Skip to content

Conversation

loeng2023
Copy link
Contributor

@loeng2023 loeng2023 commented Feb 15, 2024

User can now specify customized content column and metadata JSON column in replace of the default "page_content" and "langchain_metadata" column.

Summary of changes:

  • MSSQLEngine
    • init_document_table can create column with non-default name content_column and metadata_json_column, deprecate store_metadata.
  • MSSQLLoader
    • Add metadata_json_column for loading document.metadata from non-default JSON metadata field.
    • raise ValueError if specified columns cannot be found in table.
  • MSSQLSaver
    • Add content_column for saving document.page_content into non-default content field.
    • Add metadata_json_column for saving document.metadata into non-default JSON metadata field.
    • Support deleting document with non-default content and JSON metadata field.
  • test_mssql_loader.py
    • test_save_doc_with_customized_metadata will use non-default content and JSON metadata column.
    • test_delete_doc_with_customized_metadata will use non-default content and JSON metadata column.

Change synced with googleapis/langchain-google-cloud-sql-mysql-python#19

@loeng2023 loeng2023 requested a review from a team as a code owner February 15, 2024 18:50
@product-auto-label product-auto-label bot added the api: cloudsql-sqlserver Issues related to the googleapis/langchain-google-cloud-sql-mssql-python API. label Feb 15, 2024
@loeng2023 loeng2023 assigned loeng2023 and kurtisvg and unassigned loeng2023 Feb 15, 2024
@loeng2023 loeng2023 requested a review from kurtisvg February 15, 2024 19:03
@averikitsch averikitsch merged commit f1d117e into main Feb 16, 2024
@averikitsch averikitsch deleted the doc-loader-update branch February 16, 2024 16:50
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
api: cloudsql-sqlserver Issues related to the googleapis/langchain-google-cloud-sql-mssql-python API.
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants