Spark SQL大数据处理并写入Elasticsearch

SparkSQL(Spark用于处理结构化数据的模块)

通过SparkSQL导入的数据可以来自MySQL数据库、Json数据、Csv数据等，通过load这些数据可以对其做一系列计算

下面通过程序代码来详细查看SparkSQL导入数据并写入到ES中：

数据集：北京市PM2.5数据

Spark版本：2.3.2

Python版本：3.5.2

mysql-connector-java-8.0.11 下载

ElasticSearch：6.4.1

Kibana：6.4.1

elasticsearch-spark-20_2.11-6.4.1.jar 下载

具体代码：

# coding: utf-8
import sys
import os

pre_current_dir = os.path.dirname(os.getcwd())
sys.path.append(pre_current_dir)
from pyspark.sql import SparkSession
from pyspark.sql.types import *
from pyspark.sql.functions import udf
from settings import ES_CONF

current_dir = os.path.dirname(os.path.realpath(__file__))

spark = SparkSession.builder.appName("weather_result").getOrCreate()


def get_health_level(value):
    """
    PM2.5对应健康级别
    :param value:
    :return:
    """
    if 0 <= value <= 50:
        return "Very Good"
    elif 50 < value <= 100:
        return "Good"
    elif 100 < value <= 150:
        return "Unhealthy for Sensi"
    elif value <= 200:
        return "Unhealthy"
    elif 200 < value <= 300:
        return "Very Unhealthy"
    elif 300 < value <= 500:
        return "Hazardous"
    elif value > 500:
        return "Extreme danger"
    else:
        return None


def get_weather_result():
    """
    获取Spark SQL分析后的数据
    :return:
    """
    # load所需字段的数据到DF
    df_2017 = spark.read.format("csv") 
        .option("header", "true") 
        .option("inferSchema", "true") 
        .load("file://{}/data/Beijing2017_PM25.csv".format(current_dir)) 
        .select("Year", "Month", "Day", "Hour", "Value", "QC Name")

    # 查看Schema
    df_2017.printSchema()

    # 通过udf将字符型health_level转换为column
    level_function_udf = udf(get_health_level, StringType())

    # 新建列healthy_level 并healthy_level分组
    group_2017 = df_2017.withColumn(
        "healthy_level", level_function_udf(df_2017['Value'])
    ).groupBy("healthy_level").count()

    # 新建列days和percentage 并计算它们对应的值
    result_2017 = group_2017.select("healthy_level", "count") 
        .withColumn("days", group_2017['count'] / 24) 
        .withColumn("percentage", group_2017['count'] / df_2017.count())
    result_2017.show()

    return result_2017


def write_result_es():
    """
    将SparkSQL计算结果写入到ES
    :return:
    """
    result_2017 = get_weather_result()
    # ES_CONF配置 ES的node和index
    result_2017.write.format("org.elasticsearch.spark.sql") 
        .option("es.nodes", "{}".format(ES_CONF['ELASTIC_HOST'])) 
        .mode("overwrite") 
        .save("{}/pm_value".format(ES_CONF['WEATHER_INDEX_NAME']))


write_result_es()
spark.stop()

将mysql-connector-java-8.0.11和elasticsearch-spark-20_2.11-6.4.1.jar放到Spark的jars目录下，提交spark任务即可。

注意:

(1) 如果提示：ClassNotFoundException Failed to find data source: org.elasticsearch.spark.sql.，则表示spark没有发现jar包，此时需重新编译pyspark：

cd /opt/spark-2.3.2-bin-hadoop2.7/python 
python3 setup.py sdist 
pip install dist/*.tar.gz

 (2) 如果提示：Multiple ES-Hadoop versions detected in the classpath; please use only one ，

　　则表示ES-Hadoop jar包有多余的,可能既有elasticsearch-hadoop，又有elasticsearch-spark，此时删除多余的jar包，重新编译pyspark 即可

执行效果：

更多源码请关注我的github， https://github.com/a342058040/Spark-for-Python ，Spark相关技术全程用python实现，持续更新