0.5.1 - Move

0.5 - Разбор нескольких моделей в одной строке
2026-03-12 17:57:13 +03:00 · 2026-03-12 17:56:49 +03:00
3 changed files with 26 additions and 130 deletions
@@ -1,16 +1,3 @@
-HYUNDAI	Accent 1.3	[G4E-A]	08.1999-	
-HYUNDAI	Accent 1.3	[G4EH]	08.2002-	 
-HYUNDAI	Accent 1.4 GL		11.2005-	
+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	[G4ED-G]	12.2002-	
-HYUNDAI	Accent 1.6 GLS		11.2005-	
-HYUNDAI	Coupe 1.6 16V	[G4ED-G]	03.2002-	
-HYUNDAI	Coupe 2.0 16V	[G4GF]	08.2002-	
-HYUNDAI	Coupe 2.0 GLS		08.2002-	
-KIA	Rio 1.6 16V	[G4ED]	03.2005-
-MERCEDES-BENZ CABRIOLET (W111) 280 SE 3.5 (111.025) [M 116.980] 	01/1969 - 12/1971
-MERCEDES-BENZ COUPE (W111) 280 SE 3.5 (111.026) [M 116.980] 	01/1969 - 12/1971
-MERCEDES-BENZ S-CLASS (W108, W109) 280 SE,SEL 3.5 (108.057, 108.058) [M 116.980]	03/1971 - 08/1972
-MERCEDES-BENZ S-CLASS (W108, W109) 300 SEL 3.5 (109.056) [M 116.981] 01/1970 - 07/1972
-MERCEDES-BENZ S-CLASS (W116) 350 SE,SEL (116.028, 116.029) [M 116.983]	08/1972 - 02/1976
-MERCEDES-BENZ SL купе (C107) 350 SLC (107.023) [M 116.982, M 116.984]	01/1972 - 02/1976
+HYUNDAI	Accent 1.6
@@ -31,7 +31,8 @@ SHOW_IO = os.environ.get("SHOW_IO", "1").lower() in ("1", "true", "yes")
 RECORD_DELIMITERS = ("\n", "\r\n", "\r")

 EXTRACT_PROMPT = """Extract brand, model, engine_code, date_range from the text.
-Return JSON object only: {"brand":"","model":"","engine_code":"","date_range":""}
+If 1 record — return JSON object: {"brand":"","model":"","engine_code":"","date_range":""}
+If 2+ records on the line — return JSON array: [{"brand":"",...},{"brand":"",...}]
 Empty fields as "". No other text."""


@@ -42,8 +43,17 @@ def prepare_records(content: str) -> list[str]:
    return [r.strip() for r in content.split(RECORD_DELIMITERS[0]) if r.strip()]


-def extract_record(client: Client, record: str) -> dict:
-    """2) Отправка промпт + 1 запись в нейронку. 3) Получение ответа."""
+def _normalize_obj(obj: dict) -> dict:
+    return {
+        "brand": obj.get("brand", ""),
+        "model": obj.get("model", ""),
+        "engine_code": obj.get("engine_code", ""),
+        "date_range": obj.get("date_range", ""),
+    }
+
+
+def extract_record(client: Client, record: str) -> list[dict]:
+    """2) Отправка промпт + 1 запись в нейронку. 3) Получение ответа (1 или более объектов)."""
    for attempt in range(1, RETRY_COUNT + 1):
        try:
            response = client.chat(
@@ -59,24 +69,21 @@ def extract_record(client: Client, record: str) -> dict:
                result = result.split("```")[1]
                if result.strip().lower().startswith("json"):
                    result = result.strip()[4:].strip()
-            obj = json.loads(result)
-            return {
-                "brand": obj.get("brand", ""),
-                "model": obj.get("model", ""),
-                "engine_code": obj.get("engine_code", ""),
-                "date_range": obj.get("date_range", ""),
-            }
-        except (json.JSONDecodeError, KeyError) as e:
+            parsed = json.loads(result)
+            if isinstance(parsed, list):
+                return [_normalize_obj(o) for o in parsed]
+            return [_normalize_obj(parsed)]
+        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}]
        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}]


 def process_file(client: Client, filepath: pathlib.Path) -> bool:
@@ -96,18 +103,18 @@ def process_file(client: Client, filepath: pathlib.Path) -> bool:

