feat: AI summary in Model Arena

Author: LuckyYC

backend/.flake8

@@ -27,8 +27,12 @@ extend-ignore =
     D415,
     # B008: Do not perform function calls in argument defaults (common in FastAPI)
     B008,
+    # B010: Do not call setattr with a constant attribute value (SQLAlchemy models need this)
+    B010,
     # F401: imported but unused (handled by isort and IDE)
     F401,
+    # F541: f-string is missing placeholders (sometimes intentional for logging)
+    F541,
     # I100: Import statements are in the wrong order (handled by isort)
     I100,
     # I101: Imported names are in the wrong order (handled by isort)
@@ -62,4 +66,4 @@ per-file-ignores =
     # Fix script can have unused imports
     fix_code_*.py:F401
 
-max-complexity = 15
+max-complexity = 20

backend/api/api_analysis.py

@@ -7,24 +7,21 @@
 from fastapi.responses import JSONResponse
 
 from model.analysis import AnalysisRequest, AnalysisResponse, GetAnalysisResponse
-from service.analysis_service import analyze_task_svc, get_analysis_svc
+from service.analysis_service import analyze_tasks_svc, get_analysis_svc
 from utils.logger import logger
 
 # Create an API router for analysis-related endpoints
 router = APIRouter()
 
 
-@router.post("/{task_id}", response_model=AnalysisResponse)
-async def analyze_task(
-    request: Request, task_id: str, analysis_request: AnalysisRequest
-):
+@router.post("", response_model=AnalysisResponse)
+async def analyze_tasks(request: Request, analysis_request: AnalysisRequest):
     """
-    Perform AI analysis on task results.
+    Perform AI analysis on task results (single or multiple tasks).
 
     Args:
         request: The incoming request.
-        task_id: The task ID to analyze (from URL path).
-        analysis_request: The analysis request (optional eval_prompt).
+        analysis_request: The analysis request containing task_ids and options.
 
     Returns:
         AnalysisResponse: The analysis result.
@@ -33,7 +30,7 @@ async def analyze_task(
         HTTPException: If the task doesn't exist or analysis fails.
     """
 
-    return await analyze_task_svc(request, task_id, analysis_request)
+    return await analyze_tasks_svc(request, analysis_request)
 
 
 @router.get("/{task_id}", response_model=GetAnalysisResponse)

backend/db/init_db.sql

@@ -65,5 +65,45 @@ CREATE TABLE `task_results` (
   KEY `idx_task_id` (`task_id`)
 ) ENGINE=InnoDB AUTO_INCREMENT=262 DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;
 
+-- ----------------------------
+-- Table structure for test_insights (AI Analysis)
+-- ----------------------------
+DROP TABLE IF EXISTS `test_insights`;
+CREATE TABLE `test_insights` (
+  `id` int(11) NOT NULL AUTO_INCREMENT,
+  `task_id` varchar(36) COLLATE utf8mb4_unicode_ci NOT NULL,
+  `eval_prompt` text COLLATE utf8mb4_unicode_ci NOT NULL,
+  `analysis_report` longtext COLLATE utf8mb4_unicode_ci NOT NULL,
+  `status` varchar(20) COLLATE utf8mb4_unicode_ci NOT NULL DEFAULT 'pending' COMMENT 'pending, processing, completed, failed',
+  `error_message` text COLLATE utf8mb4_unicode_ci,
+  `created_at` datetime NOT NULL DEFAULT CURRENT_TIMESTAMP,
+  `updated_at` datetime NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
+  PRIMARY KEY (`id`),
+  UNIQUE KEY `task_id` (`task_id`),
+  KEY `idx_status` (`status`),
+  KEY `idx_created_at` (`created_at`)
+) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;
+
+-- ----------------------------
+-- Table structure for analysis_jobs (Background Analysis Jobs)
+-- ----------------------------
+DROP TABLE IF EXISTS `analysis_jobs`;
+CREATE TABLE `analysis_jobs` (
+  `id` varchar(36) COLLATE utf8mb4_unicode_ci NOT NULL,
+  `task_ids` text COLLATE utf8mb4_unicode_ci NOT NULL COMMENT 'JSON string of task IDs',
+  `analysis_type` int(11) NOT NULL COMMENT '0=single task, 1=multiple tasks',
+  `language` varchar(10) COLLATE utf8mb4_unicode_ci NOT NULL DEFAULT 'en',
+  `eval_prompt` text COLLATE utf8mb4_unicode_ci,
+  `status` varchar(20) COLLATE utf8mb4_unicode_ci NOT NULL DEFAULT 'pending' COMMENT 'pending, processing, completed, failed',
+  `result_data` longtext COLLATE utf8mb4_unicode_ci COMMENT 'JSON string of analysis result',
+  `error_message` text COLLATE utf8mb4_unicode_ci,
+  `created_at` datetime NOT NULL DEFAULT CURRENT_TIMESTAMP,
+  `updated_at` datetime NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
+  PRIMARY KEY (`id`),
+  KEY `idx_status` (`status`),
+  KEY `idx_created_at` (`created_at`),
+  KEY `idx_analysis_type` (`analysis_type`)
+) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;
+
 -- 最后重新启用外键检查
-SET FOREIGN_KEY_CHECKS = 1;
+SET FOREIGN_KEY_CHECKS = 1;

