From 15b7cd6193408266ec28c69568749155bb5a249a Mon Sep 17 00:00:00 2001
From: ReeInk <hi@ree.ink>
Date: Sat, 29 Apr 2023 18:10:27 +0800
Subject: [PATCH 1/5] feat: build docker image automatically

---
 .github/workflows/docker-image.yml | 45 ++++++++++++++++++++++++++++++
 1 file changed, 45 insertions(+)
 create mode 100644 .github/workflows/docker-image.yml

diff --git a/.github/workflows/docker-image.yml b/.github/workflows/docker-image.yml
new file mode 100644
index 0000000..112a608
--- /dev/null
+++ b/.github/workflows/docker-image.yml
@@ -0,0 +1,45 @@
+# https://docs.github.com/en/actions/publishing-packages/publishing-docker-images#publishing-images-to-github-packages
+name: Create and publish a Docker image
+
+on:
+  push:
+    branches:
+      - 'master'
+    tags:
+      - 'v*'
+
+env:
+  REGISTRY: ghcr.io
+  IMAGE_NAME: ${{ github.repository }}
+
+jobs:
+  build-and-push-image:
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      packages: write
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v3
+
+      - name: Log in to the Container registry
+        uses: docker/login-action@v2
+        with:
+          registry: ${{ env.REGISTRY }}
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Extract metadata (tags, labels) for Docker
+        id: meta
+        uses: docker/metadata-action@v4
+        with:
+          images: ${{ env.REGISTRY }}/${{ env.IMAGE_NAME }}
+
+      - name: Build and push Docker image
+        uses: docker/build-push-action@v4
+        with:
+          context: .
+          push: true
+          tags: ${{ steps.meta.outputs.tags }}
+          labels: ${{ steps.meta.outputs.labels }}

From 601c36e6079e44c17567d0bec7917cbb7edfb874 Mon Sep 17 00:00:00 2001
From: ReeInk <hi@ree.ink>
Date: Sat, 29 Apr 2023 19:53:43 +0800
Subject: [PATCH 2/5] =?UTF-8?q?=E5=8A=A0=E8=BD=BD=E7=8E=AF=E5=A2=83?=
 =?UTF-8?q?=E5=8F=98=E9=87=8F=E4=BD=9C=E4=B8=BA=E9=85=8D=E7=BD=AE?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 config.py | 47 +++++++++++++++++++++++++++++------------------
 1 file changed, 29 insertions(+), 18 deletions(-)

diff --git a/config.py b/config.py
index 30c388e..ea1bb0c 100644
--- a/config.py
+++ b/config.py
@@ -1,8 +1,11 @@
+from os import environ
+
 # [step 1]>> 例如： API_KEY = "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" （此key无效）
-API_KEY = "sk-此处填API密钥"    # 可同时填写多个API-KEY，用英文逗号分割，例如API_KEY = "sk-openaikey1,sk-openaikey2,fkxxxx-api2dkey1,fkxxxx-api2dkey2"
+API_KEY = environ.get("GPT_ACADEMIC_API_KEY") or "sk-此处填API密钥"    
+# 可同时填写多个API-KEY，用英文逗号分割，例如API_KEY = "sk-openaikey1,sk-openaikey2,fkxxxx-api2dkey1,fkxxxx-api2dkey2"
 
 # [step 2]>> 改为True应用代理，如果直接在海外服务器部署，此处不修改
-USE_PROXY = False
+USE_PROXY = environ.get("GPT_ACADEMIC_USE_PROXY") == "true" or False
 if USE_PROXY:
     # 填写格式是 [协议]://  [地址] :[端口]，填写之前不要忘记把USE_PROXY改成True，如果直接在海外服务器部署，此处不修改
     # 例如    "socks5h://localhost:11284"
@@ -13,62 +16,70 @@ if USE_PROXY:
     # 代理网络的地址，打开你的*学*网软件查看代理的协议(socks5/http)、地址(localhost)和端口(11284)
     proxies = {
         #          [协议]://  [地址]  :[端口]
-        "http":  "socks5h://localhost:11284",
-        "https": "socks5h://localhost:11284",
+        "http":  environ.get("GPT_ACADEMIC_HTTP_PROXY") or "socks5h://localhost:11284",
+        "https": environ.get("GPT_ACADEMIC_HTTPS_PROXY") or "socks5h://localhost:11284",
     }
 else:
     proxies = None
 
