使用Hive UDF和GeoIP库为Hive加入IP识别功能
Hive是基于Hadoop的数据管理系统，作为分析人员的即时分析工具和ETL等工作的执行引擎，对于如今的大数据管理与分析、处理有着非常大的 意义。GeoIP是一套IP映射数据库，它定时更新，并且提供了各种语言的API，非常适合在做地域相关数据分析时的一个数据源。
Precondition:通过 IP 地址获得用户的地理位置信息
也就是根据用户的IP，通过IP数据库查询获得信息。
一般IP数据库中，
每条记录的基本结构：
IP地址段（起始、结束），以及对应的信息数据
一般包含的信息：国家、区域（省/州）、城市、街道、经纬度、ISP提供商等信息
因为IP数据库随着时间经常变化（不过一段时间内变化很小），所以需要有人经常维护和更新。这个数据也不可能完全准确、也不可能覆盖全。这是maxmind的城市准确度 http://www.maxmind.com/app/city_accuracy 。
因为没有权威的数据组织机构，且经常有变化。各家数据供应商，基本上做着做着就形成自己的一套数据了。
目前，国内用的比较有名的是“纯真IP数据库”，国外常用的是 maxmind、ip2location。
IP数据库是否收费：收费、免费都有。一般有人维护的数据往往都是收费的，准确率和覆盖率会稍微高一些。
质量方面：
一般包含的信息：国家、区域（省/州）、城市、街道、经纬度、ISP提供商等信息
因为IP数据库随着时间经常变化（不过一段时间内变化很小），所以需要有人经常维护和更新。这个数据也不可能完全准确、也不可能覆盖全。这是maxmind的城市准确度 http://www.maxmind.com/app/city_accuracy 。
因为没有权威的数据组织机构，且经常有变化。各家数据供应商，基本上做着做着就形成自己的一套数据了。
目前，国内用的比较有名的是“纯真IP数据库”，国外常用的是 maxmind、ip2location。
IP数据库是否收费：收费、免费都有。一般有人维护的数据往往都是收费的，准确率和覆盖率会稍微高一些。
质量方面：
- 主要概念是准确率和覆盖率。
- 记录数据总条数。纯真现在是38万条（2010年07月30日更新）
- 是否有人维护。
- 数据库更新频率：每月、每周。数据库会定期更新的，maxmind开源版是每月更新一次。
查询形式：
- 本地，将IP数据库下载到本地使用，查询效率高、性能好。常用在统计分析方面。具体形式又分为：
- 内存查询：将全部数据直接加载到内存中，便于高性能查询。或者二进制的数据文件本身就是经过优化的索引文件，可以直接对文件做查询。
- 数据库查询：将数据导入到数据库，再用数据库查询。效率没有内存查询快。
- 远程（web service或ajax），调用远程第三方服务。查询效率自然比较低，一般用在网页应用中。
查询的本质：输入一个IP，找到其所在的IP段，一般都是采用二分搜索实现的。
是否提供API：有的IP数据库提供API，支持多语言（java、javascript、C#等），这样你就不用自己直接分析数据格式、整理、写查询代码了。
是否提供经纬度：纯真IP数据库不提供经纬度，Maxmind提供,如果做地图应用，一般是需要经纬度的。
而UDF是Hive提供的用户自定义函数的接口，通过实现它可以扩展Hive目前已有的内置函数。而为Hive加入一个IP映射函数，我们只需要简单地在UDF中调用GeoIP的Java API即可。
GeoIP的数据文件可以从这里下载：http://www.maxmind.com/download/geoip/database/，由于需 要国家和城市的信息，我这里下载的是http://www.maxmind.com/download/geoip/database /GeoLiteCity.dat.gz
GeoIP的各种语言的API可以从这里下载：http://www.maxmind.com/download/geoip/api/
操作Steps如下：
Step 1：Hive所需添加的IP地址信息识别UDF函数如下：
package org.hadoop.hive.additionalUDF; import java.io.File; import java.io.IOException; import org.apache.hadoop.hive.ql.exec.UDF; import com.maxmind.geoip.Location; import com.maxmind.geoip.LookupService; import com.maxmind.geoip.regionName; import com.maxmind.geoip.timeZone; import java.util.regex.*; public class IPToCC extends UDF { private static LookupService cl = null; private static String ipPattern = "\\d+\\.\\d+\\.\\d+\\.\\d+"; private static String ipNumPattern = "\\d+"; static LookupService getLS(String dbfile) throws IOException{ //String sep = System.getProperty("file.separator"); //String dir = "/home/landen/UntarFile/GeoIP"; //String dbfile = dir + sep + "GeoLiteCity.dat"; //String dbfile = "GeoLiteCity.dat"; if(new File(dbfile).exists()) { if(cl == null) { cl = new LookupService(dbfile,LookupService.GEOIP_MEMORY_CACHE); } } return cl; } /** * @param str like "114.43.181.143" * */ public String evaluate(String str,String ipDBInfo) { try { Location l1 = null; Matcher mIP = Pattern.compile(ipPattern).matcher(str); Matcher mIPNum = Pattern.compile(ipNumPattern).matcher(str); if(mIP.matches()) l1 = getLS(ipDBInfo).getLocation(str); else if(mIPNum.matches()) l1 = getLS(ipDBInfo).getLocation(Long.parseLong(str)); /*System.out.println("countryCode: " + l1.countryCode + "\n countryName: " + l1.countryName + "\n region: " + l1.region + "\n regionName: " + regionName.regionNameByCode(l1.countryCode, l1.region) + "\n city: " + l1.city + "\n latitude: " + l1.latitude + "\n longitude: " + l1.longitude + "\n timezone: " + timeZone.timeZoneByCountryAndRegion(l1.countryCode, l1.region));*/ return String.format("%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s",l1.countryCode,l1.countryName,l1.region,regionName.regionNameByCode(l1.countryCode, l1.region),l1.city,l1.latitude,l1.longitude,timeZone.timeZoneByCountryAndRegion(l1.countryCode, l1.region)); } catch(Exception e) { e.printStackTrace(); if(cl != null) cl.close(); return null; } } public static void main(String[] args) { String dbfile = "GeoLiteCity.dat"; IPToCC ipTocc = new IPToCC(); String ipAdress = "221.12.10.218"; System.out.println(ipTocc.evaluate(ipAdress,dbfile)); } }
Step 2.将以上程序和GeoIP的API程序，一起打成JAR包IPToCC.jar，和数据文件（GeoLiteCity.dat）一起放到Hive所在的服务器的一个位置。然后可以按照以下两种方式将以上资源添加到Hive中：
1> 打开Hive执行以下语句：
landen@Master:~/UntarFile/hive-0.10.0$ bin/hive
WARNING: org.apache.hadoop.metrics.jvm.EventCounter is deprecated. Please use org.apache.hadoop.log.metrics.EventCounter in all the log4j.properties files.
Logging initialized using configuration in jar:file:/home/landen/UntarFile/hive-0.10.0/lib/hive-common-0.10.0.jar!/hive-log4j.properties
Hive history file=/home/landen/UntarFile/hive-0.10.0/logs/hive_job_log_landen_201312081638_1930432077.txt
hive (default)> use stuchoosecourse;
OK
Time taken: 5.251 seconds
hive (stuchoosecourse)> add file /home/landen/UntarFile/GeoIP/GeoLiteCity.dat;
Added resource: /home/landen/UntarFile/GeoIP/GeoLiteCity.dat
hive (stuchoosecourse)> add jar /home/landen/UntarFile/hive-0.10.0/lib/IPTocc.jar;
Added /home/landen/UntarFile/hive-0.10.0/lib/IPTocc.jar to class path
Added resource: /home/landen/UntarFile/hive-0.10.0/lib/IPTocc.jar
hive (stuchoosecourse)> create temporary function IP4Tocc as 'org.hadoop.hive.additionalUDF.IPToCC';
OK
Time taken: 0.107 seconds
2> 在启动hive shell命令前，在$HIVE_HOME/conf目录下添加.hiverc文件，然后添加如下内容：
add file /home/landen/UntarFile/GeoIP/GeoLiteCity.dat;
add jar /home/landen/UntarFile/hive-0.10.0/lib/IPTocc.jar;
create temporary function IP4Tocc as 'org.hadoop.hive.additionalUDF.IPToCC';
当启动hive shell命令后，hive会将加载.hiverc文件内容并添加到全局内容中，便于client使用
Step 3：Hive测试内容如下：
hive (stuchoosecourse)> select * from ipidentifier;
OK
ipadress
221.12.10.218
60.180.248.201
125.111.251.118
Time taken: 0.099 seconds
hive (stuchoosecourse)> select IP4Tocc(ipadress,'./GeoLiteCity.dat') from ipidentifier;
Total MapReduce jobs = 1
Launching Job 1 out of 1
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_201312042044_0020, Tracking URL = http://Master:50030/jobdetails.jsp?jobid=job_201312042044_0020
Kill Command = /home/landen/UntarFile/hadoop-1.0.4/libexec/../bin/hadoop job -kill job_201312042044_0020
Hadoop job information for Stage-1: number of mappers: 1; number of reducers: 0
2013-12-08 20:54:10,276 Stage-1 map = 0%, reduce = 0%
2013-12-08 20:54:18,308 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 2.55 sec
