date

#!/usr/bin/perl -w
# Usage
# jwc2ical [class_num]
# Examples: 	jwc2ical 0903101
# 		jwc2ical 0903102 > 0903102.ics

use strict;
use LWP;
use HTML::Tree;
use HTML::TreeBuilder;
use Encode;
use utf8;

my $browser = LWP::UserAgent->new;
my $class_num = $ARGV[0];
my $date_url = 'http://jwc.hit.edu.cn/jwc/displaytwo.asp?ID=202';
my @num_name = qw{ 一 二 三 四 五 六 七 八 九 十 十一 十二};
my @num_syb = ( "\x{2474}", "\x{2475}", "\x{2476}", "\x{2477}", "\x{2478}", "\x{2479}",
	"\x{247A}", "\x{247B}", "\x{247C}", "\x{247D}", "\x{247E}", "\x{247F}", "\x{2480}",
	"\x{2481}", "\x{2482}", "\x{2483}", "\x{2484}", "\x{2485}", "\x{2486}", "\x{2487}");

my $date_table = $browser->get( $date_url);
&check( $date_table, $date_url);
# Now for date table.

my $date_tree = HTML::TreeBuilder->new_from_content( decode( 'gbk', $date_table->content));

# Only the head is needed.
my $real_date = $date_tree->look_down( '_tag', 'thead');

# Get years out.
my $year_list = $real_date->look_down( '_tag', 'tr', sub { $_[0]->as_text =~ m/哈尔滨工业大学/});
my $year = $year_list->as_text;
$year =~ s/.*([\d]{4}).*/$1/;
#print "year is $year.\n";

my @month;
my @month_length;

my %look_for;
foreach ( qw( 月 周 星期一 星期日))
{
	my $name = $_;
	$look_for{$name} = sub {
		bless $_[0], "HTML::Element";
		($_[0]->content_list)[0]->as_text eq $name
	};
}

# Get months out.
my $month_list = $real_date->look_down( '_tag', 'tr', $look_for{ 月});
#print $month_list->as_text, "\n";

foreach ( ($month_list->content_list)[1..($month_list->content_list-1)])
{
	push @month, &find( [ @num_name], $_->as_text);
	push @month_length, defined $_->attr( 'colspan') ? $_->attr( 'colspan') : 1;
}
#print "@month_length\n";

# Get weeks out.
my $week_list = $real_date->look_down( '_tag', 'tr', $look_for{ 周});
my @weeks;
my @weeks_length;

foreach ( ( $week_list->content_list)[1..($week_list->content_list-1)])
{
	my $num = &find( [ @num_syb], $_->as_text);
	push @weeks, $num ? $num + 100 : $_->as_text;
	push @weeks_length, defined $_->attr( 'colspan') ? $_->attr( 'colspan') : 1;
}
#print "@weeks", "\n";

foreach ( 1..@month_length)
{
	$month_length[$_] += $month_length[$_-1];
}

foreach( 1..@weeks_length)
{
	$weeks_length[$_] += $weeks_length[$_-1];
}

die "Not so useful date html ". ( scalar $weeks_length[-1]) . " and $month_length[-1]", "\n"
	if $weeks_length[-1] != $month_length[-1];

# Get dates of mondays and sundays out.
my @monday_dates;
my $monday_list = $real_date->look_down( '_tag', 'tr', $look_for{ 星期一});

foreach ( ( $monday_list->content_list)[1..($monday_list->content_list-1)])
{
	push @monday_dates, $_->as_text;
}

my @sunday_dates;
my $sunday_list = $real_date->look_down( '_tag', 'tr', $look_for{ 星期日});

foreach ( ( $sunday_list->content_list)[1..($sunday_list->content_list-1)])
{
	push @sunday_dates, $_->as_text;
}

# We only need the first monday here.
my @first_monday;
my $month = $month[0];

# I gussed it!! This might be WRONG!!
if ( $monday_dates[0] != $sunday_dates[0] - 6 && $monday_dates[0] > 28)
{
	# Some wrong date are present, egg pain.
	--$year, $month = 12 if ( --$month == 0);
}
# Normal case
print "$year-$month-$monday_dates[0]\n";
exit;

sub find
{
	my $num = $_[0];
	foreach ( 1..@$num)
	{
		return $_ if ( $_[1] eq ${$num}[$_-1]);
	}
	return undef;
}

sub check
{
	my $response = $_[0];
	my $url = $_[1];
	die "$url error: ", $response->status_line unless $response->is_success;
	die "Weird content type at $url -- ", $response->content_type
		unless $response->content_type eq 'text/html';
#	print $response->content;

	return $response;
}