 # [step 3]>> 多线程函数插件中，默认允许多少路线程同时访问OpenAI。Free trial users的限制是每分钟3次，Pay-as-you-go users的限制是每分钟3500次
 # 一言以蔽之：免费用户填3，OpenAI绑了信用卡的用户可以填 16 或者更高。提高限制请查询：https://platform.openai.com/docs/guides/rate-limits/overview
-DEFAULT_WORKER_NUM = 3
+DEFAULT_WORKER_NUM = int(environ.get("GPT_ACADEMIC_DEFAULT_WORKER_NUM") or 3)
 
 
 # [step 4]>> 以下配置可以优化体验，但大部分场合下并不需要修改
 # 对话窗的高度
-CHATBOT_HEIGHT = 1115
+CHATBOT_HEIGHT = int(environ.get("GPT_ACADEMIC_CHATBOT_HEIGHT") or 1115)
 
 # 代码高亮
-CODE_HIGHLIGHT = True
+CODE_HIGHLIGHT = environ.get("GPT_ACADEMIC_CODE_HIGHLIGHT") != "false" # 默认为True
 
 # 窗口布局
-LAYOUT = "LEFT-RIGHT"  # "LEFT-RIGHT"（左右布局） # "TOP-DOWN"（上下布局）
-DARK_MODE = True  # "LEFT-RIGHT"（左右布局） # "TOP-DOWN"（上下布局）
+LAYOUT = environ.get("GPT_ACADEMIC_LAYOUT") or "LEFT-RIGHT"  # "LEFT-RIGHT"（左右布局） # "TOP-DOWN"（上下布局）
+DARK_MODE = environ.get("GPT_ACADEMIC_DARK_MODE") == "true"  # 是否启用暗色模式，默认为False
 
 # 发送请求到OpenAI后，等待多久判定为超时
-TIMEOUT_SECONDS = 30
+TIMEOUT_SECONDS = int(environ.get("GPT_ACADEMIC_TIMEOUT_SECONDS") or 30) 
 
 # 网页的端口, -1代表随机端口
-WEB_PORT = -1
+WEB_PORT = int(environ.get("GPT_ACADEMIC_WEB_PORT") or -1)
 
 # 如果OpenAI不响应（网络卡顿、代理失败、KEY失效），重试的次数限制
-MAX_RETRY = 2
+MAX_RETRY = int(environ.get("GPT_ACADEMIC_MAX_RETRY") or 2)
 
 # OpenAI模型选择是（gpt4现在只对申请成功的人开放，体验gpt-4可以试试api2d）
-LLM_MODEL = "gpt-3.5-turbo" # 可选 ↓↓↓
+LLM_MODEL = environ.get("GPT_ACADEMIC_LLM_MODEL") or "gpt-3.5-turbo" # 可选 ↓↓↓
 AVAIL_LLM_MODELS = ["gpt-3.5-turbo", "api2d-gpt-3.5-turbo", "gpt-4", "api2d-gpt-4", "chatglm", "newbing"]
 
 # 本地LLM模型如ChatGLM的执行方式 CPU/GPU
-LOCAL_MODEL_DEVICE = "cpu" # 可选 "cuda"
+LOCAL_MODEL_DEVICE = environ.get("GPT_ACADEMIC_LOCAL_MODEL_DEVICE") or "cpu" # 可选 "cuda"
 
 # 设置gradio的并行线程数（不需要修改）
-CONCURRENT_COUNT = 100
+CONCURRENT_COUNT = int(environ.get("GPT_ACADEMIC_CONCURRENT_COUNT") or 100)
 
 # 设置用户名和密码（不需要修改）（相关功能不稳定，与gradio版本和网络都相关，如果本地使用不建议加这个）
 # [("username", "password"), ("username2", "password2"), ...]
 AUTHENTICATION = []
+# 如果设置`GPT_ACADEMIC_AUTHENTICATION`环境变量，格式为 "username,password;username2,password2;..."
+if environ.get("GPT_ACADEMIC_AUTHENTICATION"):
+    for i in environ.get("GPT_ACADEMIC_AUTHENTICATION").split(";"):
+        AUTHENTICATION.append(tuple(i.split(",")))
 
