|
|
@@ -0,0 +1,243 @@
|
|
|
+package com.aliyun.odps.spark.examples.makedata
|
|
|
+
|
|
|
+
|
|
|
+import com.aliyun.odps.TableSchema
|
|
|
+import com.aliyun.odps.data.Record
|
|
|
+import com.aliyun.odps.spark.examples.myUtils.{MyDateUtils, MyHdfsUtils, ParamUtils}
|
|
|
+import examples.dataloader.{OfflineVlogShareLRFeatureExtractorV1, OfflineVlogShareLRFeatureExtractorV2}
|
|
|
+import org.apache.hadoop.io.compress.GzipCodec
|
|
|
+import org.apache.spark.sql.SparkSession
|
|
|
+
|
|
|
+import java.util
|
|
|
+import scala.collection.JavaConversions._
|
|
|
+import scala.collection.mutable.ArrayBuffer
|
|
|
+import com.alibaba.fastjson.JSON
|
|
|
+import com.alibaba.fastjson.JSONObject
|
|
|
+
|
|
|
+
|
|
|
+object makedata_07_rosData {
|
|
|
+ def main(args: Array[String]) {
|
|
|
+ val spark = SparkSession
|
|
|
+ .builder()
|
|
|
+ .appName(this.getClass.getName)
|
|
|
+ .getOrCreate()
|
|
|
+ val sc = spark.sparkContext
|
|
|
+
|
|
|
+ // 1 读取参数
|
|
|
+ val param = ParamUtils.parseArgs(args)
|
|
|
+ val partitionPrefix = param.getOrElse("partitionPrefix", "dt=")
|
|
|
+ val beginStr = param.getOrElse("beginStr", "20230101")
|
|
|
+ val endStr = param.getOrElse("endStr", "20230101")
|
|
|
+ val readPath = param.getOrElse("readPath", "/dw/recommend/model/00_sample_data/")
|
|
|
+ val savePath = param.getOrElse("savePath", "/dw/recommend/model/01_str_data/")
|
|
|
+ val featureVersion = param.getOrElse("featureVersion", "v2")
|
|
|
+ val ifRepart = param.getOrElse("ifRepart", "100").toInt
|
|
|
+ val labelVersion = param.getOrElse("labelVersion", "v1")
|
|
|
+
|
|
|
+
|
|
|
+
|
|
|
+ // 3 循环执行数据生产
|
|
|
+ val dateRange = MyDateUtils.getDateRange(beginStr, endStr)
|
|
|
+ for (date <- dateRange) {
|
|
|
+ val partition = partitionPrefix + date
|
|
|
+ println("执行partiton:" + partition)
|
|
|
+ var hdfsPath = readPath + "/" + partition
|
|
|
+
|
|
|
+ // 4 过滤保留分享样本
|
|
|
+ val dataFilter = sc.textFile(hdfsPath).map(r=>{
|
|
|
+ val rList = r.split("\t")
|
|
|
+ val logKeyStr = rList(0)
|
|
|
+ val labelStr = rList(1)
|
|
|
+ val feaStr = rList(2)
|
|
|
+ val logTs = logKeyStr.split(":")(2)
|
|
|
+ val labelJson = JSON.parseObject(labelStr)
|
|
|
+ val feaJson = JSON.parseObject(feaStr)
|
|
|
+ val is_share = if (labelJson.containsKey("is_share")) labelJson.getString("is_share") else "0"
|
|
|
+ (logTs, feaJson, labelJson, is_share)
|
|
|
+ }).filter(_._4.equals("1"))
|
|
|
+
|
|
|
+ // 5 label处理
|
|
|
+ val dataTrain = labelVersion match {
|
|
|
+ case "v2" => dataFilter.flatMap({
|
|
|
+ case (logTs, feaJson, labelJson, _) =>
|
|
|
+ val res = ArrayBuffer[(String, JSONObject)]()
|
|
|
+ val hour = feaJson.getString("ctx_hour").toInt
|
|
|
+ val expTs = logTs.toLong / 1000
|
|
|
+ hour match {
|
|
|
+ case 23 => res
|
|
|
+ case _ =>
|
|
|
+ res.add(("0", feaJson))
|
|
|
+ val is_return = if (labelJson.containsKey("is_return")) labelJson.getString("is_return") else "0"
|
|
|
