dbfetch

#!/usr/local/bin/perl -- # -*-Perl-*-

=head1 NAME

dbfetch - generic CGI program to retrieve biological database entries
in various formats and styles (using SRS)

=head1 SYNOPSIS

  # URL examples:

  # prints the interactive page with the HTML form
  http://www.ebi.ac.uk/cgi-bin/dbfetch

  # for backward compatibility, implements <ISINDEX>
  # single entry queries defaulting to EMBL sequence database
  http://www.ebi.ac.uk/cgi-bin/dbfetch?J00231

  # retrieves one or more entries in default format
  # and default style (html)
  # returns nothing for IDs which are not valid
  http://www.ebi.ac.uk/cgi-bin/dbfetch?id=J00231.1,hsfos,bum

  # retrieve entries in fasta format without html tags
  http://www.ebi.ac.uk/cgi-bin/dbfetch?format=fasta&style=raw&id=J00231,hsfos,bum

  # retrieve a raw Ensembl entry
  http://www.ebi.ac.uk/cgi-bin/dbfetch?db=ensembl&style=raw&id=AL122059


=head1 DESCRIPTION

This program generates a page allowing a web user to retrieve database
entries from a local SRS in two styles: html and raw. Other
database engines can be used to implement te same interfase.

At this stage, on unique identifier queries are supported. Free text
searches returning more than one entry per query term are not in these
specs.

In its default setup, type one or more EMBL accession numbers
(e.g. J00231), entry name (e.g. BUM) or sequence version into the
seach dialog to retieve hypertext linked enties.

Note that for practical reasons only the first 50 identifiers
submitted are processed.

Additional input is needed to change the sequence format or suppress
the HTML tags.  The styles are html and raw. In future there might be
additional styles (e.g. xml). Currently XML is a 'raw' format used by
Medline. Each style is implemented as a separate subroutine.

=head1 MAINTANENCE

A new database can be added simply by adding a new entry in the global hash
%IDS. Additionally, if the database defines new formats add an entry for
each of them into the hash %IDMATCH.  After modifying the hash, run this
script from command line for some sanity checks with parameter debug set to
true (e.g. dbfetch debug=1 ).

Finally, the user interface needs to be updated in the L<print_prompt>
subroutine.

=head1 VERSIONS

Version 3 uses EBI SRS server 6.1.3. That server is able to merge release
and update libraries automatically which makes this script simpler. The
other significant change is the way sequence versions are indexed. They
used to be indexed together with the string accession
(e.g. 'J00231.1'). Now they are indexed as integers (e.g. '1').

Version 3.1 changes the command line interface. To get the debug
information use attribute 'debug' set to true. Also, it uses File::Temp
module to create temporary files securely.

Version 3.2 fixes fasta format parsing to get the entry id.

Version 3.3. Adds RefSeq to the database list.

Version 3.4. Make this compliant to BioFetch specs.

=head1 AUTHOR - Heikki Lehvaslaiho

Email:  heikki-at-bioperl-dot-org

=cut

# Let the code begin...

$VERSION = '3.4';
$DATE = '28 Jan 2002';

use CGI "standard";
#use POSIX;
use CGI::Carp qw/ fatalsToBrowser /;
use File::Temp qw/ tempfile tempdir /;
use strict;
no strict "refs";

use constant MAXIDS => 50;
use constant TMPDIR => '/usr/tmp';

use vars qw( $VERSION $DATE %DBS  %STYLES $RWGETZ $RGETZ  %IDMATCH %IDLIST $XEMBL $FH );

BEGIN {

    # paths to SRS binaries
    $RWGETZ = '/ebi/srs/srs/bin/osf_5/wgetz -e';
    $RGETZ = '/ebi/srs/srs/bin/osf_5/getz -e';
    $XEMBL = "cd /ebi/www/pages/cgi-bin/xembl/; ./XEMBL.pl";
    #$EMBOSSDIR = '/ebi/services/pkgs/emboss/bin';