backend/model/analysis.py

@@ -3,10 +3,13 @@
 Copyright (c) 2025, All Rights Reserved.
 """
 
-from typing import Optional, Union
+from datetime import datetime
+from typing import List, Optional, Union
+from uuid import uuid4
 
 from pydantic import BaseModel, Field
-from sqlalchemy import Column, DateTime, Integer, String, Text, func
+from sqlalchemy import Column, DateTime, Integer, String, Text
+from sqlalchemy.sql import func
 
 from db.mysql import Base
 
@@ -17,60 +20,103 @@ class TaskAnalysis(Base):
     """
 
     __tablename__ = "test_insights"
-    id = Column(Integer, primary_key=True, index=True)
-    task_id = Column(String(40), nullable=False, unique=True)
+
+    id = Column(Integer, primary_key=True, autoincrement=True)
+    task_id = Column(String(36), nullable=False, unique=True)
     eval_prompt = Column(Text, nullable=False)
     analysis_report = Column(Text, nullable=False)
-    status = Column(String(20), nullable=False, default="completed")
+    status = Column(
+        String(20), nullable=False, default="pending"
+    )  # pending, processing, completed, failed
     error_message = Column(Text, nullable=True)
-    created_at = Column(DateTime, server_default=func.now())
-    updated_at = Column(DateTime, server_default=func.now(), onupdate=func.now())
+    created_at = Column(DateTime, nullable=False, server_default=func.now())
+    updated_at = Column(
+        DateTime, nullable=False, server_default=func.now(), onupdate=func.now()
+    )
+
+
+class AnalysisJob(Base):
+    """
+    SQLAlchemy model for tracking background analysis jobs.
+    """
+
+    __tablename__ = "analysis_jobs"
+
+    id = Column(String(36), primary_key=True, default=lambda: str(uuid4()))
+    task_ids = Column(Text, nullable=False)  # JSON string of task IDs
+    analysis_type = Column(Integer, nullable=False)  # 0=single, 1=multiple
+    language = Column(String(10), nullable=False, default="en")
+    eval_prompt = Column(Text, nullable=True)
+    status = Column(
+        String(20), nullable=False, default="pending"
+    )  # pending, processing, completed, failed
+    result_data = Column(Text, nullable=True)  # JSON string of analysis result
+    error_message = Column(Text, nullable=True)
+    created_at = Column(DateTime, nullable=False, server_default=func.now())
+    updated_at = Column(
+        DateTime, nullable=False, server_default=func.now(), onupdate=func.now()
+    )
 
 
 class AnalysisRequest(BaseModel):
     """
-    Request model for AI analysis.
+    Request model for AI analysis (single or multiple tasks).
+    """
+
+    task_ids: List[str] = Field(..., description="List of task IDs to analyze")
+    eval_prompt: Optional[str] = Field(
+        None, description="Custom evaluation prompt for analysis"
+    )
+    language: Optional[str] = Field("en", description="Language for analysis report")
+    background: Optional[bool] = Field(False, description="Process in background")
+
 