 # 重新URL重新定向，实现更换API_URL的作用（常规情况下，不要修改!!）
 # （高危设置！通过修改此设置，您将把您的API-KEY和对话隐私完全暴露给您设定的中间人！）
 # 格式 {"https://api.openai.com/v1/chat/completions": "在这里填写重定向的api.openai.com的URL"} 
 # 例如 API_URL_REDIRECT = {"https://api.openai.com/v1/chat/completions": "https://ai.open.com/api/conversation"}
 API_URL_REDIRECT = {}
+# 如果设置`GPT_ACADEMIC_API_URL_REDIRECT`环境变量，格式为 "https://api.openai.com/v1/chat/completions,https://ai.open.com/api/conversation;..."
+if environ.get("GPT_ACADEMIC_API_URL_REDIRECT"):
+    for i in environ.get("GPT_ACADEMIC_API_URL_REDIRECT").split(";"):
+        API_URL_REDIRECT[i.split(",")[0]] = i.split(",")[1]
 
 # 如果需要在二级路径下运行（常规情况下，不要修改!!）（需要配合修改main.py才能生效!）
-CUSTOM_PATH = "/"
+CUSTOM_PATH = environ.get("GPT_ACADEMIC_CUSTOM_PATH") or "/"
 
 # 如果需要使用newbing，把newbing的长长的cookie放到这里
-NEWBING_STYLE = "creative"  # ["creative", "balanced", "precise"]
-NEWBING_COOKIES = """
+NEWBING_STYLE = environ.get("GPT_ACADEMIC_NEWBING_STYLE") or "creative"  # ["creative", "balanced", "precise"]
+NEWBING_COOKIES = environ.get("GPT_ACADEMIC_NEWBING_COOKIES") or """
 your bing cookies here
 """
\ No newline at end of file

From f187a23dc13985052842604e8f761dd4c5574f2a Mon Sep 17 00:00:00 2001
From: ReeInk <hi@ree.ink>
Date: Sun, 30 Apr 2023 14:34:35 +0800
Subject: [PATCH 3/5] =?UTF-8?q?Revert=20"=E5=8A=A0=E8=BD=BD=E7=8E=AF?=
 =?UTF-8?q?=E5=A2=83=E5=8F=98=E9=87=8F=E4=BD=9C=E4=B8=BA=E9=85=8D=E7=BD=AE?=
 =?UTF-8?q?"?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

This reverts commit 601c36e6079e44c17567d0bec7917cbb7edfb874.
---
 config.py | 47 ++++++++++++++++++-----------------------------
 1 file changed, 18 insertions(+), 29 deletions(-)

diff --git a/config.py b/config.py
index ea1bb0c..30c388e 100644
--- a/config.py
+++ b/config.py
@@ -1,11 +1,8 @@
-from os import environ
-
 # [step 1]>> 例如： API_KEY = "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" （此key无效）
-API_KEY = environ.get("GPT_ACADEMIC_API_KEY") or "sk-此处填API密钥"    
-# 可同时填写多个API-KEY，用英文逗号分割，例如API_KEY = "sk-openaikey1,sk-openaikey2,fkxxxx-api2dkey1,fkxxxx-api2dkey2"
+API_KEY = "sk-此处填API密钥"    # 可同时填写多个API-KEY，用英文逗号分割，例如API_KEY = "sk-openaikey1,sk-openaikey2,fkxxxx-api2dkey1,fkxxxx-api2dkey2"
 
 # [step 2]>> 改为True应用代理，如果直接在海外服务器部署，此处不修改
-USE_PROXY = environ.get("GPT_ACADEMIC_USE_PROXY") == "true" or False
+USE_PROXY = False
 if USE_PROXY:
     # 填写格式是 [协议]://  [地址] :[端口]，填写之前不要忘记把USE_PROXY改成True，如果直接在海外服务器部署，此处不修改
     # 例如    "socks5h://localhost:11284"
@@ -16,70 +13,62 @@ if USE_PROXY:
     # 代理网络的地址，打开你的*学*网软件查看代理的协议(socks5/http)、地址(localhost)和端口(11284)
     proxies = {
         #          [协议]://  [地址]  :[端口]
-        "http":  environ.get("GPT_ACADEMIC_HTTP_PROXY") or "socks5h://localhost:11284",
-        "https": environ.get("GPT_ACADEMIC_HTTPS_PROXY") or "socks5h://localhost:11284",
+        "http":  "socks5h://localhost:11284",
+        "https": "socks5h://localhost:11284",
     }
 else:
     proxies = None
 
