PySpark操作Hive的常用语句函数封装包
2017-07-21 本文已影响0人
小甜瓜Melon
目的:将hive常用的查看函数进行封装。
#!/usr/bin/env python
# _*_ coding:utf-8 _*_
# Standard libraries
import sys
import os
import time
# PyData stack
import pandas as pd
import numpy as np
import matplotlib
import matplotlib.pyplot as plt
#matplotlib.use('Agg')
# Spark Libraries
from pyspark import SparkContext, SparkConf
# Spark sql Libraries
from pyspark.sql import HiveContext
from pyspark.sql import SQLContext
from pyspark.sql import Row
from pyspark.sql.functions import *
from pyspark.sql.types import *
import pyspark.sql as sprksql
# Spark ml Libraries
from pyspark.ml.classification import LogisticRegression, LogisticRegressionModel
from pyspark.ml.tuning import CrossValidator, ParamGridBuilder
from pyspark.ml.evaluation import BinaryClassificationEvaluator
from pyspark.ml.feature import StringIndexer, VectorIndexer, OneHotEncoder, VectorAssembler
from pyspark.ml import Pipeline
# from utils import *
# from utils.plotting import *
# Magic!
from IPython.core.display import display, HTML, Markdown
display(HTML("<style>.container { width:90% !important; }</style>"))
# 设置jupyter中显示的格式
pd.options.display.max_columns = 1000
pd.options.display.max_rows = 1000
# 等价于pd.set_option('display.max_columns', 100)
hiveContext = HiveContext(sc)
#--*--*----*--*----*--*----*--*----*--*----*--*----*--*----*--*----*--*----*--*----*--*--#
"""利用Python查看单个数据库中的所有表"""
def scanTable(db_name):
sql = "use %s"%db_name
hiveContext.sql(sql)
tables = hiveContext.sql("show tables").collect()
for i in xrange(len(tables)):
print tables[i][0]
# scanTable("source_data")
"""查看单个表中的数据示例"""
# 参数说明:number代表查看几条数据
def scanData(db_name, table_name, number):
sql_scan = "select * from %s.%s limit %d"%(db_name, table_name, number)
return hiveContext.sql(sql_scan).toPandas().T
# scanData("source_data", "users_basic", 3)
# hiveContext.sql("select * from user_profile_project.user_pro_address_auto limit 2").toPandas().T
"""计算单个表中所有字段的记录数"""
def countColumnsNums(db_name,table_name):
print u"当前数据库为%s"%db_name
sql = "SELECT COUNT(*) FROM %s.%s" %(db_name, table_name)
# globals()[table_name + '_number'] = hiveContext.sql(sql)
print u"当前表名称为%s,总记录数为:"%table_name, hiveContext.sql(sql).collect()[0][0]
sql1 = """SHOW COLUMNS FROM %s.%s"""%(db_name, table_name)
col_name = hiveContext.sql(sql1).collect()
print u"开始计算每一列的记录数"
for i in xrange(len(col_name)):
sql2 = "SELECT COUNT(%s) FROM %s.%s" %(col_name[i][0], db_name, table_name)
print col_name[i][0], hiveContext.sql(sql2).collect()[0][0], '''''''',\
u"%s表里边一共包含%s个column,已经计算完第%s个column, 列名称为%s" %(table_name, len(col_name), i+1, col_name[i][0])
print u"计算结束!"
"""计算所有表中的记录数"""
def hiveCountTables(db_name):
sql = "use %s"%db_name
hiveContext.sql(sql)
showtables = hiveContext.sql("show tables")
table_name = showtables.collect()
#print u"表名称"
#print table_name
#print '*-' * 40 + '*'
print u"开始统计"
if type(table_name) == list:
for i in xrange(len(table_name)):
# sql = str("SELECT * FROM tb_source_data.%s" %table_name[i][0])
sql = "SELECT count(*) FROM source_data.%s" %table_name[i][0]
#print sql
# locals()[str(table_name[i][0])] = hiveContext.sql(sql).toPandas()
# globals()[str(table_name[i][0])] = hiveContext.sql(sql).toPandas()
globals()[table_name[i][0] + '_number'] = hiveContext.sql(sql)
#print table_name[i][0]
print table_name[i][0] + '_number',hiveContext.sql(sql).collect()[0][0]
#print u"该database里边一共包含%s个table,正在计算第%s个table, 表名称为%s" %(len(table_name), i+1, table_name[i][0])
#print '*-' * 40 + '*'
print u"计算结束!"
else:
print "showtables is not a list"
# hiveCountTables("source_data")
"""计算所有表的总记录数和所有字段的记录数"""
def hiveCountTablesColumns(db_name):
sql = "use %s"%db_name
hiveContext.sql(sql)
showtables = hiveContext.sql("show tables")
table_name = showtables.collect()
#print u"表名称"
#print table_name
#print '*-' * 40 + '*'
print u"开始统计"
if type(table_name) == list:
for i in xrange(len(table_name)):
sql = "SELECT COUNT(*) FROM source_data.%s" %table_name[i][0]
globals()[table_name[i][0] + '_number'] = hiveContext.sql(sql)
print table_name[i][0] + '_number',hiveContext.sql(sql).collect()[0][0]
sql1 = "SHOW COLUMNS FROM %s" %table_name[i][0]
col_name = hiveContext.sql(sql1).collect()
for j in xrange(len(col_name)):
sql2 = "SELECT COUNT(%s) FROM %s" %(col_name[j][0],table_name[i][0])
print col_name[j][0], hiveContext.sql(sql2).collect()[0][0]
print u"该database里边一共包含%s个table,已经计算完第%s个table, 表名称为%s" %(len(table_name), i+1, table_name[i][0])
print '*-' * 40 + '*'
print u"计算结束!"
else:
print "showtables is not a list"
# hiveCountTablesColumns("source_data")