我正在尝试将csv文件读入数据帧。我知道我的数据帧的架构应该是什么,因为我知道我的csv文件。另外我使用spark csv包来读取文件。我试着像下面那样指定架构。
val pagecount = sqlContext.read.format("csv")
.option("delimiter"," ").option("quote","")
.option("schema","project: string ,article: string ,requests: integer ,bytes_served: long")
.load("dbfs:/databricks-datasets/wikipedia-datasets/data-001/pagecounts/sample/pagecounts-20151124-170000")
但是当我检查我创建的数据框架的模式时,似乎已经采用了自己的模式。我做错了吗?如何制作火花来挑选我提到的架构?
> pagecount.printSchema
root
|-- _c0: string (nullable = true)
|-- _c1: string (nullable = true)
|-- _c2: string (nullable = true)
|-- _c3: string (nullable = true)
答案 0 :(得分:32)
请尝试以下操作,您无需指定架构。当你将inferSchema设为true时,它应该从你的csv文件中获取它。
val pagecount = sqlContext.read.format("csv")
.option("delimiter"," ").option("quote","")
.option("header", "true")
.option("inferSchema", "true")
.load("dbfs:/databricks-datasets/wikipedia-datasets/data-001/pagecounts/sample/pagecounts-20151124-170000")
如果要手动指定架构,则需要执行以下操作
import org.apache.spark.sql.types._
val customSchema = StructType(Array(
StructField("project", StringType, true),
StructField("article", StringType, true),
StructField("requests", IntegerType, true),
StructField("bytes_served", DoubleType, true)))
val pagecount = sqlContext.read.format("csv")
.option("delimiter"," ").option("quote","")
.option("header", "true")
.schema(customSchema)
.load("dbfs:/databricks-datasets/wikipedia-datasets/data-001/pagecounts/sample/pagecounts-20151124-170000")
答案 1 :(得分:8)
我在分析中使用Arunakiran Nulu提供的解决方案(参见代码)。尽管它能够为列分配正确的类型,但返回的所有值都是null
。以前,我尝试过选项.option("inferSchema", "true")
并在数据框中返回正确的值(尽管类型不同)。
val customSchema = StructType(Array(
StructField("numicu", StringType, true),
StructField("fecha_solicitud", TimestampType, true),
StructField("codtecnica", StringType, true),
StructField("tecnica", StringType, true),
StructField("finexploracion", TimestampType, true),
StructField("ultimavalidacioninforme", TimestampType, true),
StructField("validador", StringType, true)))
val df_explo = spark.read
.format("csv")
.option("header", "true")
.option("delimiter", "\t")
.option("timestampFormat", "yyyy/MM/dd HH:mm:ss")
.schema(customSchema)
.load(filename)
结果
root
|-- numicu: string (nullable = true)
|-- fecha_solicitud: timestamp (nullable = true)
|-- codtecnica: string (nullable = true)
|-- tecnica: string (nullable = true)
|-- finexploracion: timestamp (nullable = true)
|-- ultimavalidacioninforme: timestamp (nullable = true)
|-- validador: string (nullable = true)
,表格是:
|numicu|fecha_solicitud|codtecnica|tecnica|finexploracion|ultimavalidacioninforme|validador|
+------+---------------+----------+-------+--------------+-----------------------+---------+
| null| null| null| null| null| null| null|
| null| null| null| null| null| null| null|
| null| null| null| null| null| null| null|
| null| null| null| null| null| null| null|
答案 2 :(得分:6)
感谢@Nulu的回答,它适用于pyspark,只需要很少的调整
new Point(x,y)
答案 3 :(得分:6)
以前的解决方案使用了自定义的StructType。
使用spark-sql 2.4.5(scala版本2.12.10),现在可以使用schema
函数将模式指定为字符串
import org.apache.spark.sql.SparkSession;
val sparkSession = SparkSession.builder()
.appName("sample-app")
.master("local[2]")
.getOrCreate();
val pageCount = sparkSession.read
.format("csv")
.option("delimiter","|")
.option("quote","")
.schema("project string ,article string ,requests integer ,bytes_served long")
.load("dbfs:/databricks-datasets/wikipedia-datasets/data-001/pagecounts/sample/pagecounts-20151124-170000")
答案 4 :(得分:5)
对于那些有兴趣在Python中执行此操作的人,这里是一个有效的版本。
customSchema = StructType([
StructField("IDGC", StringType(), True),
StructField("SEARCHNAME", StringType(), True),
StructField("PRICE", DoubleType(), True)
])
productDF = spark.read.load('/home/ForTesting/testProduct.csv', format="csv", header="true", sep='|', schema=customSchema)
testProduct.csv
ID|SEARCHNAME|PRICE
6607|EFKTON75LIN|890.88
6612|EFKTON100HEN|55.66
希望这会有所帮助。
答案 5 :(得分:4)
以下是使用自定义架构的方法,一个完整的演示:
$> shell代码,
echo "
Slingo, iOS
Slingo, Android
" > game.csv
Scala代码:
import org.apache.spark.sql.types._
val customSchema = StructType(Array(
StructField("game_id", StringType, true),
StructField("os_id", StringType, true)
