Обработка ошибок, показаны проблемные записи

lists/01-example-short.txt

@@ -1,3 +1,4 @@
+CHEVROLET	Tacuma 1.6	[A16DMS]	09.2000- DAEWOO	Rezzo/Tacuma 1.6	[A16DMS]	09.2000- \n DAEWOO	Espero 1.5 16V	[A15MF]	02.1995-	
 HYUNDAI	Accent 1.3	[G4E-A]	08.1999- рандомный \n text 	HYUNDAI	Accent 1.3	[G4EH]	08.2002-	 вфы вфы  HYUNDAI	Accent 1.4 GL		11.2005-	
 HYUNDAI	Accent 1.5	[G4EB]	08.2002- 
 HYUNDAI	Accent 1.6

parse.py

@@ -43,13 +43,17 @@ def prepare_records(content: str) -> list[str]:
     return [r.strip() for r in content.split(RECORD_DELIMITERS[0]) if r.strip()]
 
 
-def _normalize_obj(obj: dict) -> dict:
-    return {
+def _normalize_obj(obj: dict, original: str) -> dict:
+    result = {
         "brand": obj.get("brand", ""),
         "model": obj.get("model", ""),
         "engine_code": obj.get("engine_code", ""),
         "date_range": obj.get("date_range", ""),
     }
+    if not result["engine_code"].strip():
+        result["has_error"] = True
+        result["original"] = original.replace("\t", " ")
+    return result
 
 
 def extract_record(client: Client, record: str) -> list[dict]:
@@ -71,19 +75,19 @@ def extract_record(client: Client, record: str) -> list[dict]:
                     result = result.strip()[4:].strip()
             parsed = json.loads(result)
             if isinstance(parsed, list):
-                return [_normalize_obj(o) for o in parsed]
-            return [_normalize_obj(parsed)]
+                return [_normalize_obj(o, record) for o in parsed]
+            return [_normalize_obj(parsed, record)]
         except (json.JSONDecodeError, KeyError):
             if attempt < RETRY_COUNT:
                 time.sleep(RETRY_DELAY)
             else:
-                return [{"brand": "", "model": "", "engine_code": "", "date_range": record}]
+                return [{"brand": "", "model": "", "engine_code": "", "date_range": record, "has_error": True, "original": record.replace("\t", " ")}]
         except Exception:
             if attempt < RETRY_COUNT:
                 time.sleep(RETRY_DELAY)
             else:
                 raise
-    return [{"brand": "", "model": "", "engine_code": "", "date_range": record}]
+    return [{"brand": "", "model": "", "engine_code": "", "date_range": record, "has_error": True, "original": record.replace("\t", " ")}]
 
 
 def process_file(client: Client, filepath: pathlib.Path) -> bool:

res/01-example-short.json

@@ -1 +1,54 @@
-[]
+[
+  {
+    "brand": "CHEVROLET",
+    "model": "Tacuma 1.6",
+    "engine_code": "[A16DMS]",
+    "date_range": "09.2000-"
+  },
+  {
+    "brand": "DAEWOO",
+    "model": "Rezzo/Tacuma 1.6",
+    "engine_code": "[A16DMS]",
+    "date_range": "09.2000-"
+  },
+  {
+    "brand": "DAEWOO",
+    "model": "Espero 1.5 16V",
+    "engine_code": "[A15MF]",
+    "date_range": "02.1995-"
+  },
+  {
+    "brand": "HYUNDAI",
+    "model": "Accent 1.3",
+    "engine_code": "[G4E-A]",
+    "date_range": "08.1999- рандомный"
+  },
+  {
+    "brand": "HYUNDAI",
+    "model": "Accent 1.3",
+    "engine_code": "[G4EH]",
+    "date_range": "08.2002-"
+  },
+  {
+    "brand": "HYUNDAI",
+    "model": "Accent 1.4 GL",
+    "engine_code": "",
+    "date_range": "11.2005-",
+    "has_error": true,
+    "original": "HYUNDAI Accent 1.3 [G4E-A] 08.1999- рандомный \\n text  HYUNDAI Accent 1.3 [G4EH] 08.2002-  вфы вфы  HYUNDAI Accent 1.4 GL  11.2005-"
+  },
+  {
+    "brand": "HYUNDAI",
+    "model": "Accent 1.5",
+    "engine_code": "[G4EB]",
+    "date_range": "08.2002-"
+  },
+  {
+    "brand": "HYUNDAI",
+    "model": "Accent",
+    "engine_code": "",
+    "date_range": "",
+    "has_error": true,
+    "original": "HYUNDAI Accent 1.6"
+  }
+]