dtflow 0.5.0__tar.gz → 0.5.3__tar.gz

dtflow-0.5.3/CHANGELOG.md

@@ -0,0 +1,19 @@
+# Changelog
+
+## [0.5.2] - 2026-01-18
+
+### Miscellaneous
+
+- Bump version to 0.5.2
+- 添加 pre-commit 配置和发版脚本
+
+## [0.5.1] - 2026-01-18
+
+### Features
+
+- 优化 sample 命令文本预览显示
+
+### Testing
+
+- 添加测试运行说明
+- 补充 tail/token-stats/validate 性能测试

{dtflow-0.5.0 → dtflow-0.5.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dtflow
-Version: 0.5.0
+Version: 0.5.3
 Summary: A flexible data transformation tool for ML training formats (SFT, RLHF, Pretrain)
 Project-URL: Homepage, https://github.com/yourusername/DataTransformer
 Project-URL: Documentation, https://github.com/yourusername/DataTransformer#readme
@@ -32,16 +32,26 @@ Requires-Dist: orjson>=3.9.0
 Requires-Dist: polars>=0.20.0
 Requires-Dist: pyyaml>=5.4.0
 Requires-Dist: rich>=10.0.0
+Requires-Dist: tiktoken>=0.5.0
 Requires-Dist: typer>=0.9.0
 Provides-Extra: converters
 Requires-Dist: datasets>=2.0.0; extra == 'converters'
 Provides-Extra: dev
 Requires-Dist: black>=21.0; extra == 'dev'
+Requires-Dist: datasets>=2.0.0; extra == 'dev'
+Requires-Dist: datasketch>=1.5.0; extra == 'dev'
 Requires-Dist: flake8>=3.9.0; extra == 'dev'
+Requires-Dist: huggingface-hub>=0.20.0; extra == 'dev'
 Requires-Dist: isort>=5.9.0; extra == 'dev'
 Requires-Dist: mypy>=0.910; extra == 'dev'
+Requires-Dist: pyarrow; extra == 'dev'
 Requires-Dist: pytest-cov>=2.12.0; extra == 'dev'
 Requires-Dist: pytest>=6.0.0; extra == 'dev'
+Requires-Dist: rich>=10.0.0; extra == 'dev'
+Requires-Dist: scikit-learn>=0.24.0; extra == 'dev'
+Requires-Dist: tiktoken>=0.5.0; extra == 'dev'
+Requires-Dist: tokenizers>=0.15.0; extra == 'dev'
+Requires-Dist: toolong>=1.5.0; extra == 'dev'
 Provides-Extra: display
 Provides-Extra: docs
 Requires-Dist: myst-parser>=0.15.0; extra == 'docs'

{dtflow-0.5.0 → dtflow-0.5.3}/dtflow/__init__.py

@@ -26,6 +26,12 @@ from .converters import (  # LLaMA-Factory 扩展; ms-swift
     to_swift_vlm,
 )
 from .core import DataTransformer, DictWrapper, TransformError, TransformErrors
+from .framework import (
+    CompatibilityResult,
+    check_compatibility,
+    detect_format,
+    export_for,
+)
 from .presets import get_preset, list_presets
 from .schema import (
     Field,
@@ -38,12 +44,6 @@ from .schema import (
     sharegpt_schema,
     validate_data,
 )
-from .framework import (
-    CompatibilityResult,
-    check_compatibility,
-    detect_format,
-    export_for,
-)
 from .storage import load_data, sample_file, save_data
 from .streaming import StreamingTransformer, load_sharded, load_stream, process_shards
 from .tokenizers import (
@@ -60,7 +60,7 @@ from .tokenizers import (
     token_stats,
 )
 
-__version__ = "0.5.0"
+__version__ = "0.5.3"
 
 __all__ = [
     # core

{dtflow-0.5.0 → dtflow-0.5.3}/dtflow/cli/common.py

@@ -57,7 +57,7 @@ def _get_file_row_count(filepath: Path) -> Optional[int]:
     return None
 
