Process a batch

What the rows of the CSV are: invoice or retention_certificate. Each type has its own columns, listed in the batches guide.

awaiting_upload until the whole file arrives at upload_url; validating while heru checks every row; ready when at least one document is valid; done when no document is invalid and every document is a duplicate, so nothing is stamped and the batch closes with its report and fires batch.done, without going through POST /batches/{id}/process; invalid when no document is valid and at least one failed validation, the file itself has an error or the upload window ended; processing after POST /batches/{id}/process; paused after POST /batches/{id}/pause; cancelling after POST /batches/{id}/cancel, or when the batch reaches 20 days unfinished, while heru stops what is pending and writes the report, then cancelled; done when every valid document has its outcome. invalid, cancelled and done are final.

When the batch was created with POST /batches. The 20 days of the batch's life count from here: see processing pace.

The defaults of the request, normalized: 16% comes back as 0.16.

The largest file upload_url takes, in bytes, compressed or not: 34359738368 (32 GiB).

Data rows in the file. Final once heru has read the whole file.

Rows checked so far. Compare it with total_rows while the batch is validating.

Rows that passed their own checks. A valid row can belong to an invalid document, which is not processed: progress is counted in documents.

Rows with at least one problem.

Documents in the file: rows that share external_id are one document. Final once heru has read the whole file.

Documents with every row valid and not stamped before. Only these are processed. It does not change after ready.

Documents with at least one invalid row. None of their rows is processed.

Documents whose external_id was already stamped. They are not stamped again and the report gives their id and uuid. A document already stamped wins over its own validation errors: it counts here, not in invalid_documents.

Valid documents with an outcome so far: succeeded + failed + cancelled + pending_review. Compare it with valid_documents while the batch is processing.

Documents already prepared and on their way to the PAC. They keep moving while the batch is paused, so this says how much is left before the counters stop.

Documents stamped.

Valid documents that failed when processed, e.g. refused by the stamping authority. The reason is in the report. A failed document can go again in a new batch.

Valid documents never processed because the batch was cancelled.

Documents with no outcome when the batch closed. They are not lost: if one gets stamped later, its webhook arrives.

Documents whose receiver data in the file differs from the data registered in heru, which is what gets stamped. Estimated during validation and final when the batch closes; the report names the columns.

The first 100 problems, in row order. error_count has the total and errors_url all of them. Final when the batch leaves validating.

Total problems found, which can be more than errors shows.