perl 关键词检索程序

来源：互联网发布：python开源运维编辑：程序博客网时间：2024/06/08 12:36

#找出语音对应的sent文件，将关键词列表中的关键词在sent文件中的时间点输出到一个文件中
#################在下面的文本中检索关键词#####################

然后你妹妹怎么打电话呀
26.1 26.36 26.36 26.48 26.48 27 27 27.24 27.24 27.86 27.86 28.12
0.0000
只能扣费打电话市话一毛二长途两毛我看有没有别的扣费的稍等
29.23 29.5 29.5 29.5 29.5 30.12 30.12 30.44 30.44 30.54 30.54 30.66 30.66 30.84 30.84 31.11 31.11 31.26 31.26 31.42 31.42 31.64 31.64 31.91 31.91 32.18 32.18 32.54 32.54 32.81 32.81 33.2
0.0030
啊
42.83 43
0.0000
只要您今日取消套餐好
44.05 44.21 44.21 44.4 44.4 44.74 44.74 45.13 45.13 45.44 45.44 45.71
0.0100
对呀五元那套餐是没有了吧
46.76 46.92 46.92 47.16 47.16 47.62 47.62 47.77 47.77 47.93 47.93 48.28 48.28 48.38 48.38 48.81 48.81 49.06
0.0020
然后身份证号身份证号记得吧
54.57 54.87 54.87 55.2 55.2 55.39 55.39 55.391 55.391 55.392 55.39 55.62 55.62 55.84
0.0000
啊行那祝您愉快
80.67 81.11 81.11 81.33 81.33 81.52 81.52 81.62 81.62 81.74 81.74 82.08
0.0000
################################################################

#!/usr/bin/perl
use Encode;
use Thread;
# use Encode;
#use threads::shared;

# $dat="测试文本";
# $str=decode("gb2312",$dat);

# print encode("gb2312",$str),"\n";

if($#ARGV < 2 )
{
print "usage:perl key_search.pl wavlist keywordlist thread_count result \n";
exit(2);
}
#find ./ -name "*.wav" > wav.lst

my($wavlist,$keywordlist,$thread_count,$keyresult)=@ARGV;

open(WAV, "<$wavlist") or die "1\n";

open(KEYWORD, "<$keywordlist") or die "3\n";
open(RESULT, ">$keyresult") or die "4\n";

my @wav = ();
my @wkeyword = ();

@wav = <WAV>;
@wkeyword = <KEYWORD>;
#print ("@wkeyword\n");

$icount=0;

#foreach my $fwav (@wav) {
#next unless defined $fwav;
# chomp $fwav;

# $threads[$tempcount]=Thread->new(\&start_thread,$fwav,\@wsent,\@wkeyword);
# $tempcount++;
# }
#foreach my $thread (@threads) {
# $thread->join();
#}

while($icount<scalar(@wav))
{ while(scalar(threads->list())<$thread_count)
{
chomp($wav[$icount]);
threads->new(\&start_thread,$wav[$icount],\@wkeyword);
$icount++;
}
foreach $thread(threads->list(threads::all))
{ if($thread->is_joinable())
{ $thread->join();
}
}
}
foreach $thread(threads->list(threads::all))
{ $thread->join();
}

sub start_thread{
local ($lwav,$nkeyword)=@_;
local @keyword=@{$nkeyword};
# print ("paramater is @keyword\n");
#print ("paramater is $lwav\n");
local $pos=0;
local $kresult;
local $count=0;
local $leap=0;
local @posarray=();
local @resultarray=();
local @tmpresult=();
local $lkeyword;
local $find=0;
local @findkey=();
#$id=Thread->self->tid();
#print (RESULT "in thread $id :\n");
# print "in thread $id :\n";
print (RESULT "$lwav\n");
$lwav=~s/\.wav/_sent.txt/;
local $lsent=$lwav;
# print " sent file is $lsent\n";

for $lkeyword (@keyword)
{
chomp($lkeyword);
#print ("line keyword is $lkeyword\n");
@segkeyword = split( /=/ , $lkeyword );
#print "segFile is @segFile\n";
$tmpkey=$segkeyword[0];
$dtmpkey=decode("gb2312",$tmpkey);
open(FSENT, "<$lsent") or die "5\n";
while($flsent=<FSENT>)
{
chomp($flsent);
$integrate=$flsent;
$integrate=~s/\s//g;
$dintegrate=decode("gb2312",$integrate);
#print ("sent line is $flsent\n");
# print ("sent line is $flsent\n");
@segsent = split( /\s/ , $flsent );
if($flsent =~ /[\x80-\xFF]+?/)
{
# print "flsent is $flsent\n";
$leap=0;
$count=0;
for my $lresult (@segsent)
{
$dresult=decode("gb2312",$lresult);
if($dresult=~/$dtmpkey/)
{
$find=1;
push @findkey, $tmpkey;
push @posarray, $count;
push @resultarray, $lresult;
#$pos=$count;
#print(RESULT "$lresult ");
}
$count++;
}

if(($dintegrate=~/$dtmpkey/)&&($find==0))
{
$count=0;
$len=length($dtmpkey);
for ($ilen=1;$ilen<$len;$ilen++)
{
$end=$len-$ilen;
$dtmpstr=substr($dtmpkey,0,$end);
#print encode("gb2312",$str),"\n";
$dstrtmp=encode("gb2312",$dtmpstr);
print ("partstr is $dstrtmp\n");
for my $lresult (@segsent)
{
$dresult=decode("gb2312",$lresult);
if($dresult=~/$dtmpstr/)
{
$find=1;
push @findkey, $tmpkey;
push @posarray, $count;
push @resultarray, $lresult;
#$pos=$count;
#print(RESULT "$lresult ");
}
$count++;
}
if($find==1)
{
last; #already find longest substring.so break the loop
}

}

}
}

if($leap==1)
{
if($find==1)
{
$inumber=@posarray;
for($i=0;$i<$inumber;$i++)
{
$pos=@posarray[$i];
#print(RESULT "@resultarray[$i] @segsent[$pos*2] @segsent[$pos*2+1] ");
#$tmpline="@resultarray[$i] @segsent[$pos*2] @segsent[$pos*2+1]";
$tmpline="@findkey[$i] @segsent[$pos*2] @segsent[$pos*2+1]";
push @tmpresult,$tmpline;
}
#for my $pos (@posarray) and for my $kresult (@resultarray)
#{
# print(RESULT "$kresult @segsent[$pos*2] @segsent[$pos*2+1] ");
#}
#print(RESULT "\n");
$find=0;
@findkey=();
@posarray=();
@resultarray=();

}

}
if($leap==2)
{
$inumber1=@tmpresult;
for($j=0;$j<$inumber1;$j++)
{
$tmp=@tmpresult[$j];
print(RESULT "$tmp $flsent\n");
}
@tmpresult=();
next;
}
$leap++;

}
close(FSENT);

}
}

close(WAV);

close(KEYWORD);
close(RESULT);

0 0