"lambda"和“bind”的初步思考

本文介绍如何利用C++中的Lambda表达式和bind函数简化find_if等算法的使用,实现灵活查找与处理vector容器中的字符串元素。
先来看个问题:在一个string的vector容器中,寻找第一个长度大于4的元素。
inline bool check_size(const string& s)
{
    return s.size() > 4;
}
int main(int argc, char *argv[])
{
    QCoreApplication a(argc, argv);
    vector<string> v{"1", "22", "333", "4444", "55555", "666666", "7777777"};
    cout << *find_if(v.begin(),v.end(),check_size) << endl;
    return a.exec();
}

很简单,只要使用find_if和一个函数就可以。可是假如我们之前不确定界定元素大小的临界值,这可怎么办呢?这是就可以使用“lambda”表达式。它的出现刚好可以弥补类型find_if这些泛型算法只能使用一元谓词的尴尬。

int main(int argc, char *argv[])
{
    QCoreApplication a(argc, argv);
    vector<string> v{"1", "22", "333", "4444", "55555", "666666", "7777777"};
    string::size_type sz = 4;
    auto c = find_if(v.begin(),v.end(),[sz](const string &s)
                           {return s.size() > sz;});
    cout << *c << endl;
    return a.exec();
}
假如我们想要直接把临界值之前的值打印输出,“lambda”可以捕获一个ostream的引用。

int main(int argc, char *argv[])
{
    QCoreApplication a(argc, argv);
    vector<string> v{"1", "22", "333", "4444", "55555", "666666", "7777777"};
    string::size_type sz = 4;
    auto &os = cout;
    auto c = find_if(v.begin(),v.end(),[&os,sz](const string &s)
                           {  os << s << " ";
                              return s.size() > sz;});
    cout << "\nthe first item over sz : " << *c << endl;
    return a.exec();
}
我们发现,当“lambda”捕获的元素多了之后,表达式就显的很臃肿,这时候回觉得还是用函数来的比较方便,那有没有可以解决函数长度参数问题的方法呢?

这里有个方法就是使用名为bind的标准库函数。

inline bool check_size(const string& s,string::size_type sz, ostream &os)
{
    os << s << " ";
    return s.size() > sz;
}
int main(int argc, char *argv[])
{
    QCoreApplication a(argc, argv);
    vector<string> v{"1", "22", "333", "4444", "55555", "666666", "7777777"};
    string::size_type sz = 4;
    auto &os = cout;
    auto c = find_if(v.begin(),v.end(),bind(check_size,_1,sz,os));
    cout << "\nthe first item over sz : " << *c << endl;
    return a.exec();
}
可是编译回报错,原因在于bind不能拷贝一个ostream参数,这时候可以使用标准库中的ref函数

inline bool check_size(const string& s,string::size_type sz, ostream &os)
{
    os << s << " ";
    return s.size() > sz;
}
int main(int argc, char *argv[])
{
    QCoreApplication a(argc, argv);
    vector<string> v{"1", "22", "333", "4444", "55555", "666666", "7777777"};
    string::size_type sz = 4;
    auto &os = cout;
    auto c = find_if(v.begin(),v.end(),bind(check_size,_1,sz,ref(os)));
    cout << "\nthe first item over sz : " << *c << endl;
    return a.exec();
}
这样就可以解决传递长度参数的问题。





#将按钮上的字调成深色,以与浅色背景区分开,发现其它有错也一并改了 import tkinter as tk from tkinter import ttk, filedialog, messagebox, scrolledtext import ollama import os import time import threading import numpy as np import matplotlib.pyplot as plt from matplotlib.backends.backend_tkagg import FigureCanvasTkAgg from matplotlib.figure import Figure import pandas as pd import seaborn as sns import PyPDF2 import docx import markdown from bs4 import BeautifulSoup import openpyxl from PIL import Image import pytesseract import io import psutil from ttkthemes import ThemedTk # 设置中文字体 plt.rcParams['font.sans-serif'] = ['SimHei'] # 用来正常显示中文标签 plt.rcParams['axes.unicode_minus'] = False # 用来正常显示负号 class RAGApplication: def __init__(self, root): self.root = root self.root.title("✨智能RAG应用系统✨") self.root.geometry("1400x900") self.root.configure(bg="#f0f0f0") # 淡灰色背景 # 设置现代主题 self.style = ttk.Style() self.style.theme_use('arc') # 使用现代主题 # 自定义样式 - 淡色调 self.style.configure('TFrame', background='#f0f0f0') self.style.configure('TLabel', background='#f0f0f0', foreground='#333333') self.style.configure('TLabelframe', background='#f0f0f0', foreground='#333333', borderwidth=1) self.style.configure('TLabelframe.Label', background='#f0f0f0', foreground='#4dabf5') # 淡蓝色标题 self.style.configure('TButton', background='#4dabf5', foreground='white', borderwidth=1) # 淡青色按钮 self.style.map('TButton', background=[('active', '#3b99e0')]) self.style.configure('TNotebook', background='#f0f0f0', borderwidth=0) self.style.configure('TNotebook.Tab', background='#e6f0ff', foreground='#333333', padding=[10, 5]) # 淡蓝色标签 self.style.map('TNotebook.Tab', background=[('selected', '#4dabf5')]) # 初始化数据 self.documents = [] self.chunks = [] self.embeddings = [] self.qa_history = [] # 模配置 self.models = { "llm": "gemma3:27b", "embedding": "bge-m3:latest" } # 默认参数 self.params = { "temperature": 0.7, "top_p": 0.9, "max_length": 2048, "num_context_docs": 3, "chunk_size": 500, "chunk_overlap": 100, "chunk_strategy": "固定大小", "separators": "\n\n\n。\n!\n?