morphik 1.2.6

Morphik Python Client

Morphik

A Python client for Morphik API that enables document ingestion, semantic search, and retrieval augmented generation capabilities.

🚨 Upgrading to v1.0

Breaking Change: list_documents() now returns a ListDocsResponse object instead of a list.

Quick Migration:

# Change this:
for doc in db.list_documents():
    process(doc)

# To this:
for doc in db.list_documents().documents:
    process(doc)

See CHANGELOG.md for full details and new features.

---

Installation

pip install morphik

Usage

The SDK provides both synchronous and asynchronous clients:

Synchronous Usage

from morphik import Morphik

# Initialize client - connects to localhost:8000 by default
db = Morphik()

# You can also use a direct HTTP(S) base URL for self-hosted deployments
# db = Morphik("http://morphik:8000")

# Or with authentication URI (for production)
# db = Morphik("morphik://owner_id:[e-mail]")

# Ingest a text document
doc = db.ingest_text(
    content="Your document content",
    metadata={"title": "Example Document"}
)

# Ingest a file
doc = db.ingest_file(
    file="path/to/document.pdf",
    metadata={"category": "reports"}
)

# Run a Morphik On-the-Fly document query
doc_query = db.query_document(
    file="path/to/document.pdf",
    prompt="Extract the parties and effective date.",
    ingestion_options={"ingest": True, "metadata": {"source": "contracts"}}
)
print(doc_query.structured_output)

# Retrieve relevant chunks
chunks = db.retrieve_chunks(
    query="Your search query",
    filters={"category": "reports"}
)

# Query with RAG
response = db.query(
    query="Summarize the key points in the document",
    filters={"category": "reports"}
)

print(response.completion)

# Migrate this app's documents into another Morphik deployment.
# Run this from a machine that can reach both source and target, such as
# inside a customer's VPN for on-prem targets.
result = db.migrate(target_uri="morphik://owner_id:[e-mail]", target_is_local=True)
print(result.created_count, result.skipped_count, result.failed_count)

Nested Folders & Folder Depth

# Create a nested folder (parents are auto-created server-side)
folder = db.create_folder(full_path="/projects/alpha/specs", description="Specs folder")

# Move or rename folder paths
moved = db.move_folder("/projects/alpha/specs", "/projects/archive/specs")
renamed = moved.rename("specs-v2")

# Scope queries to a path and include descendants with folder_depth=-1
chunks = folder.retrieve_chunks(query="design notes", folder_depth=-1)
docs = db.list_documents(folder_name="/projects/alpha", folder_depth=-1)

# List only the fields you need. The server reads and returns just those columns, so
# the full document text is never downloaded — fast for large corpora.
for doc in db.list_documents(fields=["metadata"]).documents:
    print(doc.external_id, doc.metadata)

Folder.full_path is exposed on folder objects, and Document.folder_path mirrors server responses for tracing scope.

Asynchronous Usage

import asyncio
from morphik.async_ import AsyncMorphik

async def main():
    # Initialize async client - connects to localhost:8000 by default
    async with AsyncMorphik() as db:

    # You can also use a direct HTTP(S) base URL for self-hosted deployments
    # async with AsyncMorphik("http://morphik:8000") as db:

    # Or with authentication URI (for production)
    # async with AsyncMorphik("morphik://owner_id:[e-mail]") as db:
        # Ingest a text document
        doc = await db.ingest_text(
            content="Your document content",
            metadata={"title": "Example Document"}
        )

        doc_query = await db.query_document(
            file="path/to/document.pdf",
            prompt="Extract the parties and effective date.",
            ingestion_options={"ingest": True, "metadata": {"source": "contracts"}}
        )
        print(doc_query.structured_output)

        # Query with RAG
        response = await db.query(
            query="Summarize the key points in the document",
        )

        print(response.completion)

# Run the async function
asyncio.run(main())

Features

• Document ingestion (text, files, directories)
• Semantic search and retrieval
• Retrieval-augmented generation (RAG)
• Morphik On-the-Fly document querying with optional ingestion follow-up
• Multi-user and multi-folder scoping
• Metadata filtering
• Document management

Development

Running Tests

To run the tests, first install the development dependencies:

pip install -r test_requirements.txt

Then run the tests:

# Run all tests (requires a running Morphik server)
pytest morphik/tests/ -v

# Run specific test modules
pytest morphik/tests/test_sync.py -v
pytest morphik/tests/test_async.py -v

# Skip tests if you don't have a running server
SKIP_LIVE_TESTS=1 pytest morphik/tests/ -v

# Specify a custom server URL for tests
MORPHIK_TEST_URL=http://custom-server:8000 pytest morphik/tests/ -v

Example Usage Script

The SDK comes with an example script that demonstrates basic usage:

# Run synchronous example
python -m morphik.tests.example_usage

# Run asynchronous example
python -m morphik.tests.example_usage --async

The example script demonstrates:

• Text and file ingestion
• Creating folders and user scopes
• Retrieving chunks and documents
• Generating completions using RAG
• Batch operations and cleanup