开源ollama_gui客户端V1.0

csx

最近DeepSeek-R1很火,就装了ollama进行本地部署,但是ollama是采用命令行方式,不习惯,所以做了这个简单的图形客户端.
本地要安装官方ollama软件,网址https://ollama.com/download,再用这个ollama_gui客户端.



打包可执行文件
https://wwjs.lanzoub.com/b00vnde9g
密码:52pj

python源码

[Python] 纯文本查看 复制代码

import tkinter as tk
from tkinter import ttk, messagebox
import requests
import threading
import json
import webbrowser
from tkinter import scrolledtext
import time
import tkinter.font as font
 
# Ollama API 地址
OLLAMA_API_URL = "http://localhost:11434/api"
 
# 定义颜色常量，使用多彩颜色方案
BG_COLOR = "#f5f5f5"  # 背景色
FG_COLOR = "#333333"  # 前景色
BUTTON_BG_COLORS = ["#00C2FE", "#6CE264", "#FF3A48", "#FF8A30","#84BDFE"]  # 按钮背景色数组
BUTTON_HOVER_BG_COLORS = ["#0087B1", "#1E7F18", "#B7000D", "#BE5200","#0059BE"]  # 按钮悬停背景色数组
BUTTON_FG_COLOR = "white"  # 按钮前景色
LABEL_BG_COLOR = "#e0e0e0"  # 标签背景色
CODE_BG_COLOR = "#f8f9fa"  # 代码块背景色
 
 
class OllamaClient:
    def __init__(self, root):
        self.root = root
        self.root.title("Ollama 本地客户端 V1.0")
        self.root.configure(bg=BG_COLOR)
         
         
        # 使窗口在桌面居中
        self.center_window()
 
        # 创建界面组件
        self.create_widgets()
        # 初始化模型列表
        self.update_model_list()
 
        # 下载取消标志位
        self.cancel_download = False
 
        # 关闭程序标志位
        self.cancel_APP = False
 
        # 绑定窗口关闭事件
        self.root.protocol("WM_DELETE_WINDOW", self.on_close)
 
    def center_window(self):
        # 获取屏幕宽度和高度
        screen_width = self.root.winfo_screenwidth()
        screen_height = self.root.winfo_screenheight()
 
        # 窗口宽度和高度
        window_width = 800
        window_height = 600
 
        # 计算窗口左上角的坐标
        x = (screen_width - window_width) // 2
        y = (screen_height - window_height) // 2
 
        # 设置窗口的位置和大小
        self.root.geometry(f"{window_width}x{window_height}+{x}+{y}")
 
    def create_widgets(self):
        # 使用 grid 布局管理器
        self.root.columnconfigure(0, weight=1)
        self.root.rowconfigure(1, weight=1)
 
        # 顶部框架，用于放置模型选择和下载相关组件
        top_frame = tk.Frame(self.root, bg=BG_COLOR)
        top_frame.grid(row=0, column=0, padx=10, pady=10, sticky="ew")
        top_frame.columnconfigure(0, weight=1)
 
        # 模型选择下拉框
        self.model_combobox = ttk.Combobox(top_frame, background=LABEL_BG_COLOR)
        self.model_combobox.grid(row=0, column=0, padx=10, sticky="w")
 
        # 打开模型库按钮
        open_library_button = tk.Button(top_frame, text="打开模型库", command=self.open_model_library,
                                        bg=BUTTON_BG_COLORS[0], fg=BUTTON_FG_COLOR,width=10, height=1)
        open_library_button.grid(row=0, column=1, padx=10)
        open_library_button.bind("<Enter>", lambda e: e.widget.config(bg=BUTTON_HOVER_BG_COLORS[0]))
        open_library_button.bind("<Leave>", lambda e: e.widget.config(bg=BUTTON_BG_COLORS[0]))
 
