v1.0.162: auto-classify backend — classify/grouping services, 7 new columns, /classify-batch, /api/groups, /apply-groups

deploy/services/grouping.py

@@ -0,0 +1,95 @@
+"""Grouping service — match classified documents into contract groups."""
+import re
+from db import documents as db_docs
+from db import contracts as db_contracts
+from db import supplements as db_supplements
+
+
+def normalize_number(num):
+    """Normalize contract number for matching: uppercase, only letters/digits."""
+    if not num:
+        return ""
+    return re.sub(r"[^A-Z0-9А-Я]", "", num.upper())
+
+
+def group_documents(batch_id):
+    """Group classified documents by contract. Returns list of groups."""
+    docs = db_docs.list_by_batch(batch_id)
+    classified = [d for d in docs if d.get("classify_status") == "classified"]
+
+    # Separate contracts and supplements
+    contracts_list = []
+    supplements_list = []
+
+    for d in classified:
+        if d.get("doc_type") == "contract":
+            contracts_list.append(d)
+        else:
+            supplements_list.append(d)
+
+    groups = []
+
+    # Each contract becomes a group
+    for c in contracts_list:
+        group = {
+            "contract_number": c.get("own_number") or c.get("filename", ""),
+            "counterparty": c.get("counterparty") or "",
+            "documents": [c],
+        }
+        # Find supplements matching this contract
+        c_norm = normalize_number(c.get("own_number"))
+        for s in supplements_list:
+            parent = normalize_number(s.get("parent_number") or "")
+            own = normalize_number(s.get("own_number") or "")
+            if parent == c_norm or own == c_norm:
+                if s not in group["documents"]:
+                    group["documents"].append(s)
+
+        # Sort by date
+        group["documents"].sort(key=lambda x: x.get("doc_date") or "")
+
+        # Remove matched supplements from the pool
+        for s in group["documents"]:
+            if s in supplements_list:
+                supplements_list.remove(s)
+
+        groups.append(group)
+
+    # Unmatched documents → "unresolved" group
+    unmatched = [s for s in supplements_list if s.get("doc_type") != "contract"]
+    unmatched += [d for d in classified if d.get("classify_status") != "classified"]
+
+    if unmatched or len(contracts_list) == 0:
+        # Put unmatched into a separate group
+        if unmatched:
+            groups.append({
+                "contract_number": "__unresolved__",
+                "counterparty": "",
+                "documents": unmatched,
+            })
+
+    return {"ok": True, "groups": groups, "total_docs": len(classified)}
+
+
+def apply_groups(batch_id, groups_data):
+    """Apply confirmed groups: create contracts + supplements."""
+    created = 0
+    for g in groups_data:
+        contract_number = g.get("contract_number", "")
+        if contract_number == "__unresolved__":
+            continue
+        counterparty = g.get("counterparty", "")
+        docs = g.get("documents", [])
+
+        # Create contract
+        c = db_contracts.insert(contract_number, counterparty)
+        contract_id = c["id"]
+
+        # Create supplements in order
+        for i, d in enumerate(docs):
+            doc_id = d.get("id")
+            supp_type = "initial" if i == 0 else "additional"
+            db_supplements.insert(contract_id, doc_id, supp_type)
+            created += 1
+
+    return {"ok": True, "created": created}