PythonSpark:需要从文件列执行配置单元查询 [英] PythonSpark: need to execute hive queries from file columns

问题描述

我有一个文件，其行如下所示(文件名:sample.csv)

I have a file with rows like below (file name: sample.csv)

Id,Query
T1012,"Select * from employee_dim limit 100"
T1212,"Select * from department_dim limit 100"
T1231,"Select dept_number,location,dept_name from locations"

我需要遍历此文件(sample.csv)并获取第二列(查询")，在hive数据库中运行它并获取结果，然后将其保存到名为T1012_result.csv的新文件中，并针对所有行执行类似操作.

I need to iterate through this file (sample.csv) and take the second column("query"), run it in hive database and get the result of this, then save it to a new file named T1012_result.csv, and similarly do it for all rows.

可以帮忙吗?

我尝试通过spark读取文件并将其转换为列表，然后使用不起作用的sparksession执行SQL查询.

I tried reading the file through spark and converting it to a list and then executing the SQL queries using sparksession which is not working .

from pyspark.sql import SparkSession,HiveContext

spark=SparkSession.builder.enableHiveSupport().getOrCreate()
spark.sql("use sample")
input=spark.read.csv("sample.csv")
#input.select('_c1').show()

import pandas as pd

a=input.toPandas().values.tolist()
for i in a :
   print i[1]
   spark.sql('pd.DataFrame(i)')

推荐答案

已更新:spark

file_path="file:///user/vikrant/inputfiles/multiquery.csv"
df=spark.read.format("com.databricks.spark.csv").option("header", "true").load(file_path)

+---+-------------------------------+
|id |query                          |
+---+-------------------------------+
|1  |select * from exampledate      |
|2  |select * from test             |
|3  |select * from newpartitiontable|
+---+-------------------------------+

def customFunction(row):
    for row in df.rdd.collect():
        item=(row[1])
        filename=(row[0])
        query=""
        query+=str(item)
        newdf=spark.sql(query)
        savedataframe(newdf,filename)

def savedataframe(newdf,filename):
    newdf.coalesce(1).write.csv("/user/dev/hadoop/external/files/file_" + filename + ".csv")

customFunction(df)

drwxr-xr-x   - vikct001 hdfs          0 2019-08-02 11:49 /user/dev/hadoop/external/files/file_1.csv
drwxr-xr-x   - vikct001 hdfs          0 2019-08-02 11:49 /user/dev/hadoop/external/files/file_2.csv
drwxr-xr-x   - vikct001 hdfs          0 2019-08-02 11:49 /user/dev/hadoop/external/files/file_3.csv

更新:使用熊猫 我在sql server上有很少的测试表，并且正如您在问题中提到的，我正在将它们读取到pandas数据框，并将查询结果保存到每个不同的文件中，并将其重命名为数据框的第一列:

Update: using pandas I have few test tables on sql server and I am reading them to pandas dataframe as you mentioned in your question and will be saving the query result to each different files with renamed as first column of your dataframe:

import pandas as pd
import pyodbc
from pandas import DataFrame


connection = pyodbc.connect('Driver={ODBC Driver 13 for SQL Server};SERVER=yourservername;DATABASE=some_db;UID=username;PWD=password')
cursor = connection.cursor()

data=[['1','select * from User_Stage_Table'],['2','select * from User_temp_Table']]
df=pd.DataFrame(data,columns=['id','query'])


def get_query(df):
    a=df.values.tolist()
    for i in a:
        query=i[1]    #reading second column value as query
        filename=i[0] #reading first column value as filename
        write_query(query,filename) #calling write_query function 

def write_query(query,filename):
    df=pd.read_sql_query(query,connection)
    df.to_csv(outfile_location+filename+".txt",sep=',',encoding='utf-8',index=None,mode='a')

get_query(df)  #calling get_query function to build the query
out_file_location='G:\Testing\OutputFile\outfile'

您将输出文件名为:

outfile1.txt#这将具有表User_Stage_Table

outfile2.txt#这将具有表User_temp_Table'

让我知道这是否可以解决您的问题或进一步面临任何问题.

Let me know if this solves your problem or face any issues further.

这篇关于PythonSpark:需要从文件列执行配置单元查询的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！