java/project/DataCleaner.java

package java01;

import java.util.List;
import java.util.regex.Matcher;
import java.util.regex.Pattern;

public class DataCleaner {
    
    public List<Job> cleanJobs(List<Job> jobs) {
        for (Job job : jobs) {
            cleanJob(job);
        }
        return jobs;
    }
    
    private void cleanJob(Job job) {
        // 清理职位标题
        if (job.getTitle() != null) {
            job.setTitle(job.getTitle().trim());
        }
        
        // 清理公司名称
        if (job.getCompany() != null) {
            job.setCompany(job.getCompany().trim());
        }
        
        // 清理薪资
        if (job.getSalary() != null) {
            job.setSalary(job.getSalary().trim());
        }
        
        // 清理地点
        if (job.getLocation() != null) {
            job.setLocation(job.getLocation().trim());
        }
        
        // 清理经验
        if (job.getExperience() != null) {
            job.setExperience(job.getExperience().trim());
        }
        
        // 清理学历
        if (job.getEducation() != null) {
            job.setEducation(job.getEducation().trim());
        }
        
        // 清理技能
        if (job.getSkills() != null) {
            job.setSkills(job.getSkills().trim());
        }
    }
    
    // 提取薪资范围的最小值
    public int extractMinSalary(String salary) {
        if (salary == null || salary.isEmpty()) {
            return 0;
        }
        
        // 匹配薪资范围，如"10K-20K"
        Pattern pattern = Pattern.compile("(\\d+)K-(\\d+)K");
        Matcher matcher = pattern.matcher(salary);
        
        if (matcher.find()) {
            return Integer.parseInt(matcher.group(1)) * 1000;
        }
        
        // 匹配固定薪资，如"15K"
        pattern = Pattern.compile("(\\d+)K");
        matcher = pattern.matcher(salary);
        if (matcher.find()) {
            return Integer.parseInt(matcher.group(1)) * 1000;
        }
        
        return 0;
    }
    
    // 提取薪资范围的最大值
    public int extractMaxSalary(String salary) {
        if (salary == null || salary.isEmpty()) {
            return 0;
        }
        
        // 匹配薪资范围，如"10K-20K"
        Pattern pattern = Pattern.compile("(\\d+)K-(\\d+)K");
        Matcher matcher = pattern.matcher(salary);
        
        if (matcher.find()) {
            return Integer.parseInt(matcher.group(2)) * 1000;
        }
        
        // 匹配固定薪资，如"15K"
        pattern = Pattern.compile("(\\d+)K");
        matcher = pattern.matcher(salary);
        if (matcher.find()) {
            return Integer.parseInt(matcher.group(1)) * 1000;
        }
        
        return 0;
    }
    
    // 提取经验年限
    public int extractExperienceYears(String experience) {
        if (experience == null || experience.isEmpty()) {
            return 0;
        }
        
        // 匹配经验年限，如"3-5年"
        Pattern pattern = Pattern.compile("(\\d+)-(\\d+)年");
        Matcher matcher = pattern.matcher(experience);
        
        if (matcher.find()) {
            return Integer.parseInt(matcher.group(1));
        }
        
        // 匹配固定经验，如"3年以上"
        pattern = Pattern.compile("(\\d+)年");
        matcher = pattern.matcher(experience);
        if (matcher.find()) {
            return Integer.parseInt(matcher.group(1));
        }
        
        return 0;
    }
}