 # [step 3]>> 多线程函数插件中，默认允许多少路线程同时访问OpenAI。Free trial users的限制是每分钟3次，Pay-as-you-go users的限制是每分钟3500次
 # 一言以蔽之：免费用户填3，OpenAI绑了信用卡的用户可以填 16 或者更高。提高限制请查询：https://platform.openai.com/docs/guides/rate-limits/overview
-DEFAULT_WORKER_NUM = int(environ.get("GPT_ACADEMIC_DEFAULT_WORKER_NUM") or 3)
+DEFAULT_WORKER_NUM = 3
 
 
 # [step 4]>> 以下配置可以优化体验，但大部分场合下并不需要修改
 # 对话窗的高度
-CHATBOT_HEIGHT = int(environ.get("GPT_ACADEMIC_CHATBOT_HEIGHT") or 1115)
+CHATBOT_HEIGHT = 1115
 
 # 代码高亮
-CODE_HIGHLIGHT = environ.get("GPT_ACADEMIC_CODE_HIGHLIGHT") != "false" # 默认为True
+CODE_HIGHLIGHT = True
 
 # 窗口布局
-LAYOUT = environ.get("GPT_ACADEMIC_LAYOUT") or "LEFT-RIGHT"  # "LEFT-RIGHT"（左右布局） # "TOP-DOWN"（上下布局）
-DARK_MODE = environ.get("GPT_ACADEMIC_DARK_MODE") == "true"  # 是否启用暗色模式，默认为False
+LAYOUT = "LEFT-RIGHT"  # "LEFT-RIGHT"（左右布局） # "TOP-DOWN"（上下布局）
+DARK_MODE = True  # "LEFT-RIGHT"（左右布局） # "TOP-DOWN"（上下布局）
 
 # 发送请求到OpenAI后，等待多久判定为超时
-TIMEOUT_SECONDS = int(environ.get("GPT_ACADEMIC_TIMEOUT_SECONDS") or 30) 
+TIMEOUT_SECONDS = 30
 
 # 网页的端口, -1代表随机端口
-WEB_PORT = int(environ.get("GPT_ACADEMIC_WEB_PORT") or -1)
+WEB_PORT = -1
 
 # 如果OpenAI不响应（网络卡顿、代理失败、KEY失效），重试的次数限制
-MAX_RETRY = int(environ.get("GPT_ACADEMIC_MAX_RETRY") or 2)
+MAX_RETRY = 2
 
 # OpenAI模型选择是（gpt4现在只对申请成功的人开放，体验gpt-4可以试试api2d）
-LLM_MODEL = environ.get("GPT_ACADEMIC_LLM_MODEL") or "gpt-3.5-turbo" # 可选 ↓↓↓
+LLM_MODEL = "gpt-3.5-turbo" # 可选 ↓↓↓
 AVAIL_LLM_MODELS = ["gpt-3.5-turbo", "api2d-gpt-3.5-turbo", "gpt-4", "api2d-gpt-4", "chatglm", "newbing"]
 
 # 本地LLM模型如ChatGLM的执行方式 CPU/GPU
-LOCAL_MODEL_DEVICE = environ.get("GPT_ACADEMIC_LOCAL_MODEL_DEVICE") or "cpu" # 可选 "cuda"
+LOCAL_MODEL_DEVICE = "cpu" # 可选 "cuda"
 
 # 设置gradio的并行线程数（不需要修改）
-CONCURRENT_COUNT = int(environ.get("GPT_ACADEMIC_CONCURRENT_COUNT") or 100)
+CONCURRENT_COUNT = 100
 
 # 设置用户名和密码（不需要修改）（相关功能不稳定，与gradio版本和网络都相关，如果本地使用不建议加这个）
 # [("username", "password"), ("username2", "password2"), ...]
 AUTHENTICATION = []
-# 如果设置`GPT_ACADEMIC_AUTHENTICATION`环境变量，格式为 "username,password;username2,password2;..."
-if environ.get("GPT_ACADEMIC_AUTHENTICATION"):
-    for i in environ.get("GPT_ACADEMIC_AUTHENTICATION").split(";"):
-        AUTHENTICATION.append(tuple(i.split(",")))
 