 
-def _format_value(value: Any, max_len: int = 80) -> str:
+def _format_value(value: Any, max_len: int = 120) -> str:
     """格式化单个值，长文本截断。"""
     if value is None:
         return "[dim]null[/dim]"
@@ -66,18 +66,22 @@ def _format_value(value: Any, max_len: int = 80) -> str:
     if isinstance(value, (int, float)):
         return f"[cyan]{value}[/cyan]"
     if isinstance(value, str):
+        half_len = max_len // 2
         # 处理多行文本
         if "\n" in value:
             lines = value.split("\n")
-            if len(lines) > 3:
-                preview = lines[0][:max_len] + f"... [dim]({len(lines)} 行)[/dim]"
-            else:
-                preview = value.replace("\n", "\\n")
-                if len(preview) > max_len:
-                    preview = preview[:max_len] + "..."
+            preview = value.replace("\n", "\\n")
+            if len(preview) > max_len:
+                # 前半 + 省略标记 + 后半
+                head = preview[:half_len]
+                tail = preview[-half_len:]
+                return f'"{head} [yellow]<<<{len(lines)}行>>>[/yellow] {tail}"'
             return f'"{preview}"'
         if len(value) > max_len:
-            return f'"{value[:max_len]}..." [dim]({len(value)} 字符)[/dim]'
+            # 前半 + 省略标记 + 后半
+            head = value[:half_len]
+            tail = value[-half_len:]
+            return f'"{head} [yellow]<<<{len(value)}字符>>>[/yellow] {tail}"'
         return f'"{value}"'
     return str(value)
 
@@ -86,7 +90,7 @@ def _format_nested(
     value: Any,
     indent: str = "",
     is_last: bool = True,
-    max_len: int = 80,
+    max_len: int = 120,
 ) -> List[str]:
     """
     递归格式化嵌套结构，返回行列表。

{dtflow-0.5.0 → dtflow-0.5.3}/dtflow/cli/stats.py

@@ -465,34 +465,65 @@ def token_stats(
         return
 
     total = len(data)
-    print(f"   共 {total} 条数据")
-    print(f"🔢 统计 Token (模型: {model}, 字段: {field})...")
+    print(f"   共 {total:,} 条数据")
 
     # 检查字段类型并选择合适的统计方法（支持嵌套路径）
     sample = data[0]
     field_value = get_field_with_spec(sample, field)
 
+    # 尝试使用 rich 进度条
     try:
-        if isinstance(field_value, list) and field_value and isinstance(field_value[0], dict):
-            # messages 格式
-            from ..tokenizers import messages_token_stats
-
-            stats_result = messages_token_stats(data, messages_field=field, model=model)
-            _print_messages_token_stats(stats_result, detailed)
-        else:
-            # 普通文本字段
-            from ..tokenizers import token_stats as compute_token_stats
-
-            stats_result = compute_token_stats(data, fields=field, model=model)
-            _print_text_token_stats(stats_result, detailed)
-    except ImportError as e:
-        print(f"错误: {e}")
-        return
-    except Exception as e:
-        print(f"错误: 统计失败 - {e}")
-        import traceback
+        from rich.progress import Progress, SpinnerColumn, TextColumn, BarColumn, TaskProgressColumn
+
+        with Progress(
+            SpinnerColumn(),
+            TextColumn("[bold blue]统计 Token"),
+            BarColumn(),
+            TaskProgressColumn(),
+            TextColumn(f"(模型: {model})"),
+        ) as progress:
+            task = progress.add_task("", total=total)
+
+            def update_progress(current: int, total_count: int):
+                progress.update(task, completed=current)
+
+            if isinstance(field_value, list) and field_value and isinstance(field_value[0], dict):
+                from ..tokenizers import messages_token_stats
+
+                stats_result = messages_token_stats(
+                    data, messages_field=field, model=model, progress_callback=update_progress
+                )
+                _print_messages_token_stats(stats_result, detailed)
+            else:
+                from ..tokenizers import token_stats as compute_token_stats
+
+                stats_result = compute_token_stats(
+                    data, fields=field, model=model, progress_callback=update_progress
+                )
+                _print_text_token_stats(stats_result, detailed)
+
+    except ImportError:
+        # 没有 rich，显示简单进度
+        print(f"🔢 统计 Token (模型: {model}, 字段: {field})...")
+        try:
+            if isinstance(field_value, list) and field_value and isinstance(field_value[0], dict):
+                from ..tokenizers import messages_token_stats
+
+                stats_result = messages_token_stats(data, messages_field=field, model=model)
+                _print_messages_token_stats(stats_result, detailed)
+            else:
+                from ..tokenizers import token_stats as compute_token_stats
 