))
val csv_df = spark.read.format("csv").schema(customSchema).load("game.csv")
csv_df.show
csv_df.orderBy(asc("game_id"), desc("os_id")).show
csv_df.createOrReplaceTempView("game_view")
val sort_df = sql("select * from game_view order by game_id, os_id desc")
sort_df.show
答案 6 :(得分:0)
这是一种选项,我们可以在加载CSV时将列名传递给数据框。
import pandas
names = ['sepal-length', 'sepal-width', 'petal-length', 'petal-width', 'class']
dataset = pandas.read_csv("C:/Users/NS00606317/Downloads/Iris.csv", names=names, header=0)
print(dataset.head(10))
输出
sepal-length sepal-width petal-length petal-width class
1 5.1 3.5 1.4 0.2 Iris-setosa
2 4.9 3.0 1.4 0.2 Iris-setosa
3 4.7 3.2 1.3 0.2 Iris-setosa
4 4.6 3.1 1.5 0.2 Iris-setosa
5 5.0 3.6 1.4 0.2 Iris-setosa
6 5.4 3.9 1.7 0.4 Iris-setosa
7 4.6 3.4 1.4 0.3 Iris-setosa
8 5.0 3.4 1.5 0.2 Iris-setosa
9 4.4 2.9 1.4 0.2 Iris-setosa
10 4.9 3.1 1.5 0.1 Iris-setosa
答案 7 :(得分:0)
// import Library
import java.io.StringReader ;
import au.com.bytecode.opencsv.CSVReader
//filename
var train_csv = "/Path/train.csv";
//read as text file
val train_rdd = sc.textFile(train_csv)
//use string reader to convert in proper format
var full_train_data = train_rdd.map{line => var csvReader = new CSVReader(new StringReader(line)) ; csvReader.readNext(); }
//declares types
type s = String
// declare case class for schema
case class trainSchema (Loan_ID :s ,Gender :s, Married :s, Dependents :s,Education :s,Self_Employed :s,ApplicantIncome :s,CoapplicantIncome :s,
LoanAmount :s,Loan_Amount_Term :s, Credit_History :s, Property_Area :s,Loan_Status :s)
//create DF RDD with custom schema
var full_train_data_with_schema = full_train_data.mapPartitionsWithIndex{(idx,itr)=> if (idx==0) itr.drop(1);
itr.toList.map(x=> trainSchema(x(0),x(1),x(2),x(3),x(4),x(5),x(6),x(7),x(8),x(9),x(10),x(11),x(12))).iterator }.toDF
答案 8 :(得分:0)
模式定义为简单字符串
如果有人对带有 date 和 时间戳 的简单字符串的模式定义感兴趣,
从终端或外壳程序创建数据文件
echo "
2019-07-02 22:11:11.000999, 01/01/2019, Suresh, abc
2019-01-02 22:11:11.000001, 01/01/2020, Aadi, xyz
" > data.csv
将架构定义为字符串
user_schema = 'timesta TIMESTAMP,date DATE,first_name STRING , last_name STRING'
读取数据
df = spark.read.csv(path='data.csv', schema = user_schema, sep=',', dateFormat='MM/dd/yyyy',timestampFormat='yyyy-MM-dd HH:mm:ss.SSSSSS')
df.show(10, False)
+-----------------------+----------+----------+---------+
|timesta |date |first_name|last_name|
+-----------------------+----------+----------+---------+
|2019-07-02 22:11:11.999|2019-01-01| Suresh | abc |
|2019-01-02 22:11:11.001|2020-01-01| Aadi | xyz |
+-----------------------+----------+----------+---------+
请注意,显式定义架构而不是让spark推断架构也可以提高spark读取性能。
答案 9 :(得分:0)
在pyspark 2.4及更高版本中,您只需使用header
参数即可设置正确的标头:
data = spark.read.csv('data.csv', header=True)
类似地,如果使用scala,则也可以使用header
参数。
答案 10 :(得分:0)
您还可以通过使用sparkSession和隐式
来做到这一点pat_ID
答案 11 :(得分:0)
如果您的 spark 版本是 3.0.1,您可以使用以下 Scala 脚本:
val df = spark.read.format("csv").option("delimiter",",").option("header",true).load("file:///LOCAL_CSV_FILE_PATH")
但是这样,所有的数据类型都会被设置为 String
。
答案 12 :(得分:-1)
我的解决方法是:
import org.apache.spark.sql.types._
val spark = org.apache.spark.sql.SparkSession.builder.
master("local[*]").
appName("Spark CSV Reader").
getOrCreate()
val movie_rating_schema = StructType(Array(
StructField("UserID", IntegerType, true),
StructField("MovieID", IntegerType, true),
StructField("Rating", DoubleType, true),
StructField("Timestamp", TimestampType, true)))
val df_ratings: DataFrame = spark.read.format("csv").
option("header", "true").
option("mode", "DROPMALFORMED").
option("delimiter", ",").
//option("inferSchema", "true").
option("nullValue", "null").
schema(movie_rating_schema).
load(args(0)) //"file:///home/hadoop/spark-workspace/data/ml-20m/ratings.csv"
val movie_avg_scores = df_ratings.rdd.map(_.toString()).
map(line => {
// drop "[", "]" and then split the str
val fileds = line.substring(1, line.length() - 1).split(",")
//extract (movie id, average rating)
(fileds(1).toInt, fileds(2).toDouble)
}).
groupByKey().
map(data => {
val avg: Double = data._2.sum / data._2.size
(data._1, avg)
})