    # RE matching the unique ID in the db entry
    # - key is the 
    # - put the id string in parenthesis
    %IDMATCH = (        #  123
		embl => 'ID   (\w+)',
		fasta => '>\w+.(\w+)',
		medlinefull => '[\n><]MedlineID. ?(\w+)',
                swissprot => 'ID   (\w+)',
		pdb => '.{62}(\w+)',
		bsml => 'DUMMY',
		agave => 'DUMMY',
		refseq => 'LOCUS       ([\w_]+)'
		);
    %DBS = (
	    embl => {
		fields => ['id', 'acc'],
		version => 'sv',          # name of the SRS field
		format => {
		    default => 'embl',
		    embl => 1,
		    fasta => 'FastaSeqs',
		    bsml => 1,
		    agave => 1
		}
	    },
	    medline => {
		fields => ['id'],
		format => {
		    default => 'medlinefull',
#		    medlineref => 'MedlineRef',
		    medlinefull => 'MedlineFull'
		    }
	    },
	    ensembl => {
		fields => ['id'],
		format => {
		    default => 'embl',
		    embl => 1,
		    fasta => 'FastaSeqs'
		    }
	    },
            swall => {
		fields => ['id', 'acc'],
		format => {
		    default => 'swissprot',
		    swissprot => 1,
		    fasta => 'FastaSeqs'
		    }
	    },
            pdb => {
		fields => ['id'],
		format => {
		    default => 'pdb',
		    pdb => '1'
		    }
	    },
            refseq => {
                fields => ['id', 'acc'],
                format => {
		    default => 'refseq',
		    refseq => 1,
		    fasta => 'FastaSeqs'
                }
            }
            #add more databases here...
	    );

    %STYLES = (
	       html => 1,
	       raw => 1
	       );

    %IDLIST = (); #redundancy check list built during the execution
}

my $q = new CGI;

# sanity checks if the script is running from command line
#  and debug parameter is set.
my $debug = protect($q->param('debug')) if $q->param('debug');
&debugging if not $q->user_agent and $debug; 

if ( $q->param('id') or $q->param('keywords')  ) {

    # pacify input strings
    my $value;
    $value = protect($q->param('id')) if $q->param('id');
    $value = protect($q->param('keywords')) if $q->param('keywords');
    my $db = lc protect($q->param('db')); # let's keep the case lower
    my $format = lc protect($q->param('format'));
    my $style =  lc protect($q->param('style'));

    # check input and set defaults
    $style ||= 'html';  # default style
    input_error($q, $style, "2 Unknown style [$style].") unless $STYLES{$style};

    $db ||= 'embl'; # default db
    input_error($q, $style, "1 Unknown database [$db].") unless $DBS{$db};

    $format ||= $DBS{$db}{format}{default}; # default format
    input_error($q, $style, "3 Format [$format] not known for database [$db]")  
        unless $DBS{$db}{format}{$format};
    $format = $DBS{$db}{format}{default} if $format eq 'default';


    # If people choose Bsml or AGAVE, DB can only be 'embl'
    input_error($q, $style, "1 Unknown database [$db].") 
        if ($format eq 'bsml' or $format eq 'agave') and $db ne 'embl';

    # If people choose Bsml or AGAVE, internal style has to be xml . Make it so.
    $style = ($format =~ /(bsml|agave)/i) ? 'xml' : $style;

    if ($style eq 'html') {
        print $q->header(-type => 'text/html', -charset => 'UTF-8');
    }
    elsif ($style eq 'raw') { 
        print "Content-Type: text/plain; charset=UTF-8\n\n";
    }
    $FH = tempfile('dbfetchXXXXXX', DIR => TMPDIR, UNLINK => 1 ); #automatic unlinking

    # Check the number of IDs
    my @ids = split (/ /, $value);
    input_error($q, $style, "6 Too many IDs [". scalar @ids. "]. Max [". MAXIDS. "] allowed.")
        if scalar @ids >  MAXIDS;

    # XEMBL cannot 'glue' single entries due to XML setup 
    #- we need to send things in one go.
    if ($style eq 'xml') {
	&xml($format, @ids);
    } else {
        my $counter;
        foreach my $id (@ids) {
            &$style($db, $id, $format);
        }
        no_entries($q, $value) if $style eq 'html' and tell($FH) == 0;
    }
    seek $FH, 0, 0;
    print '<pre>' if $style eq 'html';
    print $_ while <$FH>;
} else {
    print_prompt($q);
}


=head2 print_prompt

 Title   : print_prompt
 Usage   :
 Function: Prints the default page with the query form
           to STDOUT (Web page)
 Args    :
 Returns :

