Copyright (c) 2026 DataLake Solutions. All rights reserved.

This source code and all related materials are proprietary to DataLake Solutions.

You may not, without prior written permission from DataLake Solutions:

• copy, modify, distribute, sublicense, sell, publish, or otherwise disclose this code;
• share this code with third parties or post it to public repositories, forums, or websites;
• use this code to create derivative works for external distribution or commercial exploitation.

Unauthorized use, disclosure, or distribution is strictly prohibited.

• Local CSV output at schemas/<schema_name>/Tables_generated/<table_name>.csv
• Schema metadata JSON at schemas/<schema_name>/instructions/

1. Create venv:
   • python -m venv .venv
2. Install:
   • python -m pip install -r requirements.txt
3. Fill .env:
   • OPENAI_API_KEY=<your_key>
   • OPENAI_MODEL=gpt-4o-mini (or any supported model)
4. Run Streamlit UI:
   • python -m streamlit run app.py

The project includes a headless API that accepts one JSON payload with the schema prompt and table definitions, then runs the same backend generation flow used by the UI.

You do not need to open the Streamlit UI.

Instead, you:

1. Start the API server.
2. Send one JSON file to the API.
3. The backend generates the data.
4. The backend validates the data.
5. The API returns the output folder and file paths.

Use the API server without --reload when generating data.

Why:

• this project creates and updates files while it runs
• --reload can restart the API in the middle of a run
• that can interrupt generation or validation

Run:

cd ..\datagenx

Run:

.\.venv\Scripts\python.exe -m uvicorn headless_api:app --host 127.0.0.1 --port 8000

After startup:

• Swagger UI: http://127.0.0.1:8000/docs
• Health check: http://127.0.0.1:8000/health

Leave this terminal open.

Run:

cd ..\datagenx

Run:

Invoke-RestMethod -Method Get -Uri "http://127.0.0.1:8000/health"

Expected result:

{
  "status": "ok"
}

Main endpoint:

• POST /generate

Payload shape:

{
  "schema_name": "Banking",
  "schema_prompt": "Generate realistic synthetic US banking data with referential integrity.",
  "replace_existing": true,
  "tables": [
    {
      "table_name": "CUSTOMERS",
      "num_entries": 1000,
      "ddl": "CREATE TABLE CUSTOMERS (\n  CUSTOMER_ID BIGINT PRIMARY KEY,\n  FULL_NAME VARCHAR(100) NOT NULL,\n  EMAIL VARCHAR(120) UNIQUE\n);",
      "instructions": "Generate realistic customer names and unique emails."
    },
    {
      "table_name": "ACCOUNTS",
      "num_entries": 1200,
      "ddl": "CREATE TABLE ACCOUNTS (\n  ACCOUNT_ID BIGINT PRIMARY KEY,\n  CUSTOMER_ID BIGINT NOT NULL,\n  ACCOUNT_NUMBER VARCHAR(20) UNIQUE,\n  FOREIGN KEY (CUSTOMER_ID) REFERENCES CUSTOMERS(CUSTOMER_ID)\n);",
      "instructions": "Every CUSTOMER_ID must exist in CUSTOMERS."
    }
  ]
}

Example payload files are available in the examples/ folder.

• Retail two-table example: examples/retail_two_table_payload.json

PowerShell:

Invoke-RestMethod -Method Post `
  -Uri "http://127.0.0.1:8000/generate" `
  -ContentType "application/json" `
  -InFile "examples\retail_two_table_payload.json"

What this command does:

• reads the JSON file from the examples folder
• sends it to the API
• runs schema creation, data generation, and validation
• returns the output location and status

cURL:

curl -X POST "http://127.0.0.1:8000/generate" \
  -H "Content-Type: application/json" \
  --data-binary "@examples/retail_two_table_payload.json"

• GET /schemas to list saved schemas
• GET /schemas/{org_id} to inspect one saved schema and its statuses

The POST /generate response includes:

• org_id
• schema_name
• output_dir
• generated_files
• validation_report_path
• log_path
• status

If status is DONE, the run completed successfully.

For the retail example, generated files will be under:

• schemas\Retail\Tables_generated\
• schemas\Retail\validations\validation_report.json
• schemas\Retail\logs\latest.log

For a non-technical user, the headless flow is:

1. Start the API.
2. Check /health.
3. Send the JSON file.
4. Wait for the response.
5. Open the generated files.

• Generated files are saved in project schemas/.
• Per schema, instruction folder contains:
  • ddl.json
  • instructions.json
  • schema_prompt.json
• UI keeps the same core generate/schema-management flow as your existing app.
• Query module and login are intentionally removed per requirement.