一时心血来潮想做一个网页图片爬虫，从网上找了好多资料，不过大多数的都看不懂，知识面太浅，历经千辛万苦，终于实现简单的完成了！
还是要感谢老前辈，体会到自学的艰辛！路还长继续努力！！！
1 using System;
2 using System.Collections.Generic;
3 using System.ComponentModel;
4 using System.Data;
5 using System.Drawing;
6 using System.IO;
7 using System.Linq;
8 using System.Net;
9 using System.Text;
10 using System.Threading;
11 using System.Threading.Tasks;
12 using System.Windows.Forms;
13 using Newtonsoft.Json;
14 using Newtonsoft.Json.Linq;
15
16 namespace 网页图片爬虫_窗体版
17 {
18 public partial class Form1 : ZHSkin.ZHForm
19 {
20 /// <summary>
21 /// 关键词
22 /// </summary>
23 private string SavePath = string.Empty;
24 private Thread thread = null;
25 public Form1()
26 {
27 InitializeComponent();
28 }
29 /// <summary>
30 /// 获取保存路径
31 /// </summary>
32 /// <param name="sender"></param>
33 /// <param name="e"></param>
34 private void btn_GetPath_Click(object sender, EventArgs e)
35 {
36 FolderBrowserDialog folderBrowser = new FolderBrowserDialog();
37 if (folderBrowser.ShowDialog() == DialogResult.OK)
38 {
39 txt_SavePath.Text = folderBrowser.SelectedPath;
40 }
41 }
42 /// <summary>
43 /// 开始获取
44 /// </summary>
45 /// <param name="sender"></param>
46 /// <param name="e"></param>
47 private void btn_Start_Click(object sender, EventArgs e)
48 {
49 string KeyWord = txt_KeyWords.Text.Trim();
50 int pageCount = (int)NUD_PageCount.Value;
51 SavePath = txt_SavePath.Text;
52 if (string.IsNullOrEmpty(KeyWord))
53 {
54 txt_ShowPath.AppendText("请输入要搜索的关键词！" + Environment.NewLine);
55 return;
56 }
57 if (string.IsNullOrEmpty(SavePath))
58 {
59 txt_ShowPath.AppendText("请选择要保存的路径！" + Environment.NewLine);
60 return;
61 }
62 if (!Directory.Exists(SavePath))//如果不存在就创建file文件夹
63 {
64 txt_ShowPath.AppendText("输入路径不正确，请核对！" + Environment.NewLine);
65 return;
66 }
67 if (!SavePath.EndsWith("\\"))
68 {
69 SavePath = SavePath + "\\";
70 }
71 btn_Stop.Enabled = true;//启用停止按钮
72 btn_Start.Enabled = false;//禁用开始按钮
73 txt_ShowPath.Clear();//清空日志
74 txt_ShowPath.AppendText("正在启动下载！" + Environment.NewLine);
75 msg.Text = "已启动下载...";
76 //启动进度条
77 ProgressBar.Enabled = true;
78 //使用多线程下载
79 thread = new Thread(() =>
80 {
81 ProcessDownload(KeyWord);
82 });
83 thread.Start();//启动下载
84 }
85 /// <summary>
86 /// 处理下载
87 /// </summary>
88 /// <param name="KeyWord"></param>
89 public void ProcessDownload(string KeyWord)
90 {
91 try
92 {
93 int pageCount = (int)NUD_PageCount.Value;//下载页数
94 //循环获取路径
95 for (int i = 0; i < pageCount; i++)
96 {
97 msg.Text = "正在下载第" + (i + 1) + "页，共" + pageCount + "页";
98 string URL = "https://image.baidu.com/search/acjson?tn=resultjson_com&ipn=rj&ct=201326592&is=&fp=result&queryWord=" + Uri.EscapeDataString(KeyWord) + "&cl=2&lm=-1&ie=utf-8&oe=utf-8&adpicid=&st=-1&z=&ic=0&word=" + Uri.EscapeDataString(KeyWord) + "&s=&se=&tab=&width=&height=&face=0&istype=2&qc=&nc=1&fr=&cg=wallpaper&pn=" + (i + 1) * 60 + "&rn=60&gsm=3c&1525422519486=";
99 HttpWebRequest request = (HttpWebRequest)WebRequest.Create(URL);
100 using (HttpWebResponse response = (HttpWebResponse)request.GetResponse())
101 {
102 if (response.StatusCode == HttpStatusCode.OK)
103 {
104 using (Stream stream = response.GetResponseStream())
105 {
106 try
107 {
108 DownloadPage(stream, i);//下载页面
109 }
110 catch (Exception ex)
111 {
112 txt_ShowPath.BeginInvoke(new Action(() =>
113 {
114 txt_ShowPath.AppendText(ex.Message + Environment.NewLine);
115 }));
116 }
117 }
118 }
119 else
120 {
121 MessageBox.Show("获取第" + i + "页失败：" + response.StatusCode);
122 }
123 }
124 }
125 msg.Text = "下载完成！下载" + pageCount + "页，共" + pageCount * 60 + "张图片";
126 ProgressBar.Value = 0;//进度条归零
127 btn_Stop.Enabled = false;//禁用停止按钮
128 btn_Start.Enabled = true;//启用开始按钮
129 btn_Stop.Text = "停止";
130 }
131 catch (Exception)
132 {
133 txt_ShowPath.BeginInvoke(new Action(() =>
134 {
135 txt_ShowPath.AppendText("网络连接失败！" + Environment.NewLine);
