#!/usr/bin/perl
# Filename:	myfilm
# Original Author:	David Ljung Madison <DaveSource.com>
# See License:	http://MarginalHacks.com/License/
# Description:	For use with the Nokia 770.  The basic idea is to 
#			browse youtube and bookmark videos you are interested in.
#			Drop to osso-xterm and run the script to batch download!
my $VERSION=  '1.02';
#
# Requires that youtube's obfuscation doesn't change drastically.
#
# youtube: http://youtube.com/watch?v=[video_id]
#   video at:  http://youtube.com/get_video.php?l=165&video_id=[video_id]&t=[??]
#   Need to find &t=[c0de] in HTML
#
#
# Also see: http://javimoya.com/blog/youtube_en.php
##################################################
# CHANGELOG
##################################################
# 2007/8/16
# ----------------
# Haxxed for the 770 by gogol.  Somebody with 
# better perl skillz should really clean it up!
#
# 1.02  2006/05/13
# ----------------
# + Updated for new youtube obfuscation (&t=[c0de])
#
# 1.01  2006/04/08
# ----------------
# + Released as myfilm
# + Now handles youtube as well as iFilm
#
# 1.00  2005/12/08
# ----------------
# + Released as diFilm
#
##################################################
#use strict;
##################################################
# Setup the variables
##################################################
my $PROGNAME = $0; $PROGNAME =~ s|.*/||;
my ($BASENAME,$PROGNAME) = ($0 =~ m|(.*)/(.+)|) ? ($1?$1:'/',$2) : ('.',$0);

# This will probably change..
my $YOUTUBE = "http://youtube.com/get_video.php?l=165&video_id=";

# How to fetch web files - pick one
# Actual Safari agent example:
# Mozilla/5.0 (Macintosh; U; PPC Mac OS X; en) AppleWebKit/416.12 (KHTML, like Gecko) Safari/416.13
my $GET = "GET --header 'user agent: Mozilla/4.0 (compatible; MSIE 6.0; ; Linux armv5tejl; U) Opera 8.02 [en_US] Maemo browser 0.4.34 N770/SU-18' ";
my $wgetter = "wget -O -";
my $GET_URL = $wgetter;

#default save dir for youtube
my $defdir = "/home/user/MyDocs/.videos"; #note no trailing slash
#bookmarks are where?
my $BMARKS = "/home/user/.bookmarks/MyBookmarks.xml";

##################################################
# Usage
##################################################
sub fatal {
	foreach my $msg (@_) { print STDERR "[$PROGNAME] ERROR:  $msg\n"; }
	exit(-1);
}

sub usage {
	foreach my $msg (@_) { print STDERR "ERROR:  $msg\n"; }
	print STDERR <<USAGE;

Usage:\t$PROGNAME [-d] <youtube url>
  Download an ifilm movie
  -o <file>         Save to file
  -bw <56,200,300>	Force a bandwidth other than your iFilm settings
  -d                Set debug mode
  -s		      Scan /home/user/.bookmarks/MyBookmarks.xml for youtube video links.
			It will then download any that don't already exist in ~/MyDocs/.videos
USAGE
	exit -1;
}

sub parse_args {
	my $opt = {};
	while (my $arg=shift(@ARGV)) {
		if ($arg =~ /^-s$/) {$opt->{bmfile} = 1; next;}
		if ($arg =~ /^-h$/) { usage(); }
		if ($arg =~ /^-d$/) { $MAIN::DEBUG=3; next; }
		if ($arg =~ /^-bg$/) { $opt->{bg} = 1; next; }
		if ($arg =~ /^-bw$/) { $opt->{bw} = shift @ARGV; next; }
		if ($arg =~ /^-o$/) { $opt->{save} = shift @ARGV; next; }
		if ($arg =~ /^-/) { usage("Unknown option: $arg"); }
		if ($arg =~ m|^http://([^/]+)/.*?([^/]+)$|) { ($opt->{url}, $opt->{dom}, $opt->{film}) = ($arg, $1, $2); next; }
		usage("Unknown film id.  I need an iFilm or youtube URL");
	}

	usage("No film URL specified.") unless $opt->{film} || $opt->{bmfile};
	$opt;
}

sub debug {
	return unless $MAIN::DEBUG;
	foreach my $msg (@_) { print STDERR "[$PROGNAME] $msg\n"; }
}

