1. 获取SparkSession
spark = SparkSession.builder.config(conf = SparkConf()).getOrCreate()
2. 获取SparkContext
1. 获取sparkSession: se = SparkSession.builder.config(conf = SparkConf()).getOrCreate()
1. 获取sparkContext: sc = se.sparkContext
2. 获取sqlContext: sq = SparkSession.builder.getOrCreate()
3. 获取DataFrame: df = sqlContext.createDataFrame(userRows)
3. 读取文件
line1 = sc.textFile("hdfs://192.168.88.128:9000/hello.txt")
rawData = sc.textFile("hdfs://192.168.88.128:9000/data/sanxi/sanxi/*.gz") 获取sanxi文件夹下所有.gz的文件
rawData = sc.textFile("file:///data/sanxi2/*.gz") spark 读取本地文件
4. filter 使用方法
1. 过滤包含指定字符的RDD
line2 = line1.filter(lambda x : "a" in x)
2. 接收一个函数, 将满足该函数的元素放入新的RDD中
def hasHWTC1AC5C088(line):
return "HWTC1AC5C088" in line
lines2 = lines.filter(hasHWTC1AC5C088("HWTC1AC5C088")) #将函数传入filter中
3. RDD 删除第一条数据
header = abc.first()
df1 = abc.filter(lambda x:x != header)
5. map 和 flatMap 使用方法 将 lambda 函数做用在每一条记录上
1)line2 = line1.map(lambda x: x.split(" "))
2)line3 = line1.map(lambda x: x+"abc") #对原数据进行任意操作, 将结果再放回给原数据
3)line4 = line1.map(lambda x: (x, 1)) 将原始数据改为 key-value形式, key为原数据, value为 1
4)line2.flatMap(lambda line: line.split(" ")) #
5)map 与 flatMap 的区别(通常用来统计单词个数示例, 必须使用flatMap来进行拆分单词)
map 具有分层, 就是每一行数据作为你一层来处理 , 结果为:
[[u'extends', u'Object'], [u'implements', u'scala.Serializable']]
flatMap 不具有分层,
[u'extends', u'Object', u'implements', u'scala.Serializable']
6)map 获取前3列数据 下例中: [:3] 表示从开头到第三个数据项, 如果是[3:] 就表示从第三项到最后
Rdd.map(lambda x: x.split(" ")[:3]) 结果:[[u'a', u'1', u'3'], [u'b', u'2', u'4'], [u'd', u'3', u'4']]
ALS 训练数据---获取指定列数据
ratingsRdd = rawRatings.map(lambda x:(x[0],x[1],x[2]) 结果为:
[(u'196', u'242', u'3'), (u'186', u'302', u'3'), (u'22', u'377', u'1')]
7) 类型转换
Rdd.map(lambda x: float(x[0])) 将第一个字段转换为 float 类型
8) 删除所有的 "" 号 replace(替换), 下列意思是将" 替换成空
df2 = df1.map(lambda x:x.replace("\"",""))
9) df2 = RDD.map(lambda x: (x[0],float(x[1]),float(x[2]))) 设置一个 key 对应 多个value,
df3 = df2.filter(lambda keyValue: keyValue[0] > 2) 操作key
df3 = df2.filter(lambda keyValue: keyValue[1] > 2) 操作第一个value
df3 = df2.filter(lambda keyValue: keyValue[2] > 2) 操作第二个value
6. RDD 类型数据 的查询方式
print(abc) 打印当前对象
type(Rdd) 获取当前对象类型
RDD.collect() 将RDD转换为数组, 结果格式为:([u'{"name":"Michael"}', u'{"name":"Andy", "age":30}', u'{"name":"Justin", "age":19}'])
RDD.count() 查看内容条数
Rdd.printSchema() 查看rdd 列
7. RDD转换操作 rdd转list
list = RDD.collect() 2) list转RDD RDD = sc.parallelize(list)
3) RDD 调用 map 函数
(1) RDD1 = RDD2.map(lambda x: x+1) #使用匿名函数操作每条数据 map(lambda x: x.split(","))字符串截取,map(lambda x: "abc"+x) 重组字符串,
(2) RDD2 = RDD1.map(addOne) #使用具名函数来操作每条数据(具名函数就是单独定义一个函数来处理数据) 如下:
def addOne(x):
return x.split(",")
print(lines.map(addOne).collect()) #调用具名函数
4. RDD 调用 filter 函数
1) intRdd.filter(lambda x: x>5) #对数字类型的 RDD 进行筛选 intRdd.filter(lambda x: x>5 and x <40) and 表示 并且 的意思, or 表示 或 的意思
2) stringRdd.filter(lambda x: "abc" in x) #筛选包含 abc 的数据
4. RDD 删除 重复 元素
1) intRdd.distinct() #去重
5. 随机将一个 RDD 通过指定比例 分为 2 个RDD
1) sRdd = stringRdd.randomSplit([0.4,0.6]) 将 stringRdd 以4:6 分为2个 RDD, 获取其中一个 RDD 的方法为: sRdd[0]
6. RDD 中 groupBy 分组计算
1) gRdd = intRdd.groupBy(lambda x: x<2) #将会分为2组, 访问第一粗: print(sorted(gRdd[0][1])), 方位第二组:print(sorted(gRdd[1][1]))
2) 分组并且取别名: gRdd = intRdd.groupBy(lambda x: "a" if(x < 2) else "b"),
(1)获取第一组信息: print(gRdd[0][0], sorted(gRdd[0][1]))
(2) 获取第二组信息: print(gRdd[1][0], sorted(gRdd[1][1])) 其中, 前半部分 gRdd[1][0] 表示获取别名 a
