xparse-client 0.2.19__py3-none-any.whl → 0.3.0b8__py3-none-any.whl

tests/unit/api/test_pipeline.py

@@ -0,0 +1,369 @@
+"""Pipeline API 测试
+
+测试 Pipeline API 的所有功能，包括：
+- parse + chunk + embed 流程
+- parse + extract 流程
+- stages 参数（PipelineStage 对象和字典格式）
+- config 参数
+- data_source 参数
+- 错误处理
+"""
+
+import httpx
+import pytest
+import respx
+
+from xparse_client import XParseClient
+from xparse_client.exceptions import ServerError, ValidationError
+from xparse_client.models import (
+    ChunkConfig,
+    ChunkStage,
+    EmbedConfig,
+    EmbedStage,
+    ExtractConfig,
+    ExtractStage,
+    ParseConfig,
+    ParseStage,
+    PipelineConfig,
+)
+
+# ============================================================================
+# 基础功能测试
+# ============================================================================
+
+
+def test_pipeline_api_exists():
+    """测试 client.pipeline 属性存在"""
+    client = XParseClient(app_id="test", secret_code="test")
+
+    assert hasattr(client, "pipeline")
+    assert client.pipeline is not None
+
+
+@respx.mock
+def test_pipeline_parse_chunk_embed(server_url, pipeline_parse_chunk_embed_response):
+    """测试 parse + chunk + embed 流程"""
+    # Mock API
+    respx.post(f"{server_url}/api/xparse/pipeline").mock(
+        return_value=httpx.Response(200, json=pipeline_parse_chunk_embed_response)
+    )
+
+    # 初始化客户端
+    client = XParseClient(app_id="test", secret_code="test", server_url=server_url)
+
+    # 创建 stages
+    stages = [
+        ParseStage(config=ParseConfig(provider="textin")),
+        ChunkStage(config=ChunkConfig(strategy="basic", max_characters=1000)),
+        EmbedStage(config=EmbedConfig(provider="qwen", model_name="text-embedding-v4")),
+    ]
+
+    # 执行 pipeline
+    result = client.pipeline.execute(
+        file=b"test pdf content",
+        filename="test.pdf",
+        stages=stages
+    )
+
+    # 验证结果
+    assert result is not None
+    assert result.job_id == "job_123"
+    assert len(result.elements) == 1
+    assert result.elements[0].embeddings is not None
+    assert len(result.elements[0].embeddings) == 1024
+    assert result.intermediate_results is not None
+    assert len(result.intermediate_results) == 2  # parse + chunk
+
+    # 验证请求包含 stages
+    request = respx.calls.last.request
+    assert b"stages" in request.content
+    assert b"parse" in request.content
+    assert b"chunk" in request.content
+    assert b"embed" in request.content
+
+
+@respx.mock
+def test_pipeline_parse_extract(server_url, pipeline_parse_extract_response):
+    """测试 parse + extract 流程"""
+    # Mock API
+    respx.post(f"{server_url}/api/xparse/pipeline").mock(
+        return_value=httpx.Response(200, json=pipeline_parse_extract_response)
+    )
+
+    # 初始化客户端
+    client = XParseClient(app_id="test", secret_code="test", server_url=server_url)
+
+    # 创建 stages
+    stages = [
+        ParseStage(config=ParseConfig(provider="textin")),
+        ExtractStage(
+            config=ExtractConfig(
+                schema_={
+                    "type": "object",
+                    "properties": {"购买方": {"type": "string"}},
+                    "required": ["购买方"]
+                }
+            )
+        ),
+    ]
+
+    # 执行 pipeline
+    result = client.pipeline.execute(
+        file=b"test invoice content",
+        filename="invoice.pdf",
+        stages=stages
+    )
+
+    # 验证结果
+    assert result is not None
+    assert result.job_id == "job_456"
+    assert result.extract_result is not None
