#!/usr/local/bin/perl
# fbgenequery2.pl   (SRS)
## dgg, dec'98, using srs fbgenes indices of .acode data

require "getopts.pl";
require "fbperls.pl";
require "fbperls_add.pl";
require "srsperls.pl";
require "fbreslist.pm";

#$debug=1;

&fbInits("noprint");
&initSRS($SRSRoot);

&fbSetDataVars("FBgn"); # ,1 == shortreport
&fbDerivedVars();
$SRSDBName = "FBgn HUgn SGgn ATgn CEgn MGgn ZFgn";
 
$selfurl= $0;
if (index($selfurl,$gopath)==0) { 
	$selfurl= substr($selfurl, length($gopath)); 
	}

$searchHints=<<SEOF
<UL>
<LI>Use 'and', 'not', and 'or' boolean terms to refine a complex search.
<LI>Use '*' at the ends (as in gen* or *gen) to match partial words.
</UL>
SEOF
;

#--------- list of all user-defined x-fields coming from forms ----------
    @xfields=('xfield1','xfield2');
    @xfieldnames=('xfieldname1','xfieldname2');
#--------- list of all fields which require special treatment ----------
    $allsymbols="symbol symbol_cs synonym synonym_cs";
#--------- list of all (form+SRS) fields with info on SRS indices -------
    ## Note: keep long names same as in the form, assign
    ## SRS short name if any ("" otherwise) and append
    ##   :digital - if field is digital (date, year etc.)
    ##   :cytoloc - if field is cytolocation (in FB style, digital)
    ##   :CV 	    - if this is CV field
    ##   :SpacedCV   -- CV field with spaces in term
    ##   :mixed   - if was indexed as both words/sentenses (spaces to _)
    ##   :hasvalue 	 - if this field has any value ("?*" in srs query)
    ##   :orvalues   - use OR| to join multiple values of field
		##   :cyto2kb  -- cytologic to kilobase location conversion, including % 100 buckets
		
    %FieldEncoder=(  
	"wild",	"",			
	"maxhit",	"",			
	"release",	"",			
    	"common_or",	"",
    	"oldquery",	"",
	"srs_pipe",	"",
	"sppsubset",	"",  
	"libs",		"", 
	"claset",	"",  
								## dgg: acc has gone in favor of id
	"acc",	 "id",  # primary id
	"id",		"id",   # primary id
	"fbid",		"id",    # primary id
	"g_acc",	"id", # primary id

	"acs",		"acs",  ## primary & 2ndary ids
	"sec_acc",	"acs", # secondary acc-like
	"a_acc",	"acs", # secondary acc-like
	#"all",		"xxx", 	## dgg,feb98, all is returning odd matches
	"query",	"xxx",	
	"all",		"all", 	## dgg,feb98, all is returning odd matches
	"query",	"all",	
	"symbol",	"sym",
	"symbol_cs",	"Smc",
	"synonym",	"syn",
	"synonym_cs",	"Snc",
	"name",		"gen",
	"allele",	"als",
	"ext_acc",	"ack",
	"hasdna", "ack:hasvalue", ## asksrs [fbclass-ack:?*]
	"phenotype",	"phe",
	"aberration",	"abr",  
	"transposon",	"trp",  
	"ref",		"ref",  
	"species",	"org",  
	"class",	"cla:CV",
	"cla",	"cla:CV",
	"funct",	"fun:CV", 
	"fun",	"fun:CV", 

	"struct",	"fun:CV", ## new for meow
	#"homology",	"fun:CV", ## new for meow
	"homology",	"hlg", ## new for meow
	#"csome",	"chn:digital", ## new for meow
	#"gmap",	"glc:digital", ## new for meow
	"csome",	"chn", ## new for meow
	"gmap",	"glc", ## new for meow

	"subcloc",	"cel:CV",  ## changed from flo
	"cel",	"cel:CV",  ## changed from flo
	"charact",	"chr:CV",  
	"chr",	"chr:CV",  
	"enzyme",	"enz:CV",  
	"enz",	"enz:CV",  

