crystal/irc.cr: finish prototyping windows tts support

1 files changed, 170 insertions, 45 deletions

diff --git a/crystal/irc.cr b/crystal/irc.cr
index 9ac7e8b..2df28fa 100644..100755
--- a/crystal/irc.cr
+++ b/crystal/irc.cr
@@ -23,36 +23,90 @@ if ENV["LOCALAPPDATA"]?
   settings["statedir"] = Path.windows( ENV["LOCALAPPDATA"] + "\\bungmobott\\state\\" ).to_s
 end
 
+settings["tempdir"] = "/tmp/bungmobott/"
+ENV["TEMP"]? && ( settings["tempdir"] = "#{ENV["TEMP"]}\\bungmobott\\" )
+
 ENV["XDG_CONFIG_HOME"]? && ( settings["configdir"] = ENV["XDG_CONFIG_HOME"] + "/bungmobott/" )
 #ENV["XDG_DATA_HOME"]? && ( settings["datadir"] = ENV["XDG_DATA_HOME"] ) # unused?
 ENV["XDG_STATE_HOME"]? && ( settings["statedir"] = ENV["XDG_STATE_HOME"] + "/bungmobott/" )
 
 Dir.mkdir_p( settings["configdir"] )
 Dir.mkdir_p( settings["statedir"] )
+Dir.mkdir_p( settings["tempdir"] )
 
 settings["home"] = Path.home.to_s
-["access_token", "channel", "channel_id", "client_id", "client_id_twitch"].each do |key|
+regextwitchuser = /^[0-9a-zA-Z_]+$/
+
+error = false
+[ "access_token", "client_id" ].each do |key|
   begin
     settings[key] = File.read( settings["configdir"] + key ).chomp
   rescue IO::Error
-    STDERR.puts "Warning: Missing " + settings["configdir"] + key
+    STDERR.puts "ERROR: Missing " + settings["configdir"] + key
+    error = true
   end
 end
 
+# enable gcloud?
+if File.exists?( settings["configdir"] + "gcloud_token" ) && ( settings["gcloud_token"] = File.read( settings["configdir"] + "gcloud_token" ).chomp )
+  gcloud = true
+else
+  gcloud = false
+  STDERR.puts "Warning: #{settings["configdir"]}gcloud_token is missing; GCS voices disabled."
+end
+
+# enable aws?
+if ! File.exists?( ENV["USERPROFILE"] + "\\.aws\\credentials" )
+  STDERR.puts "Warning: #{ENV["USERPROFILE"]}\\.aws\\credentials is missing; AWS voices disabled."
+  aws = false
+elsif ! Process.find_executable( "aws.exe" )
+  STDERR.puts "Warning: aws.exe is missing; AWS voices disabled."
+  aws = false
+else
+  aws = true
+end
+
+client = Twitcr::Client.new( settings )
+
+# derive channel_id from channel or vice versa
+if (
+     File.exists?( settings["configdir"] + "channel"    ) &&
+     ( settings["channel"]    = File.read( settings["configdir"] + "channel" ).chomp )
+   ) || (
+     File.exists?( settings["configdir"] + "channel_id" ) &&
+     ( settings["channel_id"] = File.read( settings["configdir"] + "channel_id" ).chomp )
+   )
+   if  ! ( settings["channel"]?    =~ regextwitchuser ) &&   settings["channel_id"]? =~ /^[0-9]+$/
+     settings["channel"]    = client.user( settings["channel_id"].to_u64 ).login
+   elsif ( settings["channel_id"]? =~ regextwitchuser ) && ! settings["channel_id"]? =~ /^[0-9]+$/
+     settings["channel_id"] = client.user( settings["channel"] ).id.to_s
+   end
+else
+  STDERR.puts "ERROR: Missing #{settings["configdir"]}channel and channel_id configuration keys."
+  error = true
+  # exit 2
+end
+if error == true
+  {% if flag?(:windows) %}
+    puts "press enter to end program"
+    gets
+  {% end %}
+  exit 1
+end
+
 if File.exists?( settings["configdir"] + "chatuser" )
   File.read( settings["configdir"] + "chat_user" ).chomp
 else
-  STDERR.puts "Warning: Missing " + settings["configdir"] + "chat_user; using configured channel instead."
+  STDERR.puts "Missing " + settings["configdir"] + "chat_user; using configured channel instead."
   settings["chat_user"] = settings["channel"]
 end
 
 obsipc = Channel( String ).new
 ircipc = Channel( Tuple( String, String ) ).new
+t2sipc = Channel( Tuple( String, String ) ).new
 
 snifflast = Int64.new(0)
 