    for i, record in enumerate(records, 1):
        print(f"  [{i}/{total}] ", end="", flush=True)
-        obj = extract_record(client, record)
-        data.append(obj)
+        objs = extract_record(client, record)
+        data.extend(objs)
        out_path.write_text(json.dumps(data, ensure_ascii=False, indent=2), encoding="utf-8")
        if SHOW_IO:
            inp = record[:60] + "…" if len(record) > 60 else record
-            out = json.dumps(obj, ensure_ascii=False)
+            out = json.dumps(objs, ensure_ascii=False)
            print(f"вход: {inp}", flush=True)
            print(f"      выход: {out}", flush=True)
        else:
            print("✓", flush=True)
    elapsed = time.perf_counter() - t0
-    print(f"  {filepath.name} — {total} записей, {elapsed:.1f}s — OK")
+    print(f"  {filepath.name} — {len(data)} объектов, {elapsed:.1f}s — OK")
    return True


@@ -1,98 +0,0 @@
-[
-  {
-    "brand": "HYUNDAI",
-    "model": "Accent 1.3",
-    "engine_code": "G4E-A",
-    "date_range": "08.1999-"
-  },
-  {
-    "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-"
-  },
-  {
-    "brand": "HYUNDAI",
-    "model": "Accent 1.5",
-    "engine_code": "G4EB",
-    "date_range": "08.2002-"
-  },
-  {
-    "brand": "HYUNDAI",
-    "model": "Accent 1.6",
-    "engine_code": "G4ED-G",
-    "date_range": "12.2002-"
-  },
-  {
-    "brand": "HYUNDAI",
-    "model": "Accent 1.6 GLS",
-    "engine_code": "",
-    "date_range": "11.2005-"
-  },
-  {
-    "brand": "HYUNDAI",
-    "model": "Coupe 1.6 16V",
-    "engine_code": "G4ED-G",
-    "date_range": "03.2002-"
-  },
-  {
-    "brand": "HYUNDAI",
-    "model": "Coupe 2.0 16V",
-    "engine_code": "G4GF",
-    "date_range": "08.2002-"
-  },
-  {
-    "brand": "HYUNDAI",
-    "model": "Coupe 2.0 GLS",
-    "engine_code": "",
-    "date_range": "08.2002-"
-  },
-  {
-    "brand": "KIA",
-    "model": "Rio 1.6 16V",
-    "engine_code": "G4ED",
-    "date_range": "03.2005-"
-  },
-  {
-    "brand": "MERCEDES-BENZ",
-    "model": "CABRIOLET (W111) 280 SE 3.5",
-    "engine_code": "111.025",
-    "date_range": "01/1969 - 12/1971"
-  },
-  {
-    "brand": "MERCEDES-BENZ",
-    "model": "COUPE (W111) 280 SE 3.5",
-    "engine_code": "111.026",
-    "date_range": "01/1969 - 12/1971"
-  },
-  {
-    "brand": "MERCEDES-BENZ",
-    "model": "S-CLASS (W108, W109) 280 SE,SEL 3.5",
-    "engine_code": "M 116.980",
-    "date_range": "03/1971 - 08/1972"
-  },
-  {
-    "brand": "MERCEDES-BENZ",
-    "model": "S-CLASS (W108, W109) 300 SEL 3.5",
-    "engine_code": "109.056 M 116.981",
-    "date_range": "01/1970 - 07/1972"
-  },
-  {
-    "brand": "MERCEDES-BENZ",
-    "model": "S-CLASS (W116) 350 SE,SEL",
-    "engine_code": "M 116.983",
-    "date_range": "08/1972 - 02/1976"
-  },
-  {
-    "brand": "MERCEDES-BENZ",
-    "model": "SL купе (C107) 350 SLC (107.023)",
-    "engine_code": "M 116.982, M 116.984",
-    "date_range": "01/1972 - 02/1976"
-  }
-]
Author	SHA1	Message	Date
atarasov	3502d3876c	0.5.1 - Move	2026-03-12 17:57:13 +03:00
atarasov	19365ce4f7	0.5 - Разбор нескольких моделей в одной строке	2026-03-12 17:56:49 +03:00