\n", "embed_batch_size": 1, "enable_stream": True, "show_progress": True, "show_visualization": True, "ocr_enabled": True } # 创建界面 self.create_ui() def create_ui(self): # 主框架 self.main_frame = ttk.Frame(self.root) self.main_frame.pack(fill=tk.BOTH, expand=True, padx=20, pady=20) # 标题 title_frame = ttk.Frame(self.main_frame) title_frame.pack(fill=tk.X, pady=(0, 20)) ttk.Label(title_frame, text="✨ 智能RAG应用系统 ✨", font=('Arial', 24, 'bold'), foreground="#4dabf5").pack(side=tk.LEFT) # 淡青色标题 # 状态指示器 status_frame = ttk.Frame(title_frame) status_frame.pack(side=tk.RIGHT) self.status_label = ttk.Label(status_frame, text="● 就绪", foreground="#28a745") # 绿色状态 self.status_label.pack(side=tk.RIGHT, padx=10) # 参数控制面板 self.create_sidebar() # 主内容区域 self.create_main_content() def create_sidebar(self): # 侧边栏框架 self.sidebar = ttk.LabelFrame(self.main_frame, text="⚙️ 参数控制面板", width=300) self.sidebar.pack(side=tk.LEFT, fill=tk.Y, padx=10, pady=10) # 大模参数 ttk.Label(self.sidebar, text="🔧 大模参数", font=('Arial', 10, 'bold'), foreground="#333333").pack(pady=(15, 5)) self.temperature = tk.DoubleVar(value=self.params["temperature"]) ttk.Label(self.sidebar, text="温度(temperature)").pack(anchor=tk.W, padx=10) temp_frame = ttk.Frame(self.sidebar) temp_frame.pack(fill=tk.X, padx=10, pady=(0, 5)) ttk.Scale(temp_frame, from_=0.0, to=2.0, variable=self.temperature, length=180, command=lambda v: self.update_param("temperature", float(v))).pack(side=tk.LEFT) self.temp_label = ttk.Label(temp_frame, text=f"{self.temperature.get():.1f}", width=5) self.temp_label.pack(side=tk.RIGHT, padx=5) self.top_p = tk.DoubleVar(value=self.params["top_p"]) ttk.Label(self.sidebar, text="Top P").pack(anchor=tk.W, padx=10) top_p_frame = ttk.Frame(self.sidebar) top_p_frame.pack(fill=tk.X, padx=10, pady=(0, 5)) ttk.Scale(top_p_frame, from_=0.0, to=1.0, variable=self.top_p, length=180, command=lambda v: self.update_param("top_p", float(v))).pack(side=tk.LEFT) self.top_p_label = ttk.Label(top_p_frame, text=f"{self.top_p.get():.2f}", width=5) self.top_p_label.pack(side=tk.RIGHT, padx=5) # RAG参数 ttk.Label(self.sidebar, text="\n🔧 RAG参数", font=('Arial', 10, 'bold'), foreground="#333333").pack(pady=(15, 5)) self.chunk_size = tk.IntVar(value=self.params["chunk_size"]) ttk.Label(self.sidebar, text="分块大小(字符)").pack(anchor=tk.W, padx=10) chunk_frame = ttk.Frame(self.sidebar) chunk_frame.pack(fill=tk.X, padx=10, pady=(0, 5)) ttk.Scale(chunk_frame, from_=100, to=2000, variable=self.chunk_size, length=180, command=lambda v: self.update_param("chunk_size", int(v))).pack(side=tk.LEFT) self.chunk_label = ttk.Label(chunk_frame, text=f"{self.chunk_size.get()}", width=5) self.chunk_label.pack(side=tk.RIGHT, padx=5) # OCR开关 self.ocr_var = tk.BooleanVar(value=self.params["ocr_enabled"]) ttk.Checkbutton(self.sidebar, text="启用OCR扫描", variable=self.ocr_var, command=lambda: self.update_param("ocr_enabled", self.ocr_var.get())).pack(pady=(15, 5), padx=10, anchor=tk.W) # 使用说明 ttk.Label(self.sidebar, text="\n📖 使用说明", font=('Arial', 10, 'bold'), foreground="#333333").pack(pady=(15, 5)) instructions = """1. 在"文档上传"页上传您的文档 2. 在"文档处理"页对文档进行分块嵌入 3. 在"问答交互"页提问并获取答案 4. 在"系统监控"页查看系统状态""" ttk.Label(self.sidebar, text=instructions, justify=tk.LEFT, background="#e6f0ff", # 淡蓝色背景 foreground="#333333", padding=10).pack(fill=tk.X, padx=10, pady=5) def create_main_content(self): # 主内容框架 self.content_frame = ttk.Frame(self.main_frame) self.content_frame.pack(side=tk.RIGHT, fill=tk.BOTH, expand=True) # 创建选项卡 self.notebook = ttk.Notebook(self.content_frame) self.notebook.pack(fill=tk.BOTH, expand=True) # 文档上传页 self.create_upload_tab() # 文档处理页 self.create_process_tab() # 问答交互页 self.create_qa_tab() # 系统监控页 self.create_monitor_tab() def create_upload_tab(self): self.upload_tab = ttk.Frame(self.notebook) self.notebook.add(self.upload_tab, text="📤 文档上传") # 标题 title_frame = ttk.Frame(self.upload_tab) title_frame.pack(fill=tk.X, pady=(10, 20)) ttk.Label(title_frame, text="📤 文档上传与管理", font=('Arial', 14, 'bold'), foreground="#4dabf5").pack(side=tk.LEFT) # 淡青色标题 # 上传区域 upload_frame = ttk.Frame(self.upload_tab) upload_frame.pack(fill=tk.X, pady=10) # 上传按钮 upload_btn = ttk.Button(upload_frame, text="📁 上传文档", command=self.upload_files, style='Accent.TButton') upload_btn.pack(side=tk.LEFT, padx=10) # 清除按钮 clear_btn = ttk.Button(upload_frame, text="🗑️ 清除所有", command=self.clear_documents) clear_btn.pack(side=tk.RIGHT, padx=10) # 文档列表 self.doc_list_frame = ttk.LabelFrame(self.upload_tab, text="📋 已上传文档") self.doc_list_frame.pack(fill=tk.BOTH, expand=True, padx=10, pady=10) # 创建带滚动条的树状视图 tree_frame = ttk.Frame(self.doc_list_frame) tree_frame.pack(fill=tk.BOTH, expand=True, padx=5, pady=5) # 创建滚动条 tree_scroll = ttk.Scrollbar(tree_frame) tree_scroll.pack(side=tk.RIGHT, fill=tk.Y) # 创建树状视图 columns = ("name", "size", "time", "type") self.doc_tree = ttk.Treeview(tree_frame, columns=columns, show="headings", yscrollcommand=tree_scroll.set, height=8) # 设置列标题 self.doc_tree.heading("name", text="文件名") self.doc_tree.heading("size", text="大小") self.doc_tree.heading("time", text="上传时间") self.doc_tree.heading("type", text="类") # 设置列宽 self.doc_tree.column("name", width=250) self.doc_tree.column("size", width=80) self.doc_tree.column("time", width=150) self.doc_tree.column("type", width=80) self.doc_tree.pack(side=tk.LEFT, fill=tk.BOTH, expand=True) tree_scroll.config(command=self.doc_tree.yview) # 