fix bug rtdmp_normal

src/main/scala/mobvista/dmp/datasource/dm/RtdmpNormal.scala

@@ -121,6 +121,10 @@ class RtdmpNormal extends CommonSparkJob with Serializable {
         val package_name: String = array(index)._4
         val country_code: String = array(index)._5
         println(inputPath)
+        val pathUri = new URI(inputPath)
+        //过滤后面这种不存在的s3路径  s3://mob-emr-test/dataplatform/rtdmp_request/2021/07/10/dsp_req/com.taobao.idlefish_bes/*/,
+        if (FileSystem.get(new URI(s"${pathUri.getScheme}://${pathUri.getHost}"), sc.hadoopConfiguration)
+          .exists(new Path(pathUri.toString.replace("*", "")))){
             inputDataRdd = inputDataRdd.union(spark.sparkContext.textFile(inputPath).map(row => {
               if (row.length == 32) {
                 DmpDailyDataInformation(row, device_type_md5, platform, package_name, country_code)
@@ -130,6 +134,11 @@ class RtdmpNormal extends CommonSparkJob with Serializable {
               }
             }
         ))
+        }else{
+            println(inputPath+" not existed!")
+            inputDataRdd = inputDataRdd.union(spark.sparkContext.emptyRDD[DmpDailyDataInformation])
+        }
+
       }
 
       val df: DataFrame = inputDataRdd.toDF().persist(StorageLevel.MEMORY_AND_DISK_SER)