Upload 114 files

ChuanhuChatbot.py

@@ -135,6 +135,7 @@ with gr.Blocks(theme=small_and_beautiful_theme) as demo:
                         label="Chuanhu Chat",
                         elem_id="chuanhu-chatbot",
                         latex_delimiters=latex_delimiters_set,
+                        sanitize_html=False,
                         # height=700,
                         show_label=False,
                         avatar_images=[config.user_avatar, config.bot_avatar],
@@ -244,7 +245,7 @@ with gr.Blocks(theme=small_and_beautiful_theme) as demo:
                             use_websearch_checkbox = gr.Checkbox(label=i18n(
                                 "使用在线搜索"), value=False, elem_classes="switch-checkbox", elem_id="gr-websearch-cb", visible=False)
                             index_files = gr.Files(label=i18n(
-                                "上传"), type="file", elem_id="upload-index-file")
+                                "上传"), type="file", file_types=[".pdf", ".docx", ".pptx", ".epub", ".xlsx", ".txt", "text", "image"], elem_id="upload-index-file")
                             two_column = gr.Checkbox(label=i18n(
                                 "双栏pdf"), value=advance_docs["pdf"].get("two_column", False))
                             summarize_btn = gr.Button(i18n("总结"))
@@ -645,6 +646,8 @@ with gr.Blocks(theme=small_and_beautiful_theme) as demo:
                                  current_model, status_display, chatbot, lora_select_dropdown, user_api_key, keyTxt], show_progress=True, api_name="get_model")
     model_select_dropdown.change(toggle_like_btn_visibility, [model_select_dropdown], [
                                  like_dislike_area], show_progress=False)
+    # model_select_dropdown.change(
+    #     toggle_file_type, [model_select_dropdown], [index_files], show_progress=False)
     lora_select_dropdown.change(get_model, [model_select_dropdown, lora_select_dropdown, user_api_key, temperature_slider,
                                 top_p_slider, systemPromptTxt, user_name, current_model], [current_model, status_display, chatbot], show_progress=True)
 
@@ -798,6 +801,7 @@ if __name__ == "__main__":
     reload_javascript()
     demo.queue(concurrency_count=CONCURRENT_COUNT).launch(
         allowed_paths=["history", "web_assets"],
+        share=share,
         auth=auth_from_conf if authflag else None,
         favicon_path="./web_assets/favicon.ico",
         inbrowser=not dockerflag,  # 禁止在docker下开启inbrowser

config_example.json

@@ -15,6 +15,8 @@
     "spark_api_key": "", // 你的 讯飞星火大模型 API Key，用于讯飞星火大模型对话模型
     "spark_api_secret": "", // 你的 讯飞星火大模型 API Secret，用于讯飞星火大模型对话模型
     "claude_api_secret":"",// 你的 Claude API Secret，用于 Claude 对话模型
+    "ernie_api_key": "",// 你的文心一言在百度云中的API Key，用于文心一言对话模型
+    "ernie_secret_key": "",// 你的文心一言在百度云中的Secret Key，用于文心一言对话模型
 
 
     //== Azure ==

modules/config.py

@@ -135,6 +135,11 @@ os.environ["SPARK_API_SECRET"] = spark_api_secret
 claude_api_secret = config.get("claude_api_secret", "")
 os.environ["CLAUDE_API_SECRET"] = claude_api_secret
 
+ernie_api_key = config.get("ernie_api_key", "")
+os.environ["ERNIE_APIKEY"] = ernie_api_key
+ernie_secret_key = config.get("ernie_secret_key", "")
+os.environ["ERNIE_SECRETKEY"] = ernie_secret_key
+
 load_config_to_environ(["openai_api_type", "azure_openai_api_key", "azure_openai_api_base_url",
                        "azure_openai_api_version", "azure_deployment_name", "azure_embedding_deployment_name", "azure_embedding_model_name"])
 