##################################################
# Deobfuscation
##################################################
sub title_embed {
	my ($opt, $url) = @_;

	debug("URL: $url");
	my $open = "$GET_URL $url";
	debug("GET: $open");
	open(FILM,"$open |") || usage("Couldn't get URL: $url\nMake sure [$GET_URL] is installed");
	
	# Find the film location
	my $dl;
	my $title;
	my $bw = $opt->{bw};
	my $embed;
	my $and_t;
	while(<FILM>) {
		$dl = $1 if /var\s+dl\s*=\s*"?(-?\d+)"?/;
		$bw = $1 if !$bw && /var\s+bw\s*=\s*"?(\d+)"?/;
		$embed = $1 if /<embed.*(http:[^>"]+)/i;
		$title = $1 if /span class="title">([^<]+)/;
		$title = $1 if /<title>(.*)<\/title>/;
		$and_t = $1 if /video_id=.*&t=([^&"]+)/;
	}
	close(FILM);
	($title,$embed,$dl,$bw,$and_t);
}


#################################################
#Bookmark Parsing
#################################################
sub getbookmarks { 

	open BMARKS, "<$BMARKS" or die "Can't open bookmarks for reading: $!";
	while (<BMARKS>) { chomp;
	     if ($_ =  m|(http://(?:www\.)?youtube\.com/watch\?v=[^"]+)|i){ push @youtube_vids, $1; }
	}
	
	print "Found ",($#youtube_vids + 1)," bookmarked videos:\n";
	
	while ($ourcount < ($#youtube_vids+1) ) {
  		print "$ourcount : $youtube_vids[$ourcount]\n";
		if ($MAIN::DEBUG == "3") {
  			@args = ($PROGNAME, "-d", $youtube_vids[$ourcount]); 
		}
		else { @args = ($PROGNAME, $youtube_vids[$ourcount]); }
		print "$args[0] $args[1] $args[2]\n";
  		$ourcount++;
  		system(@args) == 0 or die "failed: $!\n";
  	}

}


##################################################
# youtube
##################################################
sub youtube {
	my ($opt, $film) = @_;
	print STDERR "Fetching youtube ID: $film\n";

	# Get name
	my ($save,undef,undef,undef,$and_t) = title_embed($opt, $opt->{url});
	fatal("No '&t=' code found for youtube [$opt->{url}]") unless $and_t;
	if ($opt->{save}) {
		$save = $opt->{save};
	} else {
		fatal("No title found for youtube [$opt->{url}]") unless $save;
		$save =~ s/youtube( - )?(.+)/$2/ig;
		print STDERR "Found film '$save'\n";
		$save =~ s/[:\s]+/_/g;
	}
	$save =~ s/(\.flv)?$/.flv/;
	$save =~ s|/|_of_|g;
	$save =~ s/[\*\?\'\"\&\,]//g;	# Get rid of troublesome characters..
	$save =~ s/\(//g; $save =~ s/\)//g;    #..including parentheses.
	debug("save: $save");
	# Get video
	my $video = $YOUTUBE.$film."&t=$and_t";
	debug("Video: $video");
	$filename = join "/", $defdir, $save;
	debug("filename: $filename");
	if (-e $filename) { 
		print "File $defdir/$save already exists, skipping.\n";

		#WOW the bookmarks xml SUCKS.  Eventually we'll delete old links...
		#open BMARKS, "<$BMARKS"; @all = <BMARKS>;
		#foreach (@all) {
		#if(m|$film|i) { print; $_ = ""; print "Erased bookmark to already downloaded file.\n";}
		#}
		#open BMARKS, ">$BMARKS"; 
		#foreach(@all) {
		#		#print;
		#			}
		#return;

}
	print "Saving to: $defdir/$save\n";
       system("$GET_URL \Q$video\E > $filename");

}

##################################################
# Main code
##################################################
sub main {
	my $opt = parse_args();

	if ($opt->{dom} =~ /youtube/i) {
		$opt->{film} = "watch?v=$opt->{film}"
			if $opt->{url} =~ s|www.youtube.com/v/|www.youtube.com/watch?v=|;
		usage("Couldn't find youtube video (URL must contain 'v=[video_id]')")
			unless $opt->{film} =~ /v=([a-z0-9_-]+)/i;
		return youtube($opt, $1);
	}

	if (defined $opt->{bmfile}) {&getbookmarks;} 
	

}
main();