+    assert "extracted_schema" in result.extract_result
+    assert result.extract_result["extracted_schema"]["购买方"] == "杭州百世网络技术有限公司"
+
+    # 验证请求包含 stages
+    request = respx.calls.last.request
+    assert b"stages" in request.content
+    assert b"parse" in request.content
+    assert b"extract" in request.content
+
+
+@respx.mock
+def test_pipeline_with_dict_stages(server_url, pipeline_parse_chunk_embed_response):
+    """测试使用字典格式的 stages"""
+    # Mock API
+    respx.post(f"{server_url}/api/xparse/pipeline").mock(
+        return_value=httpx.Response(200, json=pipeline_parse_chunk_embed_response)
+    )
+
+    # 初始化客户端
+    client = XParseClient(app_id="test", secret_code="test", server_url=server_url)
+
+    # 使用字典格式的 stages
+    stages = [
+        {"type": "parse", "config": {"provider": "textin"}},
+        {"type": "chunk", "config": {"strategy": "basic"}},
+        {"type": "embed", "config": {"provider": "qwen"}},
+    ]
+
+    # 执行 pipeline
+    result = client.pipeline.execute(
+        file=b"test",
+        filename="test.pdf",
+        stages=stages
+    )
+
+    # 验证结果
+    assert result is not None
+    assert result.job_id == "job_123"
+
+
+# ============================================================================
+# 参数测试
+# ============================================================================
+
+
+@respx.mock
+def test_pipeline_with_config(server_url, pipeline_parse_chunk_embed_response):
+    """测试 config 参数"""
+    # Mock API
+    respx.post(f"{server_url}/api/xparse/pipeline").mock(
+        return_value=httpx.Response(200, json=pipeline_parse_chunk_embed_response)
+    )
+
+    # 初始化客户端
+    client = XParseClient(app_id="test", secret_code="test", server_url=server_url)
+
+    # 创建 stages 和 config
+    stages = [
+        ParseStage(config=ParseConfig(provider="textin")),
+        ChunkStage(config=ChunkConfig(strategy="basic")),
+    ]
+    config = PipelineConfig(include_intermediate_results=True)
+
+    # 执行 pipeline
+    result = client.pipeline.execute(
+        file=b"test",
+        filename="test.pdf",
+        stages=stages,
+        config=config
+    )
+
+    # 验证结果
+    assert result is not None
+
+    # 验证请求包含 config
+    request = respx.calls.last.request
+    assert b"config" in request.content
+    assert b"include_intermediate_results" in request.content
+
+
+@respx.mock
+def test_pipeline_with_data_source(server_url, pipeline_parse_chunk_embed_response):
+    """测试 data_source 参数"""
+    # Mock API
+    respx.post(f"{server_url}/api/xparse/pipeline").mock(
+        return_value=httpx.Response(200, json=pipeline_parse_chunk_embed_response)
+    )
+
+    # 初始化客户端
+    client = XParseClient(app_id="test", secret_code="test", server_url=server_url)
+
+    # 创建 stages 和 data_source
+    stages = [
+        ParseStage(config=ParseConfig(provider="textin")),
+    ]
+    data_source = {
+        "url": "s3://bucket/file.pdf",
+        "protocol": "s3"
+    }
+
+    # 执行 pipeline
+    result = client.pipeline.execute(
+        file=b"test",
+        filename="test.pdf",
+        stages=stages,
+        data_source=data_source
+    )
+
+    # 验证结果
+    assert result is not None
+
+    # 验证请求包含 data_source
+    request = respx.calls.last.request
+    assert b"data_source" in request.content
+    assert b"s3://bucket/file.pdf" in request.content
+
+
+# ============================================================================
+# 错误处理测试
+# ============================================================================
+
+
+@respx.mock
