java 解析csv

opencsv

java读取csv的类库主要有两种，opencsv和javacsv，研究发现，javacsv最后一次更新是2014-12-10，很久不维护了。opencsv是apache的项目，并且至今仍在维护，所以决定使用opencsv。

---

csv

csv文件，全名 comma separated values，默认以逗号分隔，是纯文本文件。虽然用excel打开后格式排版了，但是那是excel对他进行了处理。用notepad或者sublime text打开能看到最原始的文本。
为了后续举例，这里编辑了一个test.csv

header1,header2,header3
1,a,10
2,b,20
3,c,30
4,d,40
5,e,50
6,f,60

---

读取方式

• 最原始的方式，逐行读取，然后操作

 CSVReader reader = new CSVReader(new InputStreamReader(new FileInputStream("test.csv"),"gbk"));
	  
	  /*
	   * 逐行读取
	   */
	  String[] strArr = null;
	  while((strArr = reader.readNext())!=null){
		  System.out.println(strArr[0]+"---"+strArr[1]+"----"+strArr[2]);
	  }
	  
	  reader.close();

---

绑定csv文件转换成bean

逐行读取操作是最原始的操作方式，opencsv提供了基于“策略”的映射，将csv绑定到bean。

策略简介

观察一下策略的继承层次

接口

• MappingStrategy
  • 映射的顶层接口
• HeaderColumnNameMappingStrategy
  • 列名映射策略，读取csv文件第一行作为header，比如header1,header2,header3,然后调用bean的setHeader1方法，setHeader2方法，setHeader3方法分别设置值，所以这种策略要求，列名与bean中的属性名完全一致，如果不一致，则值为空，不会出错。使用注解时，注解名字必须与csv中列名一致。
• ColumnPositionMappingStrategy
  • 列位置映射策略，他没有header的概念，所以会输出取所有行。在columnMapping数组中指定bean的属性，第一个值对应csv的第一列，第二个值对应csv的第二类…
• HeaderColumnNameTranslateMappingStrategy
  • 列头名字翻译映射策略,与HeaderColumnNameMappintStrategy相比，bean的属性名可以与csv列头不一样。通过指定map来映射。

---

具体映射用法

Java POJO类

public class SimpleBeanInfo {

	private String header1;
	

	private String header2;
	
	private String header3;

	public String getHeader1() {
		return header1;
	}

	@Override
	public String toString() {
		return "SimpleBeanInfo [header1=" + header1 + ", header2=" + header2
				+ ", header3=" + header3 + "]";
	}

	public void setHeader1(String header1) {
		this.header1 = header1;
	}

	public String getHeader2() {
		return header2;
	}

	public void setHeader2(String header2) {
		this.header2 = header2;
	}

	public String getHeader3() {
		return header3;
	}

	public void setHeader3(String header3) {
		this.header3 = header3;
	}

---

基于列索引的映射

• 通俗点来说就是列位置映射，csv文件中的列位置对应到bean中的列

• 非注解方式

CSVReader reader = new CSVReader(new InputStreamReader(new FileInputStream("test.csv"),"gbk"));
	  /*
	   *基于列位置，映射成类 
	   */
	   //csv文件中的第一列对应类的header，第二列对应类的header2，第三列对应类的header3
	  String[] columnMapping={"header1","header2","header3"};
	  ColumnPositionMappingStrategy<SimpleBeanInfo> mapper = new ColumnPositionMappingStrategy<SimpleBeanInfo>();
	  mapper.setColumnMapping(columnMapping);
	  mapper.setType(SimpleBeanInfo.class);
	  /* */ 
	  CsvToBean<SimpleBeanInfo>  csvToBean = new CsvToBean<SimpleBeanInfo>();

	  List<SimpleBeanInfo> list = csvToBean.parse(mapper, reader);
	  
	  for(SimpleBeanInfo e : list){
		  System.out.println(e);
	  }
		   
	}

• 注解方式

public class SimpleBeanInfo {

	@CsvBindByPosition(position=0)
	private String header1;
	

	@CsvBindByPosition(position=1)
	private String header2;
	
	@CsvBindByPosition(position=2)
	private String header3;
}

CSVReader reader = new CSVReader(new InputStreamReader(new FileInputStream("test.csv"),"gbk"));
 ColumnPositionMappingStrategy<SimpleBeanInfo> mapper = new ColumnPositionMappingStrategy<SimpleBeanInfo>();
	  mapper.setType(SimpleBeanInfo.class);
	  CsvToBean<SimpleBeanInfo>  csvToBean = new CsvToBean<SimpleBeanInfo>();
		 
	  List<SimpleBeanInfo> list = csvToBean.parse(mapper, reader);
	  
	  for(SimpleBeanInfo e : list){
		  System.out.println(e);
	  }

---

基于列名的映射

• 非注解方式

 CSVReader reader = new CSVReader(new InputStreamReader(new FileInputStream("test.csv"),"gbk"));
	 
	  /* */ 
	  
	  HeaderColumnNameMappingStrategy<SimpleBeanInfo> mapper = new
			  HeaderColumnNameMappingStrategy<SimpleBeanInfo>();
	  mapper.setType(SimpleBeanInfo.class);
	  CsvToBean<SimpleBeanInfo>  csvToBean = new CsvToBean<SimpleBeanInfo>();
	 
	  List<SimpleBeanInfo> list = csvToBean.parse(mapper, reader);
	  
	  for(SimpleBeanInfo e : list){
		  System.out.println(e);
	  }

• 注解方式

public class SimpleBeanInfo {

	@CsvBindByName(column="header1")
	private String header1;
	
