1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
package mobvista.dmp.datasource.age_gender
import java.net.URI
import java.util.regex.Pattern
import mobvista.dmp.common.CommonSparkJob
import mobvista.dmp.format.TextMultipleOutputFormat
import org.apache.commons.cli.Options
import org.apache.hadoop.fs.{FileSystem, Path}
import org.apache.hadoop.io.Text
import org.apache.spark.sql.{Row, SaveMode, SparkSession}
import org.apache.spark.storage.StorageLevel
import scala.collection.mutable.ArrayBuffer
class MergeInstallGenderLR extends CommonSparkJob with Serializable {
private val wellSplit: Pattern = Pattern.compile("#")
val TAB_DELIMITER = "\t"
override protected def run(args: Array[String]): Int = {
val commandLine = commParser.parse(options, args)
if (!checkMustOption(commandLine)) {
printUsage(options)
return -1
} else {
printOptions(commandLine)
}
val gender_output = commandLine.getOptionValue("gender_output")
val gender_bin_output = commandLine.getOptionValue("gender_bin_output")
val org_gender_bin_output = commandLine.getOptionValue("org_gender_bin_output")
val calc_gender_bin_output = commandLine.getOptionValue("calc_gender_bin_output")
val date = commandLine.getOptionValue("date")
val dt_yesterday = commandLine.getOptionValue("dt_yesterday")
val ga_date = commandLine.getOptionValue("ga_date")
val other_date = commandLine.getOptionValue("other_date")
val parallelism = commandLine.getOptionValue("parallelism")
val spark = SparkSession.builder()
.appName("MergeInstallGenderLR")
.config("spark.rdd.compress", "true")
.config("spark.io.compression.codec", "snappy")
.config("spark.sql.broadcastTimeout", "2400")
.config("spark.sql.autoBroadcastJoinThreshold", "209715200")
.config("spark.sql.orc.filterPushdown", "true")
.config("spark.sql.warehouse.dir", "s3://mob-emr-test/spark-warehouse")
.config("spark.serializer", "org.apache.spark.serializer.KryoSerializer")
.enableHiveSupport()
.getOrCreate()
val sc = spark.sparkContext
FileSystem.get(new URI(s"s3://mob-emr-test"), sc.hadoopConfiguration).delete(new Path(gender_output), true)
FileSystem.get(new URI(s"s3://mob-emr-test"), sc.hadoopConfiguration).delete(new Path(gender_bin_output), true)
FileSystem.get(new URI(s"s3://mob-emr-test"), spark.sparkContext.hadoopConfiguration).delete(new Path(org_gender_bin_output), true)
FileSystem.get(new URI(s"s3://mob-emr-test"), spark.sparkContext.hadoopConfiguration).delete(new Path(calc_gender_bin_output), true)
try {
spark.udf.register("check_deviceId", mobvista.dmp.common.MobvistaConstant.checkDeviceId _)
val ods_gender_sql = Constant.ods_gender_sql.replace("@date", date).replace("@dt_yesterday", dt_yesterday)
// .replace("@check_deviceId", "check_deviceId(device_id)")
spark.sql(ods_gender_sql).createOrReplaceTempView("ods_gender_tab")
val install_list_v2_gender = Constant.dmp_install_list_gender_lr_sql.replace("@date", date)
.replace("@ga_date", ga_date)
.replace("@other_date", other_date)
// .replace("@check_deviceId", "check_deviceId(device_id)")
spark.sql(install_list_v2_gender).createOrReplaceTempView("dmp_install_list_v2")
val sql = "select t1.device_id,t1.device_type,t1.label,t2.package_names,case when t1.label in('m','f') then 'org' when t2.package_names is not null and t1.label not in('m','f') then 'calc' else 'none' end as label_type " +
"from ods_gender_tab t1 join dmp_install_list_v2 t2 on(lower(t1.device_id) = lower(t2.device_id) and t1.device_type = t2.device_type)" +
"group by t1.device_id,t1.device_type,t1.label,t2.package_names"
val dmp_device_gender = spark.sql(sql).persist(StorageLevel.MEMORY_AND_DISK)
dmp_device_gender.coalesce(parallelism.toInt)
.write.mode(SaveMode.Overwrite)
.option("orc.compress", "zlib")
.orc(gender_output)
dmp_device_gender.filter(line => !"none".equalsIgnoreCase(line.getAs[String]("label_type")))
.rdd
.flatMap(buildResult(_, gender_bin_output))
.coalesce(parallelism.toInt, true)
.saveAsNewAPIHadoopFile(org_gender_bin_output, classOf[Text], classOf[Text], classOf[TextMultipleOutputFormat], sc.hadoopConfiguration)