modules/models/ERNIE.py

@@ -0,0 +1,96 @@
+from ..presets import *
+from ..utils import *
+
+from .base_model import BaseLLMModel
+
+
+class ERNIE_Client(BaseLLMModel):
+    def __init__(self, model_name, api_key, secret_key) -> None:
+        super().__init__(model_name=model_name)
+        self.api_key = api_key
+        self.api_secret = secret_key
+        if None in [self.api_secret, self.api_key]:
+            raise Exception("请在配置文件或者环境变量中设置文心一言的API Key 和 Secret Key")
+
+        if self.model_name == "ERNIE-Bot-turbo":
+            self.ERNIE_url = "https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/eb-instant?access_token="
+        elif self.model_name == "ERNIE-Bot":
+            self.ERNIE_url = "https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions?access_token="
+        elif self.model_name == "ERNIE-Bot-4":
+            self.ERNIE_url = "https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions_pro?access_token="
+
+    def get_access_token(self):
+        """
+        使用 AK，SK 生成鉴权签名（Access Token）
+        :return: access_token，或是None(如果错误)
+        """
+        url = "https://aip.baidubce.com/oauth/2.0/token?client_id=" + self.api_key + "&client_secret=" + self.api_secret + "&grant_type=client_credentials"
+
+        payload = json.dumps("")
+        headers = {
+            'Content-Type': 'application/json',
+            'Accept': 'application/json'
+        }
+
+        response = requests.request("POST", url, headers=headers, data=payload)
+
+        return response.json()["access_token"]
+    def get_answer_stream_iter(self):
+        url = self.ERNIE_url + self.get_access_token()
+        system_prompt = self.system_prompt
+        history = self.history
+        if system_prompt is not None:
+            history = [construct_system(system_prompt), *history]
+
+        # 去除history中 history的role为system的
+        history = [i for i in history if i["role"] != "system"]
+
+        payload = json.dumps({
+            "messages":history,
+            "stream": True
+        })
+        headers = {
+            'Content-Type': 'application/json'
+        }
+
+        response = requests.request("POST", url, headers=headers, data=payload, stream=True)
+
+        if response.status_code == 200:
+            partial_text = ""
+            for line in response.iter_lines():
+                if len(line) == 0:
+                    continue
+                line = json.loads(line[5:])
+                partial_text += line['result']
+                yield partial_text
+        else:
+            yield STANDARD_ERROR_MSG + GENERAL_ERROR_MSG
+
+
+    def get_answer_at_once(self):
+        url = self.ERNIE_url + self.get_access_token()
+        system_prompt = self.system_prompt
+        history = self.history
+        if system_prompt is not None:
+            history = [construct_system(system_prompt), *history]
+
+        # 去除history中 history的role为system的
+        history = [i for i in history if i["role"] != "system"]
+
+        payload = json.dumps({
+            "messages": history,
+            "stream": True
+        })
+        headers = {
+            'Content-Type': 'application/json'
+        }
+
+        response = requests.request("POST", url, headers=headers, data=payload, stream=True)
+
+        if response.status_code == 200:
+
+            return str(response.json()["result"]),len(response.json()["result"])
+        else:
+            return "获取资源错误", 0
+
+

modules/models/OpenAI.py

@@ -26,7 +26,7 @@ class OpenAIClient(BaseLLMModel):
         user_name=""
     ) -> None:
         super().__init__(
-            model_name=MODEL_METADATA[model_name]["model_name"],
+            model_name=model_name,
             temperature=temperature,
             top_p=top_p,
             system_prompt=system_prompt,

modules/models/OpenAIVision.py

@@ -32,7 +32,7 @@ class OpenAIVisionClient(BaseLLMModel):
         user_name=""
     ) -> None:
         super().__init__(
-            model_name=MODEL_METADATA[model_name]["model_name"],
+            model_name=model_name,
             temperature=temperature,
             top_p=top_p,
             system_prompt=system_prompt,

modules/models/base_model.py

@@ -78,7 +78,7 @@ def get_action_description(text):
     action_name = json_dict['action']
     action_input = json_dict['action_input']
     if action_name != "Final Answer":
-        return f'<!-- S O PREFIX --><p class="agent-prefix">{action_name}: {action_input}\n\n</p><!-- E O PREFIX -->'
+        return f'<!-- S O PREFIX --><p class="agent-prefix">{action_name}: {action_input}\n</p><!-- E O PREFIX -->'
     else:
         return ""
 
@@ -148,6 +148,7 @@ class ModelType(Enum):
     Claude = 14
     Qwen = 15
     OpenAIVision = 16
+    ERNIE = 17
 