7. 使用 union 进行并集运算, intersection 进行并集运算
1)intRdd1.union(intRdd2) 如: intRdd1 为 1, 3, 1 intRdd2 为 1, 2, 3, 4 则结果为: 1,3,1,1,2,3,4
2)intRdd1.intersection(intRdd2) 计算 2 个RDD 的交集
3)intRdd3.subtract(intRdd1) 计算 2 个 Rdd 的差集, 此例表示 intRdd3中有, 但在intRdd1中没有
4)intRdd1.cartesian(intRdd2) 计算 笛卡尔积
8. RDD 动作运算
[1] 读取元素
1) first() 查看RDD 第一条数据
2) take(2) 获取第二条数据
3) takeOrdered(3) 从小到大排序取出前 3 条数据
4) intRdd3.takeOrdered(6,key=lambda x: -x) 从大道小排序, 取出前6条数据
[2] 统计功能
1) intRdd1.stats() 统计 intRdd1, 结果为:(count: 5, mean: 5.0, stdev: 2.82842712475, max: 9, min: 1)
mean表示平均值, stdev 表示标准差
2)intRdd3.min() 最新值,
3)intRdd3.max() 最大值
4)intRdd3.stdev() 标准差
5)intRdd3.count() 数据条数
6)intRdd3.sum() 求和
7)intRdd3.mean() 平均值
9. RDD key-value 基本转换运算
1)kvRdd1 = sc.parallelize([(1, 4),(2, 5),(3, 6),(4, 7)]) 创建RDD key-value 源数据
结果为: [(1, 4), (2, 5), (3, 6), (4, 7)]
2)kvRdd1.keys() 获取全部 key 的值
3)kvRdd1.values() 获取全部 values 的值
4)kvRdd1.filter(lambda keyValue: keyValue[0] > 2) 过滤 key > 2 的数据
5)kvRdd1.filter(lambda keyValue: keyValue[1] >5) 过滤 value > 5 的数据
6)kvRdd1.mapValues(lambda x: x*x) 对每一条 value 进行运算
7)kvRdd1.sortByKey() 按照 key 从小到大 进行排序
8)kvRdd1.sortByKey(ascending=False) 按照 key 从大到小进行排序
9)kvRdd3.reduceByKey(lambda x, y:x+y) 将 key 相同的键的值合并相加
10. 多个 RDD key-value 的转换运算
1) join
intK1 = sc.parallelize([(1,5),(2,6),(3,7),(4,8),(5,9)]) intK2 = sc.parallelize([(3,30),(2,20),(6,60)]) intK1.join(intK2) join结果为: [(2, (6, 20)), (3, (7, 30))]
2)leftJoin
intK1.leftOuterJoin(intK2).collect() leftJoin结果为:
[(2, (6, 20)), (4, (8, None)), (1, (5, None)), (3, (7, 30)), (5, (9, None))]
3)rightJoin rightJoin 结果为:
intK1.rithtOuterJoin(intK2).collect()
[(2, (6, 20)), (6, (None, 60)), (3, (7, 30))]
4)subtractByKey 从 intK1 中删除 与 intK2 相同 key-value
intK1.subtractByKey(intK2) 结果为:
[(4, 8), (1, 5), (5, 9)]
11. key-value 动作 运算
1) intK1.first() 获取第一项数据
2) intK1.collect() 获取所有项数据
3) intK1.take(2) 获取前二项数据
4) intK1.first()[0] 获取第一项数据的 key
5) intK1.first()[1] 获取第一项数据的 value
例如: 一条记录结果为 [(2, (6, 20)), (4, (8, None)), (1, (5, None)), (3, (7, 30)), (5, (9, None))](leftJoin结果)
想要获取第一条记录的 6 , 可以使用: intK1.leftOuterJoin(intK2).first()[1][0] [1] 表示获取第一条记录的value, [0] 表示
从 value 中再获取第一项值 6
6) intK3.countByKey() 计算 RDD 中每一个 Key 值得项数, 例如
[(1, 2), (2, 3), (2, 5), (2, 8), (5, 10)] 源数据
defaultdict(, {1: 1, 2: 1, 3: 1, 4: 1, 5: 1}) 结果值 bcFruitMap.value 进行转换 从而获取编号对应的名称
7) KV = intK3.collectAsMap() 将 key-value 转换为 key-value的字典
{1: 2, 2: 8, 5: 10} 结果为
例如, 如果要获取 8 这个value, 就使用 KV[2] 就可以获取得到
8) intK3.lookup(2) 查找 key 为 2 的所有value 值, 如果想要再进行统计计算, 就将结果再进行转换为 RDD 进行统计计算
9) 广播变量
1> kvFrult = sc.parallelize([(1, "apple"),(2, "orange"),(3, "grape")]) 创建key-value 对照表
2> fruitMap = kvFrult.collectAsMap() 转换为 map 字典
3> bcFruitMap = sc.broadcast(fruitMap) 创建广播变量
4> fruitIds = sc.parallelize([2,4,1,3]) 创建编号 RDD
5> fruitNames = fruitIds.map(lambda x: bcFruitMap.value[x]) 使用
10) 通过累加器来计算总和
intRdd = sc.parallelize([1,2,44,2,11,22]) 源数据
total = sc.accumulator(0.0) 定义一个double类型的累加器, 来计算总和
num = sc.accumulator(0) 定义一个int类型的累加器, 来计算数量
intRdd.foreach(lambda l: [total.add(l), num.add(1)]) 通过foreach 循环来统计
total.value 获取总和
num.value 获取个数
avg = total.vaue/num.value 获取平均值
11) RDD 持久化
1.书221 页面, 设置持久化等级列表
2.intRdd1.persist() 设置持久化
2.intRdd1.persist(StorageLevel.MEMORY_AND_DISK) 设置存储等级
4.intRdd1.is_cached 查看是否持久化
12) RDD.saveAsTextFile("hdfs://192.168.88.128:9000/data/result.txt") 将结果保存成文件
