Grain API Wrapper

A Python wrapper for the Grain Workspace API to fetch transcripts and save them as JSON files.

Setup

1. Install dependencies:

pip install -r requirements.txt

2. Set your Grain API token as an environment variable:

export GRAIN_API_TOKEN="your-token-here"

Usage

The script starts in test mode by default, which fetches only the first recording:

python grain_api_wrapper.py

To fetch all recordings, edit the script and set test_mode = False in the main() function.

Features

• Streaming pagination: Processes recordings page by page without loading all into memory
• Resume capability: Automatically saves progress after each page; can resume from crashes
• Duplicate detection: Skips recordings that have already been downloaded
• Date-based organization: Creates a directory tree organized by year/month/day

Output Structure

The script creates a recordings/ directory with subdirectories organized by date:

recordings/
├── 2024/
│   ├── 01/
│   │   ├── 15/
│   │   │   └── {recording_id}_{title}.json
│   │   └── 20/
│   │       └── {recording_id}_{title}.json
│   └── 02/
│       └── 03/
│           └── {recording_id}_{title}.json
└── 2025/
    └── 06/
        └── 16/
            └── {recording_id}_{title}.json

Each JSON file contains:

• ID, title, URL, source, datetime
• List of participants with names and emails
• Full transcript in JSON format

Resume from Interruption

If the script is interrupted, it saves progress to .cursor_state.json. On the next run, it will ask if you want to resume from where it left off.

API Notes

• The script includes rate limiting (1 second between recordings) to be respectful of the API
• Transcripts are fetched using the transcript_format=json parameter