        # 下载模型按钮
        self.download_button = tk.Button(top_frame, text="下载模型", command=self.download_model,
                                         bg=BUTTON_BG_COLORS[1], fg=BUTTON_FG_COLOR,width=10, height=1)
        self.download_button.grid(row=0, column=2, padx=10)
        self.download_button.bind("<Enter>", lambda e: e.widget.config(bg=BUTTON_HOVER_BG_COLORS[1]))
        self.download_button.bind("<Leave>", lambda e: e.widget.config(bg=BUTTON_BG_COLORS[1]))
 
        # 删除模型按钮
        self.delete_button = tk.Button(top_frame, text="删除模型", command=self.delete_model,
                                       bg=BUTTON_BG_COLORS[2], fg=BUTTON_FG_COLOR,width=10, height=1)
        self.delete_button.grid(row=0, column=3, padx=10)
        self.delete_button.bind("<Enter>", lambda e: e.widget.config(bg=BUTTON_HOVER_BG_COLORS[2]))
        self.delete_button.bind("<Leave>", lambda e: e.widget.config(bg=BUTTON_BG_COLORS[2]))
 
        # 下载进度条及相关信息框架
        self.download_frame = tk.Frame(self.root, bg=BG_COLOR)
        self.download_frame.grid(row=2, column=0, padx=10, pady=5, sticky="ew")
        self.download_frame.grid_forget()
 
        self.download_progress = ttk.Progressbar(self.download_frame, orient="horizontal", length=300, mode="determinate")
        self.download_progress.pack(side=tk.LEFT, padx=10)
 
        self.download_info_label = tk.Label(self.download_frame, text="", bg=LABEL_BG_COLOR, fg=FG_COLOR)
        self.download_info_label.pack(side=tk.LEFT, padx=10)
 
        # 创建字体对象
        title_font = font.Font(family="黑体", size=12)
 
        # 聊天记录显示框，使用 scrolledtext 增加滚动条
        self.chat_text = scrolledtext.ScrolledText(self.root, width=80, height=20, bg=LABEL_BG_COLOR, fg=FG_COLOR)
        self.chat_text.grid(row=1, column=0, padx=10, pady=10, sticky="nsew")
        self.chat_text.bind("<Button-3>", self.show_chat_context_menu)
        self.chat_text.tag_configure("title", font=title_font)
        self.chat_text.tag_configure("code", background=CODE_BG_COLOR)
 
        # 输入框使用 scrolledtext
        self.input_text = scrolledtext.ScrolledText(self.root, width=80, height=5, bg=LABEL_BG_COLOR, fg=FG_COLOR)
        self.input_text.grid(row=3, column=0, padx=10, pady=10, sticky="ew")
        self.input_text.bind("<Button-3>", self.show_input_context_menu)
         
        # 创建一个新的框架来放置发送按钮和清除按钮
        button_frame = tk.Frame(self.root, bg=BG_COLOR)
        button_frame.grid(row=4, column=0, padx=10, pady=10, sticky="e")
 
        # 发送按钮
        self.send_button = tk.Button(button_frame, text="发送", command=self.send_message,
                                     bg=BUTTON_BG_COLORS[3], fg=BUTTON_FG_COLOR,width=10, height=1)
        self.send_button.grid(row=0, column=0, padx=10)
        self.send_button.bind("<Enter>", lambda e: e.widget.config(bg=BUTTON_HOVER_BG_COLORS[3]))
        self.send_button.bind("<Leave>", lambda e: e.widget.config(bg=BUTTON_BG_COLORS[3]))
 