文档统计 self.doc_stats_frame = ttk.Frame(self.upload_tab) self.doc_stats_frame.pack(fill=tk.X, pady=10, padx=10) stats_style = ttk.Style() stats_style.configure('Stats.TLabel', background='#e6f0ff', foreground='#333333', padding=5) # 淡蓝色背景 ttk.Label(self.doc_stats_frame, text="📊 文档统计:", style='Stats.TLabel').pack(side=tk.LEFT, padx=5) self.doc_count_label = ttk.Label(self.doc_stats_frame, text="0", style='Stats.TLabel') self.doc_count_label.pack(side=tk.LEFT, padx=5) ttk.Label(self.doc_stats_frame, text="总字符数:", style='Stats.TLabel').pack(side=tk.LEFT, padx=5) self.char_count_label = ttk.Label(self.doc_stats_frame, text="0", style='Stats.TLabel') self.char_count_label.pack(side=tk.LEFT, padx=5) ttk.Label(self.doc_stats_frame, text="总页数:", style='Stats.TLabel').pack(side=tk.LEFT, padx=5) self.page_count_label = ttk.Label(self.doc_stats_frame, text="0", style='Stats.TLabel') self.page_count_label.pack(side=tk.LEFT, padx=5) def clear_documents(self): if not self.documents: return if messagebox.askyesno("确认", "确定要清除所有文档吗?"): self.documents = [] self.update_doc_list() # ================== 文件读取函数 ================== def read_pdf(self, filepath): """读取PDF文件内容,支持扫描版OCR""" content = "" pages = 0 try: with open(filepath, 'rb') as f: reader = PyPDF2.PdfReader(f) num_pages = len(reader.pages) pages = num_pages for page_num in range(num_pages): page = reader.pages[page_num] text = page.extract_text() # 如果是扫描版PDF,使用OCR识别 if not text.strip() and self.params["ocr_enabled"]: try: # 获取页面图像 images = page.images if images: for img in images: image_data = img.data image = Image.open(io.BytesIO(image_data)) text += pytesseract.image_to_string(image, lang='chi_sim+eng') except Exception as e: print(f"OCR处理失败: {str(e)}") content += text + "\n" except Exception as e: print(f"读取PDF失败: {str(e)}") return content, pages def read_docx(self, filepath): """读取Word文档内容""" content = "" pages = 0 try: doc = docx.Document(filepath) for para in doc.paragraphs: content += para.text + "\n" pages = len(doc.paragraphs) // 50 + 1 # 估算页数 except Exception as e: print(f"读取Word文档失败: {str(e)}") return content, pages def read_excel(self, filepath): """读取Excel文件内容,优化内存使用""" content = "" pages = 0 try: # 使用openpyxl优化大文件读取 wb = openpyxl.load_workbook(filepath, read_only=True) for sheet_name in wb.sheetnames: content += f"\n工作表: {sheet_name}\n" sheet = wb[sheet_name] for row in sheet.iter_rows(values_only=True): row_content = " | ".join([str(cell) if cell is not None else "" for cell in row]) content += row_content + "\n" pages = len(wb.sheetnames) except Exception as e: print(f"读取Excel文件失败: {str(e)}") return content, pages def read_md(self, filepath): """读取Markdown文件内容""" content = "" pages = 0 try: with open(filepath, 'r', encoding='utf-8') as f: html = markdown.markdown(f.read()) soup = BeautifulSoup(html, 'html.parser') content = soup.get_text() pages = len(content) // 2000 + 1 # 估算页数 except Exception as e: print(f"读取Markdown文件失败: {str(e)}") return content, pages def read_ppt(self, filepath): """读取PPT文件内容(简化版)""" content = "" pages = 0 try: # 实际应用中应使用python-pptx库 # 这里仅作演示 content = f"PPT文件内容提取: {os.path.basename(filepath)}" pages = 10 # 假设有10页 except Exception as e: print(f"读取PPT文件失败: {str(e)}") return content, pages def upload_files(self): filetypes = [ ("文本文件", "*.txt"), ("PDF文件", "*.pdf"), ("Word文件", "*.docx *.doc"), ("Excel文件", "*.xlsx *.xls"), ("Markdown文件", "*.md"), ("PPT文件", "*.pptx *.ppt"), ("所有文件", "*.*") ] filenames = filedialog.askopenfilenames(title="选择文档", filetypes=filetypes) if filenames: self.status_label.config(text="● 正在上传文档...", foreground="#ffc107") # 黄色状态 total_pages = 0 for filename in filenames: try: ext = os.path.splitext(filename)[1].lower() if ext == '.txt': with open(filename, 'r', encoding='utf-8') as f: content = f.read() pages = len(content) // 2000 + 1 elif ext == '.pdf': content, pages = self.read_pdf(filename) elif ext in ('.docx', '.doc'): content, pages = self.read_docx(filename) elif ext in ('.xlsx', '.xls'): content, pages = self.read_excel(filename) elif ext == '.md': content, pages = self.read_md(filename) elif ext in ('.pptx', '.ppt'): content, pages = self.read_ppt(filename) else: messagebox.showwarning("警告", f"不支持的文件类: {ext}") continue # 处理字符编码问题 if not isinstance(content, str): try: content = content.decode('utf-8') except: content = content.decode('latin-1', errors='ignore') self.documents.append({ "name": os.path.basename(filename), "content": content, "size": len(content), "upload_time": time.strftime("%Y-%m-%d %H:%M:%S"), "type": ext.upper().replace(".", ""), "pages": pages }) total_pages += pages # 更新文档列表 self.update_doc_list() except Exception as e: messagebox.showerror("错误", f"无法读取文件 {filename}: {str(e)}") self.status_label.config(text=f"● 上传完成! 