refactor: 优化环境变量获取逻辑，确保默认值处理

- 在多个文件中更新环境变量的获取方式，使用 `or` 语法简化代码，确保在未设置环境变量时使用默认值。
- 移除 `.env.template` 中与 LightRAG 相关的环境变量配置，简化配置文件。
- 更新 `pyproject.toml` 中的项目描述，提供更清晰的项目定位。
- 在文档中增加对图片上传响应格式的详细说明，提升用户理解。

.env.template

@@ -26,10 +26,6 @@ MYSQL_DATABASE=database_name
 MYSQL_PORT=3306
 MYSQL_CHARSET=utf8mb4
 
-# region lightrag
-LIGHTRAG_LLM_PROVIDER=
-LIGHTRAG_LLM_NAME=
-# endregion lightrag
 
 
 # region neo4j

docs/latest/advanced/agents-config.md

@@ -148,4 +148,20 @@ MYSQL_CHARSET=utf8mb4
 
 智能体会自动识别多模态消息并将其传递给支持图片的模型。如果模型不支持图片，会自动忽略图片内容，只处理文本部分。系统会将图片转换为符合模型要求的格式（通常是 base64 编码的 JPEG 或 PNG），确保与主流多模态模型兼容。
 
-目前仅支持上传单个图片，图片直接以 base64 存储在数据库
+目前仅支持上传单个图片，图片以 base64 编码形式存储在数据库。系统会自动处理图片的格式转换和压缩，并生成缩略图以优化性能。
+
+### 图片上传响应格式
+
+```json
+{
+  "success": true,
+  "image_content": "<base64编码的原始图片数据>",
+  "thumbnail_content": "<base64编码的缩略图数据>",
+  "width": 1024,
+  "height": 768,
+  "format": "JPEG",
+  "mime_type": "image/jpeg"
+}
+```
+
+系统会将图片信息与用户查询一同传递给支持多模态的模型，并自动适配模型要求的格式。

docs/latest/intro/knowledge-base.md

@@ -26,10 +26,6 @@ LightRAG 知识库可在知识库详情中可视化，但不支持在侧边栏
 在 Neo4j 的检索中可以看到，实际上 LightRAG 的节点和边依然是和知识图谱本身构建在了同一个 Neo4j 数据库中，但是使用了特殊的 tag 做区分。这点在后面介绍知识图谱的时候也会额外说明。
 
 
-系统默认使用 `siliconflow` 的 `Qwen/Qwen3-30B-A3B-Instruct-2507` 模型进行图谱构建。可通过环境变量自定义图谱构建模型：
-
-<<< @/../.env.template#lightrag{bash}
-
 
 ## 文档管理
 

docs/latest/intro/model-config.md

@@ -104,14 +104,14 @@ DEFAULT_CHAT_MODEL_PROVIDERS: dict[str, ChatModelProvider] = {
 }
 ```
 
-### 3. 配置环境变量
+### 2. 配置环境变量
 
 在 `.env` 文件中添加对应的环境变量：
 ```env
 CUSTOM_API_KEY_ENV_NAME=your_api_key_here
 ```
 
-### 4. 重新部署
+### 3. 重新部署
 
 ```bash
 docker compose restart api-dev

pyproject.toml

@@ -1,7 +1,7 @@
 [project]
 name = "yuxi-know"
 version = "0.4.0.dev"
-description = "Add your description here"
+description = "基于大模型的智能知识库与知识图谱智能体开发平台，融合了 RAG 技术与知识图谱技术，基于 LangGraph v1 + Vue.js + FastAPI + LightRAG 架构构建"
 readme = "README.md"
 requires-python = ">=3.11"
 dependencies = [

