Perl分析xml文件并生成csv数据文件，同时写入数据库

#!/usr/bin/perl

use XML::Simple;
use Data::Dumper;
use DBI;
use Encode;
use strict;
use encoding "gbk";

#判断是否传递XML文件参数
if (open(MYFILE, "$ARGV[0]")) {
    # here's what to do if the file opened successfully
    print "\n\"$ARGV[0]\" : file exist! parse beginning!\n";
    #获得EMS名称
    my @ems = split(/\_/,$ARGV[0]);
    print "@ems[1]\n";
    
    #获得XML文件数据
    my $data = XMLin($ARGV[0]);
    #print Dumper($data);

    #打印文件时间戳
    print "Timestamp = $data->{'timestamp'}\n\n";

    my $i = 0;
    #判断是否存在下一条网元记录
    print "appending to csv...\n";
    while( $data->{'output'}->{'TopoMgr.NEAdditionalInfoType-array'}->{'TopoMgr.NEAdditionalInfoType'}->[$i] )    
    {
        print "\@ record num = $i\n";
        
        my $text = "$data->{'output'}->{'TopoMgr.NEAdditionalInfoType-array'}->{'TopoMgr.NEAdditionalInfoType'}->[$i]->{'neName'}\n";
        
        print $text;
        
        #将分析出的数据追加到FILE末尾
        open(APPENDFILE, ">>perlne.dat");

        #转为utf-8后不报字符串过长异常，不转不影响效果
        Encode::_utf8_off($text); 
        print APPENDFILE ($text);
        close(APPENDFILE);
        
        # Connect to target DB
        #my $dbh=DBI->connect("DBI:mysql:test:192.168.1.14", 'root', 'root');
        my $dbh = DBI->connect( 'DBI:Oracle:ora10g','res_col','res_col');

        #$dbh->do("SET character_set_client = 'utf-8'");
        #$dbh->do("SET character_set_connection = 'utf-8'");
        #$dbh->do("SET character_set_results= 'utf-8'");
        #$dbh->do("SET names  'gbk'");
        #Encode::_utf8_off($nename); 

        # Insert one row 
        my $rows = $dbh->do("INSERT INTO neinfo(text) VALUES ('$text')");
        
        $dbh->disconnect();
        
        $i ++;
    }
    
    close(MYFILE);
    # query 
    my $dbh = DBI->connect( 'DBI:Oracle:ora10g','res_col','res_col');
    my $sqr = $dbh->prepare("SELECT * FROM neinfo");
    $sqr->execute();

    while ( my @row=$sqr->fetchrow_array() )
    {
           #print join('\t', @row)."\n";
           print join(',', @row)."\n";
    }

    $sqr->finish();
    $dbh->disconnect();
    
    print 'END...';
    
}unless (open (MYFILE, "$ARGV[0]")) {
    die ("cannot open input file $ARGV[0]\n");
    close(MYFILE);
}

 

xml如下

<?xml version="1.0" encoding="GBK"?>
<outputlist timestamp="20090917232423">
<output>
<TopoMgr.NEAdditionalInfoType-array>
  <TopoMgr.NEAdditionalInfoType>
    <neId>134217729</neId>
    <neName>测试1 [1-1]</neName>
  </TopoMgr.NEAdditionalInfoType>
  <TopoMgr.NEAdditionalInfoType>
    <neId>134217731</neId>
    <neName>测试2 [1-2]</neName>
  </TopoMgr.NEAdditionalInfoType>
</TopoMgr.NEAdditionalInfoType-array>
</output>
</outputlist>

 

 

运行perl脚本时需要跟一个文件名的参数