mokaccino 0.8.0

Percolator - Match transient documents against a set of queries.

About Mokaccino

This is a Python binding for https://crates.io/crates/mokaccino.

Mokaccino is a Percolator.

A Percolator is a component that allows the matching of a stream of documents (for instance representing events) against a set of queries (representing specific interests in events).

Install/Usage

Install via pip compatible tools as usual.

Example Usage (taken from a test):

def test_percolator():
    p = Percolator()
    assert p is not None
    qids = [
        p.add_query(Query.from_kv("name", "sausage")), # Use automated query IDs
        p.add_query_id(Query.from_kprefix("name", "amaz"), 42), # Or provide your own application query IDs
        p.add_query(Query.from_kgt("price", 12)),
        p.add_query(Query.from_kv("name", "sausage") | Query.from_kgt("price", 12)),
    ]

    assert p.percolate_list(Document()) == []
    assert p.percolate_list(Document().with_value("name", "burger")) == []
    assert p.percolate_list(Document().with_value("name", "sausage")) == [qids[0], qids[3]]
    assert p.percolate_list(Document().with_value("name", "amaz")) == [qids[1]]
    assert p.percolate_list(Document().with_value("name", "amazing")) == [qids[1]]
    assert p.percolate_list(Document().with_value("name", "amazon")) == [qids[1]]
    assert p.percolate_list(Document().with_value("price", "12")) == []
    assert p.percolate_list(Document().with_value("price", "13")) == [qids[2], qids[3]]
    assert p.percolate_list(
        Document().with_value("price", "13").with_value("name", "amazed")
    ) == [qids[1], qids[2], qids[3]]

Lat,Long,Distance Geography support

You can match queries based on a location within a certain distance of a point.

    # Matches documents whose "location" field is within 1000m of the given lat/lng
    q = Query.from_latlng_within("location", 37.7749, -122.4194, 1000)

    # Alternatively, you can use query parsing too:
    q = Query.parse("location LLWITHIN 37.7749,-122.4194,1000")

The document field "location" can be either:

• A valid H3 cell index (hexadecimal string).
• A comma-separated string "lat,lng" (e.g. "37.7749,-122.4194").

H3 Geography support

You can match documents based on their location within an H3 cell.

    # Matches documents whose "location" field is inside the given H3 cell
    # (or inside any child cell)
    q = Query.from_h3in("location", "85283473fffffff")

    # Or using parsing:
    q = Query.parse("location H3IN 85283473fffffff")

Using query parsing.

If you plan to have free formed queries, you can use query parsing to build queries:

   qids = [
        p.add_query(Query.parse("name:sausage")),
        p.add_query(Query.parse("name:amaz*")),
        p.add_query(Query.parse("price>12")),
        p.add_query(Query.parse("name:sausage OR price>12")),
    ]

Query parsing works as you expect, with boolean AND, OR and NOT and ( ) to work around precedence.

Non-word values can be enclosed in "s (for example field:"non word value") and the escape character is \.

More extensive documentation will be provided, but in the meanwhile, have a look at the unit tests, which cover everything you can do with this:

https://github.com/jeteve/mokaccino_py/tree/main/tests

Development

This uses uv/uvx to handle the python side of things

1. Prepare the venv

uv venv --python 3.13
uv sync --extra dev
source .venv/bin/activate

2. Compile everything using maturin

# To regenerate the mokaccino.pyi stub:
cargo run -F stub-gen --bin stub_gen

uvx maturin develop

3. Run some examples or unit tests

uv sync --extra dev
pytest
uv run examples/...

In development, loop through steps 2 and 3:

uvx maturin develop && pytest

This is developed at https://github.com/jeteve/mokaccino_py