利用正则表达式抓取Email地址

这是一个利用正则表达式获取本地储存的文件（txt doc html）或网页中得Email地址。在原理上接近网络爬虫。

但是我在做测试得时候，不能获取openOffice的文件中的地址。应该是编码问题。可是openOffice号称自己

完全支持MS Office 得文件，那编码应该式兼容的啊。而MS的doc文件在试验中式成功的。这个我问题我以后再看看吧。

其中的英语是在太面，但是我得坚持用英语写。

1. /*
2. *Thisprogramwaswriteforcatchingemailaddresses
3. *fromadocumentthatcontainslotsofthem.
4. */
6. importjava.io.BufferedReader;
7. importjava.io.FileNotFoundException;
8. importjava.io.FileReader;
9. importjava.io.IOException;
10. importjava.io.InputStream;
11. importjava.io.InputStreamReader;
12. importjava.net.HttpURLConnection;
13. importjava.net.MalformedURLException;
14. importjava.net.URL;
15. importjava.util.regex.Matcher;
16. importjava.util.regex.Pattern;
18. /**
19. *@authorSancho_lai
20. *
21. */
22. publicclassReadFileFromLocalAndWeb{
24. publicvoidtest(){
25. /**
26. *thispartgetemailaddressfromlocaldoc
27. */
28. System.out.println("********************GetEmailAddressFromLocaldoc********************");
29. try{
30. /*
31. *hereyoucaninputwhateveraddressofdocthatcontainstheemailaddressesyouwant,
32. */
33. BufferedReaderbr=newBufferedReader(newFileReader("E://workspace//Project_For_ToyTest//doc//EmailAddress.txt"));
34. //BufferedReaderbr=newBufferedReader(newFileReader("E://workspace//Project_For_ToyTest//doc//NewOpenDocumentText.odt"));
35. Stringline="";
37. while((line=br.readLine())!=null){
38. parser(line);
39. }
40. }catch(FileNotFoundExceptione){
41. e.printStackTrace();
42. }catch(IOExceptione){
43. e.printStackTrace();
44. }
45. /**
46. *thispartgetemailaddressfromweb
47. */
48. System.out.println("********************GetEmailAddressFromtheWeb********************");
49. Stringline="";
50. InputStreamurlStream;
51. try{
52. URLurl=newURL("http://www.8885.net/bbs/thread-189746-1-3.html");
53. HttpURLConnectionconnection=(HttpURLConnection)url.openConnection();
54. urlStream=connection.getInputStream();
55. BufferedReaderbr=newBufferedReader(newInputStreamReader(urlStream,"gbk"));
56. while((line=br.readLine())!=null){
57. parser(line);
58. }
59. }catch(MalformedURLExceptione){
60. e.printStackTrace();
61. }catch(IOExceptione){
62. e.printStackTrace();
63. }finally{
64. }
65. }
67. privatevoidparser(Stringline){
68. //thisistheemailaddresspattern.
69. Patternp=Pattern.compile("[//w[.-]]+@[//w[.-]]+//.[//w]+");
70. Matcherm=p.matcher(line);
71. while(m.find()){
72. System.out.println(m.group());
73. }
74. }
76. publicstaticvidemain(String[]args){
77. ReadFileFromLocalAndWebt=newReadFileFromLocalAndWeb();
78. t.test();
79. }
80. }