#!/usr/bin/perl

#(3/3/2016): Convert BigQuery geo export from Visual GKG into GeoJSON for mapping in CartoDB...

open(FILE, "./IN.csv");
while(<FILE>) {
    ($docurl, $imgurl, $geo) = split/,/, $_, 3; $geo=~s/\"//g;
    $imgurl=~s/[\"\']//g; $docurl=~s/[\"\']//g;
    $imgurl=~s/\\/\//g; #fix some odd images...
    foreach $rec (split/<RECORD>/, $geo) {
    	$name = ''; $latlong = ''; $lat = ''; $long = '';
	($name, $confidence, $id, $latlong) = split/<FIELD>/, $rec;
	($lat, $long) = split/,/, $latlong; $lat = sprintf("%0.6f", $lat); $long = sprintf("%0.6f", $long); $latlong = "$long, $lat";
	$name=~s/[^A-Za-z\- ]//g;
	$name=~s/[\"\']//g;
	#if ($LOCS_CNT{$latlong} < 50) { $LOCS_HTML{$latlong} .= "<A HREF=\"$docurl\" TARGET=\"_blank\">Article</A><BR>"; } #article textual links
	if ($LOCS_CNT{$latlong} < 50) { $LOCS_HTML{$latlong} .= "<A HREF=\"$docurl\" TARGET=\"_blank\"><IMG SRC=\"$imgurl\" WIDTH=50 HEIGHT=50></A> "; } #article thumbnail links
	if ($LOCS_IMG{$latlong} eq '') { $LOCS_IMG{$latlong} = $imgurl; }
	if ($LOCS_NAME{$latlong} eq '') { $LOCS_NAME{$latlong} = $name; }
	$LOCS_CNT{$latlong}++;
    }
    
}
close(FILE);

open(OUT, ">OUT.geojson");
print OUT "{ \"type\": \"FeatureCollection\", \"features\": [\n\n";
foreach $latlong (keys %LOCS_HTML) {
    if ($LOCS_NAME{$latlong} eq '') { $LOCS_NAME{$latlong} = 'Untitled'; }
    $LOCS_HTML{$latlong}=~s/"/\\"/g; #JSON escape quotes in our HTML...
    if ($WROTEROWS > 0) { print OUT ",\n"; }
    print OUT "{ \"type\": \"Feature\", \"geometry\": { \"type\": \"Point\", \"coordinates\": [$latlong] }, \"properties\": { \"name\": \"$LOCS_NAME{$latlong}\", \"numarts\": $LOCS_CNT{$latlong}, \"centerimage\": \"$LOCS_IMG{$latlong}\", \"Articles\": \"$LOCS_HTML{$latlong}\" } }";
    $WROTEROWS++;
}
print OUT "\n\n] } ";
close(OUT);
print "Distinct Locations: $WROTEROWS\n";