xparse-client 0.3.0b1__tar.gz → 0.3.0b6__tar.gz

{xparse_client-0.3.0b1/xparse_client.egg-info → xparse_client-0.3.0b6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: xparse-client
-Version: 0.3.0b1
+Version: 0.3.0b6
 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
@@ -213,17 +213,17 @@ schema = {
         "author": {"type": "string", "description": "作者"},
         "date": {"type": "string", "description": "日期"}
     },
-    "required": ["title"]
+    "required": ["title", "author", "date"]
 }
 
 with open("document.pdf", "rb") as f:
     result = client.extract.extract(
         file=f,
         filename="document.pdf",
-        config=ExtractConfig(schema=schema)
+        extract_config=ExtractConfig(schema=schema)
     )
 
-print(result.extracted_data)
+print(result.result)
 ```
 
 ### 3. 本地批处理

{xparse_client-0.3.0b1 → xparse_client-0.3.0b6}/README.md

@@ -164,17 +164,17 @@ schema = {
         "author": {"type": "string", "description": "作者"},
         "date": {"type": "string", "description": "日期"}
     },
-    "required": ["title"]
+    "required": ["title", "author", "date"]
 }
 
 with open("document.pdf", "rb") as f:
     result = client.extract.extract(
         file=f,
         filename="document.pdf",
-        config=ExtractConfig(schema=schema)
+        extract_config=ExtractConfig(schema=schema)
     )
 
-print(result.extracted_data)
+print(result.result)
 ```
 
 ### 3. 本地批处理

{xparse_client-0.3.0b1 → xparse_client-0.3.0b6}/example/2_async_job.py

@@ -165,7 +165,7 @@ def example_3_error_handling():
 
         except Exception as e:
             print(f"\n⏸️  捕获到超时异常: {type(e).__name__}")
-            print(f"  这是正常的，演示了如何处理超时情况")
+            print("  这是正常的，演示了如何处理超时情况")
 
     finally:
         test_file.unlink(missing_ok=True)

{xparse_client-0.3.0b1 → xparse_client-0.3.0b6}/example/4_advanced_workflow.py

@@ -91,12 +91,12 @@ def example_1_config_file():
     XParseClient.from_env()
 
     # 演示如何从配置构建 source 和 destination（不实际创建）
-    print(f"\n✅ 配置加载成功!")
-    print(f"  - Source 配置:")
+    print("\n✅ 配置加载成功!")
+    print("  - Source 配置:")
     print(f"      目录: {loaded_config['source']['directory']}")
     print(f"      模式: {loaded_config['source']['pattern']}")
     print(f"      递归: {loaded_config['source']['recursive']}")
-    print(f"  - Destination 配置:")
+    print("  - Destination 配置:")
     print(f"      输出目录: {loaded_config['destination']['output_dir']}")
 
     # 构建 stages

{xparse_client-0.3.0b1 → xparse_client-0.3.0b6}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "xparse-client"
-version = "0.3.0b1"
+version = "0.3.0b6"
 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-0.3.0b6}/tests/unit/models/test_pipeline_stages.py

@@ -89,8 +89,8 @@ def test_extract_stage_with_config():
     assert stage.type == "extract"
     assert stage.config.generate_citations is True
 
-    # 序列化检查
-    dumped = stage.model_dump()
+    # 序列化检查（使用 by_alias=True 来确保 schema_ 序列化为 schema）
+    dumped = stage.model_dump(by_alias=True)
     assert "schema" in dumped["config"]
     assert dumped["config"]["schema"] == schema
 

{xparse_client-0.3.0b1 → xparse_client-0.3.0b6}/xparse_client/_base.py

@@ -6,7 +6,7 @@
 from __future__ import annotations
 
 from abc import ABC
-from typing import TYPE_CHECKING, Any, TypeVar
+from typing import TYPE_CHECKING, Any, Protocol, TypeVar, cast
 
 import httpx
 
@@ -14,7 +14,16 @@ if TYPE_CHECKING:
     from ._config import SDKConfiguration
     from ._http import HTTPClient
 
-T = TypeVar("T")
+
+class PydanticModel(Protocol):
+    """Pydantic 模型协议"""
+
+    @classmethod
+    def model_validate(cls, obj: Any) -> Any:
+        ...
+
+
+T = TypeVar("T", bound=PydanticModel)
 
 
 class BaseAPI(ABC):  # noqa: B024
@@ -130,7 +139,7 @@ class BaseAPI(ABC):  # noqa: B024
         if isinstance(data, dict) and "data" in data:
             data = data["data"]
 