+ if ("1".equals(is_return)) {
|
|
|
+ if (labelJson.containsKey("return_mid_ts_list")){
|
|
|
+ labelJson.getString("return_mid_ts_list").split(",")
|
|
|
+ .map(r => r.split(":")(1).toLong / 1000)
|
|
|
+ .foreach(ts => {
|
|
|
+ if (ts - expTs < 3600) {
|
|
|
+ res.add(("1", feaJson))
|
|
|
+ }
|
|
|
+ })
|
|
|
+ }
|
|
|
+ }
|
|
|
+ res
|
|
|
+ }
|
|
|
+ })
|
|
|
+ case _ => dataFilter.map({
|
|
|
+ case (logTs, feaJson, labelJson, _) =>
|
|
|
+ val is_return = if (labelJson.containsKey("is_return")) labelJson.getString("is_return") else "0"
|
|
|
+ (is_return, feaJson)
|
|
|
+ })
|
|
|
+ }
|
|
|
+ // 6 特征选择
|
|
|
+ val data = dataTrain.map{
|
|
|
+ case (is_return, feaJson) =>
|
|
|
+ if ("v1".equals(featureVersion)) {
|
|
|
+ val feaSet = Set(
|
|
|
+ "ctx_week", "ctx_hour", "ctx_region", "ctx_city",
|
|
|
+ "machineinfo_brand", "machineinfo_model", "machineinfo_platform", "machineinfo_system",
|
|
|
+ "u_1day_exp_cnt", "u_1day_click_cnt", "u_1day_share_cnt", "u_1day_return_cnt",
|
|
|
+ "u_3day_exp_cnt", "u_3day_click_cnt", "u_3day_share_cnt", "u_3day_return_cnt",
|
|
|
+ "total_time", "play_count_total",
|
|
|
+ "i_1day_exp_cnt", "i_1day_click_cnt", "i_1day_share_cnt", "i_1day_return_cnt",
|
|
|
+ "i_3day_exp_cnt", "i_3day_click_cnt", "i_3day_share_cnt", "i_3day_return_cnt",
|
|
|
+ "u_1day_ctr", "u_1day_str", "u_1day_rov", "u_1day_ros",
|
|
|
+ "u_3day_ctr", "u_3day_str", "u_3day_rov", "u_3day_ros",
|
|
|
+ "i_1day_ctr", "i_1day_str", "i_1day_rov", "i_1day_ros",
|
|
|
+ "i_3day_ctr", "i_3day_str", "i_3day_rov", "i_3day_ros"
|
|
|
+ )
|
|
|
+ val feaMap = new util.HashMap[String, String]()
|
|
|
+ feaSet.foreach(r => {
|
|
|
+ if (feaJson.containsKey(r)) {
|
|
|
+ feaMap.put(r, feaJson.getString(r))
|
|
|
+ }
|
|
|
+ })
|
|
|
+ val bytesFeatureExtractor = new OfflineVlogShareLRFeatureExtractorV1()
|
|
|
+ bytesFeatureExtractor.makeFeature4String(feaMap)
|
|
|
+ val featureMap = bytesFeatureExtractor.featureMap
|
|
|
+ (is_return, featureMap.entries().map(r => r.getValue.getIdentifier + ":1").mkString("\t"))
|
|
|
+
|
|
|
+ } else if ("v2".equals(featureVersion)) {
|
|
|
+ val feaSet = Set(
|
|
|
+ "ctx_week", "ctx_hour", "ctx_region", "ctx_city",
|
|
|
+ "machineinfo_brand", "machineinfo_model", "machineinfo_platform", "machineinfo_system",
|
|
|
+ "u_1day_exp_cnt", "u_1day_click_cnt", "u_1day_share_cnt", "u_1day_return_cnt",
|
|
|
+ "u_3day_exp_cnt", "u_3day_click_cnt", "u_3day_share_cnt", "u_3day_return_cnt",
|
|
|
+ "total_time", "play_count_total",
|
|
|
+ "i_1day_exp_cnt", "i_1day_click_cnt", "i_1day_share_cnt", "i_1day_return_cnt",
|
|
|
+ "i_3day_exp_cnt", "i_3day_click_cnt", "i_3day_share_cnt", "i_3day_return_cnt",
|
|
|
+ "u_1day_ctr", "u_1day_str", "u_1day_rov", "u_1day_ros",
|
|
|
+ "u_3day_ctr", "u_3day_str", "u_3day_rov", "u_3day_ros",
|
|
|
+ "i_1day_ctr", "i_1day_str", "i_1day_rov", "i_1day_ros",
|
|
|
+ "i_3day_ctr", "i_3day_str", "i_3day_rov", "i_3day_ros",
|
|
|
+
|
|
|