	
	@CsvBindByName(column="header2")
	private String header2;
	
	@CsvBindByName(column="header3")
	private String header3;
}

CSVReader reader = new CSVReader(new InputStreamReader(new FileInputStream("test.csv"),"gbk"));

	  HeaderColumnNameMappingStrategy<SimpleBeanInfo> mapper = new
			  HeaderColumnNameMappingStrategy<SimpleBeanInfo>();
	  mapper.setType(SimpleBeanInfo.class);
	  CsvToBean<SimpleBeanInfo>  csvToBean = new CsvToBean<SimpleBeanInfo>();
	 
	  List<SimpleBeanInfo> list = csvToBean.parse(mapper, reader);
	  
	  for(SimpleBeanInfo e : list){
		  System.out.println(e);
	  }

---

基于列名转换映射

 CSVReader reader = new CSVReader(new InputStreamReader(new FileInputStream("test.csv"),"gbk"));
	  /*
	   * 基于列名转换，映射成类
	  */
	  HeaderColumnNameTranslateMappingStrategy<SimpleBeanInfo> mapper = 
			  new HeaderColumnNameTranslateMappingStrategy<SimpleBeanInfo>();
	  mapper.setType(SimpleBeanInfo.class);
	  
	  Map<String,String> columnMapping = new HashMap<String,String>();
	  columnMapping.put("header1", "header1");//csv中的header1对应bean的header1
	  columnMapping.put("header2", "header2");
	  columnMapping.put("header3", "header3");
	  mapper.setColumnMapping(columnMapping);
	  
	  CsvToBean<SimpleBeanInfo>  csvToBean = new CsvToBean<SimpleBeanInfo>();
		 
	  List<SimpleBeanInfo> list = csvToBean.parse(mapper, reader);
	  
	  for(SimpleBeanInfo e : list){
		  System.out.println(e);
	  }

---

过滤器

opencsv提供了过滤器，可以过滤某些行，比如page header、page footer等

• 所有的过滤器必须实现CsvToBeanFilter 接口

public class MyCsvToBeanFilter implements CsvToBeanFilter {
		
		public boolean allowLine(String[] line) {
		    //过滤第一列值等于1的行
			if("1".equals(line[0])){
				return false;
			}
	 		return true;
	 	}

}

 MyCsvToBeanFilter filter = new MyCsvToBeanFilter();
 List<SimpleBeanInfo> list = csvToBean.parse(mapper, reader,filter);

---

转化器

类中的属性不一定都是字符串，比如数字、日期等，但是我们从csv中获取到的都是字符串，这种情况就应该使用转化器。
这里定义一个SimpleBeanConverter，继承AbstractBeanField

public class SimpleBeanFieldConverter extends AbstractBeanField<SimpleBeanInfo> {

	@Override
	protected Object convert(String value) throws CsvDataTypeMismatchException,
			CsvRequiredFieldEmptyException, CsvConstraintViolationException {
		Field f = getField();
		if("date".equals(f.getName())){
			try {
				return new SimpleDateFormat("yyyy-MM-dd").parse(value);
			} catch (ParseException e) {
				e.printStackTrace();
			}
		}
		return null;
	}
}

test.csv添加一列header4

header1,header2,header3,header4
1,a,10,2016-05-01
2,b,20,2016-05-02
3,c,30,2016-05-03
4,d,40,2016-05-04
5,e,50,2016-05-05
6,f,60,2016-05-06

SimpleBeanInfo添加属性

@CsvCustomBindByPosition(position=3,converter=SimpleBeanFieldConverter.class)
private Date date;

输出结果
由于ColumnPositionMappingStrategy会连header行也解析，所以第一行会打印异常信息。我们看到header4列已经转换为日期。如果不只一个列需要转换怎么办？在相应的属性上添加注解(@CsvCustomBindByPosition或@CsvCustomBindByName)，然后在convert(Object value)中扩展即可

java.text.ParseException: Unparseable date: "header4"
	at java.text.DateFormat.parse(DateFormat.java:357)
	at test_maven.SimpleBeanFieldConverter.convert(SimpleBeanFieldConverter.java:24)
	at com.opencsv.bean.AbstractBeanField.setFieldValue(AbstractBeanField.java:70)
	at com.opencsv.bean.CsvToBean.processField(CsvToBean.java:245)
	at com.opencsv.bean.CsvToBean.processLine(CsvToBean.java:220)
	at com.opencsv.bean.CsvToBean.processLine(CsvToBean.java:189)
	at com.opencsv.bean.CsvToBean.parse(CsvToBean.java:166)
	at com.opencsv.bean.CsvToBean.parse(CsvToBean.java:133)
	at test_maven.TestCSV.main(TestCSV.java:46)
SimpleBeanInfo [header1=header1, header2=header2, header3=header3, date=null]
SimpleBeanInfo [header1=2, header2=b, header3=20, date=Mon May 02 00:00:00 CST 2016]
SimpleBeanInfo [header1=3, header2=c, header3=30, date=Tue May 03 00:00:00 CST 2016]
SimpleBeanInfo [header1=4, header2=d, header3=40, date=Wed May 04 00:00:00 CST 2016]
SimpleBeanInfo [header1=5, header2=e, header3=50, date=Thu May 05 00:00:00 CST 2016]
SimpleBeanInfo [header1=6, header2=f, header3=60, date=Fri May 06 00:00:00 CST 2016]

公众号

欢迎关注个人公众号