scripts/rename_milvus_collections.py

@@ -19,8 +19,8 @@ def rename_and_resolve_duplicates():
     Connects to Milvus, renames collections from 'kb_kb_' to 'kb_',
     and resolves duplicates by keeping the collection with more rows.
     """
-    milvus_uri = os.getenv("MILVUS_URI", "http://localhost:19530")
+    milvus_uri = os.getenv("MILVUS_URI") or "http://localhost:19530"
-    milvus_token = os.getenv("MILVUS_TOKEN", "")
+    milvus_token = os.getenv("MILVUS_TOKEN") or ""
     connection_alias = "rename_script"
 
     try:

src/agents/common/models.py

@@ -22,7 +22,7 @@ def load_chat_model(fully_specified_name: str, **kwargs) -> BaseChatModel:
 
     env_var = model_info.env
 
-    api_key = os.getenv(env_var, env_var)
+    api_key = os.getenv(env_var) or env_var
 
     base_url = get_docker_safe_url(model_info.base_url)
 

src/agents/common/toolkits/mysql/tools.py

@@ -30,9 +30,9 @@ def get_connection_manager() -> MySQLConnectionManager:
             "user": os.getenv("MYSQL_USER"),
             "password": os.getenv("MYSQL_PASSWORD"),
             "database": os.getenv("MYSQL_DATABASE"),
-            "port": int(os.getenv("MYSQL_PORT", "3306")),
+            "port": int(os.getenv("MYSQL_PORT") or "3306"),
             "charset": "utf8mb4",
-            "description": os.getenv("MYSQL_DATABASE_DESCRIPTION", "默认 MySQL 数据库"),
+            "description": os.getenv("MYSQL_DATABASE_DESCRIPTION") or "默认 MySQL 数据库",
         }
         # 验证配置完整性
         required_keys = ["host", "user", "password", "database"]

src/config/app.py

@@ -120,7 +120,7 @@ class Config(BaseModel):
 
     def _setup_paths(self):
         """设置配置文件路径"""
-        self.save_dir = os.getenv("SAVE_DIR", self.save_dir)
+        self.save_dir = os.getenv("SAVE_DIR") or self.save_dir
         self._config_file = Path(self.save_dir) / "config" / "base.toml"
         self._config_file.parent.mkdir(parents=True, exist_ok=True)
 
@@ -169,7 +169,7 @@ class Config(BaseModel):
     def _handle_environment(self):
         """处理环境变量和运行时状态"""
         # 处理模型目录
-        self.model_dir = os.environ.get("MODEL_DIR", self.model_dir)
+        self.model_dir = os.environ.get("MODEL_DIR") or self.model_dir
         if self.model_dir:
             if os.path.exists(self.model_dir):
                 logger.debug(f"Model directory ({self.model_dir}) contains: {os.listdir(self.model_dir)}")

src/knowledge/implementations/lightrag.py

@@ -8,15 +8,13 @@ from lightrag.utils import EmbeddingFunc, setup_logger
 from neo4j import GraphDatabase
 from pymilvus import connections, utility
 
+from src import config
 from src.knowledge.base import KnowledgeBase
 from src.knowledge.indexing import process_file_to_markdown, process_url_to_markdown
 from src.knowledge.utils.kb_utils import get_embedding_config, prepare_item_metadata
 from src.utils import hashstr, logger
 from src.utils.datetime_utils import shanghai_now
 
-LIGHTRAG_LLM_PROVIDER = os.getenv("LIGHTRAG_LLM_PROVIDER", "siliconflow")
-LIGHTRAG_LLM_NAME = os.getenv("LIGHTRAG_LLM_NAME", "zai-org/GLM-4.5-Air")
-
 
 class LightRagKB(KnowledgeBase):
     """基于 LightRAG 的知识库实现"""
@@ -53,8 +51,8 @@ class LightRagKB(KnowledgeBase):
         """删除数据库，同时清除Milvus和Neo4j中的数据"""
         # Drop Milvus collection
         try:
-            milvus_uri = os.getenv("MILVUS_URI", "http://localhost:19530")
+            milvus_uri = os.getenv("MILVUS_URI") or "http://localhost:19530"
-            milvus_token = os.getenv("MILVUS_TOKEN", "")
+            milvus_token = os.getenv("MILVUS_TOKEN") or ""
             connection_alias = f"lightrag_{hashstr(db_id, 6)}"
 
             connections.connect(alias=connection_alias, uri=milvus_uri, token=milvus_token)
@@ -73,9 +71,9 @@ class LightRagKB(KnowledgeBase):
             logger.error(f"Failed to drop Milvus collection {db_id}: {e}")
 
