#!/usr/bin/perl -w

use CGI qw(:standard);
use CGI::Carp qw(warningsToBrowser fatalsToBrowser);
use strict;

#DECIPHER ALL THE SEARCH PARAMETERS IN THIS NEXT CODE BLOCK
my $title = param("title");
my $pub = param("publisher");
my $issn = param("issn");
my $sort = param("sort");
my @subs = ("Agriculture","Biology","Business","Chemistry","Computer Science","Economics","Education","Engineering","Geology","History","Humanities","Law","Mathematics","Medicine","Physics","Psychology","Social Science","Nocat");
my @subjects = ();
foreach my $sub (@subs){
	push @subjects, $sub if (param($sub) eq "on");
}
@subjects = @subs if (@subjects == 0);
my @vals = ("good","medium","bad");
my @values = ();
foreach my $v (@vals){
	push @values, $v if (param($v) eq "on");
}
@values = @vals if (@values == 0);

#DECIPHER THE ACTUAL SEARCH KEYWORDS AND PHRASES FROM THE TEXT THAT WAS TYPED
my @titlewords = splitFields($title);
my @pubwords = splitFields($pub);

#THE SEARCH KEYWORDS WERE FORMATTED INCORRECTLY, SO SAY AN ERROR MESSAGE AND EXIT
if ($titlewords[0] eq "unmatchedquotation" || $pubwords[0] eq "unmatchedquotation"){
	print "<i>Error: Invalid search string. Possible unmatched quotation.<\/i><p><a href=\"http:\/\/www.journalprices.com\/\">Back to Search</a>";
	exit;
}

#THE SEARCH KEYWORDS WILL BE USED IN REGEXES LATER, SO ESCAPE ALL THE PUNCTUATION AND
#SUBSTITUTE PERL REGEX FOR THE SYNTAX ALLOWED ON THE SEARCH PAGE (* AND ?).
foreach my $t (@titlewords,@pubwords,$issn){
	$t =~ s/([\$\\\.\|\(\)\[\]\*\+\?\{\}\@\-\'\"\/\`\~\,\<])/\\$1/ig;
	$t =~ s/\\\?/\./g;
	$t =~ s/\\\*/.\*/g;
}

#ISSN IS HANDLED BY ITSELF LATER SO IT NEEDS TO BE .* TO RETURN ALL RESULTS WITH BLANK SEARCH FIELD
$issn = ".*" if ($issn eq "");

#THE DATABASE SHOULD BE SAVED AS DATA.TXT IN THE CGI-BIN WITH THIS PROGRAM
open(my $IN, "data.txt") or die "Can't open data.\n";
my @journals = <$IN>;

#IN DATA.TXT THERE ARE 17 TAB-DELMITED FIELDS FOR EACH JOURNAL. THIS ALIASES THE FIELD NUMBER
#WITH THE FIELD NAME FOR READABILITY, ETC. SEVERAL FIELDS AREN'T USED IN THE WEBSITE, THESE ARE
#PRIVATE INFORMATION, LIKE THE LISTED PRICE, INTERPRETED PRICE, CPI AND OTHER INDICES SINCE 1980, ETC.
my $titleindex = 0;
my $pubindex = 2;
my $issnindex = 1;
my $subindex = 3;
my $ppaindex = 6;
my $ppcindex = 7;
my $cpiindex = 8;
my $ypindex = 4;
my $cpiavgindex = 9;
my $valueindex = 10;
my $profitindex = 5;

#BUILD AN ARRAY OF THE JOURNALS THAT PASS THE SEARCH CRITERIA
my @results = ();
foreach my $jrn (@journals){
	chomp $jrn;
	my @fields = split(/\t/,$jrn);
	#MAKE SURE:

	# AT LEAST ONE OF THE SUBJECTS BEING SEARCHED IS LISTED IN THE SUBJECT FIELD OF THE JOURNAL
	# ALL OF THE TITLE KEYWORDS/PHRASES ARE IN THE TITLE
	# ALL OF THE PUBLISHER KEYWORDS/PHRASES ARE IN THE PUBLISHER FIELD
	# THE ISSN (REGEX) THAT WAS SEARCHED FOR MATCHES THE ISSN FIELD
	# THE VALUE FIELD IS ONE OF THE CATEGORIES SEARCHED FOR
	#IF YES, THEN ADD THIS JOURNAL TO THE RESULTS
	if ( (!(noneIn($fields[$subindex],@subjects))) && (allIn($fields[$titleindex],@titlewords)) && (allIn($fields[$pubindex],@pubwords)) && (" $fields[$issnindex] " =~ / $issn /i) && !(noneIn($fields[$valueindex],@values)) ) {
		push @results, $jrn;
	}
}

