java词频统计——web版支持

Posted 2020-08-14

需求概要：

1.把程序迁移到web平台，通过用户上传TXT的方式接收文件。

2.用户直接输入要统计的文本，服务器返回结果

3.在页面上给出链接 (如果有封皮、作者、字数、页数等信息更佳)或表格，展示经典英文小说词频统计结果；

4.支持用户自定义单词分隔符；

5.词汇范围对比（额外项）。

分析和设计：

1.创建web工程，利用servlet上传文件的技术实现用户向服务器上传文件。页面设置表单类型为enctype="multipart/form-data"，创建文件上传文本框<input type="file" id="upfilename" name="upfilename" value="" />，服务器端使用Part p = request.getPart("upfilename");获取上传的文件，然后写入到指定地址即可。

2.直接分析用户post到服务器的内容，为了使用原有的api，可以将输入内容写到文件中，再进行分析。

3.页面展示统计结果

4.用户可以输入自定义的分隔符和设置显示统计结果前10行（可修改）。需要修改原词频统计的有效字符函数。

5.暂时不考虑

部分代码实现：

表单实现

 1     <div align="center" id="txtform">
 2         <form action="upload" method="post" enctype="multipart/form-data">
 3             <input type="file" id="upfilename" name="upfilename" value="" /> 自定义分隔符<input
 4                 type="text" id="splitter" name="splitter"> <input
 5                 type="submit" id="submit" value="上传" />
 6         </form>
 7     </div>
 8     <div align="center" id="txtform">
 9         <form action="wordcount" method="post">
10             <div align="center">待统计内容</div>
11             <textarea name="content" id="content"
12                 style="width: 700px; height: 200px;"></textarea>
13             <br> 统计前<select id="num" name="num">
14                 <option value="10">10</option>
15                 <option value="20">20</option>
16                 <option value="0">所有</option>
17             </select>项 <br>自定义分隔符<input type="text" id="splitter" name="splitter"> <input
18                 type="submit" value="提交" /> <input type="button"
19                 onclick="if(confirm(‘确认重置？‘)){reset()}" value="重置">
20         </form>
21     </div>

文件上传：

 1         request.setCharacterEncoding("UTF-8");
 2         response.setCharacterEncoding("UTF-8");
 3         response.setContentType("text/html");
 4         PrintWriter out = response.getWriter();
 5         byte b[] = new byte[2048];
 6         @SuppressWarnings("unused")
 7         int len = 0;
 8         Part p = request.getPart("upfilename");
 9         if(p==null){
10             System.out.println("p == null");
11         }
12         String splitter = request.getParameter("splitter");
13         InputStream in = p.getInputStream();
14         String name = ""+System.currentTimeMillis();
15         FileWriter fr = new FileWriter("D:\\upload\\" + name+".txt");
16         while ((len = in.read(b)) > 0) {
17             fr.write(new String(b));
18         }
19         fr.close();
20         out.println("uploaded");
21         response.sendRedirect("wordcount?id="+name+"&splitter"+splitter);
22         out.flush();
23         out.close();

