fix: harden extraction batch handling

2026-10-31 19:06:34 +03:00
parent 9c73a730d1
commit 79de42da96
6 changed files with 79 additions and 21 deletions
@@ -184,8 +184,10 @@ def _build_dossiers_for_batch(batch_id):
    records = []
    for document in documents:
        passport = passport_by_id.get(document["id"])
-        if not passport:
-            passport = {"document_id": document["id"], "doc_class": "unknown"}
+        # A missing or failed passport says extraction produced no usable evidence;
+        # do not misreport that technical failure as a missing contract number.
+        if not passport or passport.get("status") == "FAILED":
+            continue
        if passport.get("doc_class") == "ignored_accounting":
            continue
        counterparty = {