googletts-tiny.agi

#!/usr/bin/env perl

#
# AGI script that uses Google's translate text to speech engine.
#
# Copyright (C) 2011 - 2015, Lefteris Zafiris <zaf.000@gmail.com>
#
# This program is free software, distributed under the terms of
# the GNU General Public License Version 2. See the COPYING file
# at the top of the source tree.
#
# -----
# Usage
# -----
# agi(googletts.agi,"text",[language],[intkey],[speed]): This will invoke the Google TTS
# engine, render the text string to speech and play it back to the user.
# If 'intkey' is set the script will wait for user input. Any given interrupt keys will
# cause the playback to immediately terminate and the dialplan to proceed to the
# matching extension (this is mainly for use in IVR, see README for examples).
# If 'speed' is set the speech rate is altered by that factor (defaults to 1.2).
#
# The script contacts google's TTS service in order to get the voice data
# which then stores in a local cache (by default /tmp/) for future use.
#
# Parameters like default language, sample rate, caching and cache dir
# can be set up by altering the following variables:
# Default langeuage: $lang
# Sample rate:       $samplerate
# Speed factor:      $speed
# Chace:             $usecache
# Chache directory:  $cachedir
# SoX Version:       $sox_ver
#
# Tiny version drops the use of mpg123 and requires MP3 format support in sox.
# It also makes the assumption that sox is in /usr/bin/ in order
# to skip detection and speed things up.
#

use warnings;
use strict;
use utf8;
use Encode qw(decode encode);
use File::Temp qw(tempfile);
use File::Copy qw(move);
use File::Path qw(mkpath);
use Digest::MD5 qw(md5_hex);
use URI::Escape;
use HTTP::Tiny;
$| = 1;

# ----------------------------- #
#   User defined parameters:    #
# ----------------------------- #
# Default language              #
my $lang = "en";

# Output speed factor           #
my $speed = 1;

# Use of cache mechanism        #
my $usecache = 1;

# Cache directory path          #
my $cachedir = "/tmp";

# Output audio sample rate      #
# Leave blank to auto-detect    #
my $samplerate = "8000";

# SoX Version                   #
# Leave blank to auto-detect    #
my $sox_ver = "14";

# Verbose debugging messages    #
my $debug = 0;

# ----------------------------- #

my %AGI;
my @text;
my $fh;
my $tmpname;
my $fexten;
my $http;
my $sox     = "/usr/bin/sox";
my $intkey  = "";
my $tmpdir  = "/tmp";
my $maxlen  = 4096;
my $timeout = 10;
my $url     = "https://translate.google.com";

# Store AGI input #
($AGI{arg_1}, $AGI{arg_2}, $AGI{arg_3}, $AGI{arg_4}) = @ARGV;
while (<STDIN>) {
	chomp;
	last if (!length);
	$AGI{$1} = $2 if (/^agi_(\w+)\:\s+(.*)$/);
}
my $name = " -- $AGI{request}:";

# Sanitising input #
$AGI{arg_1} = decode('utf8', $AGI{arg_1});
for ($AGI{arg_1}) {
	s/[\\|*~<>^\(\)\[\]\{\}[:cntrl:]]/ /g;
	s/\s+/ /g;
	s/^\s|\s$//g;
	fatal_log("No text passed for synthesis.") if (!length);
	# Split input to comply with google tts requirements #
	$_ .= "." unless (/^.+[.,?!:;]$/);
	@text = /.{1,150}[.,?!:;]|.{1,150}\s/g;
}
my $lines = @text;

# Setting language, interrupt keys and speed rate #
if (length($AGI{arg_2})) {
	if ($AGI{arg_2} =~ /^[a-zA-Z]{2}(-[a-zA-Z]{2,6})?$/) {
		$lang = $AGI{arg_2};
	} else {
		console_log("Invalid language setting. Using default.");
	}
}

if (length($AGI{arg_3})) {
	$intkey = "0123456789#*" if ($AGI{arg_3} eq "any");
	$intkey = $AGI{arg_3}    if ($AGI{arg_3} =~ /^[0-9*#]+$/);
}

if (length($AGI{arg_4})) {
	$speed = $AGI{arg_4} if ($AGI{arg_4} =~ /^\d+(\.\d+)?$/);
}

