#!/usr/bin/perl

use strict;
use warnings;

use File::Basename qw/basename/;
use Getopt::Long;
use POSIX qw/strftime/;
use Text::Wrap;
use Time::Local qw/timelocal/;

my $FORECAST_FILENAME = 'homework-14-forecasts.txt';
my $ACTUALS_FILENAME = 'homework-14-actuals.txt';
my $TWELVE_HOURS = 12 * 60 * 60;

my $below_count = 0;
my $above_count = 0;
my $good_count = 0;


# Main =========================================================================

GetOptions('test' => \&run_tests);

chomp(my @observations = read_file($ACTUALS_FILENAME));
my $actuals_ref = analyze_observations(@observations);

chomp(my @forecasts = read_file($FORECAST_FILENAME));
print analyze_forecasts(\@forecasts, $actuals_ref);
print analyze_counts(64);

exit(0);


# Subroutines ==================================================================

# read_file(filename)
#
# Reads the contents of the given filename and returns them as a single scalar
# or as a list of lines, depending on context.  Dies upon failure.
sub read_file {
    my $filename = shift;

    # Read file contents into array
    open(my $filehandle, '<', $filename)
        or die "Could not open '$filename': $!\n";
    my @lines = <$filehandle>;
    close($filehandle);

    # Choose the correct return format
    return @lines             if wantarray;
    return join('', @lines)   if defined wantarray;
    return;
}


# analyze_observations(hourly observations) => hash
#
# Processes each HOURLY line from the observations file and builds a simple data
# structure of DAILY actual highs and lows.  Returns a reference to a hash that
# contains data like this:
#
# {
#   '2011-07-29' => { 'max' => 86.500, 'min' => 71.213, 'obs' => 24 },
#   '2011-07-30' => { 'max' => 88.115, 'min' => 72.102, 'obs' => 24 },
#   ...
# }
#
# Notes: The definition of a "day" is unusual: The observations for a particular
# date run from noon on that day through the 11 a.m. hour of the following day.
# Thus, the 'min' key in the inner hash is most likely the minimum temperature
# of the following day.  Also, the 'obs' key is the number of observations
# counted for that day; 24 are required for a complete record of a day.

sub analyze_observations {
    my %actuals;
    foreach my $hourly_observation (@_) {
        my ($date, $hour, $min, $max) = split(/\t/, $hourly_observation);
        my ($year, $month, $day) = split(/-/, $date);
        my $timestamp = timelocal(0, 30, $hour, $day, $month - 1, $year);
        my $date_of_record = strftime('%Y-%m-%d', localtime($timestamp - $TWELVE_HOURS));
        if (exists $actuals{$date_of_record}) {
            $actuals{$date_of_record}{'min'} = $min if $min < $actuals{$date_of_record}{'min'};
            $actuals{$date_of_record}{'max'} = $max if $max > $actuals{$date_of_record}{'max'};
            $actuals{$date_of_record}{'obs'} += 1;
        } else {
            $actuals{$date_of_record} = {'min' => $min, 'max' => $max, 'obs' => 1};
        }
    }

    return \%actuals;
}


# analyze_forecasts(forecast data, actuals data) => list
#
# Main analysis subroutine: Goes through the forecast days and builds a report
# for each one.  The report may indicate that not enough observation data is
# available, or else describes and compares the forecast and actuals; most of
# the latter is handled by a separate subroutine.  Returns a string containing
# the entire report.

sub analyze_forecasts {
    my ($forecasts_ref, $actuals_ref) = @_;

    my $report = "WEATHER REPORT\n\n";
    $report .= "DATE               HIGHS                   LOWS (early next AM)\n";
    $report .= "================   =====================   =====================\n";

    foreach my $forecast (sort @{$forecasts_ref}) {
        my ($date, $time, $forecast_highs, $forecast_lows) = split(/\t/, $forecast);

        # Date
        my ($year, $month, $day) = split('-', $date);
        my $weekday = substr(strftime('%A', localtime(timelocal(0, 0, 12, $day, $month - 1, $year - 1900))), 0, 3);
        $report .= "$date ($weekday)   ";

        # Check for observations
        unless (exists $actuals_ref->{$date}) {
            $report .= "---------- no weather observations ----------\n";
            next;
        }
        if ($actuals_ref->{$date}{'obs'} < 24) {
            $report .= "-------- only $actuals_ref->{$date}{'obs'} weather observations --------\n";
            next;
        }

        $report .= format_cell($forecast_highs, $actuals_ref->{$date}{'max'});
        $report .= '   ';
        $report .= format_cell($forecast_lows, $actuals_ref->{$date}{'min'});

        $report .= "\n";
    }

    return $report;
}


# format_cell(forecast, actual) => report fragment
#
# For a day, takes the forecast string (e.g., "UPPER 80S"), and the actual
# corresponding high or low temperature, and builds the table cell that
# describes the forecast, actual, and the accuracy of the forecast.
#
# Also, this subroutine increments the global counts for forecast accuracy --
# $below_count, $above_count, and $good_count -- as it compares forecasts and
# actual observations.
#
# Returns the report fragment string.  Example return values:
#
# format_cell('UPPER 80S', 85.021) => '87-89 => 85.0 (below)'
# format_cell('AROUND 70', 69.557) => '69-71 => 69.6 (good) '

sub format_cell {
    my ($forecast, $actual) = @_;
    my ($forecast_min, $forecast_max) = compute_forecast_range($forecast);

    # WRITE THIS SUBROUTINE!

    return $report;
}


# compute_forecast_range(forecast) => (minimum, maximum)
#
# Takes a string forecast (e.g., "UPPER 80S") and determines the actual forecast
# temperature range; returns the minimum and maximum temperatures of the range.
# Examples:
#
# AROUND 80 => (79, 81)
# LOWER 80S => (81, 83)
# MID 80S   => (84, 86)
# UPPER 80S => (87, 89)

sub compute_forecast_range {
    my $forecast_string = shift;

    # WRITE THIS SUBROUTINE!

    return ($minimum_temperature, $maximum_temperature);
}


# analyze_counts() => report
#
# Analyzes the counts of forecast accuracy and returns a multiline string that
# contains a textual report on the counts.  The result string is wrapped to the
# given column width.
#
# Example return value (wrapped to 64 columns):
#
# Of the 8 forecasts, the actual temperatures were below the
# forecast 3 times, above the forecast 1 time, and accurate 4
# times.  Overall accuracy was 50%.


sub analyze_counts {
    my $column = shift;
    my $report = '';

    # WRITE THIS SUBROUTINE!

    return $report;
}


# Tests ========================================================================

sub run_tests {
    require Test::More;
    Test::More->import;
    plan(tests => 2);

    is((compute_forecast_range("UPPER 80S"))[0], 87, 'upper 80s minimum');
    is((compute_forecast_range("UPPER 80S"))[1], 89, 'upper 80s maximum');

    exit(0);
}