+def test_pipeline_invalid_stages_error(server_url, business_error_response):
+    """测试无效 stages 错误（业务 code 400）"""
+    # Mock API 返回业务错误
+    respx.post(f"{server_url}/api/xparse/pipeline").mock(
+        return_value=httpx.Response(200, json=business_error_response)
+    )
+
+    client = XParseClient(app_id="test", secret_code="test", server_url=server_url)
+    stages = [
+        {"type": "invalid_stage", "config": {}}
+    ]
+
+    # 应该抛出 ValidationError
+    with pytest.raises(ValidationError) as exc_info:
+        client.pipeline.execute(
+            file=b"test",
+            filename="test.pdf",
+            stages=stages
+        )
+
+    assert "参数错误" in str(exc_info.value)
+
+
+@respx.mock
+def test_pipeline_server_error(server_url):
+    """测试服务器错误（HTTP 500）"""
+    # Mock API 返回 500 错误
+    respx.post(f"{server_url}/api/xparse/pipeline").mock(
+        return_value=httpx.Response(500, json={"message": "服务器内部错误"})
+    )
+
+    client = XParseClient(app_id="test", secret_code="test", server_url=server_url)
+    stages = [
+        ParseStage(config=ParseConfig(provider="textin")),
+    ]
+
+    # 应该抛出 ServerError
+    with pytest.raises(ServerError) as exc_info:
+        client.pipeline.execute(
+            file=b"test",
+            filename="test.pdf",
+            stages=stages
+        )
+
+    assert "服务器" in str(exc_info.value)
+
+
+# ============================================================================
+# 中间结果测试
+# ============================================================================
+
+
+@respx.mock
+def test_pipeline_intermediate_results(server_url, pipeline_parse_chunk_embed_response):
+    """测试中间结果返回"""
+    # Mock API
+    respx.post(f"{server_url}/api/xparse/pipeline").mock(
+        return_value=httpx.Response(200, json=pipeline_parse_chunk_embed_response)
+    )
+
+    # 初始化客户端
+    client = XParseClient(app_id="test", secret_code="test", server_url=server_url)
+
+    # 创建 stages
+    stages = [
+        ParseStage(config=ParseConfig(provider="textin")),
+        ChunkStage(config=ChunkConfig(strategy="basic")),
+        EmbedStage(config=EmbedConfig(provider="qwen")),
+    ]
+
+    # 执行 pipeline
+    result = client.pipeline.execute(
+        file=b"test",
+        filename="test.pdf",
+        stages=stages
+    )
+
+    # 验证中间结果
+    assert result.intermediate_results is not None
+    assert len(result.intermediate_results) == 2
+
+    # 验证 parse stage 结果
+    parse_result = result.intermediate_results[0]
+    assert parse_result["stage"] == "parse"
+    assert "elements" in parse_result
+
+    # 验证 chunk stage 结果
+    chunk_result = result.intermediate_results[1]
+    assert chunk_result["stage"] == "chunk"
+    assert "elements" in chunk_result
+
+
+@respx.mock
+def test_pipeline_stats(server_url, pipeline_parse_chunk_embed_response):
+    """测试 stats 信息"""
+    # Mock API
+    respx.post(f"{server_url}/api/xparse/pipeline").mock(
+        return_value=httpx.Response(200, json=pipeline_parse_chunk_embed_response)
+    )
+
+    # 初始化客户端
+    client = XParseClient(app_id="test", secret_code="test", server_url=server_url)
+
+    # 创建 stages
+    stages = [
+        ParseStage(config=ParseConfig(provider="textin")),
+        ChunkStage(config=ChunkConfig(strategy="basic")),
+        EmbedStage(config=EmbedConfig(provider="qwen")),
+    ]
+
+    # 执行 pipeline
+    result = client.pipeline.execute(
+        file=b"test",
+        filename="test.pdf",
+        stages=stages
+    )
+
+    # 验证 stats
+    assert result.stats is not None