# Check cache path size: dir length + md5 + file extension #
if ($usecache) {
	if ((length($cachedir) + 32 + 6) < $maxlen) {
		mkpath("$cachedir") unless (-d "$cachedir");
	} else {
		console_log("Cache path size exceeds limit. Disabling cache.");
		$usecache = 0;
	}
}

# Answer channel if not already answered #
print "CHANNEL STATUS\n";
my @result = checkresponse();
if ($result[0] == 4) {
	print "ANSWER\n";
	@result = checkresponse();
	if ($result[0] != 0) {
		fatal_log("Failed to answer channel.");
	}
}

# Setting filename extension according to sample rate. #
if    (!$samplerate)         { ($fexten, $samplerate) = detect_format(); }
elsif ($samplerate == 12000) { $fexten = "sln12"; }
elsif ($samplerate == 16000) { $fexten = "sln16"; }
elsif ($samplerate == 32000) { $fexten = "sln32"; }
elsif ($samplerate == 44100) { $fexten = "sln44"; }
elsif ($samplerate == 48000) { $fexten = "sln48"; }
else                         { ($fexten, $samplerate) = ("sln", 8000); }

for (my $i = 0; $i < $lines; $i++) {
	my $filename;
	my $res;
	my $len = length($text[$i]);
	my $line = encode('utf8', $text[$i]);
	$line =~ s/^\s+|\s+$//g;
	next if (length($line) == 0);
	if ($debug) {
		console_log("Text passed for synthesis: $line",
			"Language: $lang, Interrupt keys: $intkey, Sample rate: $samplerate",
			"Speed: $speed, Caching: $usecache, Cache dir: $cachedir"
		);
	}
	if ($usecache) {
		$filename = md5_hex("$line.$lang.$speed");

		# Stream file from cache if it exists #
		if (-r "$cachedir/$filename.$fexten") {
			console_log("File already in cache.") if ($debug);
			$res = playback("$cachedir/$filename", $intkey);
			die  if ($res < 0);
			last if ($res > 0);
			next;
		}
	}

	# Hnadle interrupts #
	$SIG{'INT'} = \&int_handler;
	$SIG{'HUP'} = \&int_handler;

	# Initialise User angent #
	if (!$http) {
		$http = HTTP::Tiny->new(
		agent      => 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/45.0.2454.101 Safari/537.36',
		timeout    => $timeout,
		verify_SSL => 1,
		);
	}

	($fh, $tmpname) = tempfile("ggl_XXXXXXXX", DIR => $tmpdir, UNLINK => 1);
	my $headers = {
		'Accept'          => '*/*',
		'Accept-Encoding' => 'identity;q=1, *;q=0',
		'Accept-Language' => 'en-US,en;q=0.5',
		'DNT'             => '1',
		'Range'           => 'bytes=0-',
		'Referer'         => 'https://translate.google.com/',
	};
	my $token = make_token($line);
	$line = uri_escape($line);
	my $req   = "$url/translate_tts?ie=UTF-8&q=$line&tl=$lang&total=$lines&idx=$i&textlen=$len&client=tw-ob&tk=$token&prev=input";
	console_log("URL passed: $req") if ($debug);

	my $response = $http->mirror($req, $tmpname, $headers,);
	fatal_log("Failed to fetch file: ", $response->{status}, $response->{reason}) unless ($response->{success});

	my @soxargs = get_sox_args($tmpname, "$tmpname.$fexten");
	system(@soxargs) == 0 or fatal_log("$sox failed: $?");

