[Feature][Connector] Add HubSpot Source Connector

[davidzollo]

Background

HubSpot is a leading marketing automation and CRM platform used by over 200,000 customers worldwide, particularly popular among small to mid-sized businesses. It provides comprehensive tools for marketing, sales, customer service, and content management.

Currently, SeaTunnel lacks native support for HubSpot as a data source, preventing users from integrating CRM and marketing data with their data warehouses and analytics platforms.

Motivation

• SMB Market Leader: HubSpot is the dominant choice for small and medium-sized businesses globally
• Marketing Automation: Critical source for marketing campaign data, lead tracking, and conversion analytics
• Multi-Object Integration: Need to sync multiple HubSpot objects (Contacts, Companies, Deals, Tickets, etc.) simultaneously
• API-Only Access: HubSpot uses REST API exclusively - no JDBC or SQL interface available
• Data-Driven Marketing: Organizations need to analyze marketing performance, customer journeys, and ROI

Proposed Solution

Implement a dedicated HubSpot Source connector using HubSpot REST API v3 with multi-object support following SeaTunnel's standard architecture:

Core Features

1. Multi-Object Support

   • Support multiple HubSpot objects in single configuration
   • Object discovery and filtering
   • Per-object configuration (properties, filters, associations)
   • Example:

   source {
     HubSpot {
       # Multi-object configuration
       object-configs = [
         {
           object_type = "contacts"
           properties = ["firstname", "lastname", "email", "company"]
           filter_groups = [{filters = [{property_name = "lifecyclestage", operator = "EQ", value = "customer"}]}]
         },
         {
           object_type = "companies"
           fetch_all_properties = true
           include_associations = true
         },
         {
           object_type = "deals"
           properties = ["dealname", "amount", "dealstage"]
         }
       ]
     }
   }

2. CRM Objects Support

   • Standard Objects: Contacts, Companies, Deals, Tickets, Products, Line Items
   • Custom Objects: User-defined objects created in HubSpot
   • Activities: Emails, Calls, Meetings, Tasks, Notes
   • Engagement Data: Email opens, clicks, form submissions, page views

3. Marketing Data

   • Campaigns: Email campaigns, ad campaigns, social media campaigns
   • Forms: Form submissions and field values
   • Landing Pages: Page analytics and conversion data
   • Lists: Contact lists and segmentation
   • Workflows: Automation workflow execution data

4. Data Extraction Modes

   • Full Snapshot: Complete object/entity extraction
   • Incremental: Based on lastModifiedDate or createDate
   • Association-Based: Extract related objects (e.g., Contacts with their Deals)

5. Authentication

   • Private App Access Token: Recommended for server-to-server integration
   • OAuth 2.0: For user-context integrations
   • API Key (Legacy): Support for existing integrations

Configuration Examples

source {
  HubSpot {
    # Authentication
    auth_type = "private_app"
    access_token = "pat-na1-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
    
    # Multi-object configuration
    object-configs = [
      {
        # Contacts configuration
        object_type = "contacts"
        extraction_mode = "incremental"
        incremental_field = "lastmodifieddate"
        start_date = "2024-01-01T00:00:00Z"
        
        # Properties to fetch
        properties = [
          "firstname", "lastname", "email", "company",
          "lifecyclestage", "createdate", "lastmodifieddate"
        ]
        
        # Filtering
        filter_groups = [
          {
            filters = [
              {property_name = "lifecyclestage", operator = "EQ", value = "customer"},
              {property_name = "createdate", operator = "GT", value = "2024-01-01"}
            ]
          }
        ]
        
        # Associations
        include_associations = true
        association_types = ["contacts_to_companies", "contacts_to_deals"]
        
        # Performance tuning
        batch_size = 100
      },
      {
        # Companies configuration
        object_type = "companies"
        extraction_mode = "incremental"
        incremental_field = "hs_lastmodifieddate"
        start_date = "2024-01-01T00:00:00Z"
        
        # Fetch all properties
        fetch_all_properties = true
        
        # Include associations
        include_associations = true
        association_types = ["companies_to_contacts", "companies_to_deals"]
      },
      {
        # Deals configuration
        object_type = "deals"
        extraction_mode = "incremental"
        incremental_field = "hs_lastmodifieddate"
        start_date = "2024-01-01T00:00:00Z"
        
        properties = [
          "dealname", "amount", "dealstage", "closedate",
          "pipeline", "createdate", "hs_lastmodifieddate"
        ]
        
        filter_groups = [
          {
            filters = [
              {property_name = "dealstage", operator = "NEQ", value = "closedlost"}
            ]
          }
        ]
      },
      {
        # Tickets configuration
        object_type = "tickets"
        extraction_mode = "full"
        
        properties = [
          "subject", "content", "hs_pipeline_stage",
          "createdate", "hs_lastmodifieddate"
        ]
      },
      {
        # Custom object configuration
        object_type = "custom_objects"
        custom_object_name = "2-12345678"  # Custom object schema ID
        
        extraction_mode = "incremental"
        incremental_field = "hs_lastmodifieddate"
        fetch_all_properties = true
        
        include_associations = true
        association_types = ["custom_to_contacts", "custom_to_companies"]
      }
    ]
    