#FIGURE OUT WHICH FIELD WE ARE SORTING THE RESULTS BY
my $sortindex = 0;
if ($sort eq "pubsort") {
	$sortindex = $pubindex;
} elsif ($sort eq "issnsort"){
	$sortindex = $issnindex;
} elsif ($sort eq "ppasort"){
	$sortindex = $ppaindex;
} elsif ($sort eq "ppcsort"){
	$sortindex = $ppcindex;
} elsif ($sort eq "cpisort"){
	$sortindex = $cpiindex;
} elsif ($sort eq "ypsort"){
	$sortindex = $ypindex;
} elsif ($sort eq "valuesort"){
	$sortindex = $cpiavgindex;
}

#THERE IS A DIFFERENT SORT FUNCTION FOR NUMERIC FIELDS THAN ALPHANUMERIC FIELDS, SO FIGURE OUT WHICH TO USE
my $numericsort = 0;
$numericsort = 1 if ($sort eq "ppcsort" || $sort eq "ppasort" || $sort eq "cpisort" || $sort eq "ypsort" || $sort eq "valuesort");

#SORT THE RESULTS ALPHABETICALLY OR NUMERICALLY
my @sorted = sort { 
	if($numericsort){
		return toNumber(getField($a,$sortindex)) <=> toNumber(getField($b,$sortindex));
	} else {
		return lc(getField($a,$sortindex)) cmp lc(getField($b,$sortindex));
	}
} @results;

#REVERSE THE LIST IF THE SEARCHER WANTS THEM IN DESCENDING ORDER
@sorted = reverse @sorted if (param("order") eq "desc");

#OUTPUT THE SEARCH RESULTS TO THE WEB BROWSER.
#THIS BLOCK DOES IT IN TEXT FORMAT IF THAT'S WHAT THE SEARCHER SELECTED
if (param("text") eq "on"){
	print "Content-type: text/plain\n\n";
	print "RESULT NUMBER\tTITLE\tISSN\tPUBLISHER\tSUBJECT\tPRICE PER ARTICLE\tPRICE PER CITATION\tCOMPOSITE PRICE INDEX\tRELATIVE PRICE INDEX\tYEAR FIRST PUBLISHED\tPROFIT STATUS\tVALUE CATEGORY\n";
	for (my $i=0; $i<=$#sorted; $i++){
		my $r = $sorted[$i];
		my @fields = split(/\t/,$r);
		my $j = $i+1;
		print "$j\t$fields[$titleindex]\t$fields[$issnindex]\t$fields[$pubindex]\t";

		#REPLACE "NOCAT" WITH "MISC" IN THE SUBJECT FIELD
		my $subjectfield = $fields[$subindex];
		$subjectfield =~ s/Nocat/Misc/ig;

		#FORMAT THE PRICE INDICES HERE
		my $ppa = $fields[$ppaindex];
		my $ppc = $fields[$ppcindex];
		my $cpi = $fields[$cpiindex];
		my $val = $fields[$cpiavgindex];
		foreach my $p ($ppa,$ppc,$cpi,$val){
			#ONLY DISPLAY TWO DECIMAL PLACES, AND DON'T PRINT THE DUMMY NUMBER "999999" BECAUSE THAT MEANS THERE WEREN'T
			# ANY CITATIONS OR ARTICLES SO THE PRICE PER ARTICLE/CITATION AND COMPOSITE INDEX COULDN'T BE CALCULATED
			my $t = index($p,".");
			$p = substr($p,0,$t+3) if ($t >= 0);
			$p = "N/A: No Articles Or Citations" if ($p == 999999);
		}
		print "$subjectfield\t$ppa\t$ppc\t$cpi\t$val\t$fields[$ypindex]\t$fields[$profitindex]\t$fields[$valueindex]\n";
	}
	exit;
}

