HTTP 文件上传原理 Java 实现

前言：

文件上传用的已经很多，java web 大概用到如下

1. Struts
2. Spring MVC  CommonsMultipartResolver
3. Commons-fileupload

  Struts/Spring MVC 实现都是基于Commons-fileupload，但背后的原理，大多数估计没有关注，最近阅读一些开源源码也发现，只有基础才是最重要的，万变不离其宗，在it领域不然会被漫天的新技术，冲昏了头，不知所措，下面开始。

 

HTTP：

1. 表单form 类似

 

<form action="/file/upload" method="post" enctype="multipart/form-data">

<input type="text" name="name"><br>

<input type="file" name="file1"><br>

<input type="file" name="file2"><br>

<input type="submit" value="提交">

</form>

 

    2. 用浏览器追踪表单提交，会发现如下

    

Accept:text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8
Content-Type:multipart/form-data; boundary=----WebKitFormBoundary4PCP0w0H0qxg16VB
Origin:http://localhost:8080
Referer:http://localhost:8080/sys/template/tem/create
Upgrade-Insecure-Requests:1
User-Agent:Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/46.0.2490.71 Safari/537.36


------WebKitFormBoundary4PCP0w0H0qxg16VB
Content-Disposition: form-data; name="id"


------WebKitFormBoundary4PCP0w0H0qxg16VB
Content-Disposition: form-data; name="name"

测试
------WebKitFormBoundary4PCP0w0H0qxg16VB
Content-Disposition: form-data; name="type"

INDEX
------WebKitFormBoundary4PCP0w0H0qxg16VB
Content-Disposition: form-data; name="layoutFile"; filename="confirm-btn.png"
Content-Type: image/png


------WebKitFormBoundary4PCP0w0H0qxg16VB
Content-Disposition: form-data; name="temFile"; filename="login.html"
Content-Type: text/html


------WebKitFormBoundary4PCP0w0H0qxg16VB--

   

 

    重要部位红色已经标注，表单提交时http 头部的 Content-Type 会有一个boundary分隔符，分隔符会分割表单提交的每项内容(也就是每个input域)，如是文件则Content-Disposition会出现一个filename,同时带上Content-Type描述文件类型，否则没有，大体的解析格式如下(为了显示观看，故意换行显示，实际上没有)

-----------分隔符\r\n
Content-Disposition: form-data; name="XX"\r\n
Content-Type: image/png\r\n
\r\n
具体内容
------------分隔符\r\n
Content-Disposition: form-data; name="XX"; filename="XX"\r\n
Content-Type: image/png\r\n
\r\n
具体内容
------------分隔符\r\n
Content-Disposition: form-data; name="XX"; filename="XX"\r\n
Content-Type: image/png\r\n
\r\n
具体内容
------------分隔符--\r\n

 

注：最后一行会多出--，例如---------------分隔符--\r\n，同时------------分隔符会比boundary=----分隔符 多--两个，总体可以理解以--boundary进行分割的

 

 

JAVA Servlet 实现：

@WebServlet(urlPatterns="/file/upload")
public class FileServlet extends HttpServlet{
    
    private static final long serialVersionUID = 1L;

    @Override
    protected void doPost(HttpServletRequest request, HttpServletResponse response) 
            throws ServletException, IOException {
        
        String contentType = request.getContentType();
        
        //文件上传(类似：Content-Type:multipart/form-data; boundary=----WebKitFormBoundary4PCP0w0H0qxg16VB)
        if(contentType != null && contentType.startsWith("multipart/form-data")){
            try {
                List<FileItem> fileItems = FileItemParse.parseForm(request);
                System.out.println(fileItems);
            } catch (Exception e) {
                e.printStackTrace();
            }
        }
        
    }
    
    
    @Override
    protected void doGet(HttpServletRequest request, HttpServletResponse response) 
            throws ServletException, IOException {
        
        request.getRequestDispatcher("/WEB-INF/views/file/upload.jsp")
                .forward(request, response);
    }
    
}

 

 

public class FileItemParse {
    
    //获取边界值
    public static String getBoundary(HttpServletRequest request) {  
        String rtnStr = null;  
        String tmpType = request.getContentType();  
        if (null != tmpType) {  
            rtnStr = tmpType.contains("boundary=") ? tmpType.split("boundary=")[1] : null;  
        }  
        return "--".concat(rtnStr); //此处应该是规范，比ContentType中多2个-
    }  

    //解析表单
    public static List<FileItem> parseForm(HttpServletRequest request) throws Exception{
        List<FileItem> fileItems = new ArrayList<FileItem>();
        
        byte[] boundaryBytes = getBoundary(request).getBytes();
        int boundaryBytesLen = boundaryBytes.length;
        