	"mincytoloc",	"clm:cytoloc",
	"maxcytoloc",	"clx:cytoloc",
	"cytoloc",	"kbl:cyto2kb", ## dgg, convert to kbl !
	"kbloc",	"kbl:digital",
	"mindate",	"dat:digital",
	"maxdate",	"dat:digital",
	"date",	"dat:digital",
	); 
	
	
foreach $FieldName ( keys(%FieldEncoder) ) { $FieldArray{$FieldName}=""; }
#----------------------- fields with special actions ---------------------
    # corrections will be performed through eval 
    %FieldActions=(  
	"id",		 '&accession_to_standard($FieldName,"gn|al")',
	"fbid",		'&accession_to_standard($FieldName,"gn|al")',
	"query",	'&check_fbid_in_query($FieldName,"gn|al","id")',
	"g_acc",  '&accession_to_standard($FieldName,"gn|al")',

	"a_acc",	'&transfer_to_other_field($FieldName,"acs")',
	"sec_acc",	'&transfer_to_other_field($FieldName,"acs")',
	"acs",		'&accession_to_standard($FieldName)',
	"sppsubset",	'&transfer_to_other_field($FieldName,"species") if($FieldArray{"species"} eq "subset")',
	"claset",	'&decode_gene_classes($FieldName)',
	"class",	'$FieldValue="" if( $FieldValue eq "all" )',
	#"libs",		'$SRSDBName = $FieldValue',
	"species",	'$FieldValue="" if( $FieldValue eq "all" )',
	"srs_pipe",	'$srs_pipe2=$FieldValue',
	); 

##
## MAIN
##

&Getopts("cdg:hilm:n:rstv:x:"); 

if ($opt_d || $ENV{'REQUEST_METHOD'} eq "POST") { $isdialog = 1; }
# if ($opt_i) { $plus = "-+i"; }
# if ($opt_l) { $lastonly= 1; }
if ($opt_m) { $maxhits = $opt_m; }
if ($opt_n) { $indexstart = $opt_n; }
# if ($opt_g) { $groupsym="-s $q$opt_g$q"; }
if ($opt_s) { $suffix .= "s"; }
# if ($opt_t) { $dotitle= 0;}
 
&SetProtocolVars();  

## HTML
if ($dohtml) { &retrieve_html_fields(); }
## GOPHER
else {
  if ($isdialog) {
    chop( $FieldArray{'query'} = <STDIN>); 
    chop( $maxhits = <STDIN>);
    chop( $isturbogo = <STDIN>);
    chop( $FieldArray{'symbol'} = <STDIN>); 
    chop( $FieldArray{'name'} = <STDIN>); 
    chop( $FieldArray{'mincytoloc'} = <STDIN>); 
    chop( $FieldArray{'maxcytoloc'} = <STDIN>); 
    chop( $FieldArray{'allele'} = <STDIN>); 
    chop( $FieldArray{'class'} = <STDIN>);
    chop( $FieldArray{'species'} = <STDIN>); 
    chop( $FieldArray{'sppsubset'} = <STDIN>); 
    chop( $FieldArray{'phenotype'} = <STDIN>); 
    chop( $FieldArray{'ref'} = <STDIN>); 
    }
  else {
    $FieldArray{'query'}= join(' ',@ARGV); 
    if( $ENV{'QUERY_STRING'} && ($ENV{'QUERY_STRING'} ne "") ) {
       $FieldArray{'query'}=&URLunescape($ENV{'QUERY_STRING'}); }
    }
  $query=$FieldArray{'query'};
  if( $query!~/FBgn/ && $query!~/species=/ && $query!~/['"]/) { 
	  $FieldArray{'species'}="Dmel"; }   #"
  }    

## reporting style options
# if ($opt_x eq "al") { $xtra='-x '.$FieldArray{'query'}; }
#elsif ($opt_x eq "f") { $xtra=$opt_x= "-sfull"; }
#elsif ($opt_x eq "c") { $xtra=$opt_x= "-scustom"; }
#elsif ($opt_x eq "s") { $xtra=$opt_x= "-sshort"; }
#elsif ($opt_x eq "r") { $xtra=$opt_x= "-sreftable"; }

$libs= $FieldArray{'libs'};
if ($libs && $libs !~ /all/) { $SRSDBName= $libs; }

#$rels= $FieldArray{'release'};
#if ($rels) {
#	if ($rels =~ /full/) { $sdb = "FBgn"; }
#	if ($rels =~ /pre/) { 
#	  $sdb .= ' ' if ($sdb);
##	  $sdb .= "FBgnpre";
#	  }
#	$SRSDBName = $sdb;
#	}

# &FixGeneQuery(); ##???
 
$query=""; # KEEP IT HERE
$query=&standard_fields_processing($query); ## this one calls SetSRSQueryPipe()

