An E2E Azure CognitiveSearch POC showing the two indexing ways of index managing (push/pull).

There are two ways of populating an index in Azure Cognitive Search:

• Push: in the push mode, an application progammatically send documents, in the JSON format, to the index via the Azure Cognitive Search REST API.
• Pull: in the pull mode, an indexer is created. An indexer is a crawler that extracts searchable content from a data source and populates a search index using field mappings. You can run indexers on demand or on a schedule, and you can also use them to apply AI enrichment to your data.

• Azure Cognitive Search
• Azure Functions
• Azure Storage account
• Azure Document Intelligence
• Azure SQL Server

A console application that creates the indexes and the indexer.

A MSSQL server that contains the documents metadata.

An Azure Function application that provides the endpoints for the indexer and the publishing app. The Pull endpoint is a custom WebAPI Skill used by the indexer to enrich the document with data from the database before publishing to the index. The Push endpoint is a queue triggered Durable function that calls the Document Intelligence API to read the content of the document using OCR an than publishes the document to the index.

A Blazor Web App that can be used to:

• publish document on the indexes
• mark documents as updated or deleted
• search the indexes