我需要创建一个Java udf，以根据其他变量中的特定条件返回字符串（CD_NOT）输出：
for each group (IDT_GCP) do :
if there is a CD_NOT == "" then return the latest CD_NOT (given by the date variable DA_PRM_CTR_ORDER )
if IDC_PSE_PCL = 1 ET IDC_CD_NOT == 0
return CD_NOT having this condition
else ORDER BY DA_PRM_CTR_ORDER ASC
if DA_PRM_CTR_ORDER are equal
ORDER BY IDT_ETT_PSE ASC
return the lowest CD_NOT
else
return the CD_NOT of the latest DA_PRM_CTR_ORDER
此函数在Pig脚本中的调用方式为：
ENEE_ENR_GCP = FOREACH (GROUP ENEE_ENR BY IDT_GCP)
{
ENEE_ENR_GROUP = ORDER ENEE_ENR BY IDT_GCP;
GENERATE
group AS IDT_GCP,
SelectionNote(ENEE_ENR_GROUP) AS CD_NOT;
};
ENEE_ENR_GCP如下所示：
"001590842","51862499", "", "1", "20170201","0"
"001590842","51862499", "X0", "1", "20150529", "1"
"001639055","51862517", "", "1", "20170201", "2"
"001639055","51862517", "G3", "1", "20170201", "2"
"001597135","51862517", "G3", "1", "20170201", "2"
"001597135","51862517", "", "1", "20170201", "2"
"002804935","00006178","G4","1","19870101","1"
"002804935","00009118","X0","1","19861201","1"
"002804935","00009957","","1","19861229","1"
"002804935","00012970","B3++","1","20100227","1"
Where :
1column : IDT_GCP
2column : IDT_ETT_PSE
3column : CD_NOT (same times it's empty)
4column : IDC_PSE_PCL
5column : DA_PRM_CTR_ORDER
6column : IDC_CD_NOT
用于获取输入并返回结果的输出和输入函数是：
public DataBag exec(Tuple input) throws IOException {
try {
if (input.get(0) == null || input.size() == 0)
return null;
// Récupération de bag à partir de Script PIG
DataBag bagFromPigScript = (DataBag) input.get(0);
// Tableau permettant de sauvegarder les tuples sous format
// string
List<Personne> listPersonnes = new ArrayList<Personne>();
// Convertir au script PIG
Tuple returnTuple = TupleFactory.getInstance().newTuple();
List<Tuple> returnTupleList = new ArrayList<Tuple>();
returnTuple.append(SelectionCodeNote((ArrayList<Personne>) listPersonnes));
returnTupleList.add(returnTuple);
return BagFactory.getInstance().newDefaultBag(returnTupleList);
} catch (Exception e) {
throw new IOException("Caught exception processing input row ", e);
}
}
@Override
public Schema outputSchema(final Schema input) {
try {
Schema.FieldSchema CD_NOT = new Schema.FieldSchema("CD_NOT", DataType.CHARARRAY);
Schema tupleSchema = new Schema();
tupleSchema.add(CD_NOT);
Schema bagSchema = new Schema(tupleSchema);
return new Schema(new Schema.FieldSchema("DATA_CD_NOT", bagSchema, DataType.BAG));
} catch (FrontendException e) {
e.printStackTrace();
return null;
}
}
我的人员班是：
private class Personne implements Comparable {
private String ID_GCP;
private String IDT_ETT_PSE;
private String CD_NOT;
private String IDC_PSE_PCL;
private String DA_PRM_CTR_ORDER;
private String IDC_CD_NOT;
public Personne (String ID_GCP, String IDT_ETT_PSE, String CD_NOT, String IDC_PSE_PCL, String DA_PRM_CTR_ORDER, String IDC_CD_NOT) {
this.ID_GCP = ID_GCP;
this.IDC_PSE_PCL = IDC_PSE_PCL;
this.IDC_CD_NOT = IDC_CD_NOT;
this.DA_PRM_CTR_ORDER = DA_PRM_CTR_ORDER;
this.IDT_ETT_PSE = IDT_ETT_PSE;
this.CD_NOT = CD_NOT;
}
public String getIDC_CD_NOT() {
return this.IDC_CD_NOT;
}
public String getID_GCP() {
return this.ID_GCP;
}
public String getIDC_PSE_PCL() {
return this.IDC_PSE_PCL;
}
public String getDA_PRM_CTR_ORDER() {
return this.DA_PRM_CTR_ORDER;
}
public String getIDT_ETT_PSE() {
return this.IDT_ETT_PSE;
}
public String getCD_NOT() {
return this.CD_NOT;
}
我的操作功能是：
CompareTo：如上所述进行比较和排序
public int compareTo(Object personne) {
int res = 0;
Personne other = (Personne) personne;
// Conversion of Dates from String to Dates
Date otherDate = converteDate(other.getDA_PRM_CTR_ORDER());
Date entreePersonne = converteDate(this.DA_PRM_CTR_ORDER);
res = entreePersonne.compareTo(otherDate);
// if there is Legality between dates
if (res == 0) {
Long entreePersonneIDT = Long.parseLong(this.getIDT_ETT_PSE());
Long otherPersonneIDT = Long.parseLong(other.getIDT_ETT_PSE());
res = entreePersonneIDT.compareTo(otherPersonneIDT);
return res;
}
return res;
}
private Date converteDate(String date) {
DateFormat df = new SimpleDateFormat("yyyyMMdd");
Date convertedDate = null;
try {
convertedDate = (Date) df.parse(date);
} catch (ParseException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
return convertedDate;
}
要将字符串转换为日期
private Date converteDate(String date) {
DateFormat df = new SimpleDateFormat("yyyyMMdd");
Date convertedDate = null;
try {
convertedDate = (Date) df.parse(date);
} catch (ParseException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
return convertedDate;
}
恢复组标识符的每个不同值
private static List DistinctValues (ArrayList<Personne> listPersonnes)
{
List outList = (List) listPersonnes.stream().map(m -> m.getIDT_GCT()).distinct().collect(Collectors.toList());
return(outList);
}
最后按照条件选择CD_NOT：
private static String SelectionCodeNote(ArrayList<Personne> listPersonnes) {
if(listPersonnes != null){
List IDT_CGCT_DISTINCT = (List) DistinctValues(listPersonnes);
ArrayList<Personne> usedList = new ArrayList<Personne>();
for(int i=0; i<IDT_CGCT_DISTINCT.size();i++)
{
String IDT_GCT_Local = (String) IDT_CGCT_DISTINCT.get(i);
usedList.stream().filter(c -> c.getIDT_GCT().equals(IDT_GCT_Local));
for(Personne personne: usedList)
{
if (personne.getIDC_PSE_PCL().equals("1") && personne.getIDC_CD_NOT().equals("0")) {
return (personne.getCD_NOT());
} else {
Collections.sort(usedList);
return (personne.getCD_NOT());
}
}
}
}
return null;
}
这是作为猪脚本输出感知的一个例子：
011274144,{()}
011274145,{()}
011274146,{()}
011274148,{()}
011274149,{()}
011274150,{()}
011274162,{()}
011274167,{()}
011274178,{()}
011274197,{()}
011274231,{()}
您可以看到第二列为空（SelectionCodeNote的输出 ）
非常感谢您