Java抓取Codeforces——针对某一次提交的源码和数据

Posted 2021-12-16 zifeiy

需要引入Jsoup依赖：

        <dependency>
          <!-- jsoup html parser library @ https://jsoup.org/ -->
          <groupId>org.jsoup</groupId>
          <artifactId>jsoup</artifactId>
          <version>1.12.1</version>
        </dependency>

Jsoup相关依赖：https://jsoup.org/
其他需要的是 FileHelper 类，参见：https://www.cnblogs.com/zifeiy/p/9224569.html
（不过我们这边可能需要将"UTF-8"改成"GBK" ~）
然后实现的类如下：

package com.zifeiy.cf_data_get.handle;

import java.io.File;
import java.io.IOException;

import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;
import org.jsoup.select.Elements;

import com.zifeiy.cf_data_get.assist.FileHelper;

public class SubmissionInfoGet 
    
    public void getSubmission(int contestId, int submissionId) throws IOException 
        String urlString = "http://codeforces.com/contest/" + contestId + "/submission/" + submissionId;
        Document doc = Jsoup.connect(urlString).get();
        System.out.println("title = " + doc.title());
        
        String dirString = "D:/codeforces/" + submissionId;
        File dir = new File(dirString);
        if (dir.exists() == false) dir.mkdirs();
        File dataDir = new File(dirString + File.separator + "data");
        if (dataDir.exists() == false) dataDir.mkdirs();
        
        // 代码
        String codeString = doc.getElementById("program-source-text").text();
        File codeFile = new File(dirString + File.separator + "std.cpp");
        FileHelper.writeFile(codeFile, codeString);
        Elements inputElements = doc.getElementsByClass("file input-view");
        Elements outputElements = doc.getElementsByClass("file answer-view");
        int sz = inputElements.size();
        int cnt = 0;
        for (int i = 0; i < sz; i ++) 
            String inputTextString = inputElements.get(i).getElementsByClass("text").get(0).text();
            String outputTextString = outputElements.get(i).getElementsByClass("text").get(0).text();
            if (inputTextString.endsWith("...") == false) 
//              System.out.println("[input]\\n" + inputTextString + "\\n[output]\\n" + outputTextString + "\\n[end]");
                cnt ++;
                File inputFile = new File(dirString + File.separator + "data" + File.separator + cnt + ".in");
                FileHelper.writeFile(inputFile, inputTextString);
                File outputFile = new File(dirString + File.separator + "data" + File.separator + cnt + ".out");
                FileHelper.writeFile(outputFile, outputTextString);
            
        
        
    
    
    
    // main for test
    public static void main(String[] args) throws IOException 
        new SubmissionInfoGet().getSubmission(1169, 54847813);
    
    

其中，contestId 和 submissionId 分别对应 比赛Id 和 提交Id 。