feat: 知识库小重构x2

scripts/import_openie.py

@@ -10,13 +10,13 @@ from time import sleep
 
 sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), "..")))
 
-from src.chat.knowledge.src.lpmmconfig import PG_NAMESPACE, global_config
-from src.chat.knowledge.src.embedding_store import EmbeddingManager
-from src.chat.knowledge.src.llm_client import LLMClient
-from src.chat.knowledge.src.open_ie import OpenIE
-from src.chat.knowledge.src.kg_manager import KGManager
+from src.chat.knowledge.lpmmconfig import PG_NAMESPACE, global_config
+from src.chat.knowledge.embedding_store import EmbeddingManager
+from src.chat.knowledge.llm_client import LLMClient
+from src.chat.knowledge.open_ie import OpenIE
+from src.chat.knowledge.kg_manager import KGManager
 from src.common.logger import get_module_logger
-from src.chat.knowledge.src.utils.hash import get_sha256
+from src.chat.knowledge.utils.hash import get_sha256
 
 
 # 添加项目根目录到 sys.path

scripts/info_extraction.py

@@ -13,11 +13,11 @@ sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), "..")))
 from rich.progress import Progress  # 替换为 rich 进度条
 
 from src.common.logger import get_module_logger
-from src.chat.knowledge.src.lpmmconfig import global_config
-from src.chat.knowledge.src.ie_process import info_extract_from_str
-from src.chat.knowledge.src.llm_client import LLMClient
-from src.chat.knowledge.src.open_ie import OpenIE
-from src.chat.knowledge.src.raw_processing import load_raw_data
+from src.chat.knowledge.lpmmconfig import global_config
+from src.chat.knowledge.ie_process import info_extract_from_str
+from src.chat.knowledge.llm_client import LLMClient
+from src.chat.knowledge.open_ie import OpenIE
+from src.chat.knowledge.raw_processing import load_raw_data
 from rich.progress import (
     BarColumn,
     TimeElapsedColumn,

scripts/raw_data_preprocessor.py

@@ -6,7 +6,7 @@ import datetime  # 新增导入
 
 sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), "..")))
 from src.common.logger_manager import get_logger
-from src.chat.knowledge.src.lpmmconfig import global_config
+from src.chat.knowledge.lpmmconfig import global_config
 
 logger = get_logger("lpmm")
 ROOT_PATH = os.path.abspath(os.path.join(os.path.dirname(__file__), ".."))

src/chat/knowledge/embedding_store.py

@@ -12,7 +12,7 @@ import faiss
 
 from .llm_client import LLMClient
 from .lpmmconfig import ENT_NAMESPACE, PG_NAMESPACE, REL_NAMESPACE, global_config
-from src.chat.knowledge.utils import get_sha256
+from .utils.hash import get_sha256
 from .global_logger import logger
 from rich.traceback import install
 from rich.progress import (
@@ -27,7 +27,7 @@ from rich.progress import (
 )
 
 install(extra_lines=3)
-ROOT_PATH = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "..", "..", ".."))
+ROOT_PATH = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "..", ".."))
 EMBEDDING_DATA_DIR = (
     os.path.join(ROOT_PATH, "data", "embedding")
     if global_config["persistence"]["embedding_data_dir"] is None

src/chat/knowledge/kg_manager.py

@@ -18,7 +18,7 @@ from rich.progress import (
 from quick_algo import di_graph, pagerank
 
 
-from src.chat.knowledge.utils import get_sha256
+from .utils.hash import get_sha256
 from .embedding_store import EmbeddingManager, EmbeddingStoreItem
 from .lpmmconfig import (
     ENT_NAMESPACE,
@@ -31,7 +31,7 @@ from .lpmmconfig import (
 
 from .global_logger import logger
 
-ROOT_PATH = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "..", "..", ".."))
+ROOT_PATH = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "..", ".."))
 KG_DIR = (
     os.path.join(ROOT_PATH, "data/rag")
     if global_config["persistence"]["rag_data_dir"] is None

src/chat/knowledge/knowledge_lib.py

@@ -1,10 +1,10 @@
-from .src.lpmmconfig import PG_NAMESPACE, global_config
-from .src.embedding_store import EmbeddingManager
-from .src.llm_client import LLMClient
-from .src.mem_active_manager import MemoryActiveManager
-from .src.qa_manager import QAManager
-from .src.kg_manager import KGManager
-from .src.global_logger import logger
+from src.chat.knowledge.lpmmconfig import PG_NAMESPACE, global_config
+from src.chat.knowledge.embedding_store import EmbeddingManager
+from src.chat.knowledge.llm_client import LLMClient
+from src.chat.knowledge.mem_active_manager import MemoryActiveManager
+from src.chat.knowledge.qa_manager import QAManager
+from src.chat.knowledge.kg_manager import KGManager
+from src.chat.knowledge.global_logger import logger
 # try:
 #     import quick_algo
 # except ImportError:

src/chat/knowledge/lpmmconfig.py

@@ -45,7 +45,7 @@ def _load_config(config, config_file_path):
     if "llm_providers" in file_config:
         for provider in file_config["llm_providers"]:
             if provider["name"] not in config["llm_providers"]:
-                config["llm_providers"][provider["name"]] = dict()
+                config["llm_providers"][provider["name"]] = {}
             config["llm_providers"][provider["name"]]["base_url"] = provider["base_url"]
             config["llm_providers"][provider["name"]]["api_key"] = provider["api_key"]
 
@@ -135,6 +135,6 @@ global_config = dict(
 # _load_config(global_config, parser.parse_args().config_path)
 # file_path = os.path.abspath(__file__)
 # dir_path = os.path.dirname(file_path)
-ROOT_PATH = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "..", "..", ".."))
+ROOT_PATH = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "..", ".."))
 config_path = os.path.join(ROOT_PATH, "config", "lpmm_config.toml")
 _load_config(global_config, config_path)

src/chat/knowledge/mem_active_manager.py

@@ -1,7 +1,7 @@
 from .lpmmconfig import global_config
 from .embedding_store import EmbeddingManager
 from .llm_client import LLMClient
-from src.chat.knowledge.utils import dyn_select_top_k
+from .utils.dyn_topk import dyn_select_top_k
 
 
 class MemoryActiveManager:

src/chat/knowledge/qa_manager.py

@@ -8,7 +8,7 @@ from .embedding_store import EmbeddingManager
 from .llm_client import LLMClient
 from .kg_manager import KGManager
 from .lpmmconfig import global_config
-from src.chat.knowledge.utils import dyn_select_top_k
+from .utils.dyn_topk import dyn_select_top_k
 
 
 MAX_KNOWLEDGE_LENGTH = 10000  # 最大知识长度