Skip to content

README.md

Latest commit

 

History

History
94 lines (59 loc) · 3.07 KB

README.md

File metadata and controls

94 lines (59 loc) · 3.07 KB
Raw

Automated Form Filling with Parse JSON

Code example showing how to use the Gaffa Browser Request API to:

  • Extract all fields from a web form using the parse_json action (structured schema output)
  • Prompt you in the terminal for values
  • Fill & submit the form in a real browser session using type + click
  • Capture a screenshot after submission

This is designed automation workflows where you want schema‑driven extraction plus a simple human‑in‑the‑loop data entry step. For more information, read the accompanying blog post.

Files

File Description
automated_form_filling.py Extracts form fields with parse_json, prompts for input, fills and submits the form, captures a screenshot

Requirements

  • Python 3.8+
  • A Gaffa API key — sign up at gaffa.dev and create your key in the API Keys section of the dashboard

Setup

1. Install dependencies

pip install requests

2. Add your API key

In automated_form_filling.py, set:

  • GAFFA_API_KEY = "your_api_key_here"

(Alternatively, you can export it as an environment variable and load it in code.)

3. (Optional) Choose a form URL

Update FORM_URL in automated_form_filling.py to the form you want to analyze and fill.

How it works

Step 1 — Extract fields (Parse JSON)

The script opens the form URL and runs actions:

  • wait for form to appear
  • parse_json with a schema:
    • form_title
    • fields[] with: label, field_name, field_type, required, placeholder

It then reads the parse_json action’s output from the response.

Step 2 — Collect values (Terminal prompt)

For each extracted field, the script prompts you for a value:

  • Required fields keep prompting until you enter something
  • Optional fields can be skipped by pressing Enter

Step 3 — Fill, submit & screenshot

The script builds a second action list:

  • wait for form
  • type into each input using the selector: [name='<field_name>']
  • click the submit button: button[type='submit']
  • capture_screenshot (fullscreen)

If present, the screenshot is printed from the capture_screenshot action output.

Usage

From this folder:

python automated_form_filling.py

You’ll see:

  • “Step 1: Analyzing form…”
  • A guided prompt to enter values
  • A summary of what will be submitted
  • A confirmation prompt (Submit this form? (y/n))
  • Submission result + screenshot output (if returned)

Notes

  • Selectors assume name attributes: Filling uses [name='...']. If your target form doesn’t use name, update the extraction schema to capture a usable selector (e.g., id, aria-label, or a stable CSS selector).
  • Modals / load delays: If the page loads slowly or shows a modal, adjust the wait action timeout or add additional waits/clicks as needed.
  • Model choice: parse_json is configured with model: "gpt-4o-mini". You can switch models depending on your accuracy/cost needs.