hermes-client-python 1.8.43

Async Python client for Hermes search server

Hermes Client

Async Python client for Hermes search server.

Installation

pip install hermes-client-python

Quick Start

import asyncio
from hermes_client_python import HermesClient

async def main():
    async with HermesClient("localhost:50051") as client:
        # Create index with SDL schema
        await client.create_index("articles", '''
            index articles {
                field title: text [indexed, stored]
                field body: text [indexed, stored]
                field score: f64 [stored]
            }
        ''')

        # Index documents
        await client.index_documents("articles", [
            {"title": "Hello World", "body": "First article", "score": 1.5},
            {"title": "Goodbye World", "body": "Last article", "score": 2.0},
        ])

        # Commit changes
        await client.commit("articles")

        # Search
        results = await client.search("articles", term=("title", "hello"), limit=10)
        for hit in results.hits:
            print(f"Doc {hit.doc_id}: score={hit.score}, fields={hit.fields}")

        # Get document by ID
        doc = await client.get_document("articles", 0)
        print(doc.fields)

        # Delete index
        await client.delete_index("articles")

asyncio.run(main())

API Reference

HermesClient

client = HermesClient(address="localhost:50051")

Connection

# Using context manager (recommended)
async with HermesClient("localhost:50051") as client:
    ...

# Manual connection
client = HermesClient("localhost:50051")
await client.connect()
# ... use client ...
await client.close()

Index Management

# Create index with SDL schema
await client.create_index("myindex", '''
    index myindex {
        field title: text [indexed, stored]
        field body: text [indexed, stored]
    }
''')

# Create index with JSON schema
await client.create_index("myindex", '''
{
    "fields": [
        {"name": "title", "type": "text", "indexed": true, "stored": true},
        {"name": "body", "type": "text", "indexed": true, "stored": true}
    ]
}
''')

# Get index info
info = await client.get_index_info("myindex")
print(f"Documents: {info.num_docs}, Segments: {info.num_segments}")

# Delete index
await client.delete_index("myindex")

Document Indexing

# Index multiple documents (batch)
indexed, errors = await client.index_documents("myindex", [
    {"title": "Doc 1", "body": "Content 1"},
    {"title": "Doc 2", "body": "Content 2"},
])

# Index single document
await client.index_document("myindex", {"title": "Doc", "body": "Content"})

# Stream documents (for large datasets)
async def doc_generator():
    for i in range(10000):
        yield {"title": f"Doc {i}", "body": f"Content {i}"}

count = await client.index_documents_stream("myindex", doc_generator())

# Commit changes (required to make documents searchable)
num_docs = await client.commit("myindex")

# Force merge segments (for optimization)
num_segments = await client.force_merge("myindex")

Searching

# Term query
results = await client.search("myindex", term=("title", "hello"), limit=10)

# Boolean query
results = await client.search("myindex", boolean={
    "must": [("title", "hello")],
    "should": [("body", "world")],
    "must_not": [("title", "spam")],
})

# With pagination
results = await client.search("myindex", term=("title", "hello"), limit=10, offset=20)

# With field loading
results = await client.search(
    "myindex",
    term=("title", "hello"),
    fields_to_load=["title", "body"]
)

# Access results
for hit in results.hits:
    print(f"Doc {hit.doc_id}: {hit.score}")
    print(f"  Title: {hit.fields.get('title')}")

print(f"Total hits: {results.total_hits}")
print(f"Took: {results.took_ms}ms")

Document Retrieval

# Get document by ID
doc = await client.get_document("myindex", doc_id=42)
if doc:
    print(doc.fields["title"])

Field Types

Type	Python Type	Description
text	str	Full-text searchable string
u64	int (>= 0)	Unsigned 64-bit integer
i64	int	Signed 64-bit integer
f64	float	64-bit floating point
bytes	bytes	Binary data
json	dict / list	JSON object (auto-serialized)
dense_vector	list[float]	Dense vector for semantic search
sparse_vector	dict	Sparse vector with indices and values

Error Handling

import grpc

try:
    await client.search("nonexistent", term=("field", "value"))
except grpc.RpcError as e:
    if e.code() == grpc.StatusCode.NOT_FOUND:
        print("Index not found")
    else:
        raise

Development

Generate protobuf stubs:

pip install grpcio-tools
python generate_proto.py

License

MIT

Timeouts / deadlines

Every RPC accepts an optional timeout (seconds) which sets a gRPC deadline; a client-wide default can be set in the constructor. On expiry the call raises grpc.aio.AioRpcError with DEADLINE_EXCEEDED.

client = HermesClient("localhost:50051", default_timeout=5.0)
results = await client.search("articles", query={...}, timeout=0.5)  # per-call override
await client.force_merge("articles", timeout=3600)  # long op, long deadline