-    Attributes:
-        task_id: The task ID to analyze.
-        language: The language for analysis prompt (en/zh).
+class AnalysisJobRequest(BaseModel):
+    """
+    Request model for starting a background analysis job.
     """
 
-    eval_prompt: Optional[str] = Field(None, description="Custom evaluation prompt")
-    language: Optional[str] = Field(
-        "en", description="Language for analysis prompt (en/zh)"
+    task_ids: List[str] = Field(..., description="List of task IDs to analyze")
+    eval_prompt: Optional[str] = Field(
+        None, description="Custom evaluation prompt for analysis"
     )
+    language: Optional[str] = Field("en", description="Language for analysis report")
 
 
 class AnalysisResponse(BaseModel):
     """
     Response model for AI analysis.
-
-    Attributes:
-        task_id: The task ID.
-        analysis_report: The AI analysis content.
-        status: The analysis status.
-        error_message: Error message if analysis failed.
-        created_at: The creation timestamp.
     """
 
-    task_id: str
+    task_ids: List[str]
     analysis_report: str
     status: str
     error_message: Optional[str] = None
     created_at: str
+    job_id: Optional[str] = Field(
+        None, description="Background job ID if processed asynchronously"
+    )
 
 
-class GetAnalysisResponse(BaseModel):
+class AnalysisJobResponse(BaseModel):
+    """
+    Response model for background analysis job status.
     """
-    Response model for getting analysis results.
 
-    Attributes:
-        data: The analysis data.
-        status: The status of the response.
-        error: An error message if the request failed, otherwise None.
+    job_id: str
+    task_ids: List[str]
+    status: str
+    result_data: Optional[str] = None
+    error_message: Optional[str] = None
+    created_at: str
+    updated_at: str
+
+
+class GetAnalysisResponse(BaseModel):
+    """
+    Response model for getting AI analysis.
     """
 
-    data: Optional[AnalysisResponse] = None
+    data: Union[AnalysisResponse, None]
     status: str
-    error: Union[str, None]
+    error: Optional[str] = None

backend/model/task.py

@@ -218,13 +218,15 @@ class ModelTaskInfo(BaseModel):
         task_id: The task ID.
         task_name: The task name.
         created_at: The creation timestamp.
+        duration: The test duration in seconds.
     """
 
     model_name: str
     concurrent_users: int
     task_id: str
     task_name: str
     created_at: str
+    duration: int
 
 
 class ComparisonRequest(BaseModel):
@@ -249,25 +251,31 @@ class ComparisonMetrics(BaseModel):
         model_name: The model name.
         concurrent_users: The number of concurrent users.
         task_name: The task name.
-        ttft: Time to first token (avg_latency in seconds).
+        duration: Test duration with 's' suffix.
+        stream_mode: Whether stream mode is enabled.
+        dataset_type: Type of dataset used.
+        first_token_latency: Time to first token (avg_latency in seconds).
+        total_time: Total time for request completion.
         total_tps: Total tokens per second.
         completion_tps: Completion tokens per second.
-        avg_total_tpr: Average total tokens per request.
-        avg_completion_tpr: Average completion tokens per request.
-        avg_response_time: Average response time.
+        avg_total_tokens_per_req: Average total tokens per request.
+        avg_completion_tokens_per_req: Average completion tokens per request.
         rps: Requests per second.
     """
 
     task_id: str
     model_name: str
     concurrent_users: int
     task_name: str
-    ttft: float
+    duration: str
+    stream_mode: bool
+    dataset_type: str
+    first_token_latency: float
+    total_time: float
     total_tps: float
     completion_tps: float
-    avg_total_tpr: float
-    avg_completion_tpr: float
-    avg_response_time: float
+    avg_total_tokens_per_req: float
+    avg_completion_tokens_per_req: float
     rps: float
 
 

backend/mypy.ini

@@ -9,6 +9,10 @@ disallow_untyped_defs = False
 disallow_incomplete_defs = False
 check_untyped_defs = False
 
+# Fix module resolution issues
+namespace_packages = True
+explicit_package_bases = True
+
 # ignore SQLAlchemy module import errors
 [mypy-sqlalchemy.*]
 ignore_missing_imports = True