app.py，neo4j_indexing.py、Hybrid_graphsearch.py里面配置信息统一从app.py导入

app.py

@@ -92,7 +92,7 @@ from default_ontology_config import (
     get_relationships,
     get_relationship_types,
 )
-from config import NEO4J_CONFIG
+from config import NEO4J_CONFIG,URL,API_URLS,API_OTHER_URLS
 converter = Doc2PDF()
 DATA_DIR = "/app/files"
 # ================== 全局资源容器 ==================
@@ -199,24 +199,11 @@ NEO4J_URI = NEO4J_CONFIG["uri"]
 NEO4J_USER = NEO4J_CONFIG["username"]
 NEO4J_PASSWORD = NEO4J_CONFIG["password"] # 不设默认值，强制要求提供
 NEO4J_DATABASE =NEO4J_CONFIG["database"]
-indexing_url = "http://192.168.1.108:9085/neo4j_indexing"  # 根据实际部署调整
+indexing_url = URL['indexing_url'] # 根据实际部署调整
-PREFIX_URL = os.getenv("KGRAG_PREFIX_URL", "http://192.168.1.108:9085")
+PREFIX_URL = URL['prefix_url']
-API_URLS = [
+API_URLS = API_URLS
-      "http://192.168.1.64:18000/analyze-pdf",
+API_other_URLS = API_OTHER_URLS
-    # "http://192.168.0.111:9977/analyze-pdf",
+
-    # "http://192.168.0.111:9978/analyze-pdf",
-    #"http://192.168.0.111:9979/analyze-pdf",
-    #"http://192.168.0.111:9980/analyze-pdf",
-    # "http://192.168.0.111:9975/analyze-pdf",
-]
-API_other_URLS = [
-      "http://192.168.1.64:18000/analyze-otherfile",
-    # "http://192.168.0.111:9977/analyze-otherfile",
-    # "http://192.168.0.111:9978/analyze-otherfile",
-    #"http://192.168.0.111:9979/analyze-otherfile",
-    #"http://192.168.0.111:9980/analyze-otherfile",
-    # "http://192.168.0.111:9975/analyze-otherfile",
-]
 _api_url_inflight = [0] * len(API_URLS)
 _api_url_lock = threading.Lock()
 

config.py

@@ -5,26 +5,25 @@ import os
 # ==================== 大模型配置 ====================
 LLM_CONFIG = {
     "model": "46-qwen3.5-35B",
-    "base_url": "https://openai.zkzdht.com/v1",
+    "base_url": "http://192.168.0.46:59800/v1",
     "api_key": "gpustack_dee9ca823290886c_5edfc86aeeeceb1e9ee5162941cb2cb5",
     "temperature": 0.1,
-    "max_tokens": 25096,
+    "max_tokens": 55096,
     "timeout": 30
 }
 
 # ==================== 嵌入模型配置 ====================
 EMBEDDING_CONFIG = {
     "model": "bge-m3",
-    "base_url": "http://embedding:9700/v1/embeddings",
+    "base_url": "http://192.168.0.46:59700/v1/embeddings",
     "api_key": "gpustack_dee9ca823290886c_5edfc86aeeeceb1e9ee5162941cb2cb5",
-    "base_url_v2": "http://embedding:9700/v1",
 }
 
 # ==================== Neo4j 图数据库配置 ====================
 NEO4J_CONFIG = {
     # 优先读取环境变量，如果没有则使用默认值
     # 注意：本地开发通常用 localhost，部署时用服务名 neo4j
-    "uri": "bolt://neo4j:7687", 
+    "uri": "neo4j://192.168.0.46:57687", 
     "username":  "neo4j",
     "password": "zdht123@",
     "database":  "neo4j", # 新增 database 配置
@@ -38,12 +37,39 @@ SEARCH_CONFIG = {
     "search_label": "Searchable",
     "vector_index_name": "global_searchable_embedding",
     "fulltext_index_name": "global_searchable_content_search",
-    "fulltext_field_index_name": "searchable_fulltext",
+    "fulltext_field_index_name": "global_searchable_fulltext_search",
+    "global_entity_embedding" : "global_entity_embedding",
+    "global_entity_content_search": "global_entity_content_search",
     "excluded_business_labels": [
         "Entity", "Chunk", "Document", "_Bloom_Perspective_", "Searchable"
     ],
     "vector_dimension": 1024,
     "jieba_pos_whitelist": ["n", "nr", "ns", "nt", "nz", "vn", "eng"],
+    "embed_batch_size" : 64,
+    "embed_thread_num"  : 4,
+    "node_fetch_page" : 10000,
 }
 
+INDEXING_ULR = "http://192.168.0.46:59085/neo4j_indexing"
 
