xparse-client 0.3.0b1__tar.gz → 0.3.0b3__tar.gz

{xparse_client-0.3.0b1/xparse_client.egg-info → xparse_client-0.3.0b3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: xparse-client
-Version: 0.3.0b1
+Version: 0.3.0b3
 Summary: 面向 Agent 和 RAG 的文档处理 Pipeline 客户端
 Author-email: INTSIG-TEXTIN <support@textin.com>
 License-Expression: MIT

{xparse_client-0.3.0b1 → xparse_client-0.3.0b3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "xparse-client"
-version = "0.3.0b1"
+version = "0.3.0b3"
 description = "面向 Agent 和 RAG 的文档处理 Pipeline 客户端"
 readme = "README.md"
 license = "MIT"

{xparse_client-0.3.0b1 → xparse_client-0.3.0b3}/xparse_client/api/local.py

@@ -28,7 +28,7 @@ from typing import TYPE_CHECKING, Callable, Literal
 
 from .._base import BaseAPI
 from ..models.local import FailedFile, WorkflowResult
-from ..models.pipeline import PipelineStage
+from ..models.pipeline import PipelineConfig, PipelineStage
 
 if TYPE_CHECKING:
     pass
@@ -55,6 +55,7 @@ class Local(BaseAPI):
         destination,  # Union[LocalDestination, S3Destination, MilvusDestination, QdrantDestination]
         stages: list[PipelineStage],
         *,
+        pipeline_config: PipelineConfig | None = None,
         progress_callback: Callable[[int, int, str], None] | None = None,
         on_error: Literal["stop", "continue", "retry"] = "stop",
         max_retries: int = 3,
@@ -75,6 +76,7 @@ class Local(BaseAPI):
             source: 数据源（本地 Connector 对象）
             destination: 输出目的地（本地 Connector 对象）
             stages: 处理阶段列表
+            pipeline_config: Pipeline 配置（包含中间结果保存配置）
             progress_callback: 进度回调函数 (current, total, message) -> None
             on_error: 错误处理策略 ("stop"|"continue"|"retry")
             max_retries: 最大重试次数（on_error="retry" 时生效）
@@ -125,20 +127,46 @@ class Local(BaseAPI):
 
                 # 调用 Pipeline API
                 from .pipeline import PipelineAPI
+
                 pipeline_api = PipelineAPI(self._config, self._http)
                 result = pipeline_api.execute(
                     file=file_bytes,
                     filename=filename,
                     stages=stages,
+                    config=pipeline_config,
                     data_source=data_source,
                 )
 
+                # 处理中间结果
+                if (
+                    pipeline_config
+                    and pipeline_config.include_intermediate_results
+                    and pipeline_config.intermediate_results_destination
+                    and hasattr(result, "intermediate_results")
+                    and result.intermediate_results
+                ):
+                    for stage_result in result.intermediate_results:
+                        stage_name = stage_result.get("stage")
+                        elements = stage_result.get("elements", [])
+                        if stage_name and elements:
+                            metadata_with_stage = {
+                                "filename": filename,
+                                "file_path": file_path,
+                                "stage": stage_name,
+                            }
+                            if result.stats and hasattr(result.stats, "record_id"):
+                                metadata_with_stage["record_id"] = result.stats.record_id
+
+                            pipeline_config.intermediate_results_destination.write(
+                                elements, metadata_with_stage
+                            )
+
                 # 准备写入数据
                 # 如果是向量数据库，写入 embeddings
-                if hasattr(result, 'elements') and result.elements:
+                if hasattr(result, "elements") and result.elements:
                     elements_data = []
                     for elem in result.elements:
-                        elem_dict = elem.model_dump() if hasattr(elem, 'model_dump') else elem
+                        elem_dict = elem.model_dump() if hasattr(elem, "model_dump") else elem
                         elements_data.append(elem_dict)
                 else:
                     elements_data = []
@@ -148,8 +176,8 @@ class Local(BaseAPI):
                     "filename": filename,
                     "file_path": file_path,
                 }
-                if hasattr(result, 'record_id'):
-                    metadata["record_id"] = result.record_id
+                if result.stats and hasattr(result.stats, "record_id"):
+                    metadata["record_id"] = result.stats.record_id
 
                 destination.write(elements_data, metadata)
 
@@ -164,11 +192,13 @@ class Local(BaseAPI):
                 else:
                     # continue: 记录失败并继续下一个文件
                     failed += 1
-                    failed_files.append(FailedFile(
-                        file_path=file_path,
-                        error=error_msg,
-                        retry_count=0,  # HTTP客户端内部已经处理了重试
-                    ))
+                    failed_files.append(
+                        FailedFile(
+                            file_path=file_path,
+                            error=error_msg,
+                            retry_count=0,  # HTTP客户端内部已经处理了重试
+                        )
+                    )
 
         # 计算总耗时
         duration = time.time() - start_time

{xparse_client-0.3.0b1 → xparse_client-0.3.0b3}/xparse_client/api/pipeline.py

@@ -120,7 +120,9 @@ class PipelineAPI(BaseAPI):
         }
 
         if config:
-            data["config"] = json.dumps(config.model_dump(), ensure_ascii=False)
+            # 只序列化 API 需要的字段（排除 intermediate_results_destination）
+            config_dict = config.model_dump(exclude={"intermediate_results_destination"})
+            data["config"] = json.dumps(config_dict, ensure_ascii=False)
         if data_source:
             data["data_source"] = json.dumps(data_source, ensure_ascii=False)
 

{xparse_client-0.3.0b1 → xparse_client-0.3.0b3}/xparse_client/models/pipeline.py

@@ -99,11 +99,13 @@ class PipelineConfig(BaseModel):
 
     Attributes:
         include_intermediate_results: 是否包含中间结果
+        intermediate_results_destination: 中间结果保存目的地（仅支持 LocalDestination）
     """
 
     include_intermediate_results: bool = False
+    intermediate_results_destination: Any = None  # LocalDestination 对象
 
-    model_config = {"extra": "allow"}
+    model_config = {"extra": "allow", "arbitrary_types_allowed": True}
 
 
 class PipelineResponse(BaseModel):

{xparse_client-0.3.0b1 → xparse_client-0.3.0b3/xparse_client.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: xparse-client
-Version: 0.3.0b1
+Version: 0.3.0b3
 Summary: 面向 Agent 和 RAG 的文档处理 Pipeline 客户端
 Author-email: INTSIG-TEXTIN <support@textin.com>
 License-Expression: MIT