共{len(filenames)}个文档", foreground="#28a745") # 绿色状态 self.page_count_label.config(text=str(total_pages)) def update_doc_list(self): # 清空现有列表 for item in self.doc_tree.get_children(): self.doc_tree.delete(item) # 添加新文档 for doc in self.documents: size_kb = doc["size"] / 1024 size_str = f"{size_kb:.1f} KB" if size_kb < 1024 else f"{size_kb / 1024:.1f} MB" self.doc_tree.insert("", tk.END, values=( doc["name"], size_str, doc["upload_time"], doc["type"] )) # 更新统计信息 self.doc_count_label.config(text=str(len(self.documents))) self.char_count_label.config(text=str(sum(d['size'] for d in self.documents))) def create_process_tab(self): self.process_tab = ttk.Frame(self.notebook) self.notebook.add(self.process_tab, text="🔧 文档处理") # 标题 title_frame = ttk.Frame(self.process_tab) title_frame.pack(fill=tk.X, pady=(10, 20)) ttk.Label(title_frame, text="🔧 文档处理与分块", font=('Arial', 14, 'bold'), foreground="#4dabf5").pack(side=tk.LEFT) # 淡青色标题 # 处理按钮 btn_frame = ttk.Frame(self.process_tab) btn_frame.pack(fill=tk.X, pady=10) process_btn = ttk.Button(btn_frame, text="🔄 处理文档", command=self.process_documents, style='Accent.TButton') process_btn.pack(side=tk.LEFT, padx=10) visualize_btn = ttk.Button(btn_frame, text="📊 更新可视化", command=self.show_visualizations) visualize_btn.pack(side=tk.LEFT, padx=10) # 主内容区域 content_frame = ttk.Frame(self.process_tab) content_frame.pack(fill=tk.BOTH, expand=True) # 左侧:可视化区域 self.visual_frame = ttk.LabelFrame(content_frame, text="📈 文档分析") self.visual_frame.pack(side=tk.LEFT, fill=tk.BOTH, expand=True, padx=10, pady=10) # 右侧:分块列表 self.chunk_frame = ttk.LabelFrame(content_frame, text="📋 分块结果") self.chunk_frame.pack(side=tk.RIGHT, fill=tk.BOTH, expand=True, padx=10, pady=10) # 创建带滚动条的树状视图 tree_frame = ttk.Frame(self.chunk_frame) tree_frame.pack(fill=tk.BOTH, expand=True, padx=5, pady=5) # 创建滚动条 tree_scroll = ttk.Scrollbar(tree_frame) tree_scroll.pack(side=tk.RIGHT, fill=tk.Y) # 创建树状视图 columns = ("doc_name", "start", "end", "content") self.chunk_tree = ttk.Treeview(tree_frame, columns=columns, show="headings", yscrollcommand=tree_scroll.set, height=15) # 设置列标题 self.chunk_tree.heading("doc_name", text="来源文档") self.chunk_tree.heading("start", text="起始位置") self.chunk_tree.heading("end", text="结束位置") self.chunk_tree.heading("content", text="内容预览") # 设置列宽 self.chunk_tree.column("doc_name", width=150) self.chunk_tree.column("start", width=80) self.chunk_tree.column("end", width=80) self.chunk_tree.column("content", width=300) self.chunk_tree.pack(side=tk.LEFT, fill=tk.BOTH, expand=True) tree_scroll.config(command=self.chunk_tree.yview) # 初始显示占位图 self.show_placeholder() def show_placeholder(self): """显示可视化占位图""" for widget in self.visual_frame.winfo_children(): widget.destroy() placeholder = ttk.Label(self.visual_frame, text="文档处理后将显示分析图表", font=('Arial', 12), foreground="#7f8c8d") placeholder.pack(expand=True, pady=50) def process_documents(self): if not self.documents: messagebox.showwarning("警告", "请先上传文档") return # 在新线程中处理文档 threading.Thread(target=self._process_documents_thread, daemon=True).start() def _process_documents_thread(self): # 显示进度条 self.progress_window = tk.Toplevel(self.root) self.progress_window.title("处理进度") self.progress_window.geometry("400x150") self.progress_window.resizable(False, False) self.progress_window.transient(self.root) self.progress_window.grab_set() self.progress_window.configure(bg="#f0f0f0") # 淡灰色背景 # 设置窗口居中 x = self.root.winfo_x() + (self.root.winfo_width() - 400) // 2 y = self.root.winfo_y() + (self.root.winfo_height() - 150) // 2 self.progress_window.geometry(f"+{x}+{y}") # 进度窗口内容 ttk.Label(self.progress_window, text="正在处理文档...", font=('Arial', 11)).pack(pady=(20, 10)) progress_frame = ttk.Frame(self.progress_window) progress_frame.pack(fill=tk.X, padx=20, pady=10) self.progress_var = tk.DoubleVar() progress_bar = ttk.Progressbar(progress_frame, variable=self.progress_var, maximum=100, length=360) progress_bar.pack() self.progress_label = ttk.Label(progress_frame, text="0%") self.progress_label.pack(pady=5) self.status_label.config(text="● 正在处理文档...", foreground="#ffc107") # 黄色状态 self.progress_window.update() try: # 分块处理 self.chunks = self.chunk_documents( self.documents, self.params["chunk_strategy"], self.params["chunk_size"], self.params["chunk_overlap"] ) # 生成嵌入 self.embeddings = self.generate_embeddings(self.chunks) # 更新UI self.root.after(0, self.update_chunk_list) self.root.after(0, self.show_visualizations) self.root.after(0, lambda: messagebox.showinfo("完成", "文档处理完成!")) self.status_label.config(text="● 文档处理完成", foreground="#28a745") # 绿色状态 except Exception as e: self.root.after(0, lambda: messagebox.showerror("错误", f"处理文档时出错: {str(e)}")) self.status_label.config(text="● 处理出错", foreground="#dc3545") # 红色状态 finally: self.root.after(0, self.progress_window.destroy) def chunk_documents(self, documents, strategy, size, overlap): chunks = [] total_docs = len(documents) for doc_idx, doc in enumerate(documents): content = doc['content'] if strategy == "固定大小": for i in range(0, len(content), size - overlap): chunk = content[i:i + size] chunks.append({ "doc_name": doc['name'], "content": chunk, "start": i, "end": min(i + size, len(content)) }) # 更新进度 progress = (doc_idx + 1) / total_docs * 100 self.progress_var.set(progress) self.progress_label.config(text=f"{int(progress)}%") self.progress_window.update() return chunks def generate_embeddings(self, chunks): """修复:单批次处理每个分块,避免API参数类错误""" embeddings = [] total_chunks = len(chunks) for idx, chunk in enumerate(chunks): try: # 修复:传递单个字符串而不是列表 response = ollama.embeddings( model=self.models['embedding'], prompt=chunk['content'] # 单个字符串 ) embeddings.append({ "chunk_id": idx, "embedding": response['embedding'], "doc_name": chunk['doc_name'] }) except Exception as e: print(f"生成嵌入时出错: {str(e)}") # 添加空嵌入占位符 embeddings.append({ "chunk_id": idx, "embedding": None, "doc_name": chunk['doc_name'] }) # 更新进度 progress = (idx + 1) / total_chunks * 100 self.progress_var.set(progress) self.progress_label.config(text=f"{int(progress)}%") self.progress_window.update() # 添加延迟避免请求过快 time.sleep(0.1) return embeddings def update_chunk_list(self): # 清空现有列表 for item in self.chunk_tree.get_children(): self.chunk_tree.delete(item) # 添加新分块 for chunk in self.chunks: preview = chunk['content'][:50] + "..." if len(chunk['content']) > 50 else chunk['content'] self.chunk_tree.insert("", tk.END, values=( chunk['doc_name'], chunk['start'], chunk['end'], preview )) def show_visualizations(self): # 清空可视化区域 for widget in self.visual_frame.winfo_children(): widget.destroy() if not self.params["show_visualization"] or not self.chunks: self.show_placeholder() return # 创建图表框架 fig = plt.Figure(figsize=(10, 8), dpi=100) fig.set_facecolor('#f0f0f0') # 淡灰色背景 # 分块大小分布 ax1 = fig.add_subplot(221) ax1.set_facecolor('#e6f0ff') # 淡蓝色背景 chunk_sizes = [len(c['content']) for c in self.chunks] sns.histplot(chunk_sizes, bins=20, ax=ax1, color='#4dabf5') # 淡青色 ax1.set_title("分块大小分布", color='#333333') ax1.set_xlabel("字符数", color='#333333') ax1.set_ylabel("数量", color='#333333') ax1.tick_params(axis='x', colors='#333333') ax1.tick_params(axis='y', colors='#333333') ax1.spines['bottom'].set_color('#333333') ax1.spines['left'].set_color('#333333') # 文档分块数量 ax2 = fig.add_subplot(222) ax2.set_facecolor('#e6f0ff') # 淡蓝色背景 doc_chunk_counts = {} for chunk in self.chunks: doc_chunk_counts[chunk['doc_name']] = doc_chunk_counts.get(chunk['doc_name'], 0) + 1 # 只显示前10个文档 doc_names = list(doc_chunk_counts.keys()) counts = list(doc_chunk_counts.values()) if len(doc_names) > 10: # 按分块数量排序,取前10 sorted_indices = np.argsort(counts)[::-1][:10] doc_names = [doc_names[i] for i in sorted_indices] counts = [counts[i] for i in sorted_indices] sns.barplot(x=counts, y=doc_names, ax=ax2, palette='Blues', orient='h') # 蓝色调 ax2.set_title("各文档分块数量", color='#333333') ax2.set_xlabel("分块数", color='#333333') ax2.set_ylabel("") ax2.tick_params(axis='x', colors='#333333') ax2.tick_params(axis='y', colors='#333333') ax2.spines['bottom'].set_color('#333333') ax2.spines['left'].set_color('#333333') # 内容词云(模拟) ax3 = fig.add_subplot(223) ax3.set_facecolor('#e6f0ff') # 淡蓝色背景 ax3.set_title("内容关键词分布", color='#333333') ax3.text(0.5, 0.5, "关键词可视化区域", horizontalalignment='center', verticalalignment='center', color='#333333', fontsize=12) ax3.axis('off') # 处理进度 ax4 = fig.add_subplot(224) ax4.set_facecolor('#e6f0ff') # 淡蓝色背景 ax4.set_title("处理进度", color='#333333') # 模拟数据 stages = ['上传', '分块', '嵌入', '完成'] progress = [100, 100, 100, 100] # 假设都已完成 ax4.barh(stages, progress, color=['#4dabf5', '#20c997', '#9b59b6', '#ffc107']) # 淡色系 ax4.set_xlim(0, 100) ax4.set_xlabel("完成百分比", color='#333333') ax4.tick_params(axis='x', colors='#333333') ax4.tick_params(axis='y', colors='#333333') ax4.spines['bottom'].set_color('#333333') ax4.spines['left'].set_color('#333333') # 调整布局 fig.tight_layout(rect=[0, 0, 1, 0.95], pad=3.0) # 添加总标题 fig.suptitle("文档分析概览", fontsize=16, color='#333333') # 在Tkinter中显示图表 canvas = FigureCanvasTkAgg(fig, master=self.visual_frame) canvas.draw() canvas.get_tk_widget().pack(fill=tk.BOTH, expand=True) def create_qa_tab(self): self.qa_tab = ttk.Frame(self.notebook) self.notebook.add(self.qa_tab, text="💬 问答交互") # 标题 title_frame = ttk.Frame(self.qa_tab) title_frame.pack(fill=tk.X, pady=(10, 20)) ttk.Label(title_frame, text="💬 问答交互", font=('Arial', 14, 'bold'), foreground="#4dabf5").pack(side=tk.LEFT) # 淡青色标题 # 主内容区域 main_frame = ttk.Frame(self.qa_tab) main_frame.pack(fill=tk.BOTH, expand=True, padx=10, pady=5) # 左侧:问答区域 left_frame = ttk.Frame(main_frame) left_frame.pack(side=tk.LEFT, fill=tk.BOTH, expand=True, padx=5, pady=5) # 问题输入 self.question_frame = ttk.LabelFrame(left_frame, text="❓ 输入问题") self.question_frame.pack(fill=tk.X, padx=5, pady=5) self.question_text = scrolledtext.ScrolledText(self.question_frame, height=8, wrap=tk.WORD, font=('Arial', 11)) self.question_text.pack(fill=tk.BOTH, expand=True, padx=5, pady=5) self.question_text.focus_set() # 提交按钮 btn_frame = ttk.Frame(left_frame) btn_frame.pack(fill=tk.X, pady=10) submit_btn = ttk.Button(btn_frame, text="🚀 提交问题", command=self.submit_question, style='Accent.TButton') submit_btn.pack(side=tk.LEFT, padx=5) clear_btn = ttk.Button(btn_frame, text="🗑️ 清除问题", command=self.clear_question) clear_btn.pack(side=tk.LEFT, padx=5) # 回答显示 self.answer_frame = ttk.LabelFrame(left_frame, text="💡 回答") self.answer_frame.pack(fill=tk.BOTH, expand=True, padx=5, pady=5) self.answer_text = scrolledtext.ScrolledText(self.answer_frame, state=tk.DISABLED, wrap=tk.WORD, font=('Arial', 11)) self.answer_text.pack(fill=tk.BOTH, expand=True, padx=5, pady=5) # 右侧:问答历史 right_frame = ttk.Frame(main_frame, width=400) # 在创建时设置宽度 right_frame.pack(side=tk.RIGHT, fill=tk.BOTH, expand=False, padx=5, pady=5) self.history_frame = ttk.LabelFrame(right_frame, text="🕒 问答历史") self.history_frame.pack(fill=tk.BOTH, expand=True, padx=5, pady=5) # 创建带滚动条的树状视图 tree_frame = ttk.Frame(self.history_frame) tree_frame.pack(fill=tk.BOTH, expand=True, padx=5, pady=5) # 创建滚动条 tree_scroll = ttk.Scrollbar(tree_frame) tree_scroll.pack(side=tk.RIGHT, fill=tk.Y) # 创建树状视图 columns = ("question", "time") self.history_tree = ttk.Treeview(tree_frame, columns=columns, show="headings", yscrollcommand=tree_scroll.set, height=20) # 设置列标题 self.history_tree.heading("question", text="问题") self.history_tree.heading("time", text="时间") # 设置列宽 self.history_tree.column("question", width=250) self.history_tree.column("time", width=120) self.history_tree.pack(side=tk.LEFT, fill=tk.BOTH, expand=True) tree_scroll.config(command=self.history_tree.yview) # 历史操作按钮 history_btn_frame = ttk.Frame(right_frame) history_btn_frame.pack(fill=tk.X, pady=10) view_btn = ttk.Button(history_btn_frame, text="👁️ 查看详情", command=lambda: self.show_history_detail(None)) view_btn.pack(side=tk.LEFT, padx=5, fill=tk.X, expand=True) clear_history_btn = ttk.Button(history_btn_frame, text="🗑️ 清除历史", command=self.clear_history) clear_history_btn.pack(side=tk.LEFT, padx=5, fill=tk.X, expand=True) # 绑定双击事件查看历史详情 self.history_tree.bind("<Double-1>", self.show_history_detail) def clear_question(self): self.question_text.delete("1.0", tk.END) def clear_history(self): if not self.qa_history: return if messagebox.askyesno("确认", "确定要清除所有问答历史吗?"): self.qa_history = [] self.update_history_list() def submit_question(self): question = self.question_text.get("1.0", tk.END).strip() if not question: messagebox.showwarning("警告", "问题不能为空") return # 在新线程中处理问题 threading.Thread(target=self._submit_question_thread, args=(question,), daemon=True).start() def _submit_question_thread(self, question): try: # 显示进度窗口 self.progress_window = tk.Toplevel(self.root) self.progress_window.title("处理中...") self.progress_window.geometry("400x150") self.progress_window.resizable(False, False) self.progress_window.transient(self.root) self.progress_window.grab_set() self.progress_window.configure(bg="#f0f0f0") # 淡灰色背景 # 设置窗口居中 x = self.root.winfo_x() + (self.root.winfo_width() - 400) // 2 y = self.root.winfo_y() + (self.root.winfo_height() - 150) // 2 self.progress_window.geometry(f"+{x}+{y}") # 进度窗口内容 ttk.Label(self.progress_window, text="正在思考中...", font=('Arial', 11)).pack(pady=(20, 10)) progress_frame = ttk.FFrame(self.progress_window) progress_frame.pack(fill=tk.X, padx=20, pady=10) self.progress_var = tk.DoubleVar() progress_bar = ttk.Progressbar(progress_frame, variable=self.progress_var, maximum=100, length=360) progress_bar.pack() self.progress_label = ttk.Label(progress_frame, text="0%") self.progress_label.pack(pady=5) self.status_label.config(text="● 正在处理问题...", foreground="#ffc107") # 黄色状态 self.progress_window.update() # 检索相关文档块 relevant_chunks = self.retrieve_relevant_chunks(question, self.params["num_context_docs"]) # 构建上下文 context = "\n\n".join([ f"文档: {c['doc_name']}\n内容: {c['content']}\n相关性: {c['similarity']:.4f}" for c in relevant_chunks ]) # 调用大模生成回答 prompt = f"""基于以下上下文,回答问题。如果答案不在上下文中,请回答"我不知道"。 上下文: {context} 问题: {question} 回答:""" # 更新进度 self.progress_var.set(50) self.progress_label.config(text="50%") self.progress_window.update() # 流式输出或一次性输出 self.root.after(0, self.answer_text.config, {'state': tk.NORMAL}) self.root.after(0, self.answer_text.delete, "1.0", tk.END) if self.params["enable_stream"]: full_response = "" for chunk in ollama.generate( model=self.models['llm'], prompt=prompt, stream=True, options={ 'temperature': self.params["temperature"], 'top_p': self.params["top_p"], 'num_ctx': self.params["max_length"] } ): full_response += chunk['response'] self.root.after(0, self.answer_text.insert, tk.END, chunk['response']) self.root.after(0, self.answer_text.see, tk.END) self.root.after(0, self.answer_text.update) # 更新进度 if len(full_response) > 0: progress = min(50 + len(full_response) / 200, 99) self.progress_var.set(progress) self.progress_label.config(text=f"{int(progress)}%") self.progress_window.update() else: response = ollama.generate( model=self.models['llm'], prompt=prompt, options={ 'temperature': self.params["temperature"], 'top_p': self.params["top_p"], 'num_ctx': self.params["max_length"] } ) full_response = response['response'] self.root.after(0, self.answer_text.insert, tk.END, full_response) # 记录问答历史 self.qa_history.append({ "question": question, "answer": full_response, "context": context, "time": time.strftime("%Y-%m-%d %H:%M:%S") }) # 更新历史列表 self.root.after(0, self.update_history_list) # 完成 self.progress_var.set(100) self.progress_label.config(text="100%") self.status_label.config(text="● 问题处理完成", foreground="#28a745") # 绿色状态 self.root.after(1000, self.progress_window.destroy) except Exception as e: self.root.after(0, lambda: messagebox.showerror("错误", f"处理问题时出错: {str(e)}")) self.root.after(0, self.progress_window.destroy) self.status_label.config(text="● 处理出错", foreground="#dc3545") # 红色状态 def retrieve_relevant_chunks(self, query, k): """修复:处理嵌入为None的情况""" # 生成查询的嵌入 query_embedding = ollama.embeddings( model=self.models['embedding'], prompt=query )['embedding'] # 注意:返回的是字典中的'embedding'字段 # 计算相似度 similarities = [] for emb in self.embeddings: # 跳过无效的嵌入 if emb['embedding'] is None: continue # 计算余弦相似度 similarity = np.dot(query_embedding, emb['embedding']) similarities.append({ 'chunk_id': emb['chunk_id'], 'similarity': similarity, 'doc_name': emb['doc_name'] }) # 按相似度排序并返回前k个 top_chunks = sorted(similarities, key=lambda x: x['similarity'], reverse=True)[:k] return [{ **self.chunks[c['chunk_id']], 'similarity': c['similarity'] } for c in top_chunks] def update_history_list(self): # 清空现有列表 for item in self.history_tree.get_children(): self.history_tree.delete(item) # 添加新历史记录 for i, qa in enumerate(reversed(self.qa_history)): # 截断长问题 question = qa["question"] if len(question) > 50: question = question[:47] + "..." self.history_tree.insert("", tk.END, values=(question, qa["time"])) def show_history_detail(self, event): selected_item = self.history_tree.selection() if not selected_item: return item = self.history_tree.item(selected_item) question = item['values'][0] # 查找对应的问答记录 for qa in reversed(self.qa_history): if qa["question"].startswith(question) or question.startswith(qa["question"][:50]): # 显示详情窗口 detail_window = tk.Toplevel(self.root) detail_window.title("问答详情") detail_window.geometry("900x700") detail_window.configure(bg='#f0f0f0') # 淡灰色背景 # 设置窗口居中 x = self.root.winfo_x() + (self.root.winfo_width() - 900) // 2 y = self.root.winfo_y() + (self.root.winfo_height() - 700) // 2 detail_window.geometry(f"+{x}+{y}") # 问题 ttk.Label(detail_window, text="问题:", font=('Arial', 12, 'bold'), foreground="#4dabf5").pack(pady=(15, 5), padx=20, anchor=tk.W) question_frame = ttk.Frame(detail_window) question_frame.pack(fill=tk.X, padx=20, pady=(0, 10)) question_text = scrolledtext.ScrolledText(question_frame, wrap=tk.WORD, height=3, font=('Arial', 11)) question_text.insert(tk.INSERT, qa["question"]) question_text.config(state=tk.DISABLED) question_text.pack(fill=tk.X) # 回答 ttk.Label(detail_window, text="回答:", font=('Arial', 12, 'bold'), foreground="#4dabf5").pack(pady=(15, 5), padx=20, anchor=tk.W) answer_frame = ttk.Frame(detail_window) answer_frame.pack(fill=tk.BOTH, expand=True, padx=20, pady=(0, 10)) answer_text = scrolledtext.ScrolledText(answer_frame, wrap=tk.WORD, font=('Arial', 11)) answer_text.insert(tk.INSERT, qa["answer"]) answer_text.config(state=tk.DISABLED) answer_text.pack(fill=tk.BOTH, expand=True) # 上下文 ttk.Label(detail_window, text="上下文:", font=('Arial', 12, 'bold'), foreground="#4dabf5").pack(pady=(15, 5), padx=20, anchor=tk.W) context_frame = ttk.Frame(detail_window) context_frame.pack(fill=tk.BOTH, expand=True, padx=20, pady=(0, 20)) context_text = scrolledtext.ScrolledText(context_frame, wrap=tk.WORD, font=('Arial', 10)) context_text.insert(tk.INSERT, qa["context"]) context_text.config(state=tk.DISABLED) context_text.pack(fill=tk.BOTH, expand=True) break def create_monitor_tab(self): self.monitor_tab = ttk.Frame(self.notebook) self.notebook.add(self.monitor_tab, text="📊 系统监控") # 标题 title_frame = ttk.Frame(self.monitor_tab) title_frame.pack(fill=tk.X, pady=(10, 20)) ttk.Label(title_frame, text="📊 系统监控", font=('Arial', 14, 'bold'), foreground="#4dabf5").pack(side=tk.LEFT) # 主内容区域 main_frame = ttk.Frame(self.monitor_tab) main_frame.pack(fill=tk.BOTH, expand=True, padx=10, pady=5) # 修复:main极值rame → main_frame # 左侧:资源监控 left_frame = ttk.Frame(main_frame) left_frame.pack(side=tk.LEFT, fill=tk.BOTH, expand=True, padx=5, pady=5) # 资源使用 self.resource_frame = ttk.LabelFrame(left_frame, text="📈 资源使用") self.resource_frame.pack(fill=tk.BOTH, expand=True, padx=5, pady=5) # CPU使用 cpu_frame = ttk.Frame(self.resource_frame) cpu_frame.pack(fill=tk.X, padx=10, pady=10) ttk.Label(cpu_frame, text="CPU使用率:").pack(side=tk.LEFT) self.cpu_value = ttk.Label(cpu_frame, text="0%", width=5) self.cpu_value.pack(side=tk.RIGHT, padx=10) self.cpu_usage = ttk.Progressbar(self.resource_frame, length=400, mode='determinate') self.cpu_usage.pack(fill=tk.X, padx=10, pady=(0, 10)) # 内存使用 mem_frame = ttk.Frame(self.resource_frame) mem_frame.pack(fill=tk.X, padx=10, pady=10) ttk.Label(mem_frame, text="内存使用率:").pack(side=tk.LEFT) self.mem_value = ttk.Label(mem_frame, text="0%", width=5) self.mem_value.pack(side=tk.RIGHT, padx=10) self.mem_usage = ttk.Progressbar(self.resource_frame, length=400, mode='determinate') self.mem_usage.pack(fill=tk.X, padx=10, pady=(0, 10)) # 磁盘使用 disk_frame = ttk.Frame(self.resource_frame) disk_frame.pack(fill=tk.X, padx=10, pady=10) ttk.Label(disk_frame, text="磁盘使用率:").pack(side=tk.LEFT) self.disk_value = ttk.Label(disk_frame, text="0%", width=5) self.disk_value.pack(side=tk.RIGHT, padx=10) self.disk_usage = ttk.Progressbar(self.resource_frame, length=400, mode='determinate') self.disk_usage.pack(fill=tk.X, padx=10, pady=(0, 10)) # 右侧:模状态 right_frame = ttk.Frame(main_frame) right_frame.pack(side=tk.RIGHT, fill=tk.BOTH, expand=True, padx=5, pady=5) # 模状态 self.model_frame = ttk.LabelFrame(right_frame, text="🤖 模状态") self.model_frame.pack(fill=tk.BOTH, expand=True, padx=5, pady=5) btn_frame = ttk.Frame(self.model_frame) btn_frame.pack(fill=tk.X, padx=10, pady=10) ttk.Button(btn_frame, text="🔄 检查模状态", command=self.check_model_status).pack() self.model_status_text = scrolledtext.ScrolledText(self.model_frame, height=15, state=tk.DISABLED, font=('Consolas', 10)) self.model_status_text.pack(fill=tk.BOTH, expand=True, padx=10, pady=(0, 10)) # 性能统计 self.perf_frame = ttk.LabelFrame(left_frame, text="⚡ 性能统计") self.perf_frame.pack(fill=tk.BOTH, expand=True, padx=5, pady=5) # 创建图表 fig = Figure(figsize=(8, 4), dpi=100) fig.set_facecolor('#f0f0f0') # 淡灰色背景 self.ax = fig.add_subplot(111) self.ax.set_facecolor('#e6f0ff') # 淡蓝色背景 self.ax.set_title("CPU使用率历史", color='#333333') self.ax.set_xlabel("时间", color='#333333') self.ax.set_ylabel("使用率(%)", color='#333333') self.ax.tick_params(axis='x', colors='#333333') self.ax.tick_params(axis='y', colors='#333333') self.ax.spines['bottom'].set_color('#333333') self.ax.spines['left'].set_color('#333333') self.cpu_history = [] self.line, = self.ax.plot([], [], color='#4dabf5', marker='o', markersize=4) # 淡青色线条 self.ax.set_ylim(0, 100) canvas = FigureCanvasTkAgg(fig, master=self.perf_frame) canvas.draw() canvas.get_tk_widget().pack(fill=tk.BOTH, expand=True, padx=10, pady=10) # 开始更新资源使用情况 self.update_resource_usage() def update_resource_usage(self): # 获取真实资源数据 cpu_percent = psutil.cpu_percent() mem_percent = psutil.virtual_memory().percent disk_percent = psutil.disk_usage('/').percent # 更新进度条 self.cpu_usage['value'] = cpu_percent self.mem_usage['value'] = mem_percent self.disk_usage['value'] = disk_percent # 更新数值标签 self.cpu_value.config(text=f"{cpu_percent}%") self.mem_value.config(text=f"{mem_percent}%") self.disk_value.config(text=f"{disk_percent}%") # 更新CPU历史图表 self.cpu_history.append(cpu_percent) if len(self.cpu_history) > 20: self.cpu_history.pop(0) self.line.set_data(range(len(self.cpu_history)), self.cpu_history) self.ax.set_xlim(0, max(10, len(self.cpu_history))) self.ax.figure.canvas.draw() # 5秒后再次更新 self.root.after(5000, self.update_resource_usage) def check_model_status(self): try: self.model_status_text.config(state=tk.NORMAL) self.model_status_text.delete("1.0", tk.END) # 添加加载动画 self.model_status_text.insert(tk.INSERT, "正在检查模状态...") self.model_status_text.update() # 模拟检查过程 time.sleep(1) # 清空并插入真实信息 self.model_status_text.delete("1.0", tk.END) llm_info = ollama.show(self.models['llm']) embed_info = ollama.show(self.models['embedding']) status_text = f"""✅ 大模信息: 名称: {self.models['llm']} 参数大小: {llm_info.get('size', '未知')} 最后使用时间: {llm_info.get('modified_at', '未知')} 支持功能: {llm_info.get('capabilities', '未知')} ✅ 嵌入模信息: 名称: {self.models['embedding']} 参数大小: {embed_info.get('size', '未知')} 最后使用时间: {embed_info.get('modified_at', '未知')} 支持功能: {embed_info.get('capabilities', '未知')} ⏱️ 最后检查时间: {time.strftime("%Y-%m-%d %H:%M:%S")} """ self.model_status_text.insert(tk.INSERT, status_text) self.model_status_text.config(state=tk.DISABLED) self.status_label.config(text="● 模状态检查完成", foreground="#28a745") # 绿色状态 except Exception as e: self.model_status_text.config(state=tk.NORMAL) self.model_status_text.delete("1.0", tk.END) self.model_status_text.insert(tk.INSERT, f"❌ 检查模状态时出错: {str(e)}") self.model_status_text.config(state=tk.DISABLED) self.status_label.config(text="● 模检查出错", foreground="#dc3545") # 红色状态 def update_param(self, param, value): self.params[param] = value # 更新标签显示 if param == "temperature": self.temp_label.config(text=f"{value:.1f}") elif param == "top_p": self.top_p_label.config(text=f"{value:.2f}") elif param == "chunk_size": self.chunk_label.config(text=f"{value}") # 运行应用程序 if __name__ == "__main__": root = ThemedTk(theme="arc") # 使用现代主题 app = RAGApplication(root) root.mainloop()
最新发布
08-07
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值