torque-http 0.5.0

Official Python client for Torque — Truespar's in-memory search engine with a Typesense v30.1 compatible API. Zero external dependencies.

torque-http

Official Python client for Torque — Truespar's in-memory search engine with a Typesense v30.1 compatible API.

• Zero external dependencies — uses only the Python standard library
• HTTP API for search, collection management, and document CRUD
• TCP binary ingest for high-throughput streaming from databases
• TQBF file upload for bulk import
• Multi-node failover with automatic retry
• Typed search parameters and response models
• Python 3.10+, MIT licensed

Installation

pip install torque-http

Quick Start

from torque_http import TorqueHttpClient, SearchParameters

client = TorqueHttpClient("http://localhost:8108", api_key="your-key")

# Search
results = client.search("products", q="laptop", query_by="title")
for hit in results.hits:
    print(hit.document["title"])

# Typed parameters
params = SearchParameters(
    q="laptop",
    query_by="title,description",
    filter_by="price:>100",
    per_page=20,
)
results = client.search("products", params)

Multi-node with failover

client = TorqueHttpClient(
    ["http://node1:8108", "http://node2:8108"],
    api_key="your-key",
    num_retries=5,
)

Collection management

schema = {
    "name": "products",
    "fields": [
        {"name": "title", "type": "string"},
        {"name": "price", "type": "float", "sort": True},
        {"name": "brand", "type": "string", "facet": True},
    ],
}
client.create_collection(schema)

client.create_document("products", {
    "id": "1", "title": "Laptop Pro", "price": 1299.99, "brand": "Acme",
})

High-throughput TCP ingest

For streaming 100K+ documents, use the binary TCP protocol:

from torque_http import TorqueIngestClient, DocumentEncoder

schema = {"fields": [
    {"name": "title", "type": "string"},
    {"name": "price", "type": "float"},
]}
encoder = DocumentEncoder(schema)

with TorqueIngestClient() as tcp:
    tcp.connect("localhost", 8109)
    tcp.start_ingest("products")
    tcp.send_batch(documents, encoder)
    tcp.commit()

TQBF bulk file import

Pre-encode a compressed binary file for maximum throughput:

from torque_http import TorqueBinaryFileWriter, DocumentEncoder

encoder = DocumentEncoder(schema)
with TorqueBinaryFileWriter("data.tqbf", encoder) as writer:
    for doc in source:
        writer.write(doc)

client.import_binary("products", "data.tqbf")

Documentation

• Torque documentation

License

MIT — see LICENSE.