##print "Content-type: text/html\n\n query=$query\n";

if( $srs_pipe2 ne "" ) { 
  ## $query= &SetSRSQueryPipe($query,$srs_pipe2); }
  $query= '(' . $query . ')' if( &separated_items($query) ); 
  $query = $query . '>' . $srs_pipe2; 
  &fbSetDataVars($srs_pipe2);  
  &fbDerivedVars();
}

QUERY:

$dowild= ($FieldArray{'wild'} eq 'on');
if ($dowild) {
 $query =~ s/(\w)]/$1\*]/g; # is this safe??
}

&SetViewVars();

## &DoSRSQuery();
$nmatches= &DoSRSAcodeQuery($query, $askdata, $dowild );

$prettyquery= &humanize_query($query); 

&fbreslist'init($prettyquery);
##$fbreslist'refineForm= &getRefineDialog($store_query,$prettyquery);

if ($nmatches < 2) {
	&fbreslist'putResultsReports( $nmatches, $askdata, 0) 
	}
else {
	&fbreslist'putResultsList( $nmatches, $askdata, 0);
	}

&KillOldFiles();
&fbexit(0);



#*************************************************************************
#
#*************************************************************************

sub getRefineDialog
{
    local($query,$humquery)=@_;
    local $s= <<REOF_BEG
<FORM ACTION="$selfurl" METHOD="POST">
<h3>Refine query or find items in related data classes:</h3>  
<B><INPUT TYPE="submit" VALUE="Refine query"></B>
<BR>
<B>$humquery</B>

<SELECT NAME="common_or">
  <OPTION SELECTED>and
  <OPTION>or
  <OPTION>but not
</SELECT>
<BR>
<SELECT NAME="xfieldname1">
     <OPTION VALUE="all" >All data
      <OPTION VALUE="synonym" SELECTED>Gene symbol/synonym
      <OPTION VALUE="name">Gene name
      <OPTION VALUE="funct">Gene function
      <OPTION VALUE="struct">Gene structure
      <OPTION VALUE="homology">Homology
      <OPTION VALUE="id">Database ID
      <OPTION VALUE="csome">Chromosome no.
      <OPTION VALUE="gmap">Map position
</SELECT>
matches <INPUT NAME="xfield1" SIZE=20>
<BR>
<TABLE BORDER=1 CELPADDING=1>
<TR>
REOF_BEG
;
    $s .= "</TR>\n</TABLE>\n";  
    $query=&URLescape($query);
    $s .= '<BR><INPUT NAME="oldquery" TYPE="hidden" VALUE="'.$query.'">';
    $s .= '<BR><INPUT NAME="maxhit" TYPE="hidden" VALUE="'.$maxhits.'">';
    $s .= '</FORM>';
    return $s;
}

#*************************************************************************
#
#*************************************************************************

sub decode_gene_classes
{
    
    local($FieldName)=@_;
    local($FieldValue,$i,$txt,$add);
    $FieldValue=$FieldArray{$FieldName};
    return if( $FieldValue eq "" );
    local(@classes)=("__", # table for decoding classes subsets for genes
	"gene",				# 1
	"fusion_gene",			# 2
	"honorary_gene",		# 3	block Dmel setting!
	"transposon",			# 4
	"foreign_gene",			# 5	block Dmel setting!
	"safe_element",			# 6
	"safe_element.h",		# 7	block Dmel setting!
	"mitochondrial_gene",		# 8
	"virus_symbiont_pathogen",	# 9
	"foreign_fusion",		# 10	block Dmel setting!
	"sire_element",			# 11
	"sire_element.h",		# 12	block Dmel setting!
	"non_coding_RNA_gene",		# 13
	"__");
    local($block_Dmel)="x3x5x7x10x12x";
    for( $i=1; $i<$#classes; $i++ ) {
	$txt='x'.$i.'x';
    	if( $FieldValue=~m|$txt| ) { 
      	    $add=$classes[$i]; 
	    if( !($FieldArray{'class'}=~m|$add|) ) {
		$FieldArray{'class'}.=' '.$add; } 
	    if( ($FieldArray{'species'} eq 'Dmel') && $block_Dmel=~m|$txt| ) {
		$FieldArray{'species'}=""; } 
    	    $FieldArray{'class'}=~s|^\s*||;
	    }
	}
    $FieldArray{$FieldName}=""; # kill it because it's done
    return;
}