        # 清空聊天记录按钮
        self.clear_button = tk.Button(button_frame, text="清空聊天", command=self.clear_chat,
                                      bg=BUTTON_BG_COLORS[4], fg=BUTTON_FG_COLOR,width=10, height=1)
        self.clear_button.grid(row=0, column=1, padx=10)
        self.clear_button.bind("<Enter>", lambda e: e.widget.config(bg=BUTTON_HOVER_BG_COLORS[4]))
        self.clear_button.bind("<Leave>", lambda e: e.widget.config(bg=BUTTON_BG_COLORS[4]))
 
    def update_model_list(self):
        try:
            response = requests.get(f"{OLLAMA_API_URL}/tags")
            if response.status_code == 200:
                models = [model["name"] for model in response.json().get("models", [])]
                self.model_combobox['values'] = models
                if models:
                    self.model_combobox.set(models[0])
        except Exception as e:
            messagebox.showerror("错误", f"获取模型列表失败: {str(e)}")
 
    def open_model_library(self):
        webbrowser.open("https://ollama.com/library")
 
    def download_model(self):
        # 创建自定义输入对话框并居中显示
        dialog = tk.Toplevel(self.root)
        dialog.title("下载模型")
        dialog.geometry("300x150")
        dialog.configure(bg=BG_COLOR)
        dialog.attributes("-topmost", True)
        # 使对话框居中
        dialog.update_idletasks()
        width = dialog.winfo_width()
        height = dialog.winfo_height()
        x = (self.root.winfo_screenwidth() // 2) - (width // 2)
        y = (self.root.winfo_screenheight() // 2) - (height // 2)
        dialog.geometry(f"{width}x{height}+{x}+{y}")
 
        tk.Label(dialog, text="请输入要下载的模型名称:", bg=BG_COLOR, fg=FG_COLOR).pack(pady=20)
        model_entry = tk.Entry(dialog, bg=LABEL_BG_COLOR, fg=FG_COLOR)
        model_entry.pack(pady=10)
 
     
 
        def start_download():
            model_name = model_entry.get()
            if model_name:
                dialog.destroy()
                self.download_frame.grid()
                self.cancel_download = False  # 重置取消标志位
                threading.Thread(target=self._download_model_thread, args=(model_name,)).start()
 
        download_btn = tk.Button(dialog, text="开始下载", command=start_download,
                                 bg=BUTTON_BG_COLORS[1], fg=BUTTON_FG_COLOR)
        download_btn.pack(pady=10)
        download_btn.bind("<Enter>", lambda e: e.widget.config(bg=BUTTON_HOVER_BG_COLORS[1]))
        download_btn.bind("<Leave>", lambda e: e.widget.config(bg=BUTTON_BG_COLORS[1]))
 
    def on_close(self):
        # 设置下载取消标志
        self.cancel_download = True
        self.cancel_APP = True
        # 关闭主窗口
        self.root.destroy()
    def _download_model_thread(self, model_name):
        try:
            data = {
                "name": model_name
            }
            response = requests.post(f"{OLLAMA_API_URL}/pull", json=data, stream=True)
            if response.status_code == 200:
                total_size = None
                downloaded_size = 0
                start_time = time.time()
                prev_downloaded = 0
                prev_progress = 0
                prev_time = start_time
                self.download_progress['value'] = 0
                for line in response.iter_lines():
                    if self.cancel_download:  # 检查取消标志位
                        messagebox.showinfo("提示", "下载已取消")
                        break
                    if line:
                        try:
                            chunk = line.decode('utf-8')
                            data = json.loads(chunk)
                            if 'total' in data:
                                total_size = data['total']
                            if 'completed' in data:
                                downloaded_size = data['completed']
                                if total_size:
                                    progress = (downloaded_size / total_size) * 100
                                    current_time = time.time()
                                    # 设置进度更新阈值和时间间隔
                                    if progress - prev_progress >= 1 or current_time - prev_time >= 0.5:
                                        elapsed_time = current_time - start_time
                                        if elapsed_time > 0:
                                            speed = (downloaded_size - prev_downloaded) / elapsed_time / 1024
                                            # 根据文件大小动态调整单位
                                            if total_size / (1024 * 1024 * 1024) >= 1:
                                                total_size_gb = total_size / (1024 * 1024 * 1024)
                                                downloaded_size_gb = downloaded_size / (1024 * 1024 * 1024)
                                                info_text = f"速度: {speed:.2f} KB/s, 已下载: {downloaded_size_gb:.2f} GB, 总大小: {total_size_gb:.2f} GB, 进度: {progress:.2f}%"
                                            else:
                                                total_size_kb = total_size / 1024
                                                downloaded_size_kb = downloaded_size / 1024
                                                info_text = f"速度: {speed:.2f} KB/s, 已下载: {downloaded_size_kb:.2f} KB, 总大小: {total_size_kb:.2f} KB, 进度: {progress:.2f}%"
                                            self.download_info_label.config(text=info_text)
                                            prev_downloaded = downloaded_size
                                        self.download_progress['value'] = progress
                                        self.root.update_idletasks()
                                        prev_progress = progress
                                        prev_time = current_time
                        except Exception as e:
                            messagebox.showerror("错误", f"解析下载进度出错: {str(e)}")
                if not self.cancel_download:
                    messagebox.showinfo("成功", "模型下载成功")
                    self.update_model_list()
            else:
                messagebox.showerror("错误", f"模型下载失败: {response.text}")
        except Exception as e:
            messagebox.showerror("错误", f"发生异常: {str(e)}")
        finally:
            self.download_frame.grid_forget()
 