=cut

sub print_prompt {
    print $q->header(),
         $q->start_html(-title => 'DB Entry Retrieval',
                        -bgcolor => 'white',
			-author => 'heikki-at-bioperl-dot-org'
			),
	 '<IMG align=middle SRC="/icons/ebibanner.gif">',
	  $q->h1('Generic DB Entry Retrieval'),
	  $q->p("This page allows you to retrieve up to ". MAXIDS .
		 " entries at the time from various up-to-date biological databases."),
	  $q->p("For EMBL, enter an  accession number (e.g. J00231) or entry name (e.g.
		 BUM) or a sequence version (e.g. J00231.1), or any combination of them
		 separated by a non-word character into your browser's search dialog.
		 SWALL examples are: fos_human, p53_human.
		 For short Ensembl entries, try : AL122059, AL031002, AL031030 .
		 'Random' Medline entry examples are: 20063307, 98276153.
		 PDB entry examples are: 100D, 1FOS. Try NM_006732 for RefSeq.
		 Only one copy of the latest version of the entry is returned."),
	  $q->hr,
	  $q->startform,
	  $q->popup_menu(-name => 'db',
			 -values => ['EMBL',
				     'SWALL',
				     'PDB',
				     'Medline',
				     'Ensembl',
				     'RefSeq'
				     ]),
	  $q->textfield(-name => 'id',
			 -size => 40,
			 -maxlength => 1000),
	  $q->popup_menu(-name => 'format',
			 -values => ['default','Fasta','bsml','agave']),
	  $q->popup_menu(-name => 'style',
			 -values => ['html','raw']),
	  $q->submit('Retrieve'),
	  $q->endform,
	  $q->hr,
	  $q->h2('Direct access'),
	  $q->p('For backward compatibility, the script defaults to EMBL:'),
	  $q->code('<A href="http://www.ebi.ac.uk/cgi-bin/dbfetch?J00231">
		     http://www.ebi.ac.uk/cgi-bin/dbfetch?J00231</a>'),
	  $q->p('but the preferred way of calling it is:'),
	  $q->code('<A href="http://www.ebi.ac.uk/cgi-bin/dbfetch?id=J00231.1,hsfos,bum">
		     http://www.ebi.ac.uk/cgi-bin/dbfetch?id=J00231.1,hsfos,bum</a>'),
	  $q->p('which can be extended to retrieve entries in alternative sequence formats
		      and other databases:'),
	  $q->code('<A href="http://www.ebi.ac.uk/cgi-bin/dbfetch?db=swall&format=fasta&id=fos_human">
		     http://www.ebi.ac.uk/cgi-bin/dbfetch?db=swall&format=fasta&id=fos_human</a>'),
	  $q->p('Set style to <code>raw</code> to retrieve plain text entries for computational purposes
                 and saving to disk:'),
	  $q->code('<A href="http://www.ebi.ac.uk/cgi-bin/dbfetch?db=medline&style=raw&id=21131735">
                    http://www.ebi.ac.uk/cgi-bin/dbfetch?db=medline&style=raw&id=21131735</a>'),
 	  $q->p('There is now the possibility to retrieve EMBL sequences formatterd into two XML standards:
                Bsml (Bioinformatic Sequence Markup Language - from 
                Labbook, Inc.) or as AGAVE (Architecture for Genomic Annotation, 
                Visualisation, and Exchange - from Labbook, Inc.). To do this, use the 
                formats \'bsml\' or \'agave\', as follows:'),
	  $q->code('<A href="http://www.ebi.ac.uk/cgi-bin/dbfetch?format=bsml&id=J00231">
                   http://www.ebi.ac.uk/cgi-bin/dbfetch?format=bsml&id=J00231</a><br>'),
	  $q->code('<A href="http://www.ebi.ac.uk/cgi-bin/dbfetch?format=agave&id=J00231">
                   http://www.ebi.ac.uk/cgi-bin/dbfetch?format=agave&id=J00231</a>'),
          $q->p("Version numbers are not supported with the XML retrieval."),
	  $q->hr,
          $q->address("Version $VERSION, $DATE, <a href=\"mailto:support\@ebi.ac.uk\">support\@ebi.ac.uk</a>"),
	  $q->end_html, "\n" ;
}

=head2 protect

 Title   : protect
 Usage   : $value = protect($q->param('id'));
 Function:

           Removes potentially dangerous characters from the input
	   string.  At the same time, converts word separators into a
	   single space character.