servlet处理：

 1     protected void doGet(HttpServletRequest request, HttpServletResponse response)
 2             throws ServletException, IOException {
 3         request.setCharacterEncoding("UTF-8");
 4         response.setCharacterEncoding("UTF-8");
 5         response.setContentType("text/html");
 6         PrintWriter out = response.getWriter();
 7         String id = request.getParameter("id");
 8         int num = 10;
 9         String filename = "D:\\upload\\" + id + ".txt";
10         WordUtil wu = WordUtilFactory.getWordUtil();
11         long start = System.currentTimeMillis();
12         String splitter = request.getParameter("splitter");
13         wu.setSplitter(splitter);
14         List<String[]> result = wu.getSortedWordGroupCountBuffered(filename, splitter);
15         int size = result.size();
16         for (int i = 0; i < (size > num ? num == 0 ? size : num : size); i++) {
17             String[] strs = result.get(i);
18             out.println(strs[1] + "  : " + strs[0] + "<br>");
19         }
20         long end = System.currentTimeMillis();
21         out.println("execution time :" + (end - start) + "ms");
22         out.flush();
23         out.close();
24     }
25 
26     protected void doPost(HttpServletRequest request, HttpServletResponse response)
27             throws ServletException, IOException {
28         request.setCharacterEncoding("UTF-8");
29         response.setCharacterEncoding("UTF-8");
30         response.setContentType("text/html");
31         PrintWriter out = response.getWriter();
32         String content = request.getParameter("content");
33         String numStr = request.getParameter("num");
34         int num = 10;
35         if (numStr != null) {
36             num = Integer.parseInt(numStr);
37         }
38         WordUtil wu = WordUtilFactory.getWordUtil();
39     
40         long start = System.currentTimeMillis();
41         String filename = "D://tmp.txt";
42 
43         FileWriter fr = new FileWriter(filename);
44         fr.write(content);
45         fr.close();
46         String splitter = request.getParameter("splitter");
47         wu.setSplitter(splitter);
48         List<String[]> result = wu.getSortedWordGroupCountBuffered(filename, splitter);
49         int size = result.size();
50         for (int i = 0; i < (size > num ? num == 0 ? size : num : size); i++) {
51             String[] strs = result.get(i);
52             out.println(strs[1] + "  : " + strs[0] + "<br>");
53         }
54         long end = System.currentTimeMillis();
55         out.println("execution time :" + (end - start) + "ms");
56         out.flush();
57         out.close();
58     }

有效字符判定（即自定义分隔符）

 1     public void setSplitter(String splitter) {
 2         char[] tmp = splitter.toCharArray();
 3         ArrayList<Character> deleted = new ArrayList<>();
 4         for(int i=0;i<tmp.length-1;i++){
 5             if(tmp[i]==‘\\‘){
 6                 char c = tmp[i+1];
 7                 if(c==‘n‘){
 8                     deleted.add(‘\n‘);
 9                 }
10                 if(c==‘r‘){
11                     deleted.add(‘\n‘);
12                 }
13                 if(c==‘t‘){
14                     deleted.add(‘\n‘);
15                 }
16                 char[] copy = new char[tmp.length-2];
17                 for(int j = 0;j <i;j++){
18                     copy[j]=tmp[j];
19                 }
20                 for(int j=i;j<tmp.length-2;j++){
21                     copy[j]=tmp[j+2];
22                 }
23                 i++;
24             }
25         }
26         split = new char[tmp.length+deleted.size()];
27         for(int i = 0;i<tmp.length;i++){
28             split[i]=tmp[i];
29         }
30         for(int i=tmp.length;i<split.length;i++){
31             split[i]=deleted.get(split.length-tmp.length-1);
32         }
33     }
34 
35     private int isCharacter(char ch, String splitter) {
36         if (split == null) {
37             if ((ch >= ‘a‘ && ch <= ‘z‘))
38                 return 1;
39             if ((ch >= ‘A‘ && ch <= ‘Z‘))
40                 return 1;
41             if (ch >= ‘0‘ && ch <= ‘9‘)
42                 return 2;
43             return 0;
44         }
45         if (split.equals("")) {
46             if ((ch >= ‘a‘ && ch <= ‘z‘))
47                 return 1;
48             if ((ch >= ‘A‘ && ch <= ‘Z‘))
49                 return 1;
50             if (ch >= ‘0‘ && ch <= ‘9‘)
51                 return 2;
52             return 0;
53         }
54         for (int i = 0; i < split.length; i++) {
55             if (ch == split[i]) {
56                 return 0;
57             }
58         }
59         if ((ch >= ‘a‘ && ch <= ‘z‘))
60             return 1;
61         if ((ch >= ‘A‘ && ch <= ‘Z‘))
62             return 1;
63         if (ch >= ‘0‘ && ch <= ‘9‘)
64             return 2;
65         return 1;
66     }

 

web版工程地址：https://git.coding.net/jx8zjs/wordcount-web.git

ssh:  [email protected]:jx8zjs/wordcount-web.git

 

console版工程地址：https://coding.net/u/jx8zjs/p/wordCount/git

ssh:  [email protected]:jx8zjs/wordCount.git