Document Search Engine in Azure Functions and Cosmos DB
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
IFilterTextReader
IFilterTextViewer
search
.gitattributes
.gitignore
LICENSE
README.md
search-azurefunc-logo.png
search.sln

README.md

doc-search-azure

Document Search Engine in Azure Functions and Cosmos DB doc-search-azure is a document search engine, written in C#/.NET, that utilizes Azure Functions and Cosmos DB.

For an overview of this project, please refer to this blog post:http://davidpallmann.blogspot.com/2019/02/a-document-search-engine-using-azure.html

doc-search-azure

Cloud Assets

In order to deploy this solution to the cloud, you will need to define the following:

  • Azure blob storage account, with a "docs" container for documents and a "site" container for a web site.
  • Azure Function App definition, with an index function (blob trigger) and a query function (HTTP trigger)
  • A CosmosDB database, with a DocCollection (partitioned on /Category) and a DocWordCollection (partitioned on /DocId)

The solution files will need updating to refer to YOUR cloud assets.

Building and Deploying

To build and deploy doc-search-azure, you need:

  • Visual Studio 2017
  • An Azure account

Before building, update the local.settings.json file with your storage account settings.

Publish the search project to your Azure Function app in the cloud. The first time you do this, download the publish profile for your Azure Function App and import it in the Publish wizard.

Using

Monitor your Function App log so you can view activity.

Upload a document to the "docs" storage account container.

The index function should run and index the document. This will create one DocCollection record and multiple DocWord records in the database.

Verify the Cosmos DB DocCollection now contains a record for the newly-uploaded document.

Go the query function in the Azure console and test it by adding a query term named search. Set search to some value that should be in the uploaded document and see if a search returns a document result.

Get the web site configured correctly--it needs to point to YOUR HTTP URL for the query function--and try running it. You will need to allow the storage container domain in the CORS setting for the Azure function.