         # Delete Neo4j data
-        neo4j_uri = os.getenv("NEO4J_URI", "bolt://localhost:7687")
+        neo4j_uri = os.getenv("NEO4J_URI") or "bolt://localhost:7687"
-        neo4j_username = os.getenv("NEO4J_USERNAME", "neo4j")
+        neo4j_username = os.getenv("NEO4J_USERNAME") or "neo4j"
-        neo4j_password = os.getenv("NEO4J_PASSWORD", "0123456789")
+        neo4j_password = os.getenv("NEO4J_PASSWORD") or "0123456789"
 
         try:
             driver = GraphDatabase.driver(neo4j_uri, auth=(neo4j_username, neo4j_password))
@@ -118,7 +116,7 @@ class LightRagKB(KnowledgeBase):
         if isinstance(metadata.get("language"), str) and metadata.get("language"):
             addon_params.setdefault("language", metadata.get("language"))
         # 默认语言从环境变量读取，默认 English
-        addon_params.setdefault("language", os.getenv("SUMMARY_LANGUAGE", "English"))
+        addon_params.setdefault("language", os.getenv("SUMMARY_LANGUAGE") or "English")
 
         # 创建工作目录
         working_dir = os.path.join(self.work_dir, db_id)
@@ -181,10 +179,8 @@ class LightRagKB(KnowledgeBase):
             model_spec = f"{llm_info['provider']}/{llm_info['model_name']}"
             logger.info(f"Using user-selected LLM: {model_spec}")
         else:
-            provider = LIGHTRAG_LLM_PROVIDER
+            model_spec = config.default_model
-            model_name = LIGHTRAG_LLM_NAME
+            logger.info(f"Using default LLM from environment: {model_spec}")
-            model_spec = f"{provider}/{model_name}"
-            logger.info(f"Using default LLM from environment: {provider}/{model_name}")
 
         model = select_model(model_spec=model_spec)
 

src/knowledge/implementations/milvus.py

@@ -40,9 +40,9 @@ class MilvusKB(KnowledgeBase):
         # Milvus 配置
         # self.milvus_host = kwargs.get('milvus_host', os.getenv('MILVUS_HOST', 'localhost'))
         # self.milvus_port = kwargs.get('milvus_port', int(os.getenv('MILVUS_PORT', '19530')))
-        self.milvus_token = kwargs.get("milvus_token", os.getenv("MILVUS_TOKEN", ""))
+        self.milvus_token = kwargs.get("milvus_token", os.getenv("MILVUS_TOKEN") or "")
-        self.milvus_uri = kwargs.get("milvus_uri", os.getenv("MILVUS_URI", "http://localhost:19530"))
+        self.milvus_uri = kwargs.get("milvus_uri", os.getenv("MILVUS_URI") or "http://localhost:19530")
-        self.milvus_db = kwargs.get("milvus_db", "yuxi_know")
+        self.milvus_db = kwargs.get("milvus_db") or "yuxi_know"
 