+URL = {
+    "indexing_url" : "http://192.168.0.46:59085/neo4j_indexing",
+    "prefix_url" : "http://192.168.0.46:59085",
+}
+
+API_URLS = [
+      "http://192.168.0.64:59988/analyze-pdf",
+    # "http://192.168.0.111:9977/analyze-pdf",
+    # "http://192.168.0.111:9978/analyze-pdf",
+    #"http://192.168.0.111:9979/analyze-pdf",
+    #"http://192.168.0.111:9980/analyze-pdf",
+    # "http://192.168.0.111:9975/analyze-pdf",
+]
+API_OTHER_URLS = [
+      "http://192.168.0.46:59988/analyze-otherfile",
+    # "http://192.168.0.111:9977/analyze-otherfile",
+    # "http://192.168.0.111:9978/analyze-otherfile",
+    #"http://192.168.0.111:9979/analyze-otherfile",
+    #"http://192.168.0.111:9980/analyze-otherfile",
+    # "http://192.168.0.111:9975/analyze-otherfile",
+]

graph_search/Hybrid_graphsearch.py

@@ -3,8 +3,10 @@ import httpx
 import json
 import asyncio
 import warnings
+import socket
 from typing import List, Set, Tuple, Dict, Optional
 import ast
+from urllib.parse import quote, urlsplit, urlunsplit
 warnings.filterwarnings("ignore", category=UserWarning, module='jieba._compat')
 import re
 import jieba
@@ -21,6 +23,12 @@ if hasattr(sys.stdout, "reconfigure"):
     sys.stdout.reconfigure(encoding="utf-8", errors="replace")
     sys.stderr.reconfigure(encoding="utf-8", errors="replace")
 sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+try:
+    from dotenv import load_dotenv
+    load_dotenv(os.path.join(os.path.dirname(os.path.dirname(os.path.abspath(__file__))), ".env"))
+    load_dotenv()
+except ImportError:
+    pass
 from config import LLM_CONFIG, EMBEDDING_CONFIG, NEO4J_CONFIG, SEARCH_CONFIG
 
 # ✅ 2. 【关键】全局清除代理环境变量，防止内网请求走代理
@@ -30,14 +38,22 @@ for _key in ['HTTP_PROXY', 'HTTPS_PROXY', 'http_proxy', 'https_proxy']:
         print(f"[Init] Removed environment variable: {_key}")
 
 
-# ================== 2. 基础配置 ==================
-# 如果 config.py 中没有 Neo4j 配置，这里暂时保留硬编码，建议也移入 config.py
 # ================== 2. 基础配置 ==================
 # 从配置文件中加载参数
-URI = os.getenv("NEO4J_URI", NEO4J_CONFIG["uri"])
+def _first_env(*names: str) -> Optional[str]:
-AUTH = (NEO4J_CONFIG["username"], NEO4J_CONFIG["password"])
+    for name in names:
-# URI = "bolt://192.168.1.164:7687"
+        value = os.getenv(name)
-# AUTH = ("neo4j", "zdht123@")
+        if value:
+            return value
+    return None
+
+
+URI = NEO4J_CONFIG["uri"]
+AUTH = (
+    NEO4J_CONFIG["username"],
+    NEO4J_CONFIG["password"],
+)
+RERANK_URL = EMBEDDING_CONFIG["base_url"]
 NAME_PROPERTY = SEARCH_CONFIG["name_property"]
 FULLTEXT_PROPERTY = SEARCH_CONFIG["fulltext_property"]
 EMBEDDING_PROPERTY = SEARCH_CONFIG["embedding_property"]
@@ -67,9 +83,9 @@ _SCHEMA_LOCK = asyncio.Lock()
 class LocalBgeM3Embeddings(Embedder):
     def __init__(self, base_url: str = None, api_key: str = None, model_name: str = None):
         # 使用默认配置，允许外部覆盖
-        self.base_url = base_url or EMBEDDING_CONFIG["base_url"]
+        self.base_url = EMBEDDING_CONFIG["base_url"]
-        self.api_key = api_key or EMBEDDING_CONFIG["api_key"]
+        self.api_key = EMBEDDING_CONFIG["api_key"]
-        self.model_name = model_name or EMBEDDING_CONFIG["model"]
+        self.model_name = EMBEDDING_CONFIG["model"]
 
     def embed_query(self, text: str) -> List[float]:
         return self.embed_documents([text])[0]
@@ -100,6 +116,7 @@ class LocalBgeM3Embeddings(Embedder):
                     
         return embedding_list
 
+
 async def openai_chat_aysnc_nothink(query: str, timeout: int = None) -> str | None:
     """异步调用 OpenAI 兼容 API"""
     global _LLM_CLIENT
@@ -164,7 +181,6 @@ def _build_retriever(sync_driver, embedder) -> HybridCypherRetriever:
         """
     )
 
