测试用例和结果

deepseek-result.json

@@ -0,0 +1,41 @@
+{
+    "provider": "deepseek",
+    "model": "deepseek-chat",
+    "choices": [
+        {
+            "index": 0,
+            "message": {
+                "role": "assistant",
+                "content": "```json\n{\n  \"table_name\": \"national_brand_sales\",\n  \"description\": \"全国品牌系统外销售数据\",\n  \"columns\": [\n    {\n      \"original_name\": \"品牌\",\n      \"standard_name\": \"brand\",\n      \"data_type\": \"string\",\n      \"db_type\": \"varchar(50)\",\n      \"java_type\": \"string\",\n      \"nullable\": true,\n      \"distinct_count_sample\": 5,\n      \"null_ratio_sample\": 0.4,\n      \"is_enum_candidate\": false,\n      \"description\": \"品牌名称\",\n      \"date_format\": null\n    },\n    {\n      \"original_name\": \"产品价类\",\n      \"standard_name\": \"price_category\",\n      \"data_type\": \"string\",\n      \"db_type\": \"varchar(10)\",\n      \"java_type\": \"string\",\n      \"nullable\": false,\n      \"distinct_count_sample\": 3,\n      \"null_ratio_sample\": 0.0,\n      \"is_enum_candidate\": true,\n      \"description\": \"产品价格分类（一类/二类/三类）\",\n      \"date_format\": null\n    },\n    {\n      \"original_name\": \"是否重点品牌"
+            }
+        }
+    ],
+    "raw": {
+        "id": "67f3cc80-38bc-4bb7-b336-48d4886722c4",
+        "object": "chat.completion",
+        "created": 1761752207,
+        "model": "deepseek-chat",
+        "choices": [
+            {
+                "index": 0,
+                "message": {
+                    "role": "assistant",
+                    "content": "```json\n{\n  \"table_name\": \"national_brand_sales\",\n  \"description\": \"全国品牌系统外销售数据\",\n  \"columns\": [\n    {\n      \"original_name\": \"品牌\",\n      \"standard_name\": \"brand\",\n      \"data_type\": \"string\",\n      \"db_type\": \"varchar(50)\",\n      \"java_type\": \"string\",\n      \"nullable\": true,\n      \"distinct_count_sample\": 5,\n      \"null_ratio_sample\": 0.4,\n      \"is_enum_candidate\": false,\n      \"description\": \"品牌名称\",\n      \"date_format\": null\n    },\n    {\n      \"original_name\": \"产品价类\",\n      \"standard_name\": \"price_category\",\n      \"data_type\": \"string\",\n      \"db_type\": \"varchar(10)\",\n      \"java_type\": \"string\",\n      \"nullable\": false,\n      \"distinct_count_sample\": 3,\n      \"null_ratio_sample\": 0.0,\n      \"is_enum_candidate\": true,\n      \"description\": \"产品价格分类（一类/二类/三类）\",\n      \"date_format\": null\n    },\n    {\n      \"original_name\": \"是否重点品牌"
+                },
+                "logprobs": null,
+                "finish_reason": "length"
+            }
+        ],
+        "usage": {
+            "prompt_tokens": 1078,
+            "completion_tokens": 256,
+            "total_tokens": 1334,
+            "prompt_tokens_details": {
+                "cached_tokens": 1024
+            },
+            "prompt_cache_hit_tokens": 1024,
+            "prompt_cache_miss_tokens": 54
+        },
+        "system_fingerprint": "fp_ffc7281d48_prod0820_fp8_kvcache"
+    }
+}