12 数据格式
1. [[u'3', u'5'], [u'4', u'6'], [u'4', u'5'], [u'4', u'2']] 拆分或截取的原始数据, 可以通过 map 中的 x[0], x[1] 来获取对应列的数据
可以通过 map 来转换为key-value 数据格式 例如: df3 = df2.map(lambda x: (x[0], x[1]))
2. key-value 数据格式
[(u'3', u'5'), (u'4', u'6'), (u'4', u'5'), (u'4', u'2')] 中每一个() 表示一组数据, 第一个表示key 第二个表示value
3)PipelinedRDD 类型表示 key-value形式数据
13 RDD类型转换
userRdd = sc.textFile("D:\data\people.json") userRdd = userRdd.map(lambda x: x.split(" ")) userRows = userRdd.map(lambda p: Row( userName = p[0], userAge = int(p[1]), userAdd = p[2], userSalary = int(p[3]) ) ) print(userRows.take(4))
结果: [Row(userAdd='shanghai', userAge=20, userName='zhangsan', userSalary=13), Row(userAdd='beijin', userAge=30, userName='lisi', userSalary=15)]
2) 创建 DataFrame
userDF = sqlContext.createDataFrame(userRows)
14. 通过sql 语句查询字段
from pyspark.conf import SparkConf from pyspark.sql.session import SparkSession from pyspark.sql.types import Row if __name__ == '__main__': spark = SparkSession.builder.config(conf = SparkConf()).getOrCreate() sc = spark.sparkContext rd = sc.textFile("D:\data\people.txt") rd2 = rd.map(lambda x:x.split(",")) people = rd2.map(lambda p: Row(name=p[0], age=int(p[1]))) peopleDF = spark.createDataFrame(people) peopleDF.createOrReplaceTempView("people") teenagers = spark.sql("SELECT name,age FROM people where name='Andy'") teenagers.show(5) # print(teenagers.rdd.collect()) teenNames = teenagers.rdd.map(lambda p: 100 + p.age).collect() for name in teenNames: print(name)
15 dateFrame,sql,json使用详细示例
# # Licensed to the Apache Software Foundation (ASF) under one or more # contributor license agreements. See the NOTICE file distributed with # this work for additional information regarding copyright ownership. # The ASF licenses this file to You under the Apache License, Version 2.0 # (the "License"); you may not use this file except in compliance with # the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # """ A simple example demonstrating basic Spark SQL features. Run with: ./bin/spark-submit examples/src/main/python/sql/basic.py """ from __future__ import print_function # $example on:init_session$ from pyspark.sql import SparkSession # $example off:init_session$ # $example on:schema_inferring$ from pyspark.sql import Row # $example off:schema_inferring$ # $example on:programmatic_schema$ # Import data types from pyspark.sql.types import * # $example off:programmatic_schema$ def basic_df_example(spark): # $example on:create_df$ # spark is an existing SparkSession df = spark.read.json("/data/people.json") # Displays the content of the DataFrame to stdout df.show() # +----+-------+ # | age| name| # +----+-------+ # |null|Michael| # | 30| Andy| # | 19| Justin| # +----+-------+ # $example off:create_df$ # $example on:untyped_ops$ # spark, df are from the previous example # Print the schema in a tree format df.printSchema() # root # |-- age: long (nullable = true) # |-- name: string (nullable = true) # Select only the "name" column df.select("name").show() # +-------+ # | name| # +-------+ # |Michael| # | Andy| # | Justin| # +-------+ # Select everybody, but increment the age by 1 df.select(df['name'], df['age'] + 1).show() # +-------+---------+ # | name|(age + 1)| # +-------+---------+ # |Michael| null| # | Andy| 31| # | Justin| 20| # +-------+---------+ # Select people older than 21 df.filter(df['age'] > 