-
 # ================== 6. Schema 解析与 N 跳标签扩展 ==================
 def parse_schema_relationships(schema_str: str) -> List[Tuple[str, str, str]]:
     pattern = re.compile(r"\(:([^)]+)\)-\[:([^\]]+)\]->\(:([^)]+)\)")
@@ -249,7 +265,7 @@ async def rerank_query(query: str, documents: List[str], top_n: int = 3) -> List
         "Authorization": f'Bearer {os.getenv("OPENAI_API_KEY", "none")}',
     }
     data = {
-        "model": "bge-rerank",
+        "model": _first_env("RERANK_MODEL", "OPENAI_RERANKER_MODEL") or "bge-rerank",
         "query": query,
         "top_n": top_n,
         "documents": documents,
@@ -257,7 +273,7 @@ async def rerank_query(query: str, documents: List[str], top_n: int = 3) -> List
     
     async with httpx.AsyncClient() as client:
         response = await client.post(
-            "http://rerank:9600/v1/rerank",
+            RERANK_URL,
             headers=headers,
             json=data,
             timeout=30.0,

neo4j_indexing.py

@@ -14,6 +14,7 @@ from neo4j_graphrag.indexes import (
 from modelsAPI.model_api import OpenaiAPI
 from dotenv import load_dotenv
 from typing import List, Set, Tuple, Dict, Optional
+from config import LLM_CONFIG, EMBEDDING_CONFIG, NEO4J_CONFIG, SEARCH_CONFIG
 
 # 加载 .env 文件中的环境变量
 load_dotenv()
@@ -26,38 +27,38 @@ if not logger.handlers:
     handler = logging.StreamHandler()
     handler.setFormatter(formatter)
     logger.addHandler(handler)
-
+    
-vector_dim = 1024  # OpenAI bge-m3 模型的嵌入向量维度
+VECTOR_DIMENSION = SEARCH_CONFIG["vector_dimension"]
-embed_batch_size = 64  # 嵌入向量计算批次大小
+vector_dim = SEARCH_CONFIG["vector_dimension"]  # OpenAI bge-m3 模型的嵌入向量维度
-EMBED_MAX_WORKERS = 4  # embedding 批次并行线程数（受下游 OpenaiAPI 限流约束，勿过大）
+embed_batch_size = SEARCH_CONFIG["embed_batch_size"]  # 嵌入向量计算批次大小
-NODE_FETCH_PAGE = 10000  # 单次拉取待生成 embedding 节点的分页大小
+EMBED_MAX_WORKERS = SEARCH_CONFIG["embed_thread_num"]  # embedding 批次并行线程数（受下游 OpenaiAPI 限流约束，勿过大）
+NODE_FETCH_PAGE = SEARCH_CONFIG["node_fetch_page"]  # 单次拉取待生成 embedding 节点的分页大小
 
 # 全局索引名称（覆盖所有节点标签）
-GLOBAL_VECTOR_INDEX_NAME = "global_entity_embedding"
+GLOBAL_VECTOR_INDEX_NAME = SEARCH_CONFIG["global_entity_embedding"]
-GLOBAL_FULLTEXT_INDEX_NAME = "global_entity_content_search"
+GLOBAL_FULLTEXT_INDEX_NAME = SEARCH_CONFIG["global_entity_content_search"]
-DEFAULT_URI = "bolt://192.168.0.46:57687"
+DEFAULT_URI = NEO4J_CONFIG["uri"]
-DEFAULT_AUTH = ("neo4j", "zdht123@")
+DEFAULT_AUTH = (
+    NEO4J_CONFIG["username"],
+    NEO4J_CONFIG["password"],
+)
 
-NAME_PROPERTY = "名称"
+NAME_PROPERTY = SEARCH_CONFIG["name_property"]
-FULLTEXT_PROPERTY = "fulltext"
+FULLTEXT_PROPERTY = SEARCH_CONFIG["fulltext_property"]
-EMBEDDING_PROPERTY = "embedding"
+EMBEDDING_PROPERTY = SEARCH_CONFIG["embedding_property"]
-SEARCH_LABEL = "Searchable"
+SEARCH_LABEL = SEARCH_CONFIG["search_label"]
 
-VECTOR_INDEX_NAME = "global_searchable_embedding"
+VECTOR_INDEX_NAME = SEARCH_CONFIG["vector_index_name"]
-FULLTEXT_INDEX_NAME = "global_searchable_content_search"
+FULLTEXT_INDEX_NAME = SEARCH_CONFIG["fulltext_index_name"]
-FULLTEXT_FIELD_INDEX_NAME = "global_searchable_fulltext_search"
+FULLTEXT_FIELD_INDEX_NAME = SEARCH_CONFIG["fulltext_field_index_name"]
 
-VECTOR_DIMENSION: int = 1024
+EXCLUDED_BUSINESS_LABELS: Set[str] = set(SEARCH_CONFIG["excluded_business_labels"])
-
-EXCLUDED_BUSINESS_LABELS: Set[str] = {
-    "Entity", "Chunk", "Document", "_Bloom_Perspective_", SEARCH_LABEL,
-}
 OLD_INDEX_NAMES: List[str] = [
     VECTOR_INDEX_NAME,
     FULLTEXT_INDEX_NAME,
     FULLTEXT_FIELD_INDEX_NAME,
-    "global_entity_embedding",
+    GLOBAL_VECTOR_INDEX_NAME,
-    "global_entity_content_search",
+    GLOBAL_FULLTEXT_INDEX_NAME,
 ]