+ // "share_uv_list_1day_6_avg", "share_uv_list_1day_6_var", "share_uv_list_1day_diff_6_avg", "share_uv_list_1day_diff_6_var",
|
|
|
+ // "return_uv_list_1day_6_avg", "return_uv_list_1day_6_var", "return_uv_list_1day_diff_6_avg", "return_uv_list_1day_diff_6_var",
|
|
|
+ // "share_uv_list_1h_6_avg", "share_uv_list_1h_6_var", "share_uv_list_1h_diff_6_avg", "share_uv_list_1h_diff_6_var",
|
|
|
+ // "return_uv_list_1h_6_avg", "return_uv_list_1h_6_var", "return_uv_list_1h_diff_6_avg", "return_uv_list_1h_diff_6_var",
|
|
|
+
|
|
|
+ "view_pv_list_1day", "view_uv_list_1day", "play_pv_list_1day", "play_uv_list_1day",
|
|
|
+ "share_pv_list_1day", "share_uv_list_1day", "return_uv_list_1day",
|
|
|
+ "p_view_uv_list_1day", "p_view_pv_list_1day", "p_return_uv_list_1day",
|
|
|
+ "share_uv_list_2day", "share_pv_list_2day", "share_uv_list_3day", "share_pv_list_3day",
|
|
|
+
|
|
|
+ "view_uv_list_1h", "view_pv_list_1h", "play_uv_list_1h", "play_pv_list_1h",
|
|
|
+ "share_uv_list_1h", "share_pv_list_1h", "return_uv_list_1h", "p_return_uv_list_1h",
|
|
|
+
|
|
|
+ "i_1day_ctr_rt", "i_1day_str_rt", "i_1day_ros_rt", "i_1day_rov_rt",
|
|
|
+ "i_1h_ctr_rt", "i_1h_str_rt", "i_1h_ros_rt", "i_1h_rov_rt"
|
|
|
+ )
|
|
|
+ val feaMap = new util.HashMap[String, String]()
|
|
|
+ feaSet.foreach(r => {
|
|
|
+ if (feaJson.containsKey(r)) {
|
|
|
+ feaMap.put(r, feaJson.getString(r))
|
|
|
+ }
|
|
|
+ })
|
|
|
+ val bytesFeatureExtractor = new OfflineVlogShareLRFeatureExtractorV2()
|
|
|
+ bytesFeatureExtractor.makeFeature4String(feaMap)
|
|
|
+ val featureMap = bytesFeatureExtractor.featureMap
|
|
|
+ (is_return, featureMap.entries().map(r => r.getValue.getIdentifier + ":1").mkString("\t"))
|
|
|
+ } else if ("v4".equals(featureVersion)) {
|
|
|
+ val feaSet = Set(
|
|
|
+ "ctx_week", "ctx_hour", "ctx_region", "ctx_city",
|
|
|
+ "machineinfo_brand", "machineinfo_model", "machineinfo_platform", "machineinfo_system",
|
|
|
+ "u_1day_exp_cnt", "u_1day_click_cnt", "u_1day_share_cnt", "u_1day_return_cnt",
|
|
|
+ "u_3day_exp_cnt", "u_3day_click_cnt", "u_3day_share_cnt", "u_3day_return_cnt",
|
|
|
+ "total_time", "play_count_total",
|
|
|
+ "i_1day_exp_cnt", "i_1day_click_cnt", "i_1day_share_cnt", "i_1day_return_cnt",
|
|
|
+ "i_3day_exp_cnt", "i_3day_click_cnt", "i_3day_share_cnt", "i_3day_return_cnt",
|
|
|
+ "u_1day_ctr", "u_1day_str", "u_1day_rov", "u_1day_ros",
|
|
|
+ "u_3day_ctr", "u_3day_str", "u_3day_rov", "u_3day_ros",
|
|
|
+ "i_1day_ctr", "i_1day_str", "i_1day_rov", "i_1day_ros",
|
|
|
+ "i_3day_ctr", "i_3day_str", "i_3day_rov", "i_3day_ros",
|
|
|
+
|
|
|
+ "i_1day_ctr_rt", "i_1day_str_rt", "i_1day_ros_rt", "i_1day_rov_rt",
|
|
|
+ "i_1h_ctr_rt", "i_1h_str_rt", "i_1h_ros_rt", "i_1h_rov_rt"
|
|
|
+ )
|
|
|
+ val feaMap = new util.HashMap[String, String]()
|
|
|
+ feaSet.foreach(r => {
|
|
|
+ if (feaJson.containsKey(r)) {
|
|
|
+ feaMap.put(r, feaJson.getString(r))
|
|
|
+ }
|
|
|
+ })
|
|
|