     @classmethod
     def get_type(cls, model_name: str):
@@ -188,6 +189,8 @@ class ModelType(Enum):
             model_type = ModelType.Claude
         elif "qwen" in model_name_lower:
             model_type = ModelType.Qwen
+        elif "ernie" in model_name_lower:
+            model_type = ModelType.ERNIE
         else:
             model_type = ModelType.LLaMA
         return model_type
@@ -210,7 +213,10 @@ class BaseLLMModel:
     ) -> None:
         self.history = []
         self.all_token_counts = []
-        self.model_name = model_name
+        if model_name in MODEL_METADATA:
+            self.model_name = MODEL_METADATA[model_name]["model_name"]
+        else:
+            self.model_name = model_name
         self.model_type = ModelType.get_type(model_name)
         try:
             self.token_upper_limit = MODEL_METADATA[model_name]["token_limit"]

modules/models/models.py

@@ -128,6 +128,9 @@ def get_model(
         elif model_type == ModelType.Qwen:
             from .Qwen import Qwen_Client
             model = Qwen_Client(model_name, user_name=user_name)
+        elif model_type == ModelType.ERNIE:
+            from .ERNIE import ERNIE_Client
+            model = ERNIE_Client(model_name, api_key=os.getenv("ERNIE_APIKEY"),secret_key=os.getenv("ERNIE_SECRETKEY"))
         elif model_type == ModelType.Unknown:
             raise ValueError(f"未知模型: {model_name}")
         logging.info(msg)

modules/presets.py

@@ -56,6 +56,7 @@ ONLINE_MODELS = [
     "GPT3.5 Turbo 16K",
     "GPT3.5 Turbo 0301",
     "GPT3.5 Turbo 0613",
+    "GPT3.5 Turbo 1106",
     "GPT4",
     "GPT4 32K",
     "GPT4 Turbo",
@@ -74,7 +75,10 @@ ONLINE_MODELS = [
     "讯飞星火大模型V3.0",
     "讯飞星火大模型V2.0",
     "讯飞星火大模型V1.5",
-    "Claude"
+    "Claude",
+    "ERNIE-Bot-turbo",
+    "ERNIE-Bot",
+    "ERNIE-Bot-4",
 ]
 
 LOCAL_MODELS = [
@@ -126,6 +130,10 @@ MODEL_METADATA = {
         "model_name": "gpt-3.5-turbo-0613",
         "token_limit": 4096,
     },
+    "GPT3.5 Turbo 1106": {
+    "model_name": "gpt-3.5-turbo-1106",
+    "token_limit": 16384,
+    },
     "GPT4": {
         "model_name": "gpt-4",
         "token_limit": 8192,
@@ -146,6 +154,18 @@ MODEL_METADATA = {
         "model_name": "Claude",
         "token_limit": 4096,
     },
+    "ERNIE-Bot-turbo": {
+        "model_name": "ERNIE-Bot-turbo",
+        "token_limit": 1024,
+    },
+    "ERNIE-Bot": {
+        "model_name": "ERNIE-Bot",
+        "token_limit": 1024,
+    },
+    "ERNIE-Bot-4": {
+        "model_name": "ERNIE-Bot-4",
+        "token_limit": 1024,
+    },
 }
 
 if os.environ.get('HIDE_LOCAL_MODELS', 'false') == 'true':

modules/utils.py

@@ -211,18 +211,16 @@ def clip_rawtext(chat_message, need_escape=True):
     hr_match = re.search(hr_pattern, chat_message, re.DOTALL)
     message_clipped = chat_message[:hr_match.start()] if hr_match else chat_message
     # second, avoid agent-prefix being escaped
-    agent_prefix_pattern = r'<!-- S O PREFIX --><p class="agent-prefix">(.*?)<\/p><!-- E O PREFIX -->'
-    agent_matches = re.findall(agent_prefix_pattern, message_clipped)
+    agent_prefix_pattern = r'(<!-- S O PREFIX --><p class="agent-prefix">.*?<\/p><!-- E O PREFIX -->)'
+    # agent_matches = re.findall(agent_prefix_pattern, message_clipped)
+    agent_parts = re.split(agent_prefix_pattern, message_clipped, flags=re.DOTALL)
     final_message = ""
-    if agent_matches:
-        agent_parts = re.split(agent_prefix_pattern, message_clipped)
-        for i, part in enumerate(agent_parts):
-            if i % 2 == 0:
-                final_message += escape_markdown(part) if need_escape else part
-            else:
-                final_message += f'<!-- S O PREFIX --><p class="agent-prefix">{part}</p><!-- E O PREFIX -->'
-    else:
-        final_message = escape_markdown(message_clipped) if need_escape else message_clipped
+    for i, part in enumerate(agent_parts):
+        if i % 2 == 0:
+            if part != "" and part != "\n":
+                final_message += f'<pre class="fake-pre">{escape_markdown(part)}</pre>' if need_escape else f'<pre class="fake-pre">{part}</pre>'
+        else:
+            final_message += part
     return final_message
 