-        traceback.print_exc()
+                stats_result = compute_token_stats(data, fields=field, model=model)
+                _print_text_token_stats(stats_result, detailed)
+        except ImportError as e:
+            print(f"错误: {e}")
+            return
+        except Exception as e:
+            print(f"错误: 统计失败 - {e}")
+            import traceback
+
+            traceback.print_exc()
 
 
 def _print_messages_token_stats(stats: Dict[str, Any], detailed: bool) -> None:
@@ -505,21 +536,39 @@ def _print_messages_token_stats(stats: Dict[str, Any], detailed: bool) -> None:
         console = Console()
 
         # 概览
+        std = stats.get("std_tokens", 0)
         overview = (
             f"[bold]总样本数:[/bold] {stats['count']:,}\n"
             f"[bold]总 Token:[/bold] {stats['total_tokens']:,}\n"
-            f"[bold]平均 Token:[/bold] {stats['avg_tokens']:,}\n"
-            f"[bold]中位数:[/bold] {stats['median_tokens']:,}\n"
+            f"[bold]平均 Token:[/bold] {stats['avg_tokens']:,} (std: {std:.1f})\n"
             f"[bold]范围:[/bold] {stats['min_tokens']:,} - {stats['max_tokens']:,}"
         )
         console.print(Panel(overview, title="📊 Token 统计概览", expand=False))
 
+        # 百分位数表格
+        table = Table(title="📈 分布统计")
+        table.add_column("百分位", style="cyan", justify="center")
+        table.add_column("Token 数", justify="right")
+        percentiles = [
+            ("Min", stats["min_tokens"]),
+            ("P25", stats.get("p25", "-")),
+            ("P50 (中位数)", stats.get("median_tokens", "-")),
+            ("P75", stats.get("p75", "-")),
+            ("P90", stats.get("p90", "-")),
+            ("P95", stats.get("p95", "-")),
+            ("P99", stats.get("p99", "-")),
+            ("Max", stats["max_tokens"]),
+        ]
+        for name, val in percentiles:
+            table.add_row(name, f"{val:,}" if isinstance(val, int) else str(val))
+        console.print(table)
+
         if detailed:
-            # 详细统计
-            table = Table(title="📋 分角色统计")
-            table.add_column("角色", style="cyan")
-            table.add_column("Token 数", justify="right")
-            table.add_column("占比", justify="right")
+            # 分角色统计
+            role_table = Table(title="📋 分角色统计")
+            role_table.add_column("角色", style="cyan")
+            role_table.add_column("Token 数", justify="right")
+            role_table.add_column("占比", justify="right")
 
             total = stats["total_tokens"]
             for role, key in [
@@ -529,22 +578,27 @@ def _print_messages_token_stats(stats: Dict[str, Any], detailed: bool) -> None:
             ]:
                 tokens = stats.get(key, 0)
                 pct = tokens / total * 100 if total > 0 else 0
-                table.add_row(role, f"{tokens:,}", f"{pct:.1f}%")
+                role_table.add_row(role, f"{tokens:,}", f"{pct:.1f}%")
 