        BufferedInputStream input = null;  
        ByteArrayOutputStream out = new ByteArrayOutputStream();  
        try {  
            input = new BufferedInputStream(request.getInputStream());  
            int tmpI = -1;  
            int tmpL = -1; 
            FileItem item = null;
            
            //跳过分界线
            input.skip(boundaryBytesLen);
            while ((tmpI = input.read()) != -1) {  
                if (tmpI == 13) {  
                    tmpL = (input.read());  
                    if (tmpL == 10) {
                        if (out.size() == 0) { //跳过空行分隔符
                            continue;
                        }
                        
                        String bufferStr = out.toString("UTF-8");
                        //Content-Disposition
                        if(bufferStr.contains("Content-Disposition:")){
                            item = new FileItem();
                            String[] tmpStr = bufferStr.split(";");
                            String nameV = tmpStr[1].split("=")[1];
                            item.setParamName(nameV.substring(1, nameV.length() - 1)); //去除"
                            
                            if(bufferStr.contains("filename")){//文件表单域
                                String filenameV = tmpStr[2].split("=")[1];
                                item.setFileName(filenameV.substring(1, filenameV.length() - 1)); //去除"
                            }else{//普通表单域
                                fetchContent(item, input, boundaryBytes);
                                fileItems.add(item);
                            }
                            out.reset();
                            continue;
                        }
                        //Content-Type
                        if(bufferStr.contains("Content-Type:")){
                            item.setMimeType(bufferStr.split(":")[1].trim());
                            fetchContent(item, input, boundaryBytes);
                            fileItems.add(item);
                            //文件存储
                            out.reset();
                            continue;
                        }
                    }  
                    out.write(tmpI);  
                    out.write(tmpL); 
                }  
                out.write(tmpI);  
            }  
        } catch (IOException ioe) {  
            ioe.printStackTrace();  
        } finally {  
            if (null != input) {  
                try {  
                    out.close();  
                    input.close();  
                } catch (IOException e) {  
                    e.printStackTrace();  
                }  
            }  
        }  
        return fileItems;  
    }
    
    //内容提取
    private static void fetchContent(FileItem item, BufferedInputStream input, byte[] boundaryBytes) throws IOException{
        input.skip(2); //跳过空行分隔符
        
        int i = -1;
        int l = -1;
        ByteArrayOutputStream tempOut = new ByteArrayOutputStream();
        byte[] tempByte = new byte[boundaryBytes.length]; 
        while((i = input.read()) != -1){
            if (13 == i) { 
                l = input.read();
                if (10 == l && isBoundary(input, boundaryBytes, tempByte)) {
                    break;
                } 
                else {  
                    tempOut.write(i);
                    tempOut.write(l);
                    if (10 == l) { //如不是分解符，则写入存储
                        tempOut.write(tempByte);
                    }
                    continue;
                }  
            }  
            tempOut.write(i); 
        }
        
        if(item.getMimeType() != null){ //文件
            //此处测试环境，故直接写入本地文件，正式应写入系统java.io.temp目录
            String url = "d:/temp/" + item.getFileName();
            File file = new File(url);
            if(!file.getParentFile().exists()){
                file.getParentFile().mkdirs();
            }
            FileOutputStream out = new FileOutputStream(file);
            out.write(tempOut.toByteArray());
            out.flush();
            out.close();
            item.setSimpleField(false);
            item.setFilePath(url);
        }
        else{
            item.setParamValue(new String(tempOut.toByteArray(), "UTF-8"));
            item.setSimpleField(true);
        }
    }
    
    private static boolean isBoundary(BufferedInputStream input, byte[] sourceBoundaryBytes, byte[] temp) throws IOException{
        int count = input.read(temp);
        
        for (int i = 0; i < count; i++) {
            if (sourceBoundaryBytes[i] != temp[i]) {
                return false;
            }
        }
        return true;
    }
    
}

 

public class FileItem {

    //file
    private String mimeType; //文件类型
    private String filePath; //存储路径
    private String fileName; //上传文件名
    
    //true:非file表单项, false:file表单项
    private boolean isSimpleField;
    
    private String paramName; 
    private String paramValue;
 
   //get set

}

 

 

以上只是一个简单的不完全实现，主要是针对HTTP 文件上传数据协议的一个解析过程，更多的可以去看Commons-fileupload源码，里面有更进一步的数据封装(例如进度条)。

 

参考文献：

http://www.ietf.org:80/rfc/rfc1867.txt

http://www.ietf.org:80/rfc/rfc2045.txt

http://blog.csdn.net/ybygjy/article/details/5869158