http_client.pl

Accessing a Mascot Server using http

#!/usr/local/bin/perl
##############################################################################
# File: http_client.pl                                                       #
# Mascot Parser toolkit example code                                         #
##############################################################################
# COPYRIGHT NOTICE                                                           #
# Copyright 1998-2015 Matrix Science Limited  All Rights Reserved.           #
#                                                                            #
##############################################################################
#     $Source: parser/examples/test_perl/http_client.pl $
#     $Author: villek@matrixscience.com $
#       $Date: 2018-07-30 16:23:53 +0100 $
#   $Revision: 1b450440f9c97e1e41d0fc6016a27d68951d4532 | MSPARSER_REL_3_0_0-2024-09-24-0-g93ebaeb4f4 $ 
##############################################################################
use strict;
use warnings;
##############################################################################
 
# required Perl packages
use lib "../bin";
use msparser;
use Getopt::Long;
use File::Basename;
 
my ($url, $cmd, $mgf, $taskID, $accession, $database, $username, $password);
my %allowedCommands = ("search" => 1, "getresults" => 1, "getseq" => 1);
$username = "";
$password = "";
 
GetOptions('url=s'      => \$url,
           'cmd=s'      => \$cmd,
           'mgf=s'      => \$mgf,
           'taskID=s'   => \$taskID,
           'accession=s'=> \$accession,
           'database=s' => \$database,
           'username=s' => \$username,
           'password=s' => \$password); 
 
if ((not defined $url or not defined $cmd or not exists($allowedCommands{$cmd}))
    or ($cmd eq 'search' and not defined $mgf)
    or ($cmd eq 'getresults' and not defined $taskID)
    or ($cmd eq 'getseq' and (not defined $accession or not defined $database))
   ) {
   print <<USAGE;
 
Usage http_client.pl --url=<URL> --cmd=<command> [options]
 
    --url         Mascot Server URL of the form 
                    http://your-server/mascot/cgi/
    --cmd         One of the following
                    --cmd=search to submit a Mascot search
                    --cmd=getresults to download the search results file
                    --cmd=getseq to save protein sequence to XML file
    --mgf         Path to an MGF peak list file
                    Required if --cmd=search
    --taskID      Mascot task ID for the search
                    Required if --cmd=getresults
    --accession   Accession string for protein
                    Required if --cmd=getseq
    --database    Mascot database name
                    Required if --cmd=getseq
    --username    Mascot Server user name
                    May be required if Mascot Security is enabled
    --password    Mascot Server password
                    May be required if Mascot Security is enabled
 
Note that the Mascot search parameters are hard coded towards the end 
of this script, and may need modifying for different peak lists.
 
USAGE
  exit 1;
}
 
# Common code for all cases
 
# Create connection settings. 
# Any settings for web server authentication or a proxy server should be set here
my $objSettings = new msparser::ms_connection_settings;
$objSettings->setUserAgent("PerlScriptTest/1.0 " . $objSettings->getUserAgent);
 
# Connect to the server
my $objHttpClient = new msparser::ms_http_client($url, $objSettings);
unless ($objHttpClient->isValid) {
  showErrorsAndExit($objHttpClient);
}
 
# Enable logging, mostly for debugging
$objHttpClient->getErrorHandler->setLoggingFile("log.txt", $msparser::ms_errs::sev_debug3);
 
# Create a Mascot security session
# If security is disabled, the empty user name and password return the 'all_secdisabledsession' session
my $objSession = new msparser::ms_http_client_session;
my $loginReturnCode = $objHttpClient->userLogin($username, $password, $objSession);
if ($loginReturnCode != $msparser::ms_http_client::L_SUCCESS and $loginReturnCode != $msparser::ms_http_client::L_SECURITYDISABLED) {
  die "Failed to login to $url as user $username. return code is $loginReturnCode\n";
} else {
  print "Logged in to $url with session: " . $objSession->sessionId() . "\n";
}
 
if ($cmd eq 'search') {
  doSearch();
}
if ($cmd eq 'getresults') {
  doGetResults();
}
if ($cmd  eq 'getseq') {
  doGetSeq();
}
 
# and finally logout
$objSession->logout();
 
 
sub doSearch {    
    
# The input to Mascot is the peak list and search parameters in MIME format
  my $httpHeader = "Content-Type: multipart/mixed; boundary=---------FormBoundary4C9ByKKVofH";
  my $prologue = "";
  while (<DATA>){
    $prologue .= $_;
  } 
  my $epilogue = "-----------FormBoundary4C9ByKKVofH--";
  
# Request a Mascot task id for the search
  my $taskID = "";
  my $objSearch = new msparser::ms_http_client_search($objSession, $taskID);
  unless ($objSearch->isValid) {
    showErrorsAndExit($objSearch);
  }
   