    def delete_model(self):
        model_name = self.model_combobox.get()
        if model_name:
            try:
                data = {
                    "model": model_name
                }
                print(f"尝试删除模型: {model_name}")  # 打印要删除的模型名称
                response = requests.delete(f"{OLLAMA_API_URL}/delete", json=data)
                print(f"请求响应状态码: {response.status_code}")  # 打印响应状态码
                print(f"请求响应内容: {response.text}")  # 打印响应内容
                if response.status_code == 200:
                    messagebox.showinfo("成功", "模型删除成功")
                    self.update_model_list()
                else:
                    messagebox.showerror("错误", f"模型删除失败: {response.text}")
            except Exception as e:
                messagebox.showerror("错误", f"发生异常: {str(e)}")
        else:
            messagebox.showwarning("警告", "请选择要删除的模型")
 
 
# 在 OllamaClient 类中修改 send_message 方法
    def send_message(self):
        message = self.input_text.get(1.0, tk.END).strip()
        if message:
            model_name = self.model_combobox.get()
            if model_name:
            # 使用线程来执行推理任务
                threading.Thread(target=self._send_message_thread, args=(message, model_name)).start()
            else:
                messagebox.showwarning("警告", "请选择要使用的模型")
 
    def _send_message_thread(self, message, model_name):
        try:
            start_time = time.time()  # 记录开始时间
            data = {
                "model": model_name,
                "prompt": message
            }
        # 使用 stream=True 开启流式响应
            response = requests.post(f"{OLLAMA_API_URL}/generate", json=data, stream=True)
            if response.status_code == 200:
                 
                self.chat_text.insert(tk.END, f"问题: {message}\n","title")
                end_time = time.time()  # 记录结束时间
                inference_time = end_time - start_time  # 计算推理时间
                self.chat_text.insert(tk.END, f"\n思考时间: {inference_time:.2f} 秒\n",)
                self.root.update_idletasks()  # 更新界面
                full_response = ""
                chunk_count = 0
                for line in response.iter_lines():
                    if self.cancel_APP:  # 检查取消标志位
                        break
                    if line:
                        try:
                            chunk = line.decode('utf-8')
                            data = json.loads(chunk)
                            part = data.get("response", "")
                            part = part.replace("<think>", "").replace("</think>", "")
                            full_response += part
                            self.chat_text.insert(tk.END, part)
                            chunk_count += 1
                            if chunk_count % 10 == 0:  # 每接收 10 个数据块更新一次界面
                                self.root.update_idletasks()
                        except Exception as e:
                            messagebox.showerror("错误", f"解析响应数据出错: {str(e)}")
                self.root.update_idletasks()  # 确保最后一次更新界面
                self.input_text.delete(1.0, tk.END)
                self.highlight_code()
            else:
                messagebox.showerror("错误", f"请求失败: {response.text}")
        except Exception as e:
            messagebox.showerror("错误", f"发生异常: {str(e)}")
 
    def clear_chat(self):
        self.chat_text.delete(1.0, tk.END)
 
    def show_chat_context_menu(self, event):
        context_menu = tk.Menu(self.root, tearoff=0)
        context_menu.add_command(label="复制", command=lambda: self.copy_chat_text())
 
