feat: wire v3 dossier API and reconciliation

site/services/extractor.py

@@ -306,10 +306,22 @@ def extract_document(document, llm_client=None, batch_id=None):
         return {"ok": False, "document_id": document_id, "error": str(error)}
 
 
-def extract_batch(batch_id, llm_client=None):
-    """Обработать все документы batch и вернуть сводку без batch-wide failure."""
+def extract_batch(batch_id, llm_client=None, progress_callback=None):
+    """Обработать документы batch и вернуть сводку без batch-wide failure.
+
+    Optional progress callback получает (completed_count, document_result) после
+    каждого документа; ошибки callback не должны останавливать обработку batch.
+    """
     documents = db_docs.list_for_extraction(batch_id)
-    results = [extract_document(document, llm_client=llm_client, batch_id=batch_id) for document in documents]
+    results = []
+    for completed, document in enumerate(documents, start=1):
+        result = extract_document(document, llm_client=llm_client, batch_id=batch_id)
+        results.append(result)
+        if progress_callback is not None:
+            try:
+                progress_callback(completed, result)
+            except Exception:
+                log.exception("Extraction progress callback failed for batch %s", batch_id)
     done = sum(1 for result in results if result["ok"])
     failed = len(results) - done
     classes = {}

site/services/timeline_builder.py

@@ -185,7 +185,39 @@ def _append_interval(dossier_id, service_item, row, section, document,
     })
 
 
-def _process_document(dossier, document, passport, items, counts):
+def _resolved_option_for(document_id, row, resolved_exceptions):
+    """Найти сохранённый выбор для точной строки и документа lineage исключения."""
+    quote = _row_quote(row)
+    for issue in resolved_exceptions:
+        if (
+            issue.get("stage") == "lineage_matching"
+            and issue.get("document_id") == document_id
+            and issue.get("source_quote") == quote
+        ):
+            resolution = issue.get("resolution")
+            if isinstance(resolution, dict):
+                option = resolution.get("selected_option")
+                return option if isinstance(option, dict) else None
+    return None
+
+
+def _reusable_service_item(identity, service_items):
+    """Найти прежнюю устойчивую identity, не включая будущие строки в matching."""
+    family = identity.get("service_family")
+    anchor = identity.get("instance_anchor") or ""
+    candidates = [item for item in service_items if item.get("service_family") == family]
+    if anchor:
+        candidates = [item for item in candidates if item.get("instance_anchor", "") == anchor]
+    else:
+        candidates = [
+            item for item in candidates
+            if item.get("canonical_name") == identity.get("canonical_name")
+        ]
+    return candidates[0] if len(candidates) == 1 else None
+
+
+def _process_document(dossier, document, passport, items, counts,
+                      resolved_exceptions=None, reusable_items=None):
     """Reconcile every section of one passport against current dossier services."""
     doc_id = document["document_id"]
     mode = passport.get("intent_mode")
@@ -216,23 +248,47 @@ def _process_document(dossier, document, passport, items, counts):
                 effective_dates.append(start)
             identity = analyze_service_row(row.get("raw_name"), charge_type)
             match = match_service_item(identity, items)
+            if match["status"] == "AMBIGUOUS":
+                selected_option = _resolved_option_for(
+                    doc_id, row, resolved_exceptions or [],
+                )
+                if selected_option:
+                    selected_id = selected_option.get("service_item_id")
+                    candidates = [item for item in match["candidates"] if item.get("id") == selected_id]
+                    if not candidates:
+                        candidates = [
+                            item for item in match["candidates"]
+                            if item.get("service_family") == selected_option.get("service_family")
+                            and item.get("instance_anchor", "") == selected_option.get("instance_anchor", "")
+                        ]
+                    if len(candidates) == 1:
+                        match = {
+                            "status": "MATCHED", "service_item": candidates[0],
+                            "candidates": candidates,
+                        }
             if match["status"] == "AMBIGUOUS":
                 _add_lineage_exception(
                     dossier["id"], dossier.get("batch_id"), doc_id,
                     "AMBIGUOUS_SERVICE_ITEM", _row_quote(row),
-                    options=[{"service_item_id": item["id"], "canonical_name": item.get("canonical_name")}
-                             for item in match["candidates"]],
+                    options=[{
+                        "service_item_id": item["id"],
+                        "service_family": item.get("service_family"),
+                        "instance_anchor": item.get("instance_anchor", ""),
+                        "canonical_name": item.get("canonical_name"),
+                    } for item in match["candidates"]],
                 )
                 counts["ambiguous"] += 1
                 continue
             service_item = match["service_item"]
             if service_item is None:
-                service_item = db_timeline.insert_service_item(
-                    dossier_id=dossier["id"],
-                    service_family=identity["service_family"],
-                    canonical_name=identity["canonical_name"] or "(empty)",
-                    instance_anchor=identity["instance_anchor"],
-                )
+                service_item = _reusable_service_item(identity, reusable_items or [])
+                if service_item is None:
+                    service_item = db_timeline.insert_service_item(
+                        dossier_id=dossier["id"],
+                        service_family=identity["service_family"],
+                        canonical_name=identity["canonical_name"] or "(empty)",
+                        instance_anchor=identity["instance_anchor"],
+                    )
                 items.append(service_item)
 
             open_intervals = [
@@ -351,7 +407,11 @@ def build_dossier_timeline(dossier_id, contract_id=None):
     if dossier is None:
         return {"ok": False, "error": "dossier not found", "dossier_id": dossier_id}
     documents = _ordered_documents(dossier_id)
+    resolved_exceptions = db_exceptions.list_exceptions(
+        dossier_id=dossier_id, status="RESOLVED",
+    )
     items = []
+    reusable_items = db_timeline.list_service_items(dossier_id)
     counts = {"created": 0, "updated": 0, "unchanged": 0, "closed": 0, "ambiguous": 0}
 
     with transaction():
@@ -360,10 +420,13 @@ def build_dossier_timeline(dossier_id, contract_id=None):
                WHERE dossier_id = %s AND stage = 'lineage_matching' AND status = 'OPEN'""",
             (dossier_id,),
         )
-        execute("DELETE FROM service_items WHERE dossier_id = %s", (dossier_id,))
+        execute("DELETE FROM timeline_intervals WHERE dossier_id = %s", (dossier_id,))
         for document in documents:
             passport = _passport_for(document)
-            _process_document(dossier, document, passport, items, counts)
+            _process_document(
+                dossier, document, passport, items, counts,
+                resolved_exceptions, reusable_items,
+            )
 
     if contract_id:
         sync_legacy_spec_current(dossier_id, contract_id)