-            console.print(table)
+            console.print(role_table)
             console.print(f"\n平均对话轮数: {stats.get('avg_turns', 0)}")
 
     except ImportError:
         # 没有 rich，使用普通打印
+        std = stats.get("std_tokens", 0)
         print(f"\n{'=' * 40}")
         print("📊 Token 统计概览")
         print(f"{'=' * 40}")
         print(f"总样本数: {stats['count']:,}")
         print(f"总 Token: {stats['total_tokens']:,}")
-        print(f"平均 Token: {stats['avg_tokens']:,}")
-        print(f"中位数: {stats['median_tokens']:,}")
+        print(f"平均 Token: {stats['avg_tokens']:,} (std: {std:.1f})")
         print(f"范围: {stats['min_tokens']:,} - {stats['max_tokens']:,}")
 
+        print(f"\n📈 百分位分布:")
+        print(f"  P25: {stats.get('p25', '-'):,}  P50: {stats.get('median_tokens', '-'):,}")
+        print(f"  P75: {stats.get('p75', '-'):,}  P90: {stats.get('p90', '-'):,}")
+        print(f"  P95: {stats.get('p95', '-'):,}  P99: {stats.get('p99', '-'):,}")
+
         if detailed:
             print(f"\n{'=' * 40}")
             print("📋 分角色统计")
@@ -566,24 +620,48 @@ def _print_text_token_stats(stats: Dict[str, Any], detailed: bool) -> None:
     try:
         from rich.console import Console
         from rich.panel import Panel
+        from rich.table import Table
 
         console = Console()
 
+        std = stats.get("std_tokens", 0)
         overview = (
             f"[bold]总样本数:[/bold] {stats['count']:,}\n"
             f"[bold]总 Token:[/bold] {stats['total_tokens']:,}\n"
-            f"[bold]平均 Token:[/bold] {stats['avg_tokens']:.1f}\n"
-            f"[bold]中位数:[/bold] {stats['median_tokens']:,}\n"
+            f"[bold]平均 Token:[/bold] {stats['avg_tokens']:.1f} (std: {std:.1f})\n"
             f"[bold]范围:[/bold] {stats['min_tokens']:,} - {stats['max_tokens']:,}"
         )
         console.print(Panel(overview, title="📊 Token 统计", expand=False))
 
+        # 百分位数表格
+        table = Table(title="📈 分布统计")
+        table.add_column("百分位", style="cyan", justify="center")
+        table.add_column("Token 数", justify="right")
+        percentiles = [
+            ("Min", stats["min_tokens"]),
+            ("P25", stats.get("p25", "-")),
+            ("P50 (中位数)", stats.get("median_tokens", "-")),
+            ("P75", stats.get("p75", "-")),
+            ("P90", stats.get("p90", "-")),
+            ("P95", stats.get("p95", "-")),
+            ("P99", stats.get("p99", "-")),
+            ("Max", stats["max_tokens"]),
+        ]
+        for name, val in percentiles:
+            table.add_row(name, f"{val:,}" if isinstance(val, int) else str(val))
+        console.print(table)
+
     except ImportError:
+        std = stats.get("std_tokens", 0)
         print(f"\n{'=' * 40}")
         print("📊 Token 统计")
         print(f"{'=' * 40}")
         print(f"总样本数: {stats['count']:,}")
         print(f"总 Token: {stats['total_tokens']:,}")
-        print(f"平均 Token: {stats['avg_tokens']:.1f}")
-        print(f"中位数: {stats['median_tokens']:,}")
+        print(f"平均 Token: {stats['avg_tokens']:.1f} (std: {std:.1f})")
         print(f"范围: {stats['min_tokens']:,} - {stats['max_tokens']:,}")
+
+        print(f"\n📈 百分位分布:")
+        print(f"  P25: {stats.get('p25', '-'):,}  P50: {stats.get('median_tokens', '-'):,}")
+        print(f"  P75: {stats.get('p75', '-'):,}  P90: {stats.get('p90', '-'):,}")
+        print(f"  P95: {stats.get('p95', '-'):,}  P99: {stats.get('p99', '-'):,}")