        # 获取点击位置的标签
        tags = self.chat_text.tag_names(tk.CURRENT)
        if "code" in tags:
            context_menu.add_command(label="复制当前代码块", command=lambda: self.copy_current_code_block())
 
        context_menu.add_command(label="复制所有代码块", command=lambda: self.copy_code_block())
        context_menu.post(event.x_root, event.y_root)
 
    def copy_chat_text(self):
        selected_text = self.chat_text.get(tk.SEL_FIRST, tk.SEL_LAST)
        if selected_text:
            self.root.clipboard_clear()
            self.root.clipboard_append(selected_text)
 
    def copy_code_block(self):
        tag_ranges = self.chat_text.tag_ranges("code")
        if tag_ranges:
            code_text = ""
            for i in range(0, len(tag_ranges), 2):
                start = tag_ranges[i]
                end = tag_ranges[i + 1]
                code_text += self.chat_text.get(start, end)
            self.root.clipboard_clear()
            self.root.clipboard_append(code_text)
 
    def copy_current_code_block(self):
        tag_ranges = self.chat_text.tag_ranges("code")
        current_pos = self.chat_text.index(tk.CURRENT)
        print(f"当前鼠标位置: {current_pos}")  # 打印当前鼠标位置用于调试
        code_text = None  # 初始化 code_text 为 None
        for i in range(0, len(tag_ranges), 2):
            start = str(tag_ranges[i])  # 将 _tkinter.Tcl_Obj 转换为字符串
            end = str(tag_ranges[i + 1])  # 将 _tkinter.Tcl_Obj 转换为字符串
            print(f"代码块范围: {start} - {end}")  # 打印代码块范围用于调试
            if start <= current_pos <= end:
                code_text = self.chat_text.get(start, end)
                break
        if code_text is not None:
            try:
                self.root.clipboard_clear()
                self.root.clipboard_append(code_text)
                print("代码块复制成功")
            except tk.TclError as e:
                print(f"剪贴板操作失败: {e}")
        else:
            print("未找到包含当前位置的代码块")
 
    def show_input_context_menu(self, event):
        context_menu = tk.Menu(self.root, tearoff=0)
        context_menu.add_command(label="复制", command=lambda: self.copy_input_text())
        context_menu.add_command(label="粘贴", command=lambda: self.paste_input_text())
        context_menu.post(event.x_root, event.y_root)
 
    def copy_input_text(self):
        selected_text = self.input_text.get(tk.SEL_FIRST, tk.SEL_LAST)
        if selected_text:
            self.root.clipboard_clear()
            self.root.clipboard_append(selected_text)
 
    def paste_input_text(self):
        clipboard_text = self.root.clipboard_get()
        if clipboard_text:
            self.input_text.insert(tk.INSERT, clipboard_text)
 
    def highlight_code(self):
        content = self.chat_text.get(1.0, tk.END)
        start_index = 0
        while True:
            start = content.find("===", start_index)
            if start == -1:
                break
            end = content.find("===", start + 3)
            if end == -1:
                break
            start_tag = f"1.0+{start}c"
            end_tag = f"1.0+{end + 3}c"
            self.chat_text.tag_add("code", start_tag, end_tag)
            print(f"代码块范围: {start_tag} - {end_tag}")  # 调试信息，打印代码块范围
            start_index = end + 3
 
 
if __name__ == "__main__":
    root = tk.Tk()
    client = OllamaClient(root)
    root.mainloop()