 # 重新URL重新定向，实现更换API_URL的作用（常规情况下，不要修改!!）
 # （高危设置！通过修改此设置，您将把您的API-KEY和对话隐私完全暴露给您设定的中间人！）
 # 格式 {"https://api.openai.com/v1/chat/completions": "在这里填写重定向的api.openai.com的URL"} 
 # 例如 API_URL_REDIRECT = {"https://api.openai.com/v1/chat/completions": "https://ai.open.com/api/conversation"}
 API_URL_REDIRECT = {}
-# 如果设置`GPT_ACADEMIC_API_URL_REDIRECT`环境变量，格式为 "https://api.openai.com/v1/chat/completions,https://ai.open.com/api/conversation;..."
-if environ.get("GPT_ACADEMIC_API_URL_REDIRECT"):
-    for i in environ.get("GPT_ACADEMIC_API_URL_REDIRECT").split(";"):
-        API_URL_REDIRECT[i.split(",")[0]] = i.split(",")[1]
 
 # 如果需要在二级路径下运行（常规情况下，不要修改!!）（需要配合修改main.py才能生效!）
-CUSTOM_PATH = environ.get("GPT_ACADEMIC_CUSTOM_PATH") or "/"
+CUSTOM_PATH = "/"
 
 # 如果需要使用newbing，把newbing的长长的cookie放到这里
-NEWBING_STYLE = environ.get("GPT_ACADEMIC_NEWBING_STYLE") or "creative"  # ["creative", "balanced", "precise"]
-NEWBING_COOKIES = environ.get("GPT_ACADEMIC_NEWBING_COOKIES") or """
+NEWBING_STYLE = "creative"  # ["creative", "balanced", "precise"]
+NEWBING_COOKIES = """
 your bing cookies here
 """
\ No newline at end of file

From e5e3e0aa43d8db8b5d731dbabcb1f44d1fed7808 Mon Sep 17 00:00:00 2001
From: ReeInk <hi@ree.ink>
Date: Sun, 30 Apr 2023 17:30:31 +0800
Subject: [PATCH 4/5] =?UTF-8?q?=E8=AF=BB=E5=8F=96=E7=8E=AF=E5=A2=83?=
 =?UTF-8?q?=E5=8F=98=E9=87=8F=E4=BD=9C=E4=B8=BA=E9=85=8D=E7=BD=AE?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 toolbox.py | 55 ++++++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 53 insertions(+), 2 deletions(-)

diff --git a/toolbox.py b/toolbox.py
index 5e42164..81ef1ed 100644
--- a/toolbox.py
+++ b/toolbox.py
@@ -3,6 +3,7 @@ import importlib
 import traceback
 import inspect
 import re
+import os
 from latex2mathml.converter import convert as tex2mathml
 from functools import wraps, lru_cache
 
@@ -517,13 +518,63 @@ def select_api_key(keys, llm_model):
     api_key = random.choice(avail_key_list) # 随机负载均衡
     return api_key
 