 Args    : scalar, string with one or more IDs or accession numbers
 Returns : scalar

=cut

sub protect {
    my ($s) = @_;
    $s =~ s![^\w\.\_]+! !g; # allow version numbers with '.' & RefSeq IDs with '_'
    $s =~ s|^\W+||;
    $s =~ s|\W+$||;
    return $s;
}

=head2 input_error

 Title   : input_error
 Usage   : input_error($q, 'html', "Error message");
 Function: Standard error message behaviour
 Args    : reference to the CGI object
           scalar, string to display on input error.
 Returns : scalar

=cut

sub input_error {
    my ($q, $style, $s) = @_;

    if ($style eq 'html' ) {
	print $q->header,
	$q->start_html(-title => 'DB Entry Retrieval: Input error', 
		       -bgcolor => 'white'
		       ),
	"<h2>ERROR in input:<h2>$s\n",
	$q->end_html, "\n";
    } else {
	print "Content-type: text/plain\n\n", "ERROR $s\n";
    }
    exit 0;
}

=head2 no_entries

 Title   : no_entries
 Usage   : no_entries($q, "Message");
 Function: Standard behaviour when no entries found
 Args    : reference to the CGI object
           scalar, string to display on input error.
 Returns : scalar

=cut

sub no_entries {
    my ($q, $value) = @_;

    print $q->start_html(-title => 'DB Entry Retrieval: Input warning',
			 -bgcolor => 'white'
			 ),
	  "<h2>Sorry, your query retrieved no entries.</h2>",
	  "Entries with [$value] where not found.",
	  "Please go back or press <a href=\"dbfetch\"><b>here</b></a> to try again",
	   $q->end_html, "\n";
    exit 0;
}


=head2 raw

 Title   : raw
 Usage   :
 Function: Retrieves a single database entry in plain text
 Args    : scalar, an ID
           scaler, format
 Returns : scalar

=cut

sub raw {
    my ($db, $value, $format) = @_;
    my ($srsq, $qdb, $entry, $id);
    my ($seqformat) = '';
    $seqformat = '-view '. $DBS{$db}{format}{$format}
       if $format ne $DBS{$db}{format}{default};

    my $version = '';
    $value =~ /(.+)\.(.+)/;
    $version = $2 if $2;
    $value = $1 if $1;

    # main db
    $qdb = $db;
    $srsq = '';
    foreach my $field (@{$DBS{$db}{fields}}) {
	$srsq .= " [$qdb-$field:$value] |";
    }
    chop $srsq;

    # if database supports versions (EMBL, GenBank, RefSeq...)
    if ($version) {
	my $vfname = $DBS{$db}{version};
	$srsq = "[$qdb-$vfname:$version] & (". $srsq. ")"
    }

#    print "rsh srs $RGETZ $seqformat $srsq\n";
    $entry = `rsh srs "$RGETZ $seqformat '$srsq'"`;

    $entry =~ s|EMBL[^\n]+\n||;
    $entry =~ s|^\s+||g;
    $entry =~ s|\s+$|\n|g;

    my $idmatch = $IDMATCH{$format};
    ($id) = $entry =~ /$idmatch/;
    # die if ID not found
    input_error(' ', 'raw', "5 ID [$value] not found in database [$db].")
	     unless $id;

#    my $tmp = substr($entry, 0, 20);
#    print "Entry:$tmp\n";
#    print  "-----id=$id---\$1=$1----idmatch=$idmatch=format=$format=\n";
#
    print $FH $entry unless $IDLIST{$id};
    $IDLIST{$id} = 1;
}

=head2 html

 Title   : html
 Usage   :
 Function: Retrieves a single database entry with HTML
           hypertext links in place. Limits retieved enties to 
           ones with correct version if the string has '.' in it.
 Args    : scalar, a UID
           scalar, format
 Returns : scalar

=cut

sub html {
    my ($db, $value, $format) = @_;
    my ($srsq, $qdb, $entry, $id, $idmatch);
    my ($seqformat) = '';
    $seqformat = '-view '. $DBS{$db}{format}{$format}
       if $format ne $DBS{$db}{format}{default};

    my $version = '';
    $value =~ /(.+)\.(.+)/;
    $version = $2 if $2;
    $value = $1 if $1;

    # SWALL plain format at EBI
    $seqformat .= ' -vn 2 ' if $db eq 'swall' or $db eq 'refseq';
    
    $qdb = $db;
    $srsq = '';
    foreach my $field (@{$DBS{$db}{fields}}) {
	$srsq .= " [$qdb-$field:$value] |";
    }
    chop $srsq;

