Fix: url path join issue. (#8013)

### What problem does this PR solve?

Close #7980

### Type of change

- [x] Bug Fix (non-breaking change which fixes an issue)

rag/llm/chat_model.py

@@ -22,6 +22,7 @@ import re
 import time
 from abc import ABC
 from typing import Any, Protocol
+from urllib.parse import urljoin
 
 import openai
 import requests
@@ -445,8 +446,7 @@ class XinferenceChat(Base):
     def __init__(self, key=None, model_name="", base_url=""):
         if not base_url:
             raise ValueError("Local llm url cannot be None")
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
         super().__init__(key, model_name, base_url)
 
 
@@ -454,8 +454,7 @@ class HuggingFaceChat(Base):
     def __init__(self, key=None, model_name="", base_url=""):
         if not base_url:
             raise ValueError("Local llm url cannot be None")
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
         super().__init__(key, model_name.split("___")[0], base_url)
 
 
@@ -463,9 +462,7 @@ class ModelScopeChat(Base):
     def __init__(self, key=None, model_name="", base_url=""):
         if not base_url:
             raise ValueError("Local llm url cannot be None")
-        base_url = base_url.rstrip("/")
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
         super().__init__(key, model_name.split("___")[0], base_url)
 
 
@@ -983,8 +980,7 @@ class LocalAIChat(Base):
 
         if not base_url:
             raise ValueError("Local llm url cannot be None")
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
         self.client = OpenAI(api_key="empty", base_url=base_url)
         self.model_name = model_name.split("___")[0]
 
@@ -1442,8 +1438,7 @@ class LmStudioChat(Base):
     def __init__(self, key, model_name, base_url):
         if not base_url:
             raise ValueError("Local llm url cannot be None")
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
         super().__init__(key, model_name, base_url)
         self.client = OpenAI(api_key="lm-studio", base_url=base_url)
         self.model_name = model_name
@@ -1542,7 +1537,7 @@ class CoHereChat(Base):
 class LeptonAIChat(Base):
     def __init__(self, key, model_name, base_url=None):
         if not base_url:
-            base_url = os.path.join("https://" + model_name + ".lepton.run", "api", "v1")
+            base_url = urljoin("https://" + model_name + ".lepton.run", "api/v1")
         super().__init__(key, model_name, base_url)
 
 
@@ -2016,6 +2011,5 @@ class GPUStackChat(Base):
     def __init__(self, key=None, model_name="", base_url=""):
         if not base_url:
             raise ValueError("Local llm url cannot be None")
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
         super().__init__(key, model_name, base_url)

rag/llm/cv_model.py

@@ -19,6 +19,7 @@ import json
 import os
 from abc import ABC
 from io import BytesIO
+from urllib.parse import urljoin
 
 import requests
 from ollama import Client
@@ -546,8 +547,7 @@ class LocalAICV(GptV4):
     def __init__(self, key, model_name, base_url, lang="Chinese"):
         if not base_url:
             raise ValueError("Local cv model url cannot be None")
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
         self.client = OpenAI(api_key="empty", base_url=base_url)
         self.model_name = model_name.split("___")[0]
         self.lang = lang
@@ -555,8 +555,7 @@ class LocalAICV(GptV4):
 
 class XinferenceCV(Base):
     def __init__(self, key, model_name="", lang="Chinese", base_url=""):
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
         self.client = OpenAI(api_key=key, base_url=base_url)
         self.model_name = model_name
         self.lang = lang
@@ -706,11 +705,9 @@ class NvidiaCV(Base):
         self.lang = lang
         factory, llm_name = model_name.split("/")
         if factory != "liuhaotian":
-            self.base_url = os.path.join(base_url, factory, llm_name)
+            self.base_url = urljoin(base_url, f"{factory}/{llm_name}")
         else:
-            self.base_url = os.path.join(
-                base_url, "community", llm_name.replace("-v1.6", "16")
-            )
+            self.base_url = urljoin(f"{base_url}/community", llm_name.replace("-v1.6", "16"))
         self.key = key
 
