spark-monog(2 数据读写)

原文：https://blog.csdn.net/u011098327/article/details/72865934

依赖：

<dependency>
            <groupId>org.mongodb.spark</groupId>
            <artifactId>mongo-spark-connector_2.11</artifactId>
            <version>${mongodb-spark.version}</version>
        </dependency>
<dependency>
            <groupId>org.apache.spark</groupId>
            <artifactId>spark-core_2.11</artifactId>
  </dependency>

<dependency>
    <groupId>org.apache.spark</groupId>
    <artifactId>spark-sql_2.11</artifactId>
</dependency>

<spark.version>2.1.1</spark.version> <mongodb-spark.version>2.0.0</mongodb-spark.version>

spark2.x向mongodb中读取写入数据，读取写入相关参数参考 https://docs.mongodb.com/spark-connector/current/configuration/#cache-configuration

从mongodb中读取数据时指定数据分区字段，分区大小提高读取效率， 当需要过滤部分数据集的情况下使用Dataset/SQL的方式filter，Mongo Connector会创建aggregation pipeline在mongodb端进行过滤，

然后再传回给spark进行优化处理

   val spark = SparkSession
      .builder
      .appName("AppMongo")
      .master("local[*]")
      .config("spark.worker.cleanup.enabled", "true")
      .config("spark.scheduler.mode", "FAIR")
      .getOrCreate()
    val inputUri="mongodb://test:pwd123456@192.168.0.1:27017/test.articles"
    val df = spark.read.format("com.mongodb.spark.sql").options(
      Map("spark.mongodb.input.uri" -> inputUri,
        "spark.mongodb.input.partitioner" -> "MongoPaginateBySizePartitioner",
        "spark.mongodb.input.partitionerOptions.partitionKey"  -> "_id",
        "spark.mongodb.input.partitionerOptions.partitionSizeMB"-> "32"))
      .load()
    val currentTimestamp = System.currentTimeMillis()
    val originDf = df.filter(df("updateTime") < currentTimestamp && df("updateTime") >= currentTimestamp - 1440 * 60 * 1000)
      .select("_id", "content", "imgTotalCount").toDF("id", "content", "imgnum")

val rdd = MongoSpark.load(sc)
val aggregatedRdd = rdd.withPipeline(Seq(Document.parse("{ $match: { test : { $gt : 5 } } }")))
println(aggregatedRdd.count)
println(aggregatedRdd.first.toJson)

spark操作mongodb的scala-api文档：https://docs.mongodb.com/spark-connector/current/scala-api/

向mongo里面写数据可以使用两种不同的方式mode=overwrite,append
overwirite 以覆盖的方式写入
append    以追加的方式写入

al outputUri="mongodb://test:pwd123456@192.168.0.1:27017/test.article_garbage" 
saveDF.write.options(Map("spark.mongodb.output.uri"-> outputUri))
          .mode("append")
          .format("com.mongodb.spark.sql")
          .save()