{dtflow-0.5.0 → dtflow-0.5.3}/dtflow/converters.py

@@ -4,7 +4,7 @@
 提供与 HuggingFace datasets 等常用格式的互转功能。
 """
 
-from typing import Any, Callable, Dict, List, Optional, Union
+from typing import Any, Callable, Dict, List, Optional
 
 
 def to_hf_dataset(data: List[Dict[str, Any]]):
@@ -143,14 +143,16 @@ def to_openai_batch(
         >>> batch_input = dt.to(to_openai_batch(model="gpt-4o"))
     """
 
-    def transform(item, idx=[0]) -> dict:
+    counter = {"idx": 0}
+
+    def transform(item) -> dict:
         messages = item.get(messages_field, []) if hasattr(item, "get") else item[messages_field]
 
         if custom_id_field:
             custom_id = item.get(custom_id_field) if hasattr(item, "get") else item[custom_id_field]
         else:
-            custom_id = f"request-{idx[0]}"
-            idx[0] += 1
+            custom_id = f"request-{counter['idx']}"
+            counter["idx"] += 1
 
         return {
             "custom_id": str(custom_id),
@@ -196,7 +198,7 @@ def to_llama_factory(
     """
 
     def transform(item) -> dict:
-        get = lambda f: (item.get(f, "") if hasattr(item, "get") else item.get(f, ""))
+        get = lambda f: item.get(f, "") if hasattr(item, "get") else getattr(item, f, "")
 
         result = {
             "instruction": get(instruction_field),
@@ -248,7 +250,7 @@ def to_axolotl(
         conversations = (
             item.get(conversations_field, [])
             if hasattr(item, "get")
-            else item.get(conversations_field, [])
+            else getattr(item, conversations_field, [])
         )
 
         # 如果已经是正确格式，直接返回
@@ -257,7 +259,9 @@ def to_axolotl(
                 return {"conversations": conversations}
 
         # 尝试从 messages 格式转换
-        messages = item.get("messages", []) if hasattr(item, "get") else item.get("messages", [])
+        messages = (
+            item.get("messages", []) if hasattr(item, "get") else getattr(item, "messages", [])
+        )
         if messages:
             role_map = {"user": "human", "assistant": "gpt", "system": "system"}
             conversations = [
@@ -312,7 +316,7 @@ def to_llama_factory_sharegpt(
     }
 
     def transform(item) -> dict:
-        get = lambda f: (item.get(f, "") if hasattr(item, "get") else item.get(f, ""))
+        get = lambda f: item.get(f, "") if hasattr(item, "get") else getattr(item, f, "")
         messages = get(messages_field) or []
 
         conversations = []
@@ -385,7 +389,7 @@ def to_llama_factory_vlm(
     """
 
     def transform(item) -> dict:
-        get = lambda f: item.get(f) if hasattr(item, "get") else item.get(f)
+        get = lambda f: item.get(f) if hasattr(item, "get") else getattr(item, f, None)
         messages = get(messages_field) or []
 
         instruction = ""
@@ -467,7 +471,7 @@ def to_llama_factory_vlm_sharegpt(
     role_map = {"user": "human", "assistant": "gpt", "system": "system"}
 
     def transform(item) -> dict:
-        get = lambda f: item.get(f) if hasattr(item, "get") else item.get(f)
+        get = lambda f: item.get(f) if hasattr(item, "get") else getattr(item, f, None)
         messages = get(messages_field) or []
 
         conversations = []
@@ -541,7 +545,7 @@ def to_swift_messages(
     """
 
     def transform(item) -> dict:
-        get = lambda f: item.get(f) if hasattr(item, "get") else item.get(f)
+        get = lambda f: item.get(f) if hasattr(item, "get") else getattr(item, f, None)
         messages = get(messages_field) or []
 