-        return model_class.model_validate(data)
+        return cast(T, model_class.model_validate(data))
 
     def _parse_list_response(
         self,
@@ -156,7 +165,7 @@ class BaseAPI(ABC):  # noqa: B024
         if not isinstance(data, list):
             data = [data]
 
-        return [model_class.model_validate(item) for item in data]
+        return [cast(T, model_class.model_validate(item)) for item in data]
 
     def _parse_raw_response(self, response: httpx.Response) -> dict[str, Any]:
         """解析原始 JSON 响应
@@ -171,9 +180,9 @@ class BaseAPI(ABC):  # noqa: B024
 
         # 处理标准响应格式
         if isinstance(data, dict) and "data" in data:
-            return data["data"]
+            return cast(dict[str, Any], data["data"])
 
-        return data
+        return cast(dict[str, Any], data)
 
 
 __all__ = ["BaseAPI"]

{xparse_client-0.3.0b1 → xparse_client-0.3.0b6}/xparse_client/_http.py

@@ -53,7 +53,8 @@ def _extract_error_message(response: httpx.Response) -> str:
                     return str(data[key])
             # 如果有嵌套的 error 对象
             if "error" in data and isinstance(data["error"], dict):
-                return data["error"].get("message", str(data["error"]))
+                error_dict = data["error"]
+                return str(error_dict.get("message", error_dict))
         return str(data)
     except Exception:
         # 非 JSON 响应，返回文本内容
@@ -106,7 +107,7 @@ def raise_for_status(response: httpx.Response) -> None:
         message = _extract_error_message(response)
         response_body = response.text[:1000] if response.text else None
 
-        common_kwargs = {
+        common_kwargs: dict[str, Any] = {
             "status_code": status_code,
             "request_id": request_id,
             "response_body": response_body,
@@ -140,24 +141,24 @@ def raise_for_status(response: httpx.Response) -> None:
                 message = data.get("message", f"业务错误 code: {code}")
                 response_body = response.text[:1000] if response.text else None
 
-                common_kwargs = {
+                error_kwargs: dict[str, Any] = {
                     "request_id": request_id,
                     "response_body": response_body,
                 }
 
                 # 根据业务 code 映射到相应异常
                 if code == 400:
-                    raise ValidationError(message, details=common_kwargs)
+                    raise ValidationError(message, details=error_kwargs)
                 elif code == 401:
-                    raise AuthenticationError(message, **common_kwargs)
+                    raise AuthenticationError(message, **error_kwargs)
                 elif code == 403:
-                    raise PermissionDeniedError(message, **common_kwargs)
+                    raise PermissionDeniedError(message, **error_kwargs)
                 elif code == 404:
-                    raise NotFoundError(message, **common_kwargs)
+                    raise NotFoundError(message, **error_kwargs)
                 elif code >= 500:
-                    raise ServerError(message, **common_kwargs)
+                    raise ServerError(message, **error_kwargs)
                 else:
-                    raise APIError(message, **common_kwargs)
+                    raise APIError(message, **error_kwargs)
     except (ValueError, KeyError):
         # 如果无法解析 JSON 或没有 code 字段，认为是成功
         pass

{xparse_client-0.3.0b1 → xparse_client-0.3.0b6}/xparse_client/api/extract.py

@@ -79,7 +79,8 @@ class Extract(BaseAPI):
             ...             "properties": {
             ...                 "invoice_number": {"type": "string"},
             ...                 "total_amount": {"type": "number"}
-            ...             }
+            ...             },
+            ...             "required": ["invoice_number", "total_amount"]
             ...         },
             ...         generate_citations=True
             ...     )
@@ -89,15 +90,11 @@ class Extract(BaseAPI):
         data = {}
 
         if parse_config:
-            data["parse_config"] = json.dumps(
-                parse_config.model_dump(), ensure_ascii=False
-            )
+            data["parse_config"] = json.dumps(parse_config.model_dump(), ensure_ascii=False)
 
         # 处理 extract_config
         if extract_config:
-            data["extract_config"] = json.dumps(
-                extract_config.model_dump(), ensure_ascii=False
-            )
+            data["extract_config"] = json.dumps(extract_config.model_dump(), ensure_ascii=False)
         else:
             raise ValueError("extract_config is required")
 
@@ -105,5 +102,4 @@ class Extract(BaseAPI):
         return self._parse_response(response, ParseResponse)
 
 
-
 __all__ = ["Extract"]