         # 连接名称
         self.connection_alias = f"milvus_{hashstr(work_dir, 6)}"

src/knowledge/utils/kb_utils.py

@@ -214,13 +214,13 @@ def get_embedding_config(embed_info: dict) -> dict:
             if hasattr(embed_info, "name"):
                 # EmbedModelInfo 对象
                 config_dict["model"] = embed_info.name
-                config_dict["api_key"] = os.getenv(embed_info.api_key, embed_info.api_key)
+                config_dict["api_key"] = os.getenv(embed_info.api_key) or embed_info.api_key
                 config_dict["base_url"] = embed_info.base_url
                 config_dict["dimension"] = embed_info.dimension
             else:
                 # 字典形式
                 config_dict["model"] = embed_info["name"]
-                config_dict["api_key"] = os.getenv(embed_info["api_key"], embed_info["api_key"])
+                config_dict["api_key"] = os.getenv(embed_info["api_key"]) or embed_info["api_key"]
                 config_dict["base_url"] = embed_info["base_url"]
                 config_dict["dimension"] = embed_info.get("dimension", 1024)
         else:

src/models/rerank.py

@@ -125,6 +125,6 @@ def get_reranker(model_id, **kwargs):
 
     model_info = config.reranker_names[model_id]
     base_url = model_info.base_url
-    api_key = os.getenv(model_info.api_key, model_info.api_key)
+    api_key = os.getenv(model_info.api_key) or model_info.api_key
     assert api_key, f"{model_info.name} api_key is required"
     return OnlineReranker(model_name=model_info.name, api_key=api_key, base_url=base_url, **kwargs)

src/plugins/mineru_parser.py

@@ -18,7 +18,7 @@ class MinerUParser(BaseDocumentProcessor):
     """MinerU 文档解析器 - 使用 HTTP API 进行文档理解和解析"""
 
     def __init__(self, server_url: str | None = None):
-        self.server_url = server_url or os.getenv("MINERU_API_URI", "http://localhost:30001")
+        self.server_url = server_url or os.getenv("MINERU_API_URI") or "http://localhost:30001"
         self.parse_endpoint = f"{self.server_url}/file_parse"
 
     def get_service_name(self) -> str:

src/plugins/paddlex_parser.py

@@ -20,7 +20,7 @@ class PaddleXDocumentParser(BaseDocumentProcessor):
     """PaddleX 文档解析器 - 使用 PP-StructureV3 进行版面解析"""
 
     def __init__(self, server_url: str | None = None):
-        self.server_url = server_url or os.getenv("PADDLEX_URI", "http://localhost:8080")
+        self.server_url = server_url or os.getenv("PADDLEX_URI") or "http://localhost:8080"
         self.base_url = self.server_url.rstrip("/")
         self.endpoint = f"{self.base_url}/layout-parsing"
 

src/storage/minio/client.py

@@ -37,9 +37,9 @@ class MinIOClient:
 
     def __init__(self):
         """初始化 MinIO 客户端"""
-        self.endpoint = os.getenv("MINIO_URI", "http://milvus-minio:9000")
+        self.endpoint = os.getenv("MINIO_URI") or "http://milvus-minio:9000"
-        self.access_key = os.getenv("MINIO_ACCESS_KEY", "minioadmin")
+        self.access_key = os.getenv("MINIO_ACCESS_KEY") or "minioadmin"
-        self.secret_key = os.getenv("MINIO_SECRET_KEY", "minioadmin")
+        self.secret_key = os.getenv("MINIO_SECRET_KEY") or "minioadmin"
         self._client = None
 
         # 设置公开访问端点

src/utils/logging_config.py

@@ -4,7 +4,7 @@ from loguru import logger as loguru_logger
 
 from src.utils.datetime_utils import shanghai_now
 
-SAVE_DIR = os.getenv("SAVE_DIR", "saves")
+SAVE_DIR = os.getenv("SAVE_DIR") or "saves"
 DATETIME = shanghai_now().strftime("%Y-%m-%d")
 LOG_FILE = f"{SAVE_DIR}/logs/yuxi-{DATETIME}.log"
 

web/package.json

@@ -1,6 +1,6 @@
 {
   "name": "yuxi-know-web",
-  "version": "0.3.0.web",
+  "version": "0.4.0.web",
   "private": true,
   "scripts": {
     "dev": "vite",