (pool_sizeis treated as 1) and the Database layer serializes all access — reads included — on _sqlite_lock. In-memory is a development/test convenience; concurrent-reader throughput is a file-database (WAL) property. WAL is unavailable for memory DBs (stays journal_mode=memory) so the PRAGMA is skipped there.

data.drivers.sqlite

Name	Type	Default	Description
`type`		—
`qualified_name`		—
`element_type`		—
`description`		—
`source_file`		—
`line_number`		—
`is_autodoc`		—
`autodoc_element`		—
`_autodoc_template`		—
`_autodoc_url_path`		—
`_autodoc_page_type`		—
`title`		—
`doc_content_hash`		—

Symbols on this page

function parse_sqlite_path

Extract the file path from a sqlite:// URL.

function is_memory_path

True when the parsed path refers to an in-memory database.

Covers:memory: and the empty path (sqlite://), plus any explicit file:...:memory:...shared-cache URI.

function _connect_target

Resolve the connection target for a parsed SQLite path.

Returns(target, uri, in_memory):

target — the string handed to sqlite3.connect.
uri—…

class SqlitePool

A small bounded pool of async SQLite connections.

Exposesacquire() / release()mirroring the asyncpg pool API so the Databasefacade can treat both…

async function create_pool

Create a small bounded pool of SQLite connections.

File-backed pools size byconfig.pool_size(minimum 1) with WAL mode per connection so SQLite serves concurrent…

parse_sqlite_path

function

def parse_sqlite_path(url: str) -> str

Extract the file path from a sqlite:// URL.

Parameters

Name	Type	Default	Description
`url`	`str`	—

is_memory_path

function

def is_memory_path(path: str) -> bool

True when the parsed path refers to an in-memory database.

Covers:memory: and the empty path (sqlite://), plus any explicit file:...:memory:...shared-cache URI.

Parameters

Name	Type	Default	Description
`path`	`str`	—

_connect_target

function

def _connect_target(path: str) -> tuple[str, bool, bool]

Resolve the connection target for a parsed SQLite path.

Returns(target, uri, in_memory):

target — the string handed to sqlite3.connect.
uri — whether uri=Truemust be passed (shared-cache memory DBs).
in_memory— whether this is an in-memory database (skips WAL).

In-memory databases collapse to a single shared connection (see module docstring), so a plain:memory:target is sufficient — every query runs on that one connection rather than a per-task private empty database.

Parameters

Name	Type	Default	Description
`path`	`str`	—

SqlitePool

class

A small bounded pool of async SQLite connections.

Exposesacquire() / release()mirroring the asyncpg pool API so the Databasefacade can treat both backends uniformly. Connections are never bound to a task — theDatabaselayer owns per-task assignment via the _current_conn ContextVar — so a connection handed out by acquire() is exclusively held until the matchingrelease().

In-memory databases (is_memory=True) hold exactly one shared connection so every task sees the same logical database; theDatabaselayer serializes all in-memory access (reads included) since a lone SQLite connection cannot safely take concurrent thread-pool dispatches.

create_pool

function async

async def create_pool(config: DatabaseConfig) -> SqlitePool

Create a small bounded pool of SQLite connections.

File-backed pools size byconfig.pool_size(minimum 1) with WAL mode per connection so SQLite serves concurrent readers alongside a single writer. In-memory databases collapse to a single shared connection (skipping WAL, which is unavailable for:memory:) so every task sees the same logical database; theDatabaselayer serializes their access.

Parameters

Name	Type	Default	Description
`config`	`DatabaseConfig`	—

View source · /home/runner/work/chirp/chirp/site/../src/chirp/data/drivers/sqlite.py:1