{xparse_client-0.3.0b1 → xparse_client-0.3.0b6}/xparse_client/api/local.py

@@ -24,11 +24,11 @@ Example:
 
 from __future__ import annotations
 
-from typing import TYPE_CHECKING, Callable, Literal
+from typing import TYPE_CHECKING, Any, 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
@@ -53,8 +53,9 @@ class Local(BaseAPI):
         self,
         source,  # Union[LocalSource, S3Source, FtpSource, SmbSource]
         destination,  # Union[LocalDestination, S3Destination, MilvusDestination, QdrantDestination]
-        stages: list[PipelineStage],
+        stages: list[PipelineStage | dict[str, Any]],
         *,
+        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 +127,56 @@ 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:
+                            # 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:
+                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 +186,8 @@ 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 +202,13 @@ 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-0.3.0b6}/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-0.3.0b6}/xparse_client/connectors/destinations/qdrant.py

@@ -90,7 +90,7 @@ class QdrantDestination(Destination, VectorDestinationMixin):
         self.dimension = dimension
         self._PointStruct = PointStruct
 
-        client_kwargs = {"url": url}
+        client_kwargs: dict[str, Any] = {"url": url}
         if api_key:
             client_kwargs["api_key"] = api_key
         if prefer_grpc:

{xparse_client-0.3.0b1 → xparse_client-0.3.0b6}/xparse_client/connectors/sources/ftp.py

@@ -90,7 +90,7 @@ class FtpSource(Source):
             SourceError: 列出文件失败
         """
         try:
-            files = []
+            files: list[str] = []
             current_dir = self.client.pwd()
 
             if self.recursive:
@@ -178,7 +178,7 @@ class FtpSource(Source):
 
         # 回退到 LIST 命令
         try:
-            lines = []
+            lines: list[str] = []
             self.client.retrlines("LIST", lines.append)
             for line in lines:
                 parts = line.split()

{xparse_client-0.3.0b1 → xparse_client-0.3.0b6}/xparse_client/connectors/sources/smb.py

@@ -125,7 +125,7 @@ class SmbSource(Source):
             SourceError: 列出文件失败
         """
         try:
-            files = []
+            files: list[str] = []
             base_path = "/" if not self.path else f"/{self.path}"
 
             self._list_recursive(base_path, base_path, files)

{xparse_client-0.3.0b1 → xparse_client-0.3.0b6}/xparse_client/models/extract.py

@@ -11,7 +11,7 @@ class ExtractConfig(BaseModel):
     """抽取配置
 
     Attributes:
-        schema: JSON Schema 定义抽取的结构
+        schema_: JSON Schema 定义抽取的结构（序列化为 schema）
         generate_citations: 是否生成引用
         stamp: 是否添加时间戳
 
@@ -23,11 +23,12 @@ class ExtractConfig(BaseModel):
         ...         "total_amount": {"type": "number"}
         ...     }
         ... }
-        >>> config = ExtractConfig(schema=schema)
+        >>> config = ExtractConfig(schema_=schema)
     """
 
-    schema: dict[str, Any] = Field(
+    schema_: dict[str, Any] = Field(
         default_factory=dict,
+        alias="schema",
         description="JSON Schema 定义抽取的结构"
     )
     generate_citations: bool = False
@@ -35,6 +36,7 @@ class ExtractConfig(BaseModel):
 
     model_config = {
         "extra": "allow",
+        "populate_by_name": True,  # 允许使用 schema_ 和 schema 两种名称
     }
 
 

{xparse_client-0.3.0b1 → xparse_client-0.3.0b6}/xparse_client/models/parse.py

@@ -63,17 +63,13 @@ class ParseResponse(BaseModel):
 
     Attributes:
         elements: 解析后的元素列表（Parse/Chunk/Embed 返回）
-        extract_result: 提取结果（Extract 返回）
+        result: 提取结果（Extract 返回）
         success_count: 成功数量
-        consume_time: 耗时
-        record_id: 记录 ID
     """
 
     elements: list[Element] = Field(default_factory=list)
-    extract_result: dict[str, Any] | None = None  # Extract API 返回
+    result: dict[str, Any] | None = None  # Extract API 返回
     success_count: int | None = None
-    consume_time: str | None = None
-    record_id: str | None = None
 
     model_config = {"extra": "allow"}
 

{xparse_client-0.3.0b1 → xparse_client-0.3.0b6}/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-0.3.0b6/xparse_client.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: xparse-client
-Version: 0.3.0b1
+Version: 0.3.0b6
 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
@@ -213,17 +213,17 @@ schema = {
         "author": {"type": "string", "description": "作者"},
         "date": {"type": "string", "description": "日期"}
     },
-    "required": ["title"]
+    "required": ["title", "author", "date"]
 }
 
 with open("document.pdf", "rb") as f:
     result = client.extract.extract(
         file=f,
         filename="document.pdf",
-        config=ExtractConfig(schema=schema)
+        extract_config=ExtractConfig(schema=schema)
     )
 
-print(result.extracted_data)
+print(result.result)
 ```
 
 ### 3. 本地批处理