This repository aims to provide some Python script files and functions (which you can use in Visual Studio Code) to quickly performs operations in Azure AI Search that are not exposed in the Azure Portal interface or to overcome problems related to private networking scenarios.
- Visual Studio Code with Phyton extensions
You can just open this repository in Visual Studio Code, find the script you want, and before running the requests make sure to replace all the variables in the beginning of the file to fit your scenario.
Pay attention to the azureAccessToken variable. To acquire an Azure access token to manage your AI Search instance, you need to run this AZ CLI command:
az account get-access-token --scope https://search.azure.com/.default --subscription <<your-subscription-id>>
Then copy the accessToken value from the output. It should begins with ey...
| Script | Description |
|---|---|
| integrated-vectorization-default.py | This script replicates the same steps that happens when you run the Integrated Vectorization wizard. It is useful when you disable Public Access from the networking blade because the button from the portal is disabled in that scenario. It creates a Data Source, Index, Skillset and Indexer |
| /azure-vision-ocr | This folder contains a Azure Function (Python 3.11) that will receive images from an AI Search skillset and submit them to Azure Vision OCR endpoint. This replaces the built-in OCR skill for scenarios where AI Search is in private networking (built-in AI skills don't work) |