推荐引擎数据导入模块的实现

# -*- coding:UTF-8 -*-
#!/usr/bin/python
# FileName:pro_env.py

#***************************************************
# 项目的路径
PROJECT_DIR = "/usr/local/EclipseProjects/MyBI"
# 项目配置文件的路径
PROJECT_CONF_DIR = PROJECT_DIR + "/conf/"
# 项目第三方库的路径
PROJECT_LIB_DIR = PROJECT_DIR + "/lib"
# 项目临时文件的路径
PROJECT_TMP_DIR = PROJECT_DIR + "/temp"
#***************************************************

# Hadoop的安装路径
HADOOP_HOME = "/usr/local/hadoop/"
# Hadoop的命令路径
HADOOP_PATH = HADOOP_HOME + "bin/"
# HIVE的安装路径
HIVE_HOME = "/opt/hive-0.9.0/"
# HIVE的命令路径
HIVE_PATH = HIVE_HOME + "bin/"
# Sqoop的安装路径
SQOOP_HOME = "/opt/Sqoop/"
# Sqoop的命令路径
SQOOP_PATH = SQOOP_HOME + "bin/"
#***************************************************

# Java的安装路径
Java_HOME = "/usr/lib/jvm/jdk1.7.0_75"

<?xml version="1.0" encoding="UTF-8"?>
<root>
    <task type="add">
        <table>ModifyRecords</table>
    </task>
</root>

<?xml version="1.0" encoding="UTF-8"?>
<root>
    <sqoop-shell type="import">
        <param key="connect">jdbc:mysql://localhost:3306/Recommend</param>
        <param key="username">${username}</param>
        <param key="password">${password}</param>
        <param key="target-dir">/user/hadoop/Recommend/$dt</param>
        <param key="query">‘select userID,movieID,pref from Recommend.ModifyRecords where modifyDate$flag"\$CONDITIONS" and $CONDITIONS‘</param>
        <param key="m">1</param>
        <param key="fields-terminated-by">‘,‘</param>
    </sqoop-shell>
</root>

# -*- coding:UTF-8 -*-
#!/usr/bin/python
# FileName:import.py
from com.utls.pro_env import PROJECT_CONF_DIR
import time
from com.utls.sqoop import SqoopUtil
import xml.etree.ElementTree as ET

# 其中dt为昨天的日期，将由调度模块传入
def resolve_conf(dt):
    
    # 获得配置文件名
    conf_file = PROJECT_CONF_DIR + "Import.xml"
    
    # 解析配置文件
    xml_tree = ET.parse(conf_file)
    # 获得task元素
    tasks = xml_tree.findall(‘./task‘)
    
    for task in tasks:
        # 获得导入类型，增量导入或者全量导入
        import_type = task.attrib["type"]
        
        # 获得表名集合
        tables = task.findall(‘./table‘)
        
        # 用来保存待执行的Sqoop命令的集合
        cmds = []

        # 迭代表名集合，解析表配置文件
        for i in range(len(tables)):
            # 表名
            table_name = tables[i].text
            # 表配置文件名
            table_conf_file = PROJECT_CONF_DIR + table_name + ".xml"
            
            # 解析表配置文件
            xmlTree = ET.parse(table_conf_file)
            
            # 获取sqoop-shell节点
            sqoopNodes = xmlTree.findall("./sqoop-shell")
            
            # 获取sqoop-shell节点
            sqoop_cmd_type = sqoopNodes[0].attrib["type"]
            # 获取
            praNodes = sqoopNodes[0].findall("./param")
            
            # 用来保存param信息的字典
            cmap = {}
            
            for i in range(len(praNodes)):
                # 获得key属性的值
                key = praNodes[i].attrib["key"]
                # 获得param标签中间的值
                value = praNodes[i].text
                # 保存到字典中
                cmap[key] = value
                
            # 首先组装成sqoop命令头
            command = "sqoop " + sqoop_cmd_type
                
            # 如果为全量导入
            if(import_type == "all"):
                # query的查询条件为<dt
                import_condition = dt
                flag = "<"
            # 如果为增量导入
            elif (import_type == "add"):
                # query的查询条件为=dt
                import_condition = dt
                flag = "="
            else:
                raise Exception
                
            # #迭代字典将param的信息拼装成字符串
            for key in cmap.keys():
                    
                value = cmap[key]
                    
                # 如果不是键值对形式的命令选项
                if(value == None or value == "" or value == " "):
                    value = ""
                    
                # 将query的CONDITIONS替换为查询条件
                if(key == "query"):
                    value = value.replace("\$CONDITIONS", import_condition)
                    value = value.replace("$flag", flag)
                        
                # 将导入分区替换为传入的时间
                if(key == "target-dir"):
                    value = value.replace("$dt", dt)
                    
                # 拼装为命令
                if key == "fields-terminated-by":
                    command += " --" + key + " " + value
                else:
                    command += " --" + key + " " + value + "\\" + "\n"
                
            # 将命令加入至待执行的命令集合
            cmds.append(command)
        
    return cmds

# Python模块的入口：main函数
if __name__ == ‘__main__‘:
    
    # 调度模块将昨天的时间传入
    dt = time.strftime("%Y-%m-%d", time.localtime(time.time()))
    # 解析配置文件，获得sqoop命令集合
    cmds = resolve_conf(dt)
    
    # 迭代集合，执行命令
    for i in range(len(cmds)):
        cmd = cmds[i]
        
        # 执行导入过程
        SqoopUtil.execute_shell(cmd)

sqoop import --username xxxx --target-dir /user/hadoop/Recommend/2015-04-26 --m 1 --connect jdbc:mysql://localhost:3306/Recommend --query ‘select userID,movieID,pref from Recommend.ModifyRecords where modifyDate="2015-04-26" and $CONDITIONS‘ --password xxxx --fields-terminated-by ‘,‘

#!/usr/bin/python
# FileName sqoop.py
# -*- coding:UTF-8 -*-
import os
class SqoopUtil(object):
    ‘‘‘
    sqoop operation
    ‘‘‘
    def __init__(self):
        pass
    
    @staticmethod
    def execute_shell(shell):
        print shell
        os.system(shell)