Skip to content

mentat-is/gulp-sdk

Repository files navigation

Gulp Python SDK

Async Python SDK for the Gulp document analysis and collaboration platform.

Features

  • Fully Async — Built on httpx and asyncio for high-performance, non-blocking I/O
  • Complete API Coverage — All major REST endpoints (operations, documents, ingestion, queries, users, collaboration)
  • WebSocket Support — Real-time ingestion progress, query results, and collaborative updates
  • Type-Safe — Full type hints, Pydantic models, and static typing support
  • Error Handling — Comprehensive exception hierarchy with HTTP status codes and response data
  • Pagination — Async iterators for large result sets
  • Retry Logic — Automatic exponential backoff on transient failures

Installation

pip install gulp-sdk

Quick Start

If you prefer a dedicated guide, see docs/quickstart.md.

import asyncio
from gulp_sdk import GulpClient

async def main():
    # Connect to Gulp server
    async with GulpClient("http://localhost:8080") as client:
        # Login
        session = await client.auth.login("user@example.com", "password")
        print(f"Logged in: {session.user_id}")

        # Create operation
        op = await client.operations.create(
            name="My Investigation",
            description="Analyze event logs"
        )
        print(f"Created operation: {op.id}")

        # Get current user
        user = await client.users.get_current()
        print(f"Current user: {user.display_name}")

asyncio.run(main())

API Reference

Documentation site

This repository includes official docs for local or remote hosting:

  • Local preview: mkdocs serve
  • Build HTML: mkdocs build

Run docs locally

pip install -r requirements-docs.txt
mkdocs serve

Documentation pages

  • Quick start: docs/quickstart.md
  • API reference: docs/api_reference.md
  • Examples: docs/examples/ (scripts)

Authentication

# Login with credentials
session = await client.auth.login("user", "password")

# Token is automatically stored
# Logout
await client.auth.logout()

Operations

# Create operation
op = await client.operations.create("Name", "Description")

# Get operation
op = await client.operations.get(op.id)

Documents

# Get document
doc = await client.documents.get(operation_id, document_id)

# Query documents (with async iteration)
async for doc in client.documents.list(operation_id):
    print(doc.content[:100])

Ingestion

# Ingest file
job = await client.ingest.file(operation_id, plugin="json", file_path="/path/to/file.json")

# Ingest raw data
job = await client.ingest.raw(operation_id, plugin="json", data={"key": "value"})

# Monitor with WebSocket
async for progress in client.ingest.stream(operation_id, req_id=job.req_id):
    print(f"Progress: {progress.percent}%")

Collaboration

# Add note
note = await client.collab.create_note(operation_id, document_id, "Important finding")

# Create link between documents
link = await client.collab.create_link(doc_id_from, doc_id_to, "related_to")

WebSocket Real-Time Updates

Auto-Managed Mode

async with GulpClient("http://localhost:8080", ws_auto_connect=True) as client:
    # WebSocket automatically connected
    
    # Subscribe to document updates
    await client.websocket.subscribe(operation_id)
    
    # Receive messages
    async for message in client.websocket:
        print(f"Update: {message.type}{message.data}")

Manual Mode

async with GulpClient("http://localhost:8080") as client:
    async with client.websocket() as ws:
        # Authenticate and subscribe
        await ws.subscribe(operation_id, req_id="ingest-123")
        
        # Receive real-time updates
        async for message in ws:
            if message.type == "WSDATA_INGEST_RAW_PROGRESS":
                print(f"Ingestion: {message.data.percent}%")
            elif message.type == "WSDATA_QUERY_DONE":
                print(f"Query complete: {len(message.data.documents)} results")

Request status: websocket vs polling

For async operations, realtime websocket monitoring is recommended; polling is a fallback.

  • WebSocket pattern: use wait_for_request_stats(client, req_id, timeout, ws_callback=...).
  • Polling: client.plugins.request_get(req_id) in a loop.

See docs/api_reference.md for details and examples.

For advanced websocket note/QUERY_DONE tracking, consult tests/integration/test_stress.py.

Error Handling

from gulp_sdk import (
    GulpClient,
    AuthenticationError,
    PermissionError,
    NotFoundError,
    ValidationError,
    GulpSDKError,
)

async with GulpClient("http://localhost:8080") as client:
    try:
        await client.auth.login("user", "pass")
    except AuthenticationError as e:
        print(f"Login failed: {e.message}")
    except GulpSDKError as e:
        print(f"SDK error: {e.message} (status: {e.status_code})")

Configuration

Environment variables are used by tests and example scripts (fixtures) to parameterize connection settings. The SDK core methods in src/gulp_sdk accept explicit arguments and do not read env vars directly.

GULP_BASE_URL=http://localhost:8080        # Server URL (default: localhost:8080)
GULP_TEST_USER=admin                       # Default test user for integration tests
GULP_TEST_PASSWORD=admin                   # Default test password for integration tests
GULP_TEST_TOKEN=                           # Optional token for test auth
GULP_REQUEST_TIMEOUT=30                    # HTTP timeout in seconds (default: 30)
GULP_WS_TIMEOUT=300                        # WebSocket timeout (default: 300)
GULP_LOG_LEVEL=INFO                        # Logging level (default: INFO)

Or programmatically:

client = GulpClient(
    base_url="http://localhost:8080",
    timeout=30.0,
    ws_auto_connect=True,
)

Examples

See docs/examples/ for complete working examples:

For additional workflows, consult the integration tests in tests/integration/ and the main gulp docs for plugin-specific ingestion and query patterns.

Testing

Run tests:

# Unit tests (no dependencies)
pytest -v -s -x tests/unit

# minimal integration tests, requires live Gulp server on localhost:8080
pytest -v -s -x tests/integration  

to run full integration test suite, look at gulp gulp testing documentation

About

Official Python SDK for Gulp

Resources

Stars

Watchers

Forks

Packages

 
 
 

Contributors

Languages