test/chat_completions_deepseek_example.py

@@ -0,0 +1,43 @@
+"""Demonstrates calling /v1/chat/completions with the DeepSeek provider."""
+
+from __future__ import annotations
+
+import asyncio
+import json
+
+import httpx
+from dotenv import load_dotenv
+
+
+load_dotenv()
+
+API_URL = "http://localhost:8000/v1/chat/completions"
+
+
+async def main() -> None:
+    payload = {
+        "provider": "deepseek",
+        "model": "deepseek-chat",
+        "messages": [
+            {
+                "role": "system",
+                "content": "角色：你是一名数据分析导入助手（Data Ingestion Analyst），擅长从原始数据抽取结构化元数据、推断字段类型、识别维度/事实属性，并输出导入建模建议（Table + JSON）。\n\n任务目标：对提供的数据（含表头或table schema与若干行样本数据）进行解析，生成一份导入分析与处理报告，指导如何将其导入为标准化表结构及 JSON 元数据定义，不要省略任何字段信息，全量输出。\n\n请从以下两个方向进行思考：\n\n方向 1：元数据识别与整理\n解析表明：根据表头、Origin Table Name、Orign File Name生成表名，表名需要有意义\n解析列名：生成标准化字段名（snake_case 或小驼峰），并给出原始列名与标准字段名映射。\n为每个字段写出中文/英文注释（若无法确定，给出“待确认”并附可能解释）。\n\n方向 2：字段数据类型与格式推断\n针对每列：输出推断数据类型（如 varchar(n) / int / bigint / tinyint / float / double / decimal(p,s) / date / datetime / text）。\n说明推断依据：样本值分布、长度范围、格式正则、是否存在空值、是否数值但含前导零等。\n指出数据质量初步观察：缺失率、是否有异常/离群值（简单规则即可）、是否需标准化（如去空格、去重、枚举值归一）。\n给出“建议处理动作”：如 trim、cast_float、cast_int、cast_double、cast_date、cast_time、cast_datetime，适用于将样本数据转换成数据库表字段兼容的格式。\n若为“可能是枚举”的字段，列出候选枚举值及占比。\n\n最终内容都输出为一个json对象，格式为（字段级与表级定义），字段含：\n{\n  \"table_name\": \"标准化后的表名\",\n  \"description\": \"表简短描述\",\n  \"columns\": [{\n    \"original_name\": \"原始名称\",\n    \"standard_name\": \"标准化后的名称: 下划线命名，大小写字母、数字、下划线\",\n    \"data_type\": \"数据类型限制为：number/string/datetime\",\n    \"db_type\": \"数据库字段类型\",\n    \"java_type\": \"java字段类型限制为: int/long/double/string/date\",\n    \"nullable\": true/false,\n    \"distinct_count_sample\": number,\n    \"null_ratio_sample\": 0.x,\n    \"is_enum_candidate\": true/false,\n    \"description\": \"字段简短描述\",\n    \"date_format\": \"转换成Date类型的pattern\"\n  }]\n}\n\n约束与风格：\n\n若信息不足，请显式指出“信息不足”并给出补充数据需求清单。\n避免武断结论，用“可能 / 候选 / 建议”字样。\n不要捏造样本未出现的值。"
+            },
+            {
+                "role": "user",
+                "content": "导入记录ID: demo-import-001\n\n表头信息:\n- 品牌\n- 产品价类\n- 是否重点品牌\n- 系统外销售量(箱)\n- 系统外销售金额(万元)\n- 同期系统外销售量(箱)\n- 同期系统外销售金额(万元)\n\n示例数据:\nCSV样本预览:\n品牌,产品价类,是否重点品牌,系统外销售量(箱),系统外销售金额(万元),同期系统外销售量(箱),同期系统外销售金额(万元)\r\n白沙,一类,重点品牌,3332.406875,64283.5593333333,3123.693375,61821.7986666667\r\nnan,二类,重点品牌,1094.4707375,3859.69366666667,869.65725,3067.00966666667\r\nnan,三类,重点品牌,3965.0457375,8388.306,4401.6714875,8802.132\r\n宝岛,一类,否,39.934375,301.617666666667,30.5975,249.399666666667\r\n长白山,一类,重点品牌,2666.53775,12360.8306666667,1916.252,9051.672\r\nnan,二类,重点品牌,2359.910025,7671.26233333333,2335.2480875,7590.791\r\nnan,三类,重点品牌,1263.293875,2826.665,1590.750875,3503.083\r\n大前门,一类,否,81.5806875,343.721333333333,114.1179875,480.809333333333\r\nnan,三类,否,226.445225,319.975666666667,254.6595125,359.894\r\n大青山,二类,否,60.73525,209.415,60.2415,207.712666666667\n\n附加结构信息:\n{\n  \"source\": \"excel\",\n  \"file_name\": \"全国品牌.xlsx\",\n  \"sheet_name\": \"Sheet1\"\n}"
+            }
+        ],
+        "temperature": 0.2,
+        "max_tokens": 256,
+    }
+
+    async with httpx.AsyncClient(timeout=httpx.Timeout(20.0)) as client:
+        response = await client.post(API_URL, json=payload)
+        response.raise_for_status()
+        data = response.json()
+        print(json.dumps(data, ensure_ascii=False, indent=2))
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

test/data_import_analysis_example.py

@@ -7,7 +7,9 @@ from pathlib import Path
 
 import httpx
 import pandas as pd
+from dotenv import load_dotenv
 
+load_dotenv()
 
 API_URL = "http://localhost:8000/v1/import/analyze"
 CALLBACK_URL = "http://localhost:8000/__mock__/import-callback"
@@ -26,9 +28,15 @@ async def main() -> None:
     payload = {
         "import_record_id": "demo-import-001",
         "rows": rows,
-        "struce": headers,
+        "headers": headers,
+        "table_schema": {
+            "source": "excel",
+            "file_name": EXCEL_PATH.name,
+            "sheet_name": sheet_name,
+        },
         "llm_model": "deepseek:deepseek-chat",
         "temperature": 0.2,
+        "max_output_tokens": 256,
         "callback_url": CALLBACK_URL,
     }
 

test/openrouter_chat_example.py

@@ -0,0 +1,44 @@
+"""Quick demo call against the unified chat endpoint using the OpenRouter provider."""
+
+from __future__ import annotations
+
+import asyncio
+
+import httpx
+from dotenv import load_dotenv
+
+load_dotenv()
+
+
+API_URL = "http://localhost:8000/v1/chat/completions"
+
+
+async def main() -> None:
+    payload = {
+        "provider": "openrouter",
+        "model": "anthropic/claude-3.5-sonnet",
+        "messages": [
+            {
+                "role": "system",
+                "content": "You are an API assistant that writes concise JSON only.",
+            },
+            {
+                "role": "user",
+                "content": "Return a JSON object describing this test invocation.",
+            },
+        ],
+        "temperature": 0.1,
+        "max_tokens": 300,
+    }
+
+    async with httpx.AsyncClient(timeout=httpx.Timeout(15.0)) as client:
+        response = await client.post(API_URL, json=payload)
+        print("Status:", response.status_code)
+        try:
+            print("Body:", response.json())
+        except ValueError:
+            print("Raw Body:", response.text)
+
+
+if __name__ == "__main__":
+    asyncio.run(main())