用Perl来Post网页数据

Apr 14	用Perl来Post网页数据
	Categories: BIO, Perl \| Tags: Perl \| Views: 2,042

我只能说，How 强大 Perl is！

我用过不同的方法来实现网页的post，但是perl是最方便最好用的！

废话不多说，看代码(这是一段实际工作中用到的代码，也可以修改一下用作其他用途，严禁做坏事：）） )：

#!/usr/bin/perl
#
# Panwen Wang, April 7th, 2009
# Submit a query to http://consurf.tau.ac.il/
# to get conservation scores of each residue.
#
 
 
use URI::Escape;
use LWP::UserAgent;
use LWP::Simple;
use HTTP::Request::Common qw(POST);
use strict;
 
my $ua = LWP::UserAgent->new;
 
if(@ARGV<1){
    print STDERR "\n $0: \n";
    print STDERR "\tPanwen Wang, April 7th, 2009";
    print STDERR "\tSubmit a query to http://consurf.tau.ac.il/ \n";
    print STDERR "\tto get conservation scores of each residue.\n";
    print STDERR "\n Parameters:\n";
    print STDERR "\t-p <pdb_ID>         pdb id in four letters\n";
    print STDERR "\t-c <chain>          chain label of given pdb\n";
    print STDERR "\t-a <algorithm>      Bayes/LikelihoodML\n";
    print STDERR "\t\tDEFAULT:Bayes\n";
    print STDERR "\t-m <MSAprogram>     MUSCLE/CLUSTALW\n";
    print STDERR "\t\tDEFAULT: MUSCLE\n";
    print STDERR "\t-d <DATABASE>       the Homologues database, SWISS-PROT/Uniprot\n";
    print STDERR "\t\tDEFAULT: SWISS-PROT\n";
    print STDERR "\t-n <MAX_NUM_HOMOL>  Max. Number of Homologues\n";
    print STDERR "\t\tDEFAULT: 50\n";
    print STDERR "\t-i <ITERATIONS>     No. of PSI-BLAST Iterations\n";
    print STDERR "\t\tDEFAULT: 1\n";
    print STDERR "\t-e <ESCORE>         PSI-BLAST E-value Cutoff\n";
    print STDERR "\t\tDEFAULT: 0.001\n";
    print STDERR "\t-s <MATRIX>         Model of substitution for proteins\n";
    print STDERR "\t\tJTT(DEFAULT)/mtREV(for mitochondrial proteins)/cpREV(for chloroplasts proteins)/WAG/Dayhoff\n";
    print STDERR "\t-l <email>          user's email\n";
    print STDERR "\t-o <output folder>  where the result files(1A0O_A.score) restore\n";
    print STDERR "\n Usage:\n";
    print STDERR "\t$0 -p 2UUY -c A -l xxx\@xxx.com -o ./conservation\n";
    print STDERR "\t$0 -p 2UUY -c A -l xxx\@xxx.com -o ./conservation -a LikelihoodML ...\n\n";
    exit;
}
 
my ($temp, $pdb_ID, $chain, $user_email, $ALGORITHM, $MSAprogram, $DATABASE, $MAX_NUM_HOMOL, $ITERATIONS, $ESCORE, $MATRIX, $out_dir);
 
while(@ARGV){
    $temp = shift;
    if($temp eq "-p") { $pdb_ID = shift; }
    elsif($temp eq "-c") { $chain = shift; }
    elsif($temp eq "-a") { $ALGORITHM = shift; }
    elsif($temp eq "-m") { $MSAprogram = shift; }
    elsif($temp eq "-d") { $DATABASE = shift; }
    elsif($temp eq "-n") { $MAX_NUM_HOMOL = shift; }
    elsif($temp eq "-i") { $ITERATIONS = shift; }
    elsif($temp eq "-e") { $ESCORE = shift; }
    elsif($temp eq "-s") { $MATRIX = shift; }
    elsif($temp eq "-l") { $user_email = shift; }
    elsif($temp eq "-o") { $out_dir = shift; }
    else {
        print STDERR "Warning: argument $temp igorned.\n";
    }
}
 
die "Error: argument '-p' is essential.\n" if(!defined($pdb_ID));
die "Error: argument '-c' is essential.\n" if(!defined($chain));
die "Error: argument '-l' is essential.\n" if(!defined($user_email));
die "Error: argument '-o' is essential.\n" if(!defined($out_dir));
$ALGORITHM = "Bayes"                       if(!defined($ALGORITHM));
$MSAprogram = "MUSCLE"                     if(!defined($MSAprogram));
$DATABASE = "SWISS-PROT"                   if(!defined($DATABASE));
$MAX_NUM_HOMOL = "50"                      if(!defined($MAX_NUM_HOMOL));
$ITERATIONS = "1"                          if(!defined($ITERATIONS));
$ESCORE = "0.001"                          if(!defined($ESCORE));
$MATRIX = "JTT"                            if(!defined($MATRIX));
 
# end of handling parameters
 
# read and encode the queries
#my $encoded_query;
##foreach my $query (@ARGV)
##    {
##    open(QUERY,$query);
##    while(<QUERY>)
##        {
##        $encoded_query = $encoded_query . uri_escape($_);
##        }
# }
 
# build the request
my $args = "pdb_ID=$pdb_ID&chain=$chain&user_email=$user_email&ALGORITHM=$ALGORITHM&MSAprogram=$MSAprogram&DATABASE=$DATABASE&MAX_NUM_HOMOL=$MAX_NUM_HOMOL&ITERATIONS=$ITERATIONS&ESCORE=$ESCORE&MATRIX=$MATRIX&send_user_mail=yes&BayesInterval=3";
 
print "Querying ...\n";
 
my $req = new HTTP::Request POST => 'http://consurf.tau.ac.il/cgi-bin/consurf.cgi';
$req->content_type('application/x-www-form-urlencoded');
$req->content($args);
 
# get the response
my $response = $ua->request($req);
 
# parse out the request id
$response->content =~ /The document has moved <a href=\"(.+)\">here/;
 
my $result_page = $1;
 
print "Waiting ...\n";
sleep 25;
 
# poll for results
for(;;){
    sleep 5;
 
    my $req = new HTTP::Request GET => $result_page;
    $response = $ua->request($req);
 
    my $content = $response->content;
    if($content =~ /ConSurf calculation is finished/){
        print "Getting result ...\n";
        my $prefix_url = $1 if($result_page =~ /(.+)output\.html/);
        my $result_file = "${prefix_url}pdb".lc($pdb_ID).".gradesPE";
        print "Saving result ...\n";
        getstore($result_file, "$out_dir/${pdb_ID}_$chain.score");
        #my $req1 = new HTTP::Request GET => $result_file;
        #my $response1 = $ua->request($req1);
        #open(OUT,">$out_dir/${pdb_ID}_$chain.score");
        #print OUT  $response1->content;
        last;
    } elsif($content =~ /For assistance please/){
        die("ERROR! ConSurf session has been terminated.\n");
    } else {
        print "Waiting ...\n";
        next;    
    }
    die("Sth. unexpected happened.\n");
} # end poll loop