    # Global settings
    max_concurrent_requests = 5
    rate_limit_per_second = 100
    request_timeout_ms = 30000
    max_retries = 3
    enable_parallel_extraction = true
  }
}

Technical Considerations

Dependencies

• HTTP Client: Use Apache HttpClient or OkHttp for REST API calls
• JSON Processing: Jackson or Gson for JSON serialization/deserialization
• OAuth Library: If supporting OAuth 2.0 authentication
• Rate Limiting: Implement token bucket or sliding window algorithm

Multi-Object Implementation

• Follow SeaTunnel's CDC connector pattern for object discovery
• Use TableId concept for HubSpot objects (portal → object → property)
• Support object filtering similar to table filtering in CDC connectors
• Parallel extraction for multiple objects
• Per-object configuration merging

API Characteristics

• Rate Limits:

  • Standard: 100 requests per 10 seconds
  • Professional/Enterprise: Higher limits (150-200 req/10s)
  • Need exponential backoff for 429 responses

• Pagination:

  • Cursor-based pagination (after parameter)
  • Maximum 100 records per page
  • Need to handle paging.next.after token

• Incremental Extraction:

  • Use lastmodifieddate or createdate properties
  • Filter by date ranges in search API
  • Store last successful timestamp in checkpoint per object

Error Handling

• 429 Too Many Requests: Exponential backoff with retry-after header
• 401/403 Authentication: Fail fast with clear error message
• 400 Bad Request: Validate property names and filter syntax
• 500 Server Errors: Retry with exponential backoff
• Network Errors: Configurable retry strategy
• Per-object error tracking: Continue processing other objects if one fails

Testing

• HubSpot Developer Account: Free tier available for testing
• Test Sandbox: HubSpot provides sandbox portals for enterprise customers
• Mock Server: Create mock API server for unit tests
• Multi-object test cases: Validate parallel extraction
• Integration Tests: Use real HubSpot account with test data

References

• HubSpot API Documentation
• CRM Objects API
• Search API
• Associations API
• Marketing Events API
• API Usage Guidelines
• SeaTunnel CDC Connector Pattern

[DanielCarter-stack]

Thanks for proposing the HubSpot Source Connector. This is a valuable addition to SeaTunnel's ecosystem.

Feasibility

The technical approach is sound. SeaTunnel has several HTTP REST API-based connectors that can serve as references:

• connector-http-base - base HTTP client with auth, pagination, and rate limiting
• connector-github - GitHub REST API with token-based auth
• connector-jira - Jira REST API connector
• connector-klaviyo - similar SaaS API pattern to HubSpot

Key Implementation Areas

1. Authentication: OAuth 2.0 or API key (patterns exist in connector-http-base)
2. API Objects: Contacts, Companies, Deals, Tickets - clarify scope (single vs multi-object connector)
3. Pagination: HubSpot uses offset-based pagination
4. Rate Limiting: tier-based limits (e.g., 100 requests/10s for Free tier)

Questions

1. Which HubSpot API objects should be supported initially?
2. Should this support incremental CDC/sync or full snapshot only?
3. Will you contribute the implementation or seeking community pickup?

@contributor for visibility

[davidzollo]

Please leave a message if anybody who'd like to implement this connector

[Suresh-Krishna-Kusuma]

Hi @davidzollo, I would like to work on this issue. Please assign it to me.

I have reviewed the detailed requirements and the API documentation. To ensure a high-quality implementation and faster review process, I propose breaking this into two phases:

Phase 1: The Foundation

• Goal: Establish the HubSpotSource structure using connector-http-base.
• Scope:
  • Single Object Support: Configuration to extract one object type (e.g., "Contacts" OR "Companies") per source.
  • Batch Mode: Full Snapshot extraction using HubSpot's cursor-based pagination.
  • Authentication: Private App Access Token (Bearer Auth).
  • Rate Limiting: Leveraging the existing connector-http-base rate limiter to handle API limits.

Phase 2: Advanced Features (Future PRs)

• Multi-object support (Parallel extraction).
• Incremental Sync (CDC based on lastmodifieddate).

I have my HubSpot Developer account ready and have successfully tested the API endpoints. I am ready to start the Phase 1 implementation immediately.