    # if database supports versions (EMBL...)
    if ($version) {
	my $vfname = $DBS{$db}{version};
	$srsq = "[$qdb-$vfname:$version] & (". $srsq. ")"
    }

#    print "rsh srs $RWGETZ $seqformat $srsq\n";
    ### '-id EBISRS' is (hopefully) a temporary addtion until SRS HTML output is fixed 
    $entry = `rsh srs "$RWGETZ $seqformat '$srsq'"`;
  
    return if $entry =~ /SRS error/;

    $entry =~ s|^Content-type:[^\n]+\n||;
    $entry =~ s|\n<A  HREF[^\n]+\n||;
    $entry =~ s|<A +HREF=\"?wgetz|<A HREF=http://srs6.ebi.ac.uk/srs6bin/cgi-bin/wgetz|g; #"\
    $entry =~ s/\+-e\"/\+-e/g; #"
    $entry =~ s|<BR>||g;
    $entry =~ s|</?pre>||g;
    $entry =~ s|\n+|\n|g;
    $entry =~ s|^\n+||g;

    $idmatch = $IDMATCH{$format};
    ($id) = $entry =~ /$idmatch/;

#    my $tmp = substr($entry, 0, 20);
#    print "Entry:$tmp\n";
#    print  "-----id=$id---\$1=$1----idmatch=$idmatch=format=$format=\n";
    print $FH $entry unless $IDLIST{$id};
    $IDLIST{$id} = 1;
}

=head2 xml

 Title   : xml
 Usage   : 
 Function: Retrieves an entry formatted as XML
 Args    : array, UID
           scalar, format 
 Returns : scalar

=cut

sub xml {
    my ($format, @ids) = @_;
    my ($entry, $id, $content, $counter, $reg);

    $content = ($ENV{'HTTP_USER_AGENT'} =~ /MSIE/) ? "Content-type: text/xml\n\n" : 
      "Content-type: text/plain\n\n";

    $entry = "--format ".(($format eq "bsml") ? "Bsml" : "sciobj") .
	     " " . join (" ", @ids);

    $entry = `rsh mercury "$XEMBL $entry"`;

    $reg = (($format eq "bsml") ? '<Sequence id=' : '<contig length-');
    $counter++ while $entry =~ /($reg)/g;

    foreach my $idl (@ids) {
	input_error($q, " ", "5 ID [$idl] not found in database [embl].")
	    if ($format eq "bsml" && $entry =~ "NOT EXIST: $idl") ||
	       ($format eq "agave" && $entry =~ "NOT FOUND: $idl")
    }

    print $FH ($content . $entry);
}

=head2 debugging

 Title   : debugging
 Usage   : 'perl dbfetch'
 Function:

           Performs sanity checks on global hash %IDS when this script
           is run from command line. %IDS holds the description of
           formats and other crusial info for each database accessible
           through the program.

           Note that hash key 'version' is not tested as it should 
           only be in sequence databases.

 Args    : none
 Returns : error messages to STDOUT

=cut

sub debugging {

    foreach my $db (keys %DBS) {
	my $status = 1;

	# field
	print "ERROR: [$db]: no SRS fields defined.".
	    " Give an array of field names?\n" and $status = 0
	    unless $DBS{$db}{fields};
	print "ERROR: [$db]: SRS fields are not defined as an array.\n" and $status = 0
	    unless ref $DBS{$db}{fields} eq 'ARRAY';

	# format
	print "ERROR: [$db]: no formats defined.\n" and $status = 0
	    unless $DBS{$db}{format};
	print "ERROR: [$db]: no default format defined.\n" and $status = 0
	    unless $DBS{$db}{format}{default};
	my $format = $DBS{$db}{format}{default};
	print "ERROR: [$db]: no format [$format] defined.".
	    " You declared it as a default and only.\n" and $status = 0
	    unless $DBS{$db}{format}{$format};
	foreach my $dbformat (keys %{$DBS{$db}{format}}) {
	    print "ERROR: [$db]: format [$format] not defined in %IDMATCH.\n"
		and $status = 0
		 unless $IDMATCH{$dbformat} or $dbformat eq 'default';
	}
	printf "%-12s%s", "[$db]", "OK\n" if $status;
    }
    exit;
}