         # 复制 messages，避免修改原数据
@@ -600,7 +604,7 @@ def to_swift_query_response(
     """
 
     def transform(item) -> dict:
-        get = lambda f: item.get(f) if hasattr(item, "get") else item.get(f)
+        get = lambda f: item.get(f) if hasattr(item, "get") else getattr(item, f, None)
 
         query = get(query_field)
         response = get(response_field)
@@ -693,7 +697,7 @@ def to_swift_vlm(
     """
 
     def transform(item) -> dict:
-        get = lambda f: item.get(f) if hasattr(item, "get") else item.get(f)
+        get = lambda f: item.get(f) if hasattr(item, "get") else getattr(item, f, None)
         messages = get(messages_field) or []
 
         result_messages = []

{dtflow-0.5.0 → dtflow-0.5.3}/dtflow/core.py

@@ -793,19 +793,29 @@ class DataTransformer:
             seed: 随机种子
 
         Returns:
-            (train, test) 两个 DataTransformer
+            (train, test) 两个 DataTransformer，各自拥有独立的血缘追踪器
         """
         data = self.shuffle(seed).data
         split_idx = int(len(data) * ratio)
 
-        # 分割后血缘追踪器各自独立
+        # 分割后血缘追踪器各自独立（使用深拷贝避免相互影响）
         tracker = self._lineage_tracker
+        train_tracker = None
+        test_tracker = None
+
         if tracker:
             tracker.record("split", {"ratio": ratio, "seed": seed}, len(self._data), len(data))
+            # 为每个子数据集创建独立的追踪器副本
+            train_tracker = tracker.copy()
+            train_tracker.record("split_part", {"part": "train", "ratio": ratio}, len(data), split_idx)
+            test_tracker = tracker.copy()
+            test_tracker.record(
+                "split_part", {"part": "test", "ratio": 1 - ratio}, len(data), len(data) - split_idx
+            )
 
         return (
-            DataTransformer(data[:split_idx], _lineage_tracker=tracker),
-            DataTransformer(data[split_idx:], _lineage_tracker=tracker),
+            DataTransformer(data[:split_idx], _lineage_tracker=train_tracker),
+            DataTransformer(data[split_idx:], _lineage_tracker=test_tracker),
         )
 
     # ============ 并行处理 ============
@@ -815,6 +825,7 @@ class DataTransformer:
         func: Callable[[Dict], Any],
         workers: Optional[int] = None,
         chunksize: int = 1000,
+        timeout: Optional[float] = None,
     ) -> List[Any]:
         """
         并行执行转换函数（使用多进程）。
@@ -825,24 +836,46 @@ class DataTransformer:
             func: 转换函数，接收原始 dict，返回转换结果
             workers: 进程数，默认为 CPU 核心数
             chunksize: 每个进程处理的数据块大小
+            timeout: 超时时间（秒），None 表示无超时
 
         Returns:
             转换后的结果列表
 
+        Raises:
+            TypeError: 如果 func 无法被 pickle（如 lambda 函数）
+            RuntimeError: 如果子进程执行出错或超时
+
         Examples:
             >>> def transform(item):
             ...     return {"id": item["id"], "text": item["text"].upper()}
             >>> results = dt.map_parallel(transform)
         """
-        from multiprocessing import Pool, cpu_count
+        from multiprocessing import Pool, TimeoutError, cpu_count
+        import pickle
 
         if not self._data:
             return []
 