-client = Twitcr::Client.new( settings )
-
 def urbandef( term : String )
   #http://api.urbandictionary.com/v0/define?term=waifu
   client = HTTP::Client.new( "api.urbandictionary.com" )
@@ -62,16 +116,7 @@ def urbandef( term : String )
   return json["list"][0]["definition"].to_s.gsub( /[\[\]]/, "" ).gsub( /\n/, "" )
 end
 
-#def ircmsg( home : String, channel : String, msg : String)
-#  if msg !~ /(\r|\n)/
-#    sock = Socket.unix
-#    sock.connect Socket::UNIXAddress.new( home + "/.irssi/twitch-socket".to_s, type: Socket::Type::DGRAM )
-#    sock.connect Socket::UNIXAddress.new( home + "/.irssi/twitch-socket".to_s )
-#    sock.puts( "#" + channel + " " + msg )
-#    sock.close
-#  end
-#end
-
+# Currently only used in flags?(:unix)
 def t2smsg( settings : Hash(String, String), msg : String)
   if File.exists?( settings["statedir"] + "/.t2s.sock" )
     sock = Socket.unix
@@ -145,7 +190,7 @@ def getvoice( settings : Hash(String, String), userdir : String, chatuser : Stri
 end
 
 # text2speech
-def t2s( settings : Hash(String, String), userdir : String, chatuser : String, text : String )
+def t2s( t2sipc : Channel, settings : Hash(String, String), userdir : String, chatuser : String, text : String )
   if ( text !~ /^ *(!|\|)/ )
     namesub, voice_setting, voice = getvoice( settings, userdir, chatuser )
     subs = Array( Tuple( Regex, String ) ){
@@ -186,27 +231,9 @@ def t2s( settings : Hash(String, String), userdir : String, chatuser : String, t
     }.each do | subtuple |
       text = text.gsub( subtuple[0], subtuple[1] )
     end
-    {% if flag?(:windows) %}
-      if ( match = voice.match( /^Microsoft-([A-Za-z]+)/ ) )
-        if ( match[1] =~ /Lili|Mary|Mike|Sam|Anna/ )
-          msttsvoice="Microsoft #{match[1]}"
-        else
-          msttsvoice="Microsoft #{match[1]} Desktop"
-        end
-        pp msttsvoice
-        p = Process.new( "powershell.exe", [ "-Command", "
-            echo test;
-            Add-Type -AssemblyName System.Speech;
-            $speak = New-Object System.Speech.Synthesis.SpeechSynthesizer;
-            $speak.SelectVoice(\"#{msttsvoice}\");
-            $speak.Speak($Input);
-            $speak.Finalize($Input);
-        "],
-      input: Process::Redirect::Pipe, output: STDOUT )
-        p.input.puts "#{namesub} #{text}"
-        p.input.close
-      end
-    {% elsif flag?(:unix) %}
+    {% if flag?(:windows) %} # send to thread
+      t2sipc.send( { voice, "#{namesub} #{text}" } )
+    {% elsif flag?(:unix) %} # send to socket
       t2smsg( settings, "#{voice} #{namesub} #{text}" )
     {% end %}
     return( voice )
@@ -238,6 +265,25 @@ def reversesource( scenes : Hash( String, Hash( String, Hash( String, Int64 | Fl
   return sourcemap
 end
 
+def playmp3file( filepath : String )
+  p = Process.new(
+    "powershell.exe",
+    [ "-Command", "#Set-PSDebug -Trace 1;
+      Add-Type -AssemblyName presentationCore;
+      $player = New-Object system.windows.media.mediaplayer;
+      $player.open(\"#{filepath}\");
+      $player.volume = .99;
+      $player.play();
+      Start-Sleep -Milliseconds 1000;
+      $duration = $player.NaturalDuration.TimeSpan.TotalMilliseconds;
+      Start-Sleep -Milliseconds ($duration - 1000 );
+    "],
+    output: STDOUT, error: STDERR
+  )
+  # https://geekeefy.wordpress.com/2016/07/19/powershellmediaplayer/ has some ideas
+  p