	# Playback and save file in cache #
	$res = playback($tmpname, $intkey);
	die if ($res < 0);
	if ($usecache) {
		console_log("Saving file $filename to cache") if ($debug);
		move("$tmpname.$fexten", "$cachedir/$filename.$fexten");
	} else {
		unlink "$tmpname.$fexten";
	}
	last if ($res > 0);
}
exit;

sub checkresponse {
	my $input = <STDIN>;
	my @values;

	chomp $input;
	if ($input =~ /^200 result=(-?\d+)\s?(.*)$/) {
		@values = ("$1", "$2");
	} else {
		$input .= <STDIN> if ($input =~ /^520-Invalid/);
		warn "$name Unexpected result: $input\n";
		@values = (-1, -1);
	}
	return @values;
}

sub playback {
	my ($file, $keys) = @_;
	my @response;

	print "STREAM FILE $file \"$keys\"\n";
	@response = checkresponse();
	if ($response[0] >= 32 && chr($response[0]) =~ /[\w*#]/) {
		console_log("Got digit chr($response[0])") if ($debug);
		print "SET EXTENSION ", chr($response[0]), "\n";
		checkresponse();
		print "SET PRIORITY 1\n";
		checkresponse();
	} elsif ($response[0] == -1) {
		console_log("Failed to play $file.");
	}
	return $response[0];
}

sub detect_format {
	# Detect the sound format used #
	my @format;
	print "GET FULL VARIABLE \${CHANNEL(audionativeformat)}\n";
	my @reply = checkresponse();
	for ($reply[1]) {
		if    (/(silk|sln)12/)                    { @format = ("sln12", 12000); }
		elsif (/(speex|slin|silk)16|g722|siren7/) { @format = ("sln16", 16000); }
		elsif (/(speex|slin|celt)32|siren14/)     { @format = ("sln32", 32000); }
		elsif (/(celt|slin)44/)                   { @format = ("sln44", 44100); }
		elsif (/(celt|slin)48/)                   { @format = ("sln48", 48000); }
		else                                      { @format = ("sln",    8000); }
	}
	return @format;
}

sub get_sox_args {
	# Set the appropiate sox cli arguments #
	my ($source_file, $dest_file) = @_;

	my @soxargs = ($sox, "-t", "mp3", $source_file, "-q", "-r", $samplerate, "-t", "raw", $dest_file);
	if ($speed != 1) {
		if ($sox_ver >= 14) {
			push(@soxargs, ("tempo", "-s", $speed));
		} else {
			push(@soxargs, ("stretch", 1 / $speed, "80"));
		}
	}
	return @soxargs;
}

# Obfuscated crap straight from Google:
# https://translate.google.com/translate/releases/twsfe_w_20151214_RC03/r/js/desktop_module_main.js
sub make_token {
	my $text = shift;
	my $time = int(time() / 3600);
	my @chars = unpack('U*', $text);
	my $stamp = $time;

	foreach (@chars) {
		$stamp = make_rl($stamp + $_, '+-a^+6');
	}
	$stamp = make_rl($stamp, '+-3^+b+-f');
	if ($stamp < 0) {
		$stamp = ($stamp & 2147483647) + 2147483648;
	}
	$stamp %= 10**6;
	return ($stamp . '.' . ($stamp ^ $time));
}

sub make_rl {
	my ($num, $str) = @_;

	for (my $i = 0; $i < length($str) - 2 ; $i += 3) {
		my $d = substr($str, $i+2, 1);
		if (ord($d) >= ord('a')) {
			$d = ord($d) - 87;
		} else {
			$d = int($d);
		}
		if (substr($str, $i+1, 1) eq '+') {
			$d = $num >> $d;
		} else {
			$d = $num << $d;
		}
		if (substr($str, $i, 1) eq '+') {
			$num = $num + $d & 4294967295;
		} else {
			$num = $num ^ $d;
		}
	}
	return $num;
}

sub console_log {
	foreach my $message (@_) {
		warn "$name $message\n";
		print "NOOP \"$name $message\"\n";
		checkresponse();
	}
}

sub fatal_log {
	console_log(@_);
	die;
}

sub int_handler {
	die "$name Interrupt signal received, terminating...\n";
}

END {
	if ($tmpname) {
		warn "$name Cleaning temp files.\n" if ($debug);
		unlink glob "$tmpname*";
	}
}