增加代码知识库；修复文档处理内容；增加API设置

rag-web-ui/backend/app/services/testing_generation_job_service.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import asyncio
+import logging
 from datetime import datetime
 from typing import Any, Dict, List
 
@@ -11,10 +12,15 @@ from app.db.session import SessionLocal
 from app.models.knowledge import Document, KnowledgeBase
 from app.models.tooling import TestingGeneration, ToolJob
 from app.services.embedding.embedding_factory import EmbeddingsFactory
+from app.services.llm.llm_factory import LLMFactory
+from app.services.model_config import ModelConfigService
 from app.services.retrieval.multi_kb_retriever import MultiKBRetriever, format_retrieval_context
 from app.services.testing_pipeline import run_testing_pipeline
 from app.services.vector_store import VectorStoreFactory
 
+logger = logging.getLogger(__name__)
+
+
 def _flatten_record(value: Dict[str, List[Dict[str, Any]]]) -> List[Dict[str, Any]]:
     items: List[Dict[str, Any]] = []
     for current in value.values():
@@ -22,8 +28,15 @@ def _flatten_record(value: Dict[str, List[Dict[str, Any]]]) -> List[Dict[str, An
     return items
 
 
-def _build_kb_vector_stores(db: Session, knowledge_bases: List[KnowledgeBase]) -> List[Dict[str, Any]]:
-    embeddings = EmbeddingsFactory.create()
+def _build_kb_vector_stores(
+    db: Session,
+    knowledge_bases: List[KnowledgeBase],
+    model_profile: Any,
+) -> List[Dict[str, Any]]:
+    if model_profile is None:
+        return []
+
+    embeddings = EmbeddingsFactory.create(model_profile=model_profile)
     kb_vector_stores: List[Dict[str, Any]] = []
 
     for kb in knowledge_bases:
@@ -47,8 +60,9 @@ def _resolve_knowledge_context(
     user_id: int,
     requirement_text: str,
     knowledge_base_id: int | None,
+    model_profile: Any,
 ) -> str:
-    if knowledge_base_id is None:
+    if knowledge_base_id is None or model_profile is None:
         return ""
 
     try:
@@ -60,7 +74,7 @@ def _resolve_knowledge_context(
             )
             .all()
         )
-        kb_vector_stores = _build_kb_vector_stores(db, knowledge_bases)
+        kb_vector_stores = _build_kb_vector_stores(db, knowledge_bases, model_profile)
         if not kb_vector_stores:
             return ""
 
@@ -143,6 +157,27 @@ def run_testing_generation_job(job_id: int, payload: Dict[str, Any]) -> None:
         source_document_name = str(payload.get("source_document_name") or job.input_file_name or "")
         source_job_id = payload.get("source_job_id")
         knowledge_base_id = payload.get("knowledge_base_id")
+        model_profile = ModelConfigService.get_active_config(db, job.user_id)
+        if model_profile is not None:
+            ModelConfigService.touch_last_used(db, model_profile)
+
+        use_model_generation = model_profile is not None
+        llm_model = None
+        if use_model_generation:
+            try:
+                llm_model = LLMFactory.create(streaming=False, model_profile=model_profile)
+            except Exception as exc:
+                logger.exception(
+                    "Testing generation LLM initialization failed for job=%s, falling back to rule-based output: %s",
+                    job_id,
+                    exc,
+                )
+                use_model_generation = False
+        else:
+            logger.info(
+                "Testing generation job=%s has no active model config; using rule-based output.",
+                job_id,
+            )
 
         job.status = "processing"
         job.started_at = datetime.utcnow()
@@ -183,6 +218,7 @@ def run_testing_generation_job(job_id: int, payload: Dict[str, Any]) -> None:
                 user_id=job.user_id,
                 requirement_text=description,
                 knowledge_base_id=knowledge_base_id,
+                model_profile=model_profile,
             )
 
             pipeline_result = run_testing_pipeline(
@@ -190,7 +226,8 @@ def run_testing_generation_job(job_id: int, payload: Dict[str, Any]) -> None:
                 requirement_type_input=req.get("requirementType"),
                 debug=False,
                 knowledge_context=knowledge_context,
-                use_model_generation=True,
+                use_model_generation=use_model_generation,
+                llm_model=llm_model,
                 max_items_per_group=12,
                 cases_per_item=2,
                 max_focus_points=6,