封装抓取网页信息的实例

1. package cn.mypic;
4. import java.io.BufferedInputStream;
5. import java.io.BufferedReader;
6. import java.io.File;
7. import java.io.FileNotFoundException;
8. import java.io.FileOutputStream;
9. import java.io.IOException;
10. import java.io.InputStreamReader;
11. import java.net.MalformedURLException;
12. import java.net.URL;
13. import java.util.regex.Matcher;
14. import java.util.regex.Pattern;
17. public class GetContentPicture {
18. //得到了图片地址并下载图片
19. public void getHtmlPicture(String httpUrl) {
20. URL url;
21. BufferedInputStream in;
22. FileOutputStream file;
23. int count; //图片文件名序号
24. FileNumber num=new FileNumber();//图片文件名序号类，num为对象
25. count=num.NumberReadFromFile();//获取图片文件序号
26. try {
27. System.out.println("获取网络图片");
28. String fileName = (String.valueOf(count)).concat(httpUrl.substring(httpUrl.lastIndexOf(".")));//图片文件序号加上图片的后缀名，后缀名用了String内的一个方法来获得
29. //httpUrl.substring(httpUrl.lastIndexOf("/"));//这样获得的文件名即是图片链接里图片的名字
30. String filePath = "d:/image/";//图片存储的位置
31. url = new URL(httpUrl);
33. in = new BufferedInputStream(url.openStream());
35. file = new FileOutputStream(new File(filePath+fileName));
36. int t;
37. while ((t = in.read()) != -1) {
38. file.write(t);
39. }
40. file.close();
41. in.close();
42. System.out.println("图片获取成功");
43. count=count+1;//图片文件序号加1
44. num.NumberWriteToFile(count);//将图片名序号保存
45. } catch (MalformedURLException e) {
46. e.printStackTrace();
47. } catch (FileNotFoundException e) {
48. e.printStackTrace();
49. } catch (IOException e) {
50. e.printStackTrace();
51. }
52. }
54. //获取网页的代码保存在String格式的Content中
55. public String getHtmlCode(String httpUrl) throws IOException {
56. String content ="";
57. URL uu = new URL(httpUrl); // 创建URL类对象
58. BufferedReader ii = new BufferedReader(new InputStreamReader(uu
59. .openStream())); // //使用openStream得到一输入流并由此构造一个BufferedReader对象
60. String input;
61. while ((input = ii.readLine()) != null) { // 建立读取循环，并判断是否有读取值
62. content += input;
63. }
64. ii.close();
65. return content;
66. }
67. //分析网页代码，找到匹配的网页图片地址
68. public void get(String url) throws IOException {
70. String searchImgReg = "(?x)(src|SRC|background|BACKGROUND)=('|\")/?(([\\w-]+/)*([\\w-]+\\.(jpg|JPG|png|PNG|gif|GIF)))('|\")";//用于在网页代码Content中查找匹配的图片链接。
71. String searchImgReg2 = "(?x)(src|SRC|background|BACKGROUND)=('|\")(http://([\\w-]+\\.)+[\\w-]+(:[0-9]+)*(/[\\w-]+)*(/[\\w-]+\\.(jpg|JPG|png|PNG|gif|GIF)))('|\")";
73. String content = this.getHtmlCode(url);//this指对象gcp，在此地调用获取网页代码，getHtmlCode方法
74. //System.out.println(content); //输出的content将是一个连续的字符串。
76. Pattern pattern = Pattern.compile(searchImgReg);//java.util.regex.Pattern
77. Matcher matcher = pattern.matcher(content); //java.util.regex.Matcher
78. while (matcher.find()) {
79. System.out.println(matcher.group(3));//输出图片链接地址到屏幕
80. // System.out.println(url);
81. this.getHtmlPicture(matcher.group(3));//对象调用getHtmlPicture从网上下载并输出图片文件到指定目录
83. }
85. pattern = Pattern.compile(searchImgReg2);
86. matcher = pattern.matcher(content);
87. while (matcher.find()) {
88. System.out.println(matcher.group(3));
89. this.getHtmlPicture(matcher.group(3));
91. }
92. // searchImgReg =
93. // "(?x)(src|SRC|background|BACKGROUND)=('|\")/?(([\\w-]+/)*([\\w-]+\\.(jpg|JPG|png|PNG|gif|GIF)))('|\")";
94. }
95. //主函数url网页的地址
96. public static void main(String[] args) throws IOException {
98. String url = "http://www.baidu.com";
99. GetContentPicture gcp = new GetContentPicture();
100. gcp.get(url);
103. }
106. }

1. package cn.mypic;
3. import java.io.*;
5. public class FileNumber{
6. //文件写
7. public void NumberWriteToFile(int x){
8. int c=0;
9. c=x;
10. File filePath=new File("d:/image");//文件名序号TXT文件保存地址
11. File f1=new File(filePath,"number.txt");
12. try{
13. FileOutputStream fout=new FileOutputStream(f1);
14. DataOutputStream out=new DataOutputStream(fout);
15. out.writeInt(c);
17. }
18. catch(FileNotFoundException e){
19. System.err.println(e);
20. }
21. catch(IOException e){
22. System.err.println(e);
23. }
25. }
26. //文件读
27. public int NumberReadFromFile(){
28. int c1 = 0;
29. File filePath=new File("d:/image");
30. File f1=new File(filePath,"number.txt");
31. try{
32. FileInputStream fin=new FileInputStream(f1);
33. DataInputStream in=new DataInputStream(fin);
34. c1=in.readInt();
35. System.out.println(c1);//输出文件内容至屏幕
36. }
37. catch(FileNotFoundException e){
38. System.err.println(e);
39. }
40. catch(IOException e){
41. System.err.println(e);
42. }
43. return c1;
44. }
45. public static void main(String args[]){
47. }
51. }