+        # 检查函数是否可 pickle
+        try:
+            pickle.dumps(func)
+        except (pickle.PicklingError, AttributeError, TypeError) as e:
+            func_name = getattr(func, "__name__", str(func))
+            raise TypeError(
+                f"函数 '{func_name}' 无法被 pickle，不能用于并行处理。"
+                f"请使用模块级函数而非 lambda 或闭包。错误: {e}"
+            ) from e
+
         workers = workers or cpu_count()
 
-        with Pool(workers) as pool:
-            results = pool.map(func, self._data, chunksize=chunksize)
+        try:
+            with Pool(workers) as pool:
+                async_result = pool.map_async(func, self._data, chunksize=chunksize)
+                results = async_result.get(timeout=timeout)
+        except TimeoutError:
+            raise RuntimeError(f"并行处理超时（{timeout}秒）")
+        except Exception as e:
+            raise RuntimeError(f"并行处理失败: {type(e).__name__}: {e}") from e
 
         return results
 
@@ -851,6 +884,7 @@ class DataTransformer:
         func: Callable[[Dict], bool],
         workers: Optional[int] = None,
         chunksize: int = 1000,
+        timeout: Optional[float] = None,
     ) -> "DataTransformer":
         """
         并行执行过滤函数（使用多进程）。
@@ -861,24 +895,46 @@ class DataTransformer:
             func: 过滤函数，接收原始 dict，返回 True 保留
             workers: 进程数，默认为 CPU 核心数
             chunksize: 每个进程处理的数据块大小
+            timeout: 超时时间（秒），None 表示无超时
 
         Returns:
             过滤后的新 DataTransformer
 
+        Raises:
+            TypeError: 如果 func 无法被 pickle（如 lambda 函数）
+            RuntimeError: 如果子进程执行出错或超时
+
         Examples:
             >>> def is_valid(item):
             ...     return len(item["text"]) > 10
             >>> filtered = dt.filter_parallel(is_valid)
         """
-        from multiprocessing import Pool, cpu_count
+        from multiprocessing import Pool, TimeoutError, cpu_count
+        import pickle
 
         if not self._data:
             return DataTransformer([])
 
+        # 检查函数是否可 pickle
+        try:
+            pickle.dumps(func)
+        except (pickle.PicklingError, AttributeError, TypeError) as e:
+            func_name = getattr(func, "__name__", str(func))
+            raise TypeError(
+                f"函数 '{func_name}' 无法被 pickle，不能用于并行处理。"
+                f"请使用模块级函数而非 lambda 或闭包。错误: {e}"
+            ) from e
+
         workers = workers or cpu_count()
 
-        with Pool(workers) as pool:
-            mask = pool.map(func, self._data, chunksize=chunksize)
+        try:
+            with Pool(workers) as pool:
+                async_result = pool.map_async(func, self._data, chunksize=chunksize)
+                mask = async_result.get(timeout=timeout)
+        except TimeoutError:
+            raise RuntimeError(f"并行处理超时（{timeout}秒）")
+        except Exception as e:
+            raise RuntimeError(f"并行处理失败: {type(e).__name__}: {e}") from e
 
         filtered = [item for item, keep in zip(self._data, mask) if keep]
         return DataTransformer(filtered)

{dtflow-0.5.0 → dtflow-0.5.3}/dtflow/lineage.py

@@ -237,6 +237,23 @@ class LineageTracker:
 
         return lineage_path
 
+    def copy(self) -> "LineageTracker":
+        """
+        创建追踪器的深拷贝。
+
+        用于 split() 等场景，确保子数据集有独立的血缘追踪。
+
+        Returns:
+            新的 LineageTracker 实例
+        """
+        import copy as copy_module
+
+        new_tracker = LineageTracker.__new__(LineageTracker)
+        new_tracker.source_path = self.source_path
+        new_tracker.source_lineage = self.source_lineage  # LineageRecord 是不可变的，可共享
+        new_tracker.operations = copy_module.deepcopy(self.operations)
+        return new_tracker
+
 
 def _sanitize_params(params: Dict[str, Any]) -> Dict[str, Any]:
     """