#IF THEY DIDN'T WANT TEXT FORMAT, NOW WE HAVE TO OUTPUT HTML HEADERS, ETC.
print header;
print start_html("Search Results");
print <<ENDHTML;
<style type="text/css">
.f1{font-family: Verdana; font-weight:bold; color:#000000; font-size: 18px; background-color:#FFFFFF}
.f2{font-family: Verdana; font-weight:bold; color:#000000; font-size: 14px; background-color:#FFFFFF}
.f3{font-family: Verdana; font-weight:normal; color:#000000; font-size: 12px; background-color:#FFFFFF}
.f3g{font-family: Verdana; font-weight:normal; color:#000000; font-size: 12px; background-color:#00FF00}
.f3m{font-family: Verdana; font-weight:normal; color:#000000; font-size: 12px; background-color:#FFFF00}
.f3b{font-family: Verdana; font-weight:normal; color:#000000; font-size: 12px; background-color:#FF0000}
.f4{font-family: Verdana; font-weight:normal; color:#000000; font-size: 14px; background-color:#FFFFFF}
.f5{font-family: Verdana; font-weight:normal; color:#000000; font-size: 11px; background-color:#FFFFFF}
</style>
<body bgcolor="#FFFFFF">
<div class="f1"><b>Search Results</b></div><br><br>
<div class="f2"><a href="http://www.journalprices.com">Back to Search</a></div><br><br>
<table border=0 class="f3">
ENDHTML

#AND THE RESULTS, IN HTML TABLE FORMAT
for (my $i=0; $i<=$#sorted; $i++){
	my $r = $sorted[$i];
	my @fields = split(/\t/,$r);
	my $j = $i+1;
	
	print "<tr><div class=\"f3\">";
	
	#FOR EXPLANATION OF THESE CALCULATIONS SEE THE PREVIOUS FOR LOOP	
	print "<b>$j. Title:</b> $fields[$titleindex] <br><b>Publisher:</b> $fields[$pubindex] <br><b>ISSN:</b> $fields[$issnindex] <br><b>Subject:</b> ";
	my $subjectfield = $fields[$subindex];
	$subjectfield =~ s/Nocat/Misc/ig;	
	my $ppa = $fields[$ppaindex];
	my $ppc = $fields[$ppcindex];
	my $cpi = $fields[$cpiindex];
	my $val = $fields[$cpiavgindex];
	foreach my $p ($ppa,$ppc,$cpi,$val){
		my $t = index($p,".");
		$p = substr($p,0,$t+3) if ($t >= 0);
		$p = "N/A: No Articles Or Citations" if ($p == 999999);
	}
	print "$subjectfield<br><b>Profit Status:</b> $fields[$profitindex]<br><b>Year First Published:</b> $fields[$ypindex]<br><b>Price per article:</b> $ppa<br><b>Price per citation:</b> $ppc<br><b>Composite Price Index:</b> $cpi<br></div>";

	#HIGHLIGHT THE CPI FIELD GREEN RED OR YELLOW DEPENDING ON THE JOURNAL VALUE
	if ($fields[$valueindex] eq "good"){
		print "<span class=\"f3g\">";
	} elsif ($fields[$valueindex] eq "bad") {
		print "<span class=\"f3b\">";
	} else {
		print "<span class=\"f3m\">";
	}
	print "<b>Relative Price Index</b> $val</span><p>";
}

#TELL THE USER THERE WERE NO SEARCH RESULTS IF THERE WEREN'T ANY
if (@sorted == 0){
	print "<div class=\"f2\"><i>Your search returned no results. Eliminate keywords or search more subject areas to broaden your search.</i></div>";
} else {
	print "<p><div class=\"f2\"><a href=\"http:\/\/www.journalprices.com\">Back to Search</a></div><br><br>";
}

#THIS JUST TAKES A STRING THAT HAS TABS IN IT DELIMITING FIELDS, AND A NUMBER, AND RETURNS THAT FIELD
sub getField {
	my @f = split(/\t/,$_[0]);
	return @f[$_[1]];
}

#THIS TAKES A STRING AND AN ARRAY OF WORDS AND RETURNS TRUE IF NONE OF THE WORDS ARE IN THE STRING
#IGNORES COMMAS (USED FOR MAKING SURE A JOURNAL IS THE RIGHT SUBJECT OR VALUE CATEGORY, SO COMMAS
#ARE THE ONLY RELEVANT PUNCTUATION)
sub noneIn {
	my $str = $_[0];
	$str =~ s/,//g;
	my @tests = @_;
	splice(@tests,0,1);
	foreach my $t (@tests){
		return 0 if (" $str " =~ / $t /i);
	}
	return 1;
}

#THIS TAKES THE SAME ARGUMENTS AS noneIn AND RETURNS TRUE IF ALL OF THE WORDS ARE IN THE STRING
#THIS OBEYS THE REGEX CONVENTIONS ALLOWED BY THE SEARCH PAGE, SO allIn IS A MISNOMER, SEE
#COMMENTS WITHIN THE FUNCTION FOR EXPLANATION
sub allIn {
	my $str = $_[0];

