Operations¶

Triggering data imports and exporting data out.

Start a sync¶

POST /api/sync

Runs the ETL immediately instead of waiting for the 06:00 cron job.

Parameter	Type	Default	Description
`source`	string	`all`	`wb`, `ozon` or `all`
`days`	integer	`30`	How far back to import

The request returns as soon as the job has been launched, not when it completes. Poll GET /api/sync/status for progress.

Response

{ "status": "started" }

If an import is already running, the request is refused rather than queued:

{ "status": "already_running", "message": "Synchronisation already in progress" }

This is still a 200. Check the status field rather than the HTTP code.

curl -X POST "http://localhost:8000/api/sync?source=wb&days=7"

Two constraints worth knowing before you rely on this

The job is killed after 15 minutes. A large historical import will exceed that. Run those from the command line with run_etl.py, where nothing times them out.

The subprocess uses absolute paths under /opt/analytics. On an installation anywhere else this endpoint fails immediately while scheduled cron syncs continue to work, because cron invokes the script directly. See Troubleshooting.

Check sync status¶

GET /api/sync/status

Current state of the import job. Takes no parameters.

Field	Type	Description
`running`	boolean	Whether a job is in progress
`status`	string	`idle`, `running (...)`, `done`, `timeout`, or `error: ...`
`started`	string \| null	ISO 8601 timestamp of when the current or last job began
`log`	array	Last 20 lines of the job's combined output

{
  "running": false,
  "status": "done",
  "started": "2026-07-28T09:14:02.317441",
  "log": ["Wildberries: done", "Ozon: done"]
}

State is held in memory, not in the database. Restarting the API service resets it to idle, and no record of past imports survives a restart. If you need an audit trail of imports, read the cron log on the server instead.

CSV exports¶

Five endpoints return a CSV file rather than JSON, each mirroring a dashboard section.

Endpoint	Contents	Parameters
`GET /api/export/finance`	Daily revenue, profit and deductions	`days` (default `30`)
`GET /api/export/profitability`	Per-product economics, full history	none
`GET /api/export/reorder`	Stock levels and reorder status	none
`GET /api/export/cards`	Daily listing statistics	`days` (default `30`)
`GET /api/export/platform`	Marketplace comparison	none

Each responds with Content-Type: text/csv and a Content-Disposition header carrying a generated filename, so a browser downloads rather than displays the result. Filenames for the windowed exports include the window, for example finance_30d.csv.

curl -OJ "http://localhost:8000/api/export/finance?days=90"

Encoding. Files are UTF-8 with a byte order mark. The mark is there so that Excel opens Cyrillic product names correctly on a double-click instead of showing mojibake. The cost is that some strict CSV parsers read it as part of the first column's name, so strip it when feeding an export into a script.

Empty results. An export with no matching rows still returns a valid CSV containing a single placeholder row, rather than an empty file or an error.