A small, AI-assisted programming sample by Nerion Dev. It uses invented tickets and a local HTTP server. This is a synthetic capability demonstration, with no client deployment or production results claimed.
The sample shows how a bounded export can retrieve both pages, retry a temporary rate limit, and stop on specific failures. It uses only Python's standard library and makes requests to its own server on 127.0.0.1. No API key, account, external service or package installation is needed.
Save demo_api_local.py alongside this file. Use an existing CPython 3.12 or newer installation. In PowerShell, replace the placeholder below with the actual path to that interpreter:
$DemoPython = 'C:\Path\To\python.exe' # Replace this placeholder.
& $DemoPython --version
& $DemoPython -I .\demo_api_local.py --output .\results\demo-001For the same checks with Python optimization enabled, use a different, new output directory:
& $DemoPython -I -O .\demo_api_local.py --output .\results\demo-002An existing output directory is refused with exit code 2. Read RESULTADO.json and report.json inside the newly created output directory. Successful execution reports PASS_LOCAL_HTTP_SYNTHETIC_ONLY and exits with code 0; a failed acceptance check exits with code 1. Check both the exit code and the receipt.
| Synthetic case | Expected observation |
|---|---|
| Two pages after a rate limit | HTTP 429, 200, 200; two tickets, two pages |
| First-page-only example versus complete export | One ticket versus two tickets |
| Unauthorized response | The export stops after one HTTP 401 |
| Persistent unavailable response | The export stops after two HTTP 503 responses |
| Invalid JSON | The export stops with a JSON decoding error |
The receipt records measured values separately from expected values, the Python runtime, source hash and output hash. Review the actual measurements rather than relying only on a PASS label.
The input contract is deliberately narrow: data / next_cursor pages, and three string fields per invented ticket. This does not demonstrate a third-party API integration, OAuth, a scheduler, a Google Sheets connection, deployment, load capacity or a repaired client system. HTTP-date Retry-After is unsupported. Production-grade atomic output replacement and filesystem syncing are outside this sample.
The source was developed and reviewed with AI assistance. The release commands above are checked on Windows using the explicit interpreter path; no registered py launcher or python command on PATH is assumed. No Linux execution is claimed. A client task would need its own agreed input contract, acceptance criteria and funded scope before implementation.