v0.1.0-a6 dbt_unified_rag

Bug Fixes (requires --full-refresh)

• Applied coalesce_cast macro to all relevant fields that are being concatenated into comment_markdown, as any concatenation in Snowflake with a null value returns null. We coalesced 'UNKNOWN' on a string field, and '1970-01-01 00:00:00' on a timestamp field to ensure Snowflake returns chunks of texts for all comments with null components.

• Fields are now coalesced in these intermediate models:

  • Hubspot
    • int_rag_hubspot__deal_comment_document:
      • email_title (string)
      • body (string)
      • comment_time (timestamp)
    • int_rag_hubspot__deal_document:
      • title (string)
      • created_on (timestamp)
  • Jira
    • int_rag_jira__issue_comment_document:
      • comment_body (string)
      • comment_time (timestamp)
    • int_rag_jira__issue_document:
      • title (string)
      • created_on (timestamp)
  • Zendesk
    • int_rag_zendesk__ticket_comment_document:
      • comment_body (string)
      • comment_time (timestamp)
    • int_rag_zendesk__ticket_document:
      • title (string)
      • created_on (timestamp)

• Corrected syntax errors for the default_variable in stg_rag_hubspot__engagement_email and stg_rag_hubspot__engagement_note.

• Updated joins to ensure engagement_deal is the base in the int_rag_hubspot__deal_comment_document CTEs.

• Added most_recent_document CTE to int_rag_*__deal_comment_documents_grouped models in Hubspot, Jira and Zendesk to correctly bring in the most_recent_chunk_update by document.

• Brought in engagement_type from the Hubspot engagement_deal source to produce proper chunk records in the rag__unified_document.

• Added filters on email and note types in int_rag_hubspot__deal_comment_document when creating email and note chunk records.

Under the Hood

• Updated Hubspot seed files to ensure proper joins on end models.

Full Changelog: v0.1.0-a5...v0.1.0-a6