Document Tags Tables

Create and manage document-level tags for entire documents

Document tags apply labels to entire documents rather than individual chunks, enabling document-level classification and filtering.

Create a Document Tags Table

Define the tags you want to apply at the document level:

# Define document tags
document_tags = [
    {
        "name": "document_type",
        "values": ["invoice", "contract", "report"]
    },
    {
        "name": "priority",
        "values": ["high", "medium", "low"]
    },
    {
        "name": "department",
        "values": ["finance", "legal", "hr"]
    }
]

# Create document tags table
doc_tags_table = warehouse.create_document_tags_table(
    table_name="doc_classifications",
    document_tags=document_tags
)

print(f"Created document tags table: {doc_tags_table.name}")

Write Document Tags

Apply tags to documents with justification and supporting evidence:

# Write document tags
tags = [
    {
        "id": "tag-1",
        "document_id": "doc-id-123",
        "tag_name": "document_type",
        "tag_value": "invoice",
        "justification": "Contains invoice number and line items",
        "verbatim": "Invoice #INV-2024-001"
    },
    {
        "id": "tag-2",
        "document_id": "doc-id-123",
        "tag_name": "priority",
        "tag_value": "high",
        "justification": "Amount exceeds $10,000",
        "verbatim": "Total: $15,250.00"
    },
    {
        "id": "tag-3",
        "document_id": "doc-id-123",
        "tag_name": "department",
        "tag_value": "finance",
        "justification": "Financial document",
        "verbatim": None
    }
]

doc_tags_table.write_data(tags)

Read Document Tags

Retrieve document tags:

# Get all document tags
all_tags = doc_tags_table.get_data()

for tag in all_tags:
    print(f"Document: {tag['document_id']}")
    print(f"Tag: {tag['tag_name']} = {tag['tag_value']}")
    print(f"Justification: {tag['justification']}")
    print("---")

# Filter by document
doc_tags = doc_tags_table.get_data(document_id="doc-id-123")

# Filter by tag name
type_tags = doc_tags_table.get_data(tag_name="document_type")

Update Document Tags

Modify existing tags:

# Update a tag value
doc_tags_table.perform_data_operation(
    operation="update",
    data=[{"tag_value": "urgent"}],
    filters={"tag_name": "priority", "document_id": "doc-id-123"}
)

# Delete tags for a document
doc_tags_table.perform_data_operation(
    operation="delete",
    data=None,
    filters={"document_id": "doc-id-old"}
)