21).show() # +---+----+ # |age|name| # +---+----+ # | 30|Andy| # +---+----+ # Count people by age df.groupBy("age").count().show() # +----+-----+ # | age|count| # +----+-----+ # | 19| 1| # |null| 1| # | 30| 1| # +----+-----+ # $example off:untyped_ops$ # $example on:run_sql$ # Register the DataFrame as a SQL temporary view df.createOrReplaceTempView("people") sqlDF = spark.sql("SELECT * FROM people") sqlDF.show() # +----+-------+ # | age| name| # +----+-------+ # |null|Michael| # | 30| Andy| # | 19| Justin| # +----+-------+ # $example off:run_sql$ # $example on:global_temp_view$ # Register the DataFrame as a global temporary view df.createGlobalTempView("people") # Global temporary view is tied to a system preserved database `global_temp` spark.sql("SELECT * FROM global_temp.people").show() # +----+-------+ # | age| name| # +----+-------+ # |null|Michael| # | 30| Andy| # | 19| Justin| # +----+-------+ # Global temporary view is cross-session spark.newSession().sql("SELECT * FROM global_temp.people").show() # +----+-------+ # | age| name| # +----+-------+ # |null|Michael| # | 30| Andy| # | 19| Justin| # +----+-------+ # $example off:global_temp_view$ def schema_inference_example(spark): # $example on:schema_inferring$ sc = spark.sparkContext # Load a text file and convert each line to a Row. lines = sc.textFile("examples/src/main/resources/people.txt") parts = lines.map(lambda l: l.split(",")) people = parts.map(lambda p: Row(name=p[0], age=int(p[1]))) # Infer the schema, and register the DataFrame as a table. schemaPeople = spark.createDataFrame(people) schemaPeople.createOrReplaceTempView("people") # SQL can be run over DataFrames that have been registered as a table. teenagers = spark.sql("SELECT name FROM people WHERE age >= 13 AND age <= 19") # The results of SQL queries are Dataframe objects. # rdd returns the content as an :class:`pyspark.RDD` of :class:`Row`. teenNames = teenagers.rdd.map(lambda p: "Name: " + p.name).collect() for name in teenNames: print(name) # Name: Justin # $example off:schema_inferring$ def programmatic_schema_example(spark): # $example on:programmatic_schema$ sc = spark.sparkContext # Load a text file and convert each line to a Row. lines = sc.textFile("examples/src/main/resources/people.txt") parts = lines.map(lambda l: l.split(",")) # Each line is converted to a tuple. people = parts.map(lambda p: (p[0], p[1].strip())) # The schema is encoded in a string. schemaString = "name age" fields = [StructField(field_name, StringType(), True) for field_name in schemaString.split()] schema = StructType(fields) # Apply the schema to the RDD. schemaPeople = spark.createDataFrame(people, schema) # Creates a temporary view using the DataFrame schemaPeople.createOrReplaceTempView("people") # SQL can be run over DataFrames that have been registered as a table. results = spark.sql("SELECT name FROM people") results.show() # +-------+ # | name| # +-------+ # |Michael| # | Andy| # | Justin| # +-------+ # $example off:programmatic_schema$ if __name__ == "__main__": # $example on:init_session$ spark = SparkSession \ .builder \ .appName("Python Spark SQL basic example") \ .config("spark.some.config.option", "some-value") \ .getOrCreate() # $example off:init_session$ basic_df_example(spark) # schema_inference_example(spark) # programmatic_schema_example(spark) spark.stop()