/* spark.sql(sql).persist(StorageLevel.MEMORY_AND_DISK).createOrReplaceTempView("dmp_device_gender")
val org_sql = "select device_id,device_type,label,package_names,label_type from dmp_device_gender"
val bin_sql = "select device_id,device_type,label,package_names,label_type from dmp_device_gender where label_type != 'none' "
val org_rdd = spark.sql(org_sql).rdd.map(line => {
val device_id = line.getAs[String]("device_id")
val device_type = line.getAs[String]("device_type")
val label = line.getAs[String]("label")
val package_names = line.getAs[String]("package_names")
val label_type = line.getAs[String]("label_type")
Row(device_id,device_type,label,package_names,label_type)
})
spark.createDataFrame(org_rdd, Constant.schema_gender_lr)
.coalesce(parallelism.toInt)
.write.mode(SaveMode.Overwrite)
.option("orc.compress", "zlib")
.orc(gender_output)
spark.sql(bin_sql)
.rdd
.flatMap(buildResult(_, gender_bin_output))
.coalesce(parallelism.toInt, true)
.saveAsNewAPIHadoopFile(org_gender_bin_output, classOf[Text], classOf[Text], classOf[TextMultipleOutputFormat], sc.hadoopConfiguration)
*/
} finally {
sc.stop()
spark.stop()
}
0
}
def buildResult(row: Row, outputPrefix: String): Array[Tuple2[Text, Text]] = {
val device_id = row.getAs[String]("device_id")
val device_type = row.getAs[String]("device_type")
val label = row.getAs[String]("label")
val package_names = row.getAs[String]("package_names")
val label_type = row.getAs[String]("label_type")
label_type match {
case "calc" => { //predict数据
val buf = new StringBuilder
buf ++= device_id
if ("idfa".equalsIgnoreCase(device_type)) {
buf ++= "#A\002"
} else {
buf ++= "#B\002"
}
buf ++= package_names.replace("#", "\001")
buf ++= "\0021"
val buffer = new ArrayBuffer[Tuple2[Text, Text]]()
buffer += Tuple2(new Text(s"${outputPrefix}/${label_type}, "), new Text(buf.toString()))
buffer.toArray
/* device_type match {
case "idfa" => {
buf ++= "#A\\002"
buf ++= package_names.replace("#","\\001")
buf ++= "\\0021"
val buffer = new ArrayBuffer[Tuple2[Text, Text]]()
buffer += Tuple2(new Text(s"${outputPrefix}/${label_type}, "), new Text(buf.toString()))
buffer.toArray
}
case _ =>{
buf ++= "#B\\002"
buf ++= package_names.replace("#","\\001")
buf ++= "\\0021"
val buffer = new ArrayBuffer[Tuple2[Text, Text]]()
buffer += Tuple2(new Text(s"${outputPrefix}/${label_type}, "), new Text(buf.toString()))
buffer.toArray
}
}*/
}
case "org" => { //train数据 只有m f
val buf = new StringBuilder
buf ++= "0\002"
val bufferArray = new ArrayBuffer[Tuple2[Text, Text]]()
if ("m".equalsIgnoreCase(label)) {
buf ++= "1\002"
} else {
buf ++= "0\002"
}
buf ++= package_names.replace("#", "\001")
buf ++= "\0021"
bufferArray += Tuple2(new Text(s"${outputPrefix}/${label_type}, "), new Text(buf.toString()))
bufferArray.toArray
/*
label match {
case "m" =>{
buf ++= "1\\002"
buf ++= package_names.replace("#","\\001")
buf ++= "\\0021"
bufferArray += Tuple2(new Text(s"${outputPrefix}/${label_type}, "), new Text(buf.toString()))
bufferArray.toArray
}
case "f" =>{
buf ++="0\\002"
buf ++= package_names.replace("#","\\001")
buf ++= "\\0021"
bufferArray += Tuple2(new Text(s"${outputPrefix}/${label_type}, "), new Text(buf.toString()))
bufferArray.toArray
}
*/
}
}
}
override protected def buildOptions(): Options = {
val options = new Options
options.addOption("gender_output", true, "[must] gender_output")
options.addOption("gender_bin_output", true, "[must] gender_bin_output")
options.addOption("org_gender_bin_output", true, "[must] org_gender_bin_output")
options.addOption("calc_gender_bin_output", true, "[must] calc_gender_bin_output")
options.addOption("date", true, "[must] date")
options.addOption("dt_yesterday", true, "[must] dt_yesterday")
options.addOption("ga_date", true, "[must] ga_date")
options.addOption("other_date", true, "[must] other_date")
options.addOption("parallelism", true, "[must] parallelism")
options
}
}
object MergeInstallGenderLR {
def main(args: Array[String]): Unit = {
new MergeInstallGenderLR().run(args)
}
}