     def describe(self, image):
@@ -799,8 +796,7 @@ class LmStudioCV(GptV4):
     def __init__(self, key, model_name, lang="Chinese", base_url=""):
         if not base_url:
             raise ValueError("Local llm url cannot be None")
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
         self.client = OpenAI(api_key="lm-studio", base_url=base_url)
         self.model_name = model_name
         self.lang = lang
@@ -810,8 +806,7 @@ class OpenAI_APICV(GptV4):
     def __init__(self, key, model_name, lang="Chinese", base_url=""):
         if not base_url:
             raise ValueError("url cannot be None")
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
         self.client = OpenAI(api_key=key, base_url=base_url)
         self.model_name = model_name.split("___")[0]
         self.lang = lang
@@ -1032,8 +1027,7 @@ class GPUStackCV(GptV4):
     def __init__(self, key, model_name, lang="Chinese", base_url=""):
         if not base_url:
             raise ValueError("Local llm url cannot be None")
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
         self.client = OpenAI(api_key=key, base_url=base_url)
         self.model_name = model_name
         self.lang = lang

rag/llm/embedding_model.py

@@ -16,6 +16,8 @@
 import logging
 import re
 import threading
+from urllib.parse import urljoin
+
 import requests
 from huggingface_hub import snapshot_download
 from zhipuai import ZhipuAI
@@ -141,8 +143,7 @@ class LocalAIEmbed(Base):
     def __init__(self, key, model_name, base_url):
         if not base_url:
             raise ValueError("Local embedding model url cannot be None")
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
         self.client = OpenAI(api_key="empty", base_url=base_url)
         self.model_name = model_name.split("___")[0]
 
@@ -322,8 +323,7 @@ class FastEmbed(DefaultEmbedding):
 
 class XinferenceEmbed(Base):
     def __init__(self, key, model_name="", base_url=""):
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
         self.client = OpenAI(api_key=key, base_url=base_url)
         self.model_name = model_name
 
@@ -598,8 +598,7 @@ class LmStudioEmbed(LocalAIEmbed):
     def __init__(self, key, model_name, base_url):
         if not base_url:
             raise ValueError("Local llm url cannot be None")
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
         self.client = OpenAI(api_key="lm-studio", base_url=base_url)
         self.model_name = model_name
 
@@ -608,8 +607,7 @@ class OpenAI_APIEmbed(OpenAIEmbed):
     def __init__(self, key, model_name, base_url):
         if not base_url:
             raise ValueError("url cannot be None")
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
         self.client = OpenAI(api_key=key, base_url=base_url)
         self.model_name = model_name.split("___")[0]
 
@@ -833,8 +831,7 @@ class GPUStackEmbed(OpenAIEmbed):
     def __init__(self, key, model_name, base_url):
         if not base_url:
             raise ValueError("url cannot be None")
-        if base_url.split("/")[-1] != "v1":
-            base_url = os.path.join(base_url, "v1")
+        base_url = urljoin(base_url, "v1")
 
         self.client = OpenAI(api_key=key, base_url=base_url)
         self.model_name = model_name

rag/llm/rerank_model.py

@@ -296,12 +296,11 @@ class NvidiaRerank(Base):
         self.model_name = model_name
 
         if self.model_name == "nvidia/nv-rerankqa-mistral-4b-v3":
-            self.base_url = os.path.join(
-                base_url, "nv-rerankqa-mistral-4b-v3", "reranking"
+            self.base_url = urljoin(base_url, "nv-rerankqa-mistral-4b-v3/reranking"
             )
 
         if self.model_name == "nvidia/rerank-qa-mistral-4b":
-            self.base_url = os.path.join(base_url, "reranking")
+            self.base_url = urljoin(base_url, "reranking")
             self.model_name = "nv-rerank-qa-mistral-4b:1"
 
         self.headers = {