CSV Template Structure (Edit existing items)

This document provides a comprehensive guide on how to structure your CSV files to update existing objects in Decube using the Export/Import feature.

Overview

The CSV update flow allows you to edit metadata for Catalog (with datasets and non-datasets), Glossary (Glossary, Category, Term), and Classification Policies. Each row corresponds to an object and specifies the fields that are being updated.

Each object type has a well-defined set of identifiers (non-editable) used to locate the object, and editable attributes that can be modified.

Identifier Fields

What are identifier fields? These are key fields used to match objects within the Decube platform. When updating existing metadata, these fields must exactly match the values in Decube to apply changes correctly.

• Catalog (Dataset): Source, Schema, Table, Column, Type

• Catalog (Non-Dataset): Source, Name, Type

• Glossary (Update): Glossary, Category, Term, Type

• Policy (Update): Policy name, Policy Classification Name/ Policy Tag

⚠️ Important Notes for Editing Existing Items

Identifiers Must Be Precise and Unchanged

• Identifiers are used to locate existing objects — these must follow strict structure and must not be edited.

• For Catalog objects:

  Source | Parent_1_name | Parent_2_name | Object_name | Type

  Example: redshift | schema_1 | table_sales | column_created_at | Column

• For Glossary:

  Parent_1 | Parent_2 | Object_name | Type

  Example: Glossary_name | Category_name | Term_name | Term

• Any change to these identifiers will result in failure to match and update the object.

Only Modifiable Fields Should Be Edited

• Do not change object names or hierarchy-related fields.

• Only update allowed attributes like Description, Tags, Owners, etc.

Single Data Source per Export

• You can only export and edit items from one data source at a time.

• This is enforced during export via the filters. Don’t manually mix sources in a single CSV to reduce possibilities of error.

Updating the Glossary

• Glossary, Category, and Term updates can be handled via single file upload.

• Ensure correct Type field is mentioned to indicate whether a row is a Glossary, Category, or Term.

• Classifications and Related Term can only be applied to Terms. For more information on what are applicable attributes to asset types, see table of Editable Attributes and Constraints.
• You must maintain the correct hierarchy using Parent_1 and Parent_2

Column Limits

• Max 3 Tags per object.

• Max 3 Data Owners.

• Descriptions have a character limit of 8,000.

• Name have a character limit of 100.

CSV Format Integrity

• Do not rename columns.

• Do not add additional column names which are not supported by the platform.

• Ensure headers match the template exactly.

Common Failure Triggers

• Using incorrect object type (e.g., writing “Dataset” instead of “Table”).

• Leaving identifier fields empty or malformed.

• Including rows for deleted or non-existent objects.

Changes Are Irreversible

• Once imported, updates are applied immediately and cannot be undone.

• Always validate your file carefully before importing.

All empty value in any editable attribute will overwrite and clear the existing value.

Editable Attributes and Constraints

Below is a breakdown of editable fields, format requirements, character limits, and constraints by object type.

1. Catalog with tables (Dataset)

Field	Required	Description	Constraints	Example	Field Type
Source	Yes	Source system name, Required for editing table and column both	Must exist in system	RedshiftPRD	Identifier/not editable
Schema	Conditional (read description)	Schema name, Required (if editing column) Optional (If editing table)	Must exist in source	public	Identifier/ not-editable
Table	Conditional (read description)	Table name, required for editing table and column both	Must exist in schema	sales	Identifier/ not-editable
Column	Conditional (read description)	Column name (only for column updates)	Optional	created_at	Identifier/ not-editable
Type	Yes	Object type: Table, Column	One of Table, Column	Table	Identifier/ not-editable
Data Owners	No	Designated data owners	Max 3, email format	[e-mail]	Editable attribute
Business Owners	No	Designated Business owners	Email format	[e-mail]	Editable Attribute
Description	No	Description text	Max 8000 characters	Some long description	Editable Attribute
Tags	No	Tags	Comma-separated, max 3	Sales,Marketing	Editable Attribute
Classification	No	Policy tags applied	Must match existing policies	PII,GDPR	Editable Attribute
Linked Terms	No	Related Glossary Terms	Format: glossary.category.term	Glossary_1.Term_1	Editable Attribute

2. Catalog (Non-Dataset: Data Job, Chart, Report)

Field	Required	Description	Constraints	Example	Field Type
Source	Yes	Source system name	Must exist in system	TableauPRD	Identifier/not editable
Name	Yes	Name of the object	Unique within source	sales_dashboard	Identifier/not editable
Type	Yes	Object type: DataJob, Chart, Report	One of DataJob, Chart, Report	Chart	Identifier/not editable
Data Owners	No	Data owners	Max 3, email format	[e-mail]	Editable Attribute
Business Owners	No	Business owners	Email format	[e-mail]	Editable Attribute
Description	No	Description text	Max 8000 characters	Dashboard for monthly sales	Editable Attribute
Tags	No	Tags	Comma-separated	Sales,Analysis	Editable Attribute
Linked Terms	No	Related Glossary Terms	Format: glossary.category.term	Glossary_1.Term_2	Editable Attribute

3. Glossary, Category and Term

The same template is used to update Glossary, Category, and Term. Identify the type of item via the Type column and fill relevant identifier fields accordingly.

Field	Required	Description	Constraints	Example	Field Type
Glossary	Yes	Glossary name	Must exist	Glossary_1	Identifier/not editable
Category	Conditional (Read description)	Category name, Required (if updating term under category or category) Not required (for adding glossary)	Must exist, Max character limit 100	Category_1	Identifier/not editable
Term	Conditional (Read description)	Term name, Required (if updating term) Not required (for adding glossary)	Must exist, Max character limit 100	Term_1	Identifier/not editable
Type	Yes	Type of object	One of Glossary, Category, Term	Term	Identifier/not editable
Data Owners	No	Designated Data owners	Max 3, email format	[e-mail]	Editable Attribute
Business Owners	No	Designated Business owners	Email format	[e-mail]	Editable Attribute
Description	Yes	Glossary, category and term level description	Max 8000 characters	Meaning of sales process term	Editable Attribute
Classifications	No	Policy tags (only applicable for Terms)	Must match existing policies	PII,GDPR	Editable Attribute
Related Terms	No	Related Terms (only applicable for term)	Must exist, Format: glossary.category.term	Glossary_1.Category.Term_2	Editable Attribute

Note:

• Classifications only be applied to Terms.

• Related Terms only apply to Terms.

• Ensure the Type field is correctly set for each row.

4. Classification Policy

Field	Required	Description	Constraints	Example	Field Type
Name	Yes	Policy name	Must exist	Data Privacy Policy	Identifier/not editable
Classification_name	Yes	Policy tag	Max 5 characters, unique	PII	Identifier/not editable
Description	No	Description	-	Protects personally identifiable info	Editable Attribute
Purpose	No	Purpose	-	Legal compliance	Editable Attribute
Stewards	No	Email(s) of policy steward	Email format	[e-mail]	Editable Attribute

This format ensures consistent structure and validation for editing metadata in bulk via CSV Export/Import. Make sure identifier fields are correct and that each row adheres to constraints to avoid import failures.