 
@@ -233,7 +231,7 @@ def convert_bot_before_marked(chat_message):
     if '<div class="md-message">' in chat_message:
         return chat_message
     else:
-        raw = f'<div class="raw-message hideM"><pre>{clip_rawtext(chat_message)}</pre></div>'
+        raw = f'<div class="raw-message hideM">{clip_rawtext(chat_message)}</div>'
         # really_raw = f'{START_OF_OUTPUT_MARK}<div class="really-raw hideM">{clip_rawtext(chat_message, need_escape=False)}\n</div>{END_OF_OUTPUT_MARK}'
 
         code_block_pattern = re.compile(r"```(.*?)(?:```|$)", re.DOTALL)
@@ -661,6 +659,15 @@ def toggle_like_btn_visibility(selected_model_name):
     else:
         return gr.update(visible=False)
 
+def get_corresponding_file_type_by_model_name(selected_model_name):
+    if selected_model_name in ["xmchat", "GPT4 Vision"]:
+        return ["image"]
+    else:
+        return [".pdf", ".docx", ".pptx", ".epub", ".xlsx", ".txt", "text"]
+
+# def toggle_file_type(selected_model_name):
+#     return gr.Files.update(file_types=get_corresponding_file_type_by_model_name(selected_model_name))
+
 def new_auto_history_filename(username):
     latest_file = get_first_history_name(username)
     if latest_file:

requirements.txt

@@ -1,3 +1,4 @@
+httpx==0.25.1
 gradio==3.43.2
 gradio_client==0.5.0
 pypinyin
@@ -31,3 +32,4 @@ websocket_client
 pydantic==1.10.8
 google-search-results
 anthropic==0.3.11
+Pillow>=10.1.0

web_assets/javascript/message-button.js

@@ -95,7 +95,11 @@ function addChuanhuButton(botElement) {
     function removeMarkdownText(message) {
         var rawDiv = message.querySelector('.raw-message');
         if (rawDiv) {
-            rawDiv.innerHTML = rawDiv.querySelector('pre')?.innerHTML || rawDiv.innerHTML;
+            // 判断pre是否存在fake-pre类，如果不存在，则为20231118之前的历史记录格式，需要转换，增加fake-pre类用于适配
+            if (!rawDiv.querySelector('pre')?.classList.contains('fake-pre')) {
+                rawDiv.innerHTML = rawDiv.innerHTML.replace(/<pre>/g, '<pre class="fake-pre">');
+            }
+            // rawDiv.innerHTML = rawDiv.querySelector('pre')?.innerHTML || rawDiv.innerHTML;
             rawDiv.classList.remove('hideM');
         }
         var mdDiv = message.querySelector('.md-message');

web_assets/javascript/updater.js

@@ -1,6 +1,7 @@
 
 var updateInfoGotten = false;
-var isLatestVersion = localStorage.getItem('isLatestVersion') || false;
+var isLatestVersion = localStorage.getItem('isLatestVersion') === "true" || false;
+var shouldCheckUpdate = false;
 
 function setUpdater() {
     const enableCheckUpdate = gradioApp().querySelector('#enableCheckUpdate_config').innerText;
@@ -10,11 +11,15 @@ function setUpdater() {
         return;
     }
 
+    if (!isLatestVersion) {
+        gradioApp().classList.add('is-outdated');
+    }
     const lastCheckTime = localStorage.getItem('lastCheckTime') || 0;
+    currentTime = new Date().getTime();
     const longTimeNoCheck = currentTime - lastCheckTime > 3 * 24 * 60 * 60 * 1000;
-    if (longTimeNoCheck && !updateInfoGotten && !isLatestVersion || isLatestVersion && !updateInfoGotten) {
-        updateLatestVersion();
-    }
+    shouldCheckUpdate = !updateInfoGotten && (!isLatestVersion && longTimeNoCheck || isLatestVersion);
+    // console.log(`shouldCheckUpdate`, shouldCheckUpdate);
+    if (shouldCheckUpdate) updateLatestVersion();
 }
 
