Gemini → Chat with PDF

Overview

This action enables you to interact with the Gemini API using a PDF file and a text question. The function accepts a reference to an already uploaded PDF file (as a file_uri) hosted on Google, and a user question, then queries Gemini’s generative content API to get a response.

Important: A file_uri must already be uploaded and available on Google before using this action. The function does not handle file uploads; it operates on an existing, accessible URI.

Inputs

Function Stack

1. Talk to Uploaded Content
   • The function receives file_uri and question as inputs.
2. API Request
   • Sends a POST request to:

https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash:generateContent?key=YOUR_API_KEY

Using the following payload structure:

{
  "contents": [
    {
      "parts": [
        {
          "file_data": { "file_uri": "<file_uri>" }
        },
        {
          "text": "<question>"
        }
      ]
    }
  ]
}

- Authenticates using `gemini_api_key`.
- Returns the API response as `gemini_api`.

3. Precondition
   • Checks: gemini_api.response.status == 200 (Proceeds only if the API response is successful.)
4. Response
   • Returns: gemini_api.response.result

Example Usage

Request

{
  "gemini_api_key": "AIzaSyD...",
  "file_uri": "https://storage.googleapis.com/path-to-your-pdf.pdf",
  "question": "Summarize the main arguments from this document."
}

Response

{
  "result": "The main arguments in this document are..."
}

Notes

• The file_uri must be a direct URL to a PDF file that is already uploaded and publicly or appropriately accessible to the Gemini API.
• The function does NOT upload files; ensure you upload your PDF first and obtain the URI.
• The question can be any prompt or inquiry about the content of the PDF.
• You must use a valid and enabled Gemini API key.

Troubleshooting

• INVALID_ARGUMENT or NOT_FOUND: Verify the file_uri is correct, accessible, and the PDF exists.
• PERMISSION_DENIED: Ensure your API key has access, and the file permissions are properly set on Google.
• REQUEST_DENIED: Check your API key and ensure billing or usage limits have not been exceeded.
• Other API errors: Refer to Gemini API documentation for more details.

References

• Gemini API: Overview \& Docs
• Gemini API: Supported Models