136 }));
137 }
138 }
139 /// <summary>
140 /// 下载页面
141 /// </summary>
142 /// <param name="stream"></param>
143 private void DownloadPage(Stream stream, int index)
144 {
145 using (StreamReader reader = new StreamReader(stream))
146 {
147 string json = reader.ReadToEnd();
148 //txt_ShowPath.AppendText(json);
149 JObject objRoot = (JObject)JsonConvert.DeserializeObject(json);
150 JArray imgs = (JArray)objRoot["data"];
151 txt_ShowPath.BeginInvoke(new Action(() =>
152 {
153 txt_ShowPath.AppendText("正在下载第" + (index + 1) + "页！" + Environment.NewLine);
154 }));
155 for (int i = 0; i < imgs.Count; i++)
156 {
157 JObject img = (JObject)imgs[i];
158 string objUrl = (string)img["middleURL"];
159 try
160 {
161
162 DownloadImage(objUrl);//下载
163 SetTextMessage(100 * i / imgs.Count);
164 }
165 catch (Exception ex)
166 {
167 txt_ShowPath.BeginInvoke(new Action(() =>
168 {
169 txt_ShowPath.AppendText(ex.Message + Environment.NewLine);
170 }));
171 }
172 }
173 txt_ShowPath.BeginInvoke(new Action(() =>
174 {
175 txt_ShowPath.AppendText("第" + (index + 1) + "页下载完成！" + Environment.NewLine);
176 }));
177 }
178 }
179 /// <summary>
180 /// 下载图片
181 /// </summary>
182 /// <param name="url"></param>
183 private void DownloadImage(string objUrl)
184 {
185 txt_ShowPath.AppendText("正在下载：" + Path.GetFileName(objUrl) + "" + Environment.NewLine);
186 //URLRefer://这个图片是从哪个页面启动下载的
187 string destFile = Path.Combine(SavePath, Path.GetFileName(objUrl));//获取保存路径
188 HttpWebRequest request = (HttpWebRequest)WebRequest.Create(objUrl);
189 request.Referer = "https://image.baidu.com/";//欺骗浏览器
190 using (HttpWebResponse response = (HttpWebResponse)request.GetResponse())
191 {
192 if (response.StatusCode == HttpStatusCode.OK)
193 {
194 using (Stream stream = response.GetResponseStream())
195 using (Stream FStream = new FileStream(destFile, FileMode.Create))
196 {
197 stream.CopyTo(FStream);
198 txt_ShowPath.AppendText("下载成功！" + Environment.NewLine);
199 }
200 }
201 else
202 {
203 throw new Exception("下载" + objUrl + "失败，错误码：" + response.StatusCode);
204 }
205 }
206 }
207 /// <summary>
208 /// 进度条代理
209 /// </summary>
210 /// <param name="ipos"></param>
211 /// <param name="vinfo"></param>
212 private delegate void SetPos(int ipos);
213 /// <summary>
214 /// 进度条更新函数
215 /// </summary>
216 /// <param name="ipos"></param>
217 /// <param name="vinfo"></param>
218 private void SetTextMessage(int ipos)
219 {
220 if (this.InvokeRequired)
221 {
222 SetPos pos = new SetPos(SetTextMessage);
223 this.Invoke(pos, new object[] { ipos });
224 }
225 else
226 {
227 this.ProgressBar.Value = Convert.ToInt32(ipos);
228 }
229 }
230 /// <summary>
231 /// 停止下载
232 /// </summary>
233 /// <param name="sender"></param>
234 /// <param name="e"></param>
235 private void btn_Stop_Click(object sender, EventArgs e)
236 {
237 if (thread != null)
238 {
239 if (btn_Stop.Text == "暂 停")
240 {
241
242 thread.Suspend();//挂起线程线程
243 txt_ShowPath.AppendText("已暂停下载！" + Environment.NewLine);
244 btn_Stop.Text = "继 续";
245 }
246 else if (btn_Stop.Text == "继 续")
247 {
248 thread.Resume();//挂起线程线程
249 txt_ShowPath.AppendText("开始下载！" + Environment.NewLine);
250 btn_Stop.Text = "暂 停";
251 }
252 }
253 }
254 /// <summary>
255 /// 窗体关闭时关闭所有线程
256 /// </summary>
257 /// <param name="sender"></param>
258 /// <param name="e"></param>
259 private void Form1_FormClosed(object sender, FormClosedEventArgs e)
260 {
261 if (thread != null)
262 {
263 thread.Abort();
264 }
265 }
266 /// <summary>
267 /// 设置滚动条
268 /// </summary>
269 /// <param name="sender"></param>
270 /// <param name="e"></param>
271 private void txt_ShowPath_TextChanged(object sender, EventArgs e)
272 {
273 txt_ShowPath.SelectionStart = txt_ShowPath.Text.Length;
274 txt_ShowPath.ScrollToCaret();
275 }
276 }
277 }
浙公网安备 33010602011771号