xueyiming 2 viikkoa sitten
vanhempi
commit
73257f8d13

+ 5 - 3
src/main/scala/com/aliyun/odps/spark/examples/makedata_ad/v20240718/makedata_ad_33_addFeatureToHive_20250708.scala

@@ -17,8 +17,8 @@ object makedata_ad_33_addFeatureToHive_20250708 {
     val project = param.getOrElse("project", "loghubods")
     val tablePart = param.getOrElse("tablePart", "64").toInt
     val beginStr = param.getOrElse("beginStr", "20250708")
-    val endStr = param.getOrElse("endStr", "20250708")
-    val input = param.getOrElse("input", "ad_easyrec_train_data_v3_sampled")
+    val endStr = param.getOrElse("endStr", "20250516")
+    val input = param.getOrElse("input", "ad_easyrec_train_data_v3_sampled_backup")
 
     val odpsOps = env.getODPS(sc)
     val dateRange = MyDateUtils.getDateRange(beginStr, endStr)
@@ -33,11 +33,13 @@ object makedata_ad_33_addFeatureToHive_20250708 {
         numPartition = tablePart
       )
 
+      val beforeDt = MyDateUtils.getPreviousDay(dt)
+      val partition1 = s"dt=$beforeDt"
       // 读取特征表,并按 mid 去重(保留最后一条)
       val featureRdd = odpsOps.readTable(
         project = project,
         table = "user_conver_ad_class_temp",
-        partition = partition,
+        partition = partition1,
         transfer = func,
         numPartition = tablePart
       )

+ 10 - 0
src/main/scala/com/aliyun/odps/spark/examples/myUtils/MyDateUtils.scala

@@ -222,6 +222,16 @@ object MyDateUtils {
     days
   }
 
+  // 计算指定日期前一天的日期
+  def getPreviousDay(dateStr: String, format: String = "yyyyMMdd"): String = {
+    val sdf = new SimpleDateFormat(format)
+    val date = sdf.parse(dateStr)
+    val calendar = Calendar.getInstance()
+    calendar.setTime(date)
+    calendar.add(Calendar.DAY_OF_YEAR, -1)
+    sdf.format(calendar.getTime)
+  }
+
   def main(args: Array[String]): Unit = {
 //    var from = DateUtils.parseDate("2019-09-01", Array[String]("yyyy-MM-dd"))
 //    var to = DateUtils.parseDate("2019-09-10", Array[String]("yyyy-MM-dd"))