+def read_single_conf_from_env(arg, default_value):
+    ENV_PREFIX = "GPT_ACADEMIC_" # 环境变量的前缀
+    env_arg = ENV_PREFIX + arg # 环境变量的KEY
+    if arg == "proxies":
+        # 对于proxies，我们使用多个环境变量来配置
+        # HTTP_PROXY: 对应http代理
+        # HTTPS_PROXY: 对应https代理
+        # ALL_PROXY: 对应http和https代理，优先级较HTTP_PROXY和HTTPS_PROXY更低
+        http_proxy = os.environ.get(ENV_PREFIX + "HTTP_PROXY") or os.environ.get("ALL_PROXY")
+        assert http_proxy is not None, f"请设置环境变量{ENV_PREFIX + 'HTTP_PROXY'}"
+        https_proxy = os.environ.get(ENV_PREFIX + "HTTPS_PROXY") or os.environ.get("ALL_PROXY")
+        assert https_proxy is not None, f"请设置环境变量{ENV_PREFIX + 'HTTPS_PROXY'}"
+        r = {
+            "http": http_proxy,
+            "https": https_proxy
+        }
+    elif arg == "AVAIL_LLM_MODELS":
+        r = []
+        # 对于AVAIL_LLM_MODELS的环境变量配置，我们允许用户使用;分隔多个模型
+        for item in os.environ[env_arg].split(";"):
+            r.append(item)
+    elif arg == "AUTHENTICATION":
+        r = []
+        # 对于AUTHENTICATION的环境变量配置，我们允许用户使用;分隔多个账号
+        # 格式为：username1:password1;username2:password2
+        for item in os.environ[env_arg].split(";"):
+            r.append(tuple(item.split(":")))
+    elif arg == "API_URL_REDIRECT":
+        r = {}
+        # 对于API_URL_REDIRECT的环境变量配置，我们允许用户使用;分隔转发地址
+        # 格式为：url1:redirect1;url2:redirect2
+        for item in os.environ[env_arg].split(";"):
+            k, v = item.split(":")
+            r[k] = v
+    elif isinstance(default_value, bool):
+        r = bool(os.environ[env_arg])
+    elif isinstance(default_value, int):
+        r = int(os.environ[env_arg])
+    elif isinstance(default_value, float):
+        r = float(os.environ[env_arg])
+    elif isinstance(default_value, str):
+        r = os.environ[env_arg]
+    else:
+        raise RuntimeError(f"[CONFIG] 环境变量{arg}不支持自动转换到{type(default_value)}类型")
+    return r
+
 @lru_cache(maxsize=128)
 def read_single_conf_with_lru_cache(arg):
     from colorful import print亮红, print亮绿, print亮蓝
+    default_r = getattr(importlib.import_module('config'), arg)
     try:
-        r = getattr(importlib.import_module('config_private'), arg)
+        r = read_single_conf_from_env(arg, default_r) # 优先获取环境变量作为配置
     except:
-        r = getattr(importlib.import_module('config'), arg)
+        try:
+            r = getattr(importlib.import_module('config_private'), arg)
+        except:
+            r = default_r
     # 在读取API_KEY时，检查一下是不是忘了改config
     if arg == 'API_KEY':
         print亮蓝(f"[API_KEY] 本项目现已支持OpenAI和API2D的api-key。也支持同时填写多个api-key，如API_KEY=\"openai-key1,openai-key2,api2d-key3\"")

From 8865b232ca68906f042a89fcdffbf3cf24907878 Mon Sep 17 00:00:00 2001
From: ReeInk <hi@ree.ink>
Date: Tue, 2 May 2023 00:12:35 +0800
Subject: [PATCH 5/5] =?UTF-8?q?=E4=BF=AE=E5=A4=8D=EF=BC=9A=E8=AF=BB?=
 =?UTF-8?q?=E5=8F=96=E7=8E=AF=E5=A2=83=E5=8F=98=E9=87=8F=E9=87=8D=E5=AE=9A?=
 =?UTF-8?q?=E5=90=91URL=E6=A0=BC=E5=BC=8F?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 toolbox.py | 12 +++++-------
 1 file changed, 5 insertions(+), 7 deletions(-)

diff --git a/toolbox.py b/toolbox.py
index 81ef1ed..20a9abb 100644
--- a/toolbox.py
+++ b/toolbox.py
@@ -545,13 +545,11 @@ def read_single_conf_from_env(arg, default_value):
         # 格式为：username1:password1;username2:password2
         for item in os.environ[env_arg].split(";"):
             r.append(tuple(item.split(":")))
-    elif arg == "API_URL_REDIRECT":
-        r = {}
-        # 对于API_URL_REDIRECT的环境变量配置，我们允许用户使用;分隔转发地址
-        # 格式为：url1:redirect1;url2:redirect2
-        for item in os.environ[env_arg].split(";"):
-            k, v = item.split(":")
-            r[k] = v
+    elif arg == "API_URL_REDIRECT": 
+        # 对于API_URL_REDIRECT的环境变量，我们允许用户使用json格式配置多个url重定向
+        # 格式为一个json字符串，例如：{"https://api.openai.com/v1/chat/completions": "https://ai.open.com/api/conversation"}
+        import json
+        r = json.loads(os.environ[env_arg])
     elif isinstance(default_value, bool):
         r = bool(os.environ[env_arg])
     elif isinstance(default_value, int):