2013-12-08 20:54:19,313 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 2.55 sec
2013-12-08 20:54:20,317 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 2.55 sec
2013-12-08 20:54:21,322 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 2.55 sec
2013-12-08 20:54:22,326 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 2.55 sec
2013-12-08 20:54:23,331 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 2.55 sec
2013-12-08 20:54:24,402 Stage-1 map = 100%, reduce = 100%, Cumulative CPU 2.55 sec
MapReduce Total cumulative CPU time: 2 seconds 550 msec
Ended Job = job_201312042044_0020
MapReduce Jobs Launched:
Job 0: Map: 1 Cumulative CPU: 2.55 sec HDFS Read: 306 HDFS Write: 188 SUCCESS
Total MapReduce CPU Time Spent: 2 seconds 550 msec
OK
_c0
CN China 02 Zhejiang Hangzhou 30.293594 120.16141 Asia/Shanghai
CN China 02 Zhejiang Wenzhou 27.999405 120.66681 Asia/Shanghai
CN China 02 Zhejiang Ningbo 29.878204 121.5495 Asia/Shanghai
hive (stuchoosecourse)> select split(IP4Tocc(ipadress,'./GeoLiteCity.dat'),'\t') from ipidentifier;
Total MapReduce jobs = 1
Launching Job 1 out of 1
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_201312042044_0021, Tracking URL = http://Master:50030/jobdetails.jsp?jobid=job_201312042044_0021
Kill Command = /home/landen/UntarFile/hadoop-1.0.4/libexec/../bin/hadoop job -kill job_201312042044_0021
Hadoop job information for Stage-1: number of mappers: 1; number of reducers: 0
2013-12-08 21:12:46,717 Stage-1 map = 0%, reduce = 0%
2013-12-08 21:12:56,764 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 4.28 sec
2013-12-08 21:12:57,768 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 4.28 sec
2013-12-08 21:12:58,772 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 4.28 sec
2013-12-08 21:12:59,775 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 4.28 sec
2013-12-08 21:13:00,778 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 4.28 sec
2013-12-08 21:13:01,782 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 4.28 sec
2013-12-08 21:13:02,786 Stage-1 map = 100%, reduce = 100%, Cumulative CPU 4.28 sec
MapReduce Total cumulative CPU time: 4 seconds 280 msec
Ended Job = job_201312042044_0021
MapReduce Jobs Launched:
Job 0: Map: 1 Cumulative CPU: 4.28 sec HDFS Read: 306 HDFS Write: 188 SUCCESS
Total MapReduce CPU Time Spent: 4 seconds 280 msec
OK
_c0
["CN","China","02","Zhejiang","Hangzhou","30.293594","120.16141","Asia/Shanghai"]
["CN","China","02","Zhejiang","Wenzhou","27.999405","120.66681","Asia/Shanghai"]
["CN","China","02","Zhejiang","Ningbo","29.878204","121.5495","Asia/Shanghai"]
Time taken: 45.037 seconds
hive (stuchoosecourse)> create table HiddenIPInfo(
> IP string,countrycode string,countryname string,region string,regionname string,city string,
> latitude string,longitude string,timezone string);
OK
Time taken: 1.828 seconds
hive (stuchoosecourse)> show tables;
OK
tab_name
hbase_stu_course
hiddenipinfo
ipidentifier
Time taken: 0.486 seconds
hive (stuchoosecourse)> describe hiddenipinfo;
OK
col_name data_type comment
