#!/usr/bin/perl

use warnings;

use strict;

use bigint;

BEGIN {

	my @deps = qw(

		Number::Nary

		LWP::Simple

		Getopt::Long

		IO::File

	);

	foreach my $dep (sort @deps) {

		eval "require $dep";

		if($@) {

			die "Cannot find module dependency $dep - Please install it via cpan\n";

		}

	}

}

use IO::File;

use Number::Nary;

use LWP::Simple;

use Getopt::Long;

$|++;

my ($enc, $dec) = n_codec( join("", 0 .. 9, 'a' .. 'z', 'A' .. 'Z') );

my ($file, $length, $start, $output, $revision_dir);

GetOptions(

	"file=s"   => \$file,

	"length=s" => \$length,

	"start=s"  => \$start,

	"output=s" => \$output,

	"revisions=s" => \$revision_dir,

);

$length ||= 10;

$start  ||= 0;

$file   ||= "ep_ids.txt";

$output ||= "archive";

$revision_dir ||= "$output/revisions";

unless(-d $output) {

	die "$output is not a valid directory\n";

}

unless(-d $revision_dir) {

	die "$revision_dir is not a valid directory\n";

}

my $i = 0;

if(-f $file) {

	print "Scanning $file for starting point...\n";

	my $last;

	my $fh = IO::File->new($file, "r");

	foreach my $line (<$fh>) {

		chomp $line;

		if( (length $line) && ($line !~ /^\s*$/) ) {

			$last = $line;

		}

	}

	$fh->close;

	undef $fh;

	if($last) {

		$last =~ s/^- //;

		if( $start ) {

			print "** Overriding the --start option, using the last recorded value in $file\n";

		}

		$start = $last;

	}

}

print <<EOT;

Using input file $file

Stopping when length cross $length

Starting with id $start

Outputting initial versions to $output

Outputting revisions to $revision_dir

EOT

if($start) {

	$i = $dec->($start);

}

$i=$i+0; # this overcomes some perl string/number strangness that angers 32bit perls

my $id = $enc->($i++);

while(length($id) <= $length) {

	my $start = "Looking at $id ...";

	my $c = get("http://etherpad.com/$id");

	unless($c) {

		until(length $c) {

			print "No content returned from http://etherpad.com/$id. Retrying in 1 second\n";

			sleep 1;

			print "Looking at $id ...";

			$c = get("http://etherpad.com/$id");

		}

	}

	if(length $c) {

		if($c =~ /var clientVars = (\{.+\})/) {

			my $js = $1;

			write_to_file("$output/$id.js", "$js\n");

			print "$start Archived\n";

			$js =~ /"initialRevisionList":\[(.+?)\]/;

			my ($revlist, @revisions);

			$revlist = $1;

			if($revlist) {

				@revisions = $revlist =~ /"id":"(.+?)"/g;

			}

			if(@revisions) {

				foreach my $rev_id (@revisions) {

					my $output_file = "$revision_dir/$id-$rev_id.js";

					my $rev_start = "** $id Revision: $rev_id ... ";

					if(-f $output_file) {

						print "$rev_start Skip\n";

					} else {

						my $js = grab_clientvars("http://etherpad.com/ep/pad/view/$id/$rev_id");

						if($js) {

							write_to_file($output_file, $js);

							print "$rev_start Archived\n";

						} else {

							print "$rev_start No content\n";

						}

					}

				}

			}

		} else {

			print "$start No\n";

		}

	} else {

		print "$start No\n";

	}

	write_to_file($file, "$id\n");

	$id = $enc->($i++);

}

#####################

sub grab_clientvars {

	my $url = shift;

	my $content = get($url); 

	my ($clientvars) = $content =~ /var clientVars = (.+?);\n/ms;

	return $clientvars if $clientvars;

	return;

}

sub write_to_file {

	my ($file, $text) = @_;

	my $fh = IO::File->new($file, '>') 

		or die "Could not open $file for writing: $!";

	print $fh $text;

	undef $fh;

	return;

}