verify_coord_data.pl #3

#!/usr/bin/perl
use strict;
use warnings;

use Date::Calc qw( check_date Date_to_Days );
use FileHandle;

# verify_coord_data.pl - script to verify Geohash coordinate data

# globals
use vars qw/ %opt $VERBOSE /;

$VERBOSE = 0;

sub usage
{
	if ( @_ ) { print STDERR ( join " ", @_ ) . "\n"; }
	print STDERR <<END_OF_USAGE;

Usage: $0 [-i <input file>] [-v]
	-i: input file to use (file format: "YYYY-MM-DD,<DOW>,<LAT>,<LON>")
		default: '-' for standard input
	-v: enable verbose logging, sent to standard error

Example:
* Verify the allcoords file in one inline command:
    \$ wget -qO - http://www.amipsychic.net/coords/allcoords.csv.bz2 |
      bzip2 -d |
      ./verify_coord_data.pl

Notes:
* You can get Geohash data from:
  http://www.amipsychic.net/geohashing.html
* This script uses Date::Calc from CPAN; you can install it via:
    \$ cpan -i Date::Calc

END_OF_USAGE
	exit 2;
}

sub logit
{
	print STDOUT ( join " ", @_ ) . "\n";
}

sub debuglog
{
	if ( $VERBOSE == 1 ) { print STDERR ( join " ", @_ ) . "\n"; }
}

sub parse_args
{
	use Getopt::Std;
	getopts( "hi:v", \%opt ) or usage( "error: invalid arguments" );

	debuglog( "validating parameters..." );
	if ( $opt{h} ) { usage(); }
	if ( ! $opt{i} ) { $opt{i} = "-"; }
	if ( $opt{i} ne "-" && ! -r $opt{i} ) { usage( "error: cannot read input file [$opt{i}]" ); }
	if ( $opt{v} ) { $VERBOSE = 1; }

	debuglog( "resulting parameters:" );
	debuglog( "opt-i: $opt{i}" );
	debuglog( "opt-v: " . ( defined $opt{v} ? $opt{v} : "undef" ) );
}

sub verify_input
{
	my $src_name = "[$opt{i}]";
	$src_name = "<standard input>" if ( $opt{i} eq "-" );

	debuglog( "src_name = [$src_name]" );

	my $output = *STDOUT;
	my $input = new FileHandle;
	my @input;
	if ( ! open $input, $opt{i} ) { die "cannot open [$src_name]: $!\n"; }

	# the max amount of data we're dealing with is reasonable to load
	debuglog( "slurping in $src_name to array..." );
	@input = <$input>;
	if ( ! close $input ) { warn "trouble closing file: $!\n"; }

	logit( "verifying Geohash data from $src_name..." );

	my( $date, $dow, $lat, $lon );
	my( $year, $month, $day );
	my( $prevdate, $prevyear, $prevmonth, $prevday );
	my( $lineno, $numerr );


	$lineno = 1;
	$numerr = 0;
	$prevdate = "";
	foreach ( @input )
	{
		chomp;
		debuglog( "line $lineno: $_" );

		if ( $_ eq "" )
		{
			logit( $numerr++ . "empty line after $prevdate (line $lineno)" );
			next;
		}

		( $date, $dow, $lat, $lon ) = split /,/;
		debuglog( "parsed: [$date] [$dow] [$lat] [$lon]" );

		( $year, $month, $day ) = split /-/, $date;

		( ! check_date( $year, $month, $day ) ) &&
			logit( ++$numerr . ": bad date [$date] (line $lineno)" ) ||
			debuglog( "date is valid" );

		if ( $prevdate ne "" )
		{
			( Date_to_Days( $year, $month, $day ) != ( Date_to_Days( $prevyear, $prevmonth, $prevday ) + 1 ) ) &&
				logit( ++$numerr . ": $date skipped days since $prevdate (line $lineno)" ) ||
				debuglog( "date is sequential since previous date" );
		}

		( $dow <= 0 ) &&
			logit( ++$numerr . ": $date dow [$dow] invalid (line $lineno)" ) ||
			debuglog( "dow is sane" );

		( ($lat <= 0) || ($lat >= 1) ) &&
			logit( ++$numerr . ": $date lat [$lat] invalid (line $lineno)" ) ||
			debuglog( "lat is sane" );

		( ($lon <= 0) || ($lon >= 1) ) &&
			logit( ++$numerr . ": $date lat [$lat] invalid (line $lineno)" ) ||
			debuglog( "lon is sane" );

		debuglog( "line $lineno is OK" );

		$prevdate = $date;
		$prevyear = $year;
		$prevmonth = $month;
		$prevday = $day;
		$lineno++;
	}

	logit( ($lineno - 1) . " lines analyzed; $numerr errors detected" );

	return $numerr;
}

parse_args();
exit verify_input();