	#REPLACE HYPHENS WITH SPACES SO EACH HALF IS MATCHED SEPARATELY, I.E.
	# "HI-THERE" MATCHES BOTH "HI" "THERE" AND "HI THERE".
	$str =~ s/[\-\/]/ /g;
	

	#IGNORE APOSTROPHES, PERIODS, COMMAS, COLONS, AND PARENTHESIS

	$str =~ s/[\'\.,:\(\)]//g;
	
	my @tests = @_;
	splice(@tests,0,1);
	foreach my $t (@tests){
		
		#REMOVE SPACE AROUND EACH OF THE SEARCH WORDS
		$t =~ s/^\s//;
		$t =~ s/\s$//;
		
		#AND REPLACE ANY PUNCTUATION THE SAME AS WE DID WITH THE SEARCH STRING ABOVE
		$t =~ s/\-/ /g;
		#$t =~ s/[\']//g;
		$t =~ s/\\[\.\'\,:\(\)]//g;
		
		#IF THE SEARCH WORD STARTS WITH "^" IT MEANS THE USER WANTS TO EXCLUDE RESULTS WITH THAT 
		#IN IT, SO RETURN FALSE IF IT IS IN THE STRING, OTHERWISE RETURN FALSE IF IT'S NOT
		if ($t =~ s/^\^//){
			return 0 if (!($t eq "") && (" $str " =~ / $t /i));
		} else {
			return 0 if (!($t eq "") && !(" $str " =~ / $t /i));
		}
	}
	return 1;
}

#THIS TAKES A STRING THAT LOOKS LIKE A NUMBER AND RETURNS THE NUMBER
#IF THE STRING DOESN'T LOOK LIKE A NUMBER, RETURNS -1 (THAT'S OK BECAUSE WHEN THIS FUNCTION IS
#USED IN THIS PROGRAM IT'S CONVENIENT TO ALWAYS GET A NUMBER BACK AND THE RESULTS IN THIS
#PROGRAM SHOULD ALWAYS BE POSITIVE SO -1 WON'T OCCUR).
sub toNumber {
	my $num = 0.0;
	my $flag = length($_[0]) - 1;
	for (my $i=0; $i<length($_[0]); $i++){
		if(substr($_[0],$i,1) =~ /(\d)/){
			$num = $num * 10 + toDigit($1);
		} elsif (substr($_[0],$i,1) =~ /\./){
			$flag = $i;
		} elsif (substr($_[0],$i,1) =~ /\,/){
			next;
		} else {
			return -1;
		}
	}
	$num = $num / (10 ** (length($_[0]) - 1 - $flag));
	return $num;
}

#THIS TAKES A STRING THAT LOOKS LIKE A SINGLE DIGIT AND RETURNS THE DIGIT
sub toDigit {
	if ($_[0] eq "0") {
		return 0;
	} elsif ($_[0] eq "1") {
		return 1;
	} elsif ($_[0] eq "2") {
		return 2;
	} elsif ($_[0] eq "3") {
		return 3;
	} elsif ($_[0] eq "4") {
		return 4;
	} elsif ($_[0] eq "5") {
		return 5;
	} elsif ($_[0] eq "6") {
		return 6;
	} elsif ($_[0] eq "7") {
		return 7;
	} elsif ($_[0] eq "8") {
		return 8;
	} elsif ($_[0] eq "9") {
		return 9;
	} else {
		return -1;
	}
}

#THIS TAKES A SEARCH STRING TYPED IN BY THE USER AND DECIPHERS THE WORDS AND PHRASES FROM IT
sub splitFields {
	my @fields = ();
	my $str = " $_[0] ";

	#FIRST GET OUT THE PHRASES THAT ARE IN QUOTATION MARKS
	while (index($str,"\"") >= 0 ){

		my $x = index($str,"\"");
		my $y = index($str,"\"",$x+1);
		return "unmatchedquotation" if ($y == -1);

		#MAKE SURE TO INCLUDE THE "^" IF THE PHRASE IS PRECEEDED BY ONE
		if (substr($str,$x-1,1) eq "^"){

			push @fields, "^" . substr($str,$x+1,$y-$x-1);
			$str = substr($str,0,$x-1) . substr($str,$y+1,length($str)-$y-1);
		} else {
			push @fields, substr($str,$x+1,$y-$x-1);
			$str = substr($str,0,$x) . substr($str,$y+1,length($str)-$y-1);
		}
	}

	#THEN JUST SPLIT THE REST APART BY WHITESPACE
	push @fields, split(/\s+/,$str);
	return @fields;
}