Extract API info from Notebooks

[halcyon57a]

Hi I am using the Fabric API to trawl through all workspaces on a tenant, identifying Python notebooks and trying to read these to see if they access an API.  If one exists, then list it out.   The Notebook I am running lists the workspaces correctly, correctly identifies whether the Notebook exists and tries to extract the API data - but it doesn't.  It just returns blanks.  Any one ever done this ? Any thoughts?

[V-yubandi-msft]

Hi @halcyon57a ,

When you have a moment, please let us know if everything is okay on your end or if you need any further assistance.

Thank you.

[V-yubandi-msft]

Hi @halcyon57a ,
Can you let us know whether your issue has been resolved or if you're still having trouble? Your feedback is important and could assist others with similar concerns.

Thank You.

[V-yubandi-msft]

Hi @halcyon57a ,

May I know if the issue has been resolved, or are you still facing any challenges? Please let us know if you need any additional details or assistance.

Thanks.

[anilgavhane]

What Might Be Going Wrong

Here are some common culprits:

1. Notebook Content Access Limitations
   • The Fabric API can list notebooks, but reading their internal code cells isn't always straightforward. If you're using the REST API to fetch notebook metadata, it may not include the actual code content unless explicitly supported.
   • Notebook Format
     • Fabric notebooks are stored in a proprietary format, not plain .ipynb. So unless you're parsing the actual notebook file correctly, you might just be getting metadata (like name, ID) but not the code itself.
     • Authentication Scope
       • If you're using a Service Principal, make sure it has the right permissions to access notebook contents. Some API endpoints may still be limited in preview and not support full content extraction.
       • Code Parsing Logic
         • Even if you're getting the code, your script needs to scan for patterns like requests.get(...), http.client, or urllib to detect API calls. If the code uses custom wrappers or indirect methods, it might slip past your filters.

           🛠️ Suggested Approach

           Here’s a more robust strategy:

           • Use the Fabric REST API to List Notebooks
           • Confirm you're hitting the right endpoint: GET /workspaces/{workspaceId}/items
           • Filter for items of type Notebook
             • Download Notebook Content
             • If possible, use the API or SDK to export the notebook or access its code cells
             • If not supported, consider using the Fabric Python experience to open notebooks and extract code manually or via automation
               • Scan for API Patterns
               • Use regex or AST parsing to look for:

                 import requests requests.get("https://api.example.com")

                 • Also check for http.client, urllib, or any custom API wrappers
                 • Community Insight
                   • You're not the only one exploring this. A similar discussion on the Fabric Community highlights the same issue—others are trying to extract API usage ...

                     If you'd like, I can help you write a Python script that scans notebook code for API calls once you have the content. Or we can explore whether using Git integration or exporting notebooks to a readable format might help. Let me know how deep you want to go.
                     
                     

                     Here’s a more robust strategy:

                     • Use the Fabric REST API to List Notebooks
                     • Confirm you're hitting the right endpoint: GET /workspaces/{workspaceId}/items
                     • Filter for items of type Notebook
                       • Download Notebook Content
                       • If possible, use the API or SDK to export the notebook or access its code cells
                       • If not supported, consider using the Fabric Python experience to open notebooks and extract code manually or via automation
                         • Scan for API Patterns
                         • Use regex or AST parsing to look for:

                           import requests
                           requests.get("https://api.example.com")

                           import requests requests.get("https://api.example.com")

                           • Also check for http.client, urllib, or any custom API wrappers

[V-yubandi-msft]

Hi @halcyon57a ,

Thank you for your post in the Fabric community, and  In addition to @tayloramy  earlier response. Fabric’s public API does allow you to retrieve complete notebook definitions, including all code cells. If you’re only getting metadata, that could be why the results appear blank you need to use the correct endpoint for full content.

You can refer to the official documentation: Manage and execute Fabric notebooks with public APIs - Microsoft Fabric | Microsoft Learn

This guide explains how to list notebooks, access their full content, and handle formats such as .ipynb, .py, and .sql. For .ipynb notebooks.

Regards,
Yugandhar.

[tayloramy]

@halcyon57a can you post the snippet of your code that is not working as expected?