 var statusObserver = new MutationObserver(function (mutationsList) {
@@ -26,6 +31,7 @@ var statusObserver = new MutationObserver(function (mutationsList) {
                     noUpdateHtml();
                     localStorage.setItem('isLatestVersion', 'true');
                     isLatestVersion = true;
+                    gradioApp().classList.remove('is-outdated');
                     enableUpdateBtns();
                 } else if (getUpdateStatus() === "failure") {
                     updatingInfoElement.innerHTML = marked.parse(updateFailure_i18n, {mangle: false, headerIds: false});
@@ -86,11 +92,16 @@ async function updateLatestVersion() {
         if (currentVersion) {
             if (latestVersion <= currentVersion) {
                 noUpdate();
+                localStorage.setItem('isLatestVersion', 'true');
+                isLatestVersion = true;
+                gradioApp().classList.remove('is-outdated');
             } else {
                 latestVersionElement.textContent = latestVersion;
                 console.log(`New version ${latestVersion} found!`);
                 if (!isInIframe) openUpdateToast();
                 gradioApp().classList.add('is-outdated');
+                localStorage.setItem('isLatestVersion', 'false');
+                isLatestVersion = false;
             }
             enableUpdateBtns();
         } else { //如果当前版本号获取失败，使用时间比较
@@ -119,6 +130,8 @@ async function updateLatestVersion() {
                     noUpdate("Local version check failed, it seems to be a local rivision. <br>But your revision is newer than the latest release.");
                     gradioApp().classList.add('is-outdated');
                     enableUpdateBtns()
+                    localStorage.setItem('isLatestVersion', 'false');
+                    isLatestVersion = false;
                 }
             }
         }

web_assets/javascript/webui.js

@@ -39,11 +39,15 @@ function showMask(obj) {
         chatbotArea.querySelector('#chatbot-input-more-area').parentNode.appendChild(mask);
     } else if (obj == "update-toast") {
         mask.classList.add('chuanhu-top-mask');
+        if (document.querySelector('.chuanhu-top-mask')) {
+            for (var i = 0; i < document.querySelectorAll('.chuanhu-top-mask').length; i++) {
+                document.querySelectorAll('.chuanhu-top-mask')[i].remove();
+            }
+        }
         document.body.appendChild(mask);
         // mask.classList.add('transparent-mask');
     }
     
-    
 
     mask.addEventListener('click', () => {
         if (obj == "box") {

web_assets/stylesheet/ChuanhuChat.css

@@ -60,6 +60,7 @@ body.popup-open {
     flex-direction: row;
     display: inline-flex;
     align-items: center;
+    position: absolute;
 }
 #description {
     text-align: center;
@@ -123,12 +124,25 @@ body.popup-open {
 }
 
 /* status-display */
-#status-display {
+#chuanhu-header > #status-display {
     display: flex;
     min-height: 2em;
     align-items: flex-end;
     justify-content: flex-end;
     transition: all 0.6s;
+    max-width: 50%;
+    height: 100%;
+    bottom: 0;
+    position: absolute;
+    
+    @media screen and (max-width: 639px) {
+        right: 16px;
+        right: max(16px, env(safe-area-inset-right));
+    }
+    @media screen and (min-width: 640px) {
+        right: 24px;
+        right: max(24px, env(safe-area-inset-right));
+    }
 }
 #status-display p {
     font-size: .85em;

web_assets/stylesheet/chatbot.css

@@ -42,7 +42,13 @@ hr.append-display {
 .agent-prefix {
     font-size: smaller; 
     opacity: 0.6;
-    padding: 6px 0 4px;
+    padding: 6px 0 12px;
+}
+.raw-message p.agent-prefix + p.agent-prefix {
+    margin-top: -1.2em !important;
+}
+.md-message p.agent-prefix + p.agent-prefix {
+    margin-top: -1.8em !important;
 }
 .agent-prefix::before {
     content: '🐯';
@@ -216,6 +222,14 @@ hr.append-display {
 .message .raw-message p {
     margin:0 !important;
 }
+.message .raw-message pre.fake-pre {
+    background: unset;
+    margin: unset;
+    font-size: unset;
+    /* font-family: unset; */
+    padding: unset;
+    white-space: inherit;
+}
 .message .raw-message {
     display: block;
     padding: 0 !important;