# Submit the search
  print "Submitting search\n";
  my $objProgress = new msparser::ms_http_helper_progress;
  if ($objSession->submitSearch($objSearch, $httpHeader, $prologue, $mgf, $epilogue, $objProgress)) {
    print "Search submitted. Task ID = " . $objSearch->searchTaskId . "\n";
    my ($success, $returnValue);
    my $returnCode = $msparser::ms_http_client_search::SS_UNKNOWN;
    my $complete = 0;
    while (!$complete) {
        ($success, $returnCode, $returnValue) = $objSearch->getStatus();
        if (!$success) {
          print "Unable to call getStatus()\n";
          exit 1;
        }
      if    ($returnCode == $msparser::ms_http_client_search::SS_UNKNOWN   ) { 
        print "Unknown search id\n";   $complete = 0; 
      } elsif ($returnCode == $msparser::ms_http_client_search::SS_ASSIGNED) { 
        print "Search not started\n";  $complete = 0; 
      } elsif ($returnCode == $msparser::ms_http_client_search::SS_QUEUED  ) { 
        print "Search queued\n";       $complete = 0; 
      } elsif ($returnCode == $msparser::ms_http_client_search::SS_RUNNING ) { 
        print "Seach running: $returnValue\% complete\n";     
      } elsif ($returnCode == $msparser::ms_http_client_search::SS_COMPLETE) { 
        print "Search complete\n";     $complete = 1; 
      } elsif ($returnCode == $msparser::ms_http_client_search::SS_ERROR   ) { 
        print "Search error : $returnValue \n"; $complete = 1;
      } elsif ($returnCode == $msparser::ms_http_client_search::SS_SEARCH_CONTROL_ERROR) {
        print "Search control  error: $returnValue\n"; $complete = 1;
      }
        sleep 1;
    }
    
  # Search complete - get the relative path to the result file
    my ($ok, $remoteFileName) = $objSearch->getResultsFileName();
    unless ($ok) {
      print "Unable to get filename for results file";
      showErrorsAndExit($objHttpClient);
    }
    
  # Display URL for search result report
    print "Search result report can be viewed at this URL:\n" 
      . $objHttpClient->baseUrl
      . "master_results_2.pl?file="
      . $remoteFileName
      . "\n";
 
  } else {
    print "Failed to submit search\n";
    showErrorsAndExit($objSession);
  }
}
 
sub doGetResults {
  my $objSearch = new msparser::ms_http_client_search($objSession, $taskID);
  unless ($objSearch->isValid) {
    showErrorsAndExit($objSearch);
  }
 
  my ($ok, $remoteFileName) = $objSearch->getResultsFileName();
  unless ($ok && $remoteFileName) {
    print "Unable to get filename for results file";
    exit -1;
  }
 
  my ($localFileName, undef, undef) = fileparse($remoteFileName);
  print "Downloading remote results file ($remoteFileName) to $localFileName\n";
  my $objProgress = new msparser::ms_http_helper_progress;
# Download the complete, uncompressed result file
  if ($objSearch->downloadResultsFile($localFileName, $objProgress)) {
    print "Downloaded remote file to $localFileName \n";
  } else {
    print "Failed to downloaded remote file to $localFileName \n";
    showErrorsAndExit($objSearch);
  }
}
 
 
sub doGetSeq {
  my $accessions = new msparser::VectorString;
  my $frames     = new msparser::vectori;
  $accessions->push($accession);
  my $localFileName = "sequence.xml";
  if ($objSession->getSequenceFile($database, $accessions, $frames, $localFileName)) {
    print "Successfully saved sequence to $localFileName \n";
  } else {
    print "Failed to get sequence or failed to save it to $localFileName \n";
    showErrorsAndExit($objSession);
  }
}
 
 
sub showErrorsAndExit {
    my ($obj) = @_;
    print "Error: ", $obj->getLastErrorString(), "\n";
 
    my $err = $obj->getErrorHandler();
    my $i;
 
    for my $i (1 .. $err->getNumberOfErrors) {
        print "Error number: ";
        print $err->getErrorNumber($i);
        print " (";
        print $err->getErrorRepeats($i) + 1;
        print " times) : ";
        print $err->getErrorString($i), "\n";
    }
 
    print "\n";
 
    exit -1;
}
 
 
__END__
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="FORMVER"
 
1.01
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="SEARCH"
 
MIS
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="COM"
 
Perl script test
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="DB"
 
SwissProt
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="CLE"
 
Trypsin/P
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="PFA"
 
1
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="QUANTITATION"
 
None
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="TAXONOMY"
 
. . . . . . . . . . . . . . . . Homo sapiens (human)
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="MODS"
 
Carbamidomethyl (C)
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="IT_MODS"
 
Oxidation (M)
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="TOL"
 
10
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="TOLU"
 
ppm
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="PEP_ISOTOPE_ERROR"
 
1
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="ITOL"
 
0.1
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="ITOLU"
 
Da
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="CHARGE"
 
2+
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="MASS"
 
Monoisotopic
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="FORMAT"
 
Mascot generic
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="INSTRUMENT"
 
ESI-TRAP
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="REPORT"
 
AUTO
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="USERNAME"
 
Mascot Parser Test
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="USEREMAIL"
 
dcreasy@matrixscience.com
-----------FormBoundary4C9ByKKVofH
Content-Disposition: form-data; name="FILE"; filename="test.mgf"