xparse-client 0.3.0b2__py3-none-any.whl → 0.3.0b4__py3-none-any.whl

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）
@@ -131,9 +133,44 @@ class Local(BaseAPI):
                     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:
+                            # 转换 elements 为字典列表
+                            elements_data = []
+                            for elem in elements:
+                                if hasattr(elem, "model_dump"):
+                                    elements_data.append(elem.model_dump())
+                                elif isinstance(elem, dict):
+                                    elements_data.append(elem)
+                                else:
+                                    elements_data.append(elem)
+
+                            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_data, 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:

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/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-0.3.0b2.dist-info → xparse_client-0.3.0b4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: xparse-client
-Version: 0.3.0b2
+Version: 0.3.0b4
 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-0.3.0b2.dist-info → xparse_client-0.3.0b4.dist-info}/RECORD

@@ -35,9 +35,9 @@ xparse_client/_http.py,sha256=OrnrkTeuG5tEpuD-a1ixLIFPdiqnVIYtuZCi3uXAzJ4,11381
 xparse_client/exceptions.py,sha256=FNozgCFaQSsdeyxE6P-3CZItkTSfFJcIQVrddHWKN0s,9521
 xparse_client/api/__init__.py,sha256=MZ3wBTDU666efOBv81MKKtc1UkjUsjzIEOpsnDUAHD4,203
 xparse_client/api/extract.py,sha256=87tH-y4sXPKW35l-d0GmZdK9RIXWq4tfmGZXYJhIN60,3011
-xparse_client/api/local.py,sha256=vFfPfdVPMYtJQK76iPfGw1DdFxNqfFiMIEPWw4jkrwo,6246
+xparse_client/api/local.py,sha256=yAEYj9a_pGra08_PTr_FaJrdUz7Pyh7HXq0Rb_6bvH0,8192
 xparse_client/api/parse.py,sha256=nvpu5kiN43QlyMW3N6vMruVADu6MdjS-TMbLFjBh-cc,5870
-xparse_client/api/pipeline.py,sha256=HscuwRUOmcdUEjzqSyAMdNkcLyK_XbF_THjNH5DXsec,4356
+xparse_client/api/pipeline.py,sha256=bPCSpWjER-7c7L0iMQ_Xz1VTe5DARMsX5XIcrayGKzw,4530
 xparse_client/api/workflows.py,sha256=29_Rf4bIH1fiJb_bWShkUm-et97ym5ymQFh5i3b5mDI,5945
 xparse_client/connectors/__init__.py,sha256=BLGxk9i8BsVCr4RMrCIh8b9cywBqXXmcDHqfF7C_FX0,1113
 xparse_client/connectors/_utils.py,sha256=Qb2OyUg8gvzyMBELx4h0GmjRSdT4d9C45v9m6NUNuMg,3515
@@ -59,10 +59,10 @@ xparse_client/models/embed.py,sha256=Tg9iqOiLZy63m4T0q2z9cq3neuQbMnqDqT55-sVA7CE
 xparse_client/models/extract.py,sha256=ZwMy0MsTt775Bq3UsZ9WlXFNCLrSnN0j1RgXQ-vFNqo,915
 xparse_client/models/local.py,sha256=hqKmyWTU_EhPk6qybtBzmbLybBBaiasTpqM4_zP-ipo,698
 xparse_client/models/parse.py,sha256=K34Fodo3emOPZJJQeu3MLjvA7HkVMxAuXk20UvO4lfQ,3297
-xparse_client/models/pipeline.py,sha256=KVmTYqyi6sbGMR9UQcEgMHFqhcSBPnoRgtGL4xp8QEA,3086
+xparse_client/models/pipeline.py,sha256=24bDzhrVotQ8St6VLEJLvG2cZF0G2AMioKI34I_hJXI,3297
 xparse_client/models/workflows.py,sha256=BivMdGOAmhP6oYLQSGAAN7yml2xb7vHHrpzwLgN_Afk,1754
-xparse_client-0.3.0b2.dist-info/licenses/LICENSE,sha256=7iuki7DyWMGB8PBzsht7PUt0YjdIcPjrcXNyUFgMJsw,1070
-xparse_client-0.3.0b2.dist-info/METADATA,sha256=QYHCW6pdcMpJKkK3GRKFNMs9comTFatvSG1jXs_5Zv4,25453
-xparse_client-0.3.0b2.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
-xparse_client-0.3.0b2.dist-info/top_level.txt,sha256=oQGc_qysOmnSAaLjwB72wH8RBHRAmxB-_qb-Uj6u56o,28
-xparse_client-0.3.0b2.dist-info/RECORD,,
+xparse_client-0.3.0b4.dist-info/licenses/LICENSE,sha256=7iuki7DyWMGB8PBzsht7PUt0YjdIcPjrcXNyUFgMJsw,1070
+xparse_client-0.3.0b4.dist-info/METADATA,sha256=f8yFWh3712-v_y5oDQDbF5NpXCZOeCOGQU_AFqZEYYk,25453
+xparse_client-0.3.0b4.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+xparse_client-0.3.0b4.dist-info/top_level.txt,sha256=oQGc_qysOmnSAaLjwB72wH8RBHRAmxB-_qb-Uj6u56o,28
+xparse_client-0.3.0b4.dist-info/RECORD,,