+    assert result.stats.original_elements == 15
+    assert result.stats.chunked_elements == 3
+    assert result.stats.embedded_elements == 3
+    # stats 包含 stages 信息（通过 extra="allow" 存储）
+    stats_dict = result.stats.model_dump()
+    assert "stages" in stats_dict
+    assert len(stats_dict["stages"]) == 3

tests/unit/api/test_workflows.py

@@ -0,0 +1,108 @@
+"""测试 Workflows API"""
+
+import pytest
+
+from xparse_client import XParseClient
+from xparse_client.models import (
+    ParseConfig,
+    ParseStage,
+    Schedule,
+    WorkflowState,
+)
+
+
+@pytest.fixture
+def client():
+    """创建测试客户端"""
+    return XParseClient(
+        app_id="test-app-id",
+        secret_code="test-secret"
+    )
+
+
+def test_workflows_api_exists(client):
+    """测试 client.workflows 属性存在"""
+    assert hasattr(client, 'workflows')
+    workflows_api = client.workflows
+    assert workflows_api is not None
+
+
+def test_workflows_api_has_create(client):
+    """测试 workflows.create 方法存在"""
+    assert hasattr(client.workflows, 'create')
+    assert callable(client.workflows.create)
+
+
+def test_workflows_api_has_list(client):
+    """测试 workflows.list 方法存在"""
+    assert hasattr(client.workflows, 'list')
+    assert callable(client.workflows.list)
+
+
+def test_workflows_api_has_get(client):
+    """测试 workflows.get 方法存在"""
+    assert hasattr(client.workflows, 'get')
+    assert callable(client.workflows.get)
+
+
+def test_workflows_api_has_update(client):
+    """测试 workflows.update 方法存在"""
+    assert hasattr(client.workflows, 'update')
+    assert callable(client.workflows.update)
+
+
+def test_workflows_api_has_delete(client):
+    """测试 workflows.delete 方法存在"""
+    assert hasattr(client.workflows, 'delete')
+    assert callable(client.workflows.delete)
+
+
+def test_workflows_api_has_run(client):
+    """测试 workflows.run 方法存在"""
+    assert hasattr(client.workflows, 'run')
+    assert callable(client.workflows.run)
+
+
+def test_create_raises_not_implemented(client):
+    """测试 create 方法抛出 NotImplementedError"""
+    with pytest.raises(NotImplementedError, match="Workflows.create 尚未实现"):
+        client.workflows.create(
+            name="test",
+            source_id="src_123",
+            destination_id="dst_456",
+            stages=[ParseStage(config=ParseConfig())],
+            schedule=Schedule(cron="0 0 * * *")
+        )
+
+
+def test_list_raises_not_implemented(client):
+    """测试 list 方法抛出 NotImplementedError"""
+    with pytest.raises(NotImplementedError, match="Workflows.list 尚未实现"):
+        client.workflows.list()
+
+
+def test_get_raises_not_implemented(client):
+    """测试 get 方法抛出 NotImplementedError"""
+    with pytest.raises(NotImplementedError, match="Workflows.get 尚未实现"):
+        client.workflows.get(workflow_id="wf_123")
+
+
+def test_update_raises_not_implemented(client):
+    """测试 update 方法抛出 NotImplementedError"""
+    with pytest.raises(NotImplementedError, match="Workflows.update 尚未实现"):
+        client.workflows.update(
+            workflow_id="wf_123",
+            state=WorkflowState.PAUSED
+        )
+
+
+def test_delete_raises_not_implemented(client):
+    """测试 delete 方法抛出 NotImplementedError"""
+    with pytest.raises(NotImplementedError, match="Workflows.delete 尚未实现"):
+        client.workflows.delete(workflow_id="wf_123")
+
+
+def test_run_raises_not_implemented(client):
+    """测试 run 方法抛出 NotImplementedError"""
+    with pytest.raises(NotImplementedError, match="Workflows.run 尚未实现"):
+        client.workflows.run(workflow_id="wf_123")