download_file_for_user1.py

# Download file for user to bot
# I apologize to future developers who have to maintain this code. 
# I can only blame the byzantine nature of the Microsoft Graph API and Teams API, 
# which stores files in whatever location it feels like, and then 
# provides a content URL that may or may not work.
def download_file_for_user(file_name, file_url, headers):
    # If file is shared in a Team channel, the file is stored in sharepoint for the Team
    # We need to decode and fetch a great deal of info about sharepoint to fetch the file
    try:
        if ".sharepoint.com/sites/" in file_url:

            # Parse SharePoint URL
            parts = file_url.split(".sharepoint.com/sites/")
            hostname = parts[0].replace("https://", "") + ".sharepoint.com"
            site_path_and_file = parts[1]
            site_name = site_path_and_file.split("/")[0]
            file_relative_path_raw = site_path_and_file[len(site_name)+1:]

            # Get the SharePoint site ID
            site_lookup_url = f"https://graph.microsoft.com/v1.0/sites/{hostname}:/sites/{site_name}"
            site_resp = requests.get(site_lookup_url, headers=headers)
            site_resp.raise_for_status()
            site_id = site_resp.json()["id"]

            # Get default document library (drive) ID
            drive_url = f"https://graph.microsoft.com/v1.0/sites/{site_id}/drive"
            drive_resp = requests.get(drive_url, headers=headers)
            drive_resp.raise_for_status()
            drive_id = drive_resp.json()["id"]

            # Try the exact file path, there are sometimes invalid unicode characters like \u202f embedded
            file_path_encoded = urllib.parse.quote(file_relative_path_raw, safe="/")
            graph_url = f"https://graph.microsoft.com/v1.0/sites/{site_id}/drives/{drive_id}/root:/{file_path_encoded}:/content"
            resp = requests.get(graph_url, headers=headers)
            if resp.status_code == 200:
                print(f"🟢 Download succeeded from SharePoint site using exact path: {graph_url}")
                return resp.content
            else:
                print(f"🚫 Exact path download failed: {resp.status_code} - {resp.text}")