+ val bytesFeatureExtractor = new OfflineVlogShareLRFeatureExtractorV2()
|
|
|
+ bytesFeatureExtractor.makeFeature4String(feaMap)
|
|
|
+ val featureMap = bytesFeatureExtractor.featureMap
|
|
|
+ (is_return, featureMap.entries().map(r => r.getValue.getIdentifier + ":1").mkString("\t"))
|
|
|
+ } else if ("v5".equals(featureVersion)) {
|
|
|
+ val feaSet = Set(
|
|
|
+ "ctx_week", "ctx_hour", "ctx_region", "ctx_city",
|
|
|
+ "machineinfo_brand", "machineinfo_model", "machineinfo_platform", "machineinfo_system",
|
|
|
+ "u_1day_exp_cnt", "u_1day_click_cnt", "u_1day_share_cnt", "u_1day_return_cnt",
|
|
|
+ "u_3day_exp_cnt", "u_3day_click_cnt", "u_3day_share_cnt", "u_3day_return_cnt",
|
|
|
+ "total_time", "play_count_total",
|
|
|
+ "i_1day_exp_cnt", "i_1day_click_cnt", "i_1day_share_cnt", "i_1day_return_cnt",
|
|
|
+ "i_3day_exp_cnt", "i_3day_click_cnt", "i_3day_share_cnt", "i_3day_return_cnt",
|
|
|
+ "u_1day_ctr", "u_1day_str", "u_1day_rov", "u_1day_ros",
|
|
|
+ "u_3day_ctr", "u_3day_str", "u_3day_rov", "u_3day_ros",
|
|
|
+ "i_1day_ctr", "i_1day_str", "i_1day_rov", "i_1day_ros",
|
|
|
+ "i_3day_ctr", "i_3day_str", "i_3day_rov", "i_3day_ros",
|
|
|
+
|
|
|
+ "view_pv_list_1day", "view_uv_list_1day", "play_pv_list_1day", "play_uv_list_1day",
|
|
|
+ "share_pv_list_1day", "share_uv_list_1day", "return_uv_list_1day",
|
|
|
+ "p_view_uv_list_1day", "p_view_pv_list_1day", "p_return_uv_list_1day",
|
|
|
+ "share_uv_list_2day", "share_pv_list_2day", "share_uv_list_3day", "share_pv_list_3day",
|
|
|
+
|
|
|
+ "view_uv_list_1h", "view_pv_list_1h", "play_uv_list_1h", "play_pv_list_1h",
|
|
|
+ "share_uv_list_1h", "share_pv_list_1h", "return_uv_list_1h", "p_return_uv_list_1h"
|
|
|
+ )
|
|
|
+ val feaMap = new util.HashMap[String, String]()
|
|
|
+ feaSet.foreach(r => {
|
|
|
+ if (feaJson.containsKey(r)) {
|
|
|
+ feaMap.put(r, feaJson.getString(r))
|
|
|
+ }
|
|
|
+ })
|
|
|
+ val bytesFeatureExtractor = new OfflineVlogShareLRFeatureExtractorV2()
|
|
|
+ bytesFeatureExtractor.makeFeature4String(feaMap)
|
|
|
+ val featureMap = bytesFeatureExtractor.featureMap
|
|
|
+ (is_return, featureMap.entries().map(r => r.getValue.getIdentifier + ":1").mkString("\t"))
|
|
|
+ } else {
|
|
|
+ (is_return, "")
|
|
|
+ }
|
|
|
+ }.filter(_._2.nonEmpty).map(r=> r._1 + "\t" + r._2)
|
|
|
+
|
|
|
+ // 7 保存数据到hdfs
|
|
|
+ hdfsPath = savePath + "/" + partition
|
|
|
+ if (hdfsPath.nonEmpty && hdfsPath.startsWith("/dw/recommend/model/")){
|
|
|
+ println("删除路径并开始数据写入:" + hdfsPath)
|
|
|
+ MyHdfsUtils.delete_hdfs_path(hdfsPath)
|
|
|
+ if (ifRepart == 0){
|
|
|
+ data.saveAsTextFile(hdfsPath, classOf[GzipCodec])
|
|
|
+ }else{
|
|
|
+ data.repartition(ifRepart).saveAsTextFile(hdfsPath, classOf[GzipCodec])
|
|
|
+ }
|
|
|
+ }else{
|
|
|
+ println("路径不合法，无法写入:" + hdfsPath)
|
|
|
+ }
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ def func(record: Record, schema: TableSchema): Record = {
|
|
|
+ record
|
|
|
+ }
|
|
|
+
|
|
|
+}
|
zhangbo