fix: harden extraction batch handling
Deploy contracts-flask / validate (push) Canceled after 0s

This commit is contained in:
Repinoid
2026-10-31 19:06:34 +03:00
parent 9c73a730d1
commit 79de42da96
6 changed files with 79 additions and 21 deletions
+42 -6
View File
@@ -25,18 +25,54 @@ def analyze_batch(batch_id, documents, extractor, dossier_builder,
progress_callback=progress_callback,
documents=documents,
)
dossier_result = dossier_builder(batch_id) if extraction.get("ok") else {"ok": False}
# Extractor-level `ok` means every input succeeded; dossier construction only
# requires at least one persisted passport and can proceed for that subset.
extraction_results = extraction.get("results") or []
total = extraction.get("total", len(documents))
done = extraction.get("done")
if done is None:
done = sum(1 for result in extraction_results if result.get("ok"))
failed = extraction.get("failed")
if failed is None:
failed = max(total - done, 0)
dossier_result = dossier_builder(batch_id) if done > 0 else {"ok": False}
dossier_ok = bool(dossier_result.get("ok"))
# Keep result usability separate from completeness so clients can continue
# with valid dossiers while still reporting failures explicitly.
status = (
"failed" if done == 0 or not dossier_ok
else "partial" if failed > 0
else "completed"
)
documents_by_id = {document.get("id"): document for document in documents}
failed_documents = []
for result in extraction_results:
if result.get("ok"):
continue
document_id = result.get("document_id")
source_document = documents_by_id.get(document_id, {})
failure = {
"document_id": document_id,
"filename": source_document.get("filename"),
"error": result.get("error") or "extraction failed",
}
if result.get("persistence_error"):
failure["persistence_error"] = result["persistence_error"]
failed_documents.append(failure)
classes = extraction.get("classes", {})
math_checks_passed = sum(
1 for result in extraction.get("results", [])
1 for result in extraction_results
if (result.get("extraction") or {}).get("math_validation", {}).get("status") == "VERIFIED"
)
return {
"ok": bool(extraction.get("ok")) and bool(dossier_result.get("ok")),
"ok": status != "failed",
"status": status,
"partial": status == "partial",
"v3": True,
"done": extraction.get("done", 0),
"total": len(documents),
"failed": extraction.get("failed", 0),
"done": done,
"total": total,
"failed": failed,
"failed_documents": failed_documents,
"classes": classes,
"ignored_accounting": classes.get("ignored_accounting", 0),
"math_checks_passed": math_checks_passed,