ip string
countrycode string
countryname string
region string
regionname string
city string
latitude string
longitude string
timezone string
Time taken: 0.33 seconds
hive (stuchoosecourse)> from(select ipadress,split(IP4Tocc(ipadress,'./GeoLiteCity.dat'),'\t') as IPInfo from ipidentifier)e
> insert overwrite table hiddenipinfo
> select e.ipadress,e.IPInfo[0] as countrycode,e.IPInfo[1] as countryname,e.IPInfo[2] as region,
> e.IPInfo[3] as regionname,e.IPInfo[4] as city,e.IPInfo[5] as latitude,e.IPInfo[6] as longitude,
> e.IPInfo[7] as timezone;
Total MapReduce jobs = 3
Launching Job 1 out of 3
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_201312042044_0023, Tracking URL = http://Master:50030/jobdetails.jsp?jobid=job_201312042044_0023
Kill Command = /home/landen/UntarFile/hadoop-1.0.4/libexec/../bin/hadoop job -kill job_201312042044_0023
Hadoop job information for Stage-1: number of mappers: 1; number of reducers: 0
2013-12-08 21:58:12,406 Stage-1 map = 0%, reduce = 0%
2013-12-08 21:58:18,449 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 1.48 sec
2013-12-08 21:58:19,454 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 1.48 sec
2013-12-08 21:58:20,458 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 1.48 sec
2013-12-08 21:58:21,462 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 1.48 sec
2013-12-08 21:58:22,466 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 1.48 sec
2013-12-08 21:58:23,470 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 1.48 sec
2013-12-08 21:58:24,474 Stage-1 map = 100%, reduce = 100%, Cumulative CPU 1.48 sec
MapReduce Total cumulative CPU time: 1 seconds 480 msec
Ended Job = job_201312042044_0023
Ended Job = 39195028, job is filtered out (removed at runtime).
Ended Job = 1695434910, job is filtered out (removed at runtime).
Moving data to: hdfs://Master:9000/home/landen/UntarFile/hive-0.10.0/warehouse/hive_2013-12-08_21-57-40_106_7083774091282915969/-ext-10000
Loading data to table stuchoosecourse.hiddenipinfo
Deleted hdfs://Master:9000/home/landen/UntarFile/hive-0.10.0/warehouse/stuchoosecourse.db/hiddenipinfo
Table stuchoosecourse.hiddenipinfo stats: [num_partitions: 0, num_files: 1, num_rows: 0, total_size: 233, raw_data_size: 0]
3 Rows loaded to hiddenipinfo
MapReduce Jobs Launched:
Job 0: Map: 1 Cumulative CPU: 1.48 sec HDFS Read: 306 HDFS Write: 233 SUCCESS
Total MapReduce CPU Time Spent: 1 seconds 480 msec
OK
ipadress countrycode countryname region regionname city latitude longitude timezone
Time taken: 45.692 seconds
hive (stuchoosecourse)> show tables;
OK
tab_name
hbase_stu_course
hiddenipinfo
ipidentifier
Time taken: 0.053 seconds
hive (stuchoosecourse)> select * from hiddenipinfo;
OK
ip countrycode countryname region regionname city latitude longitude timezone
221.12.10.218 CN China 02 Zhejiang Hangzhou 30.293594 120.16141 Asia/Shanghai
60.180.248.201 CN China 02 Zhejiang Wenzhou 27.999405 120.66681 Asia/Shanghai
125.111.251.118 CN China 02 Zhejiang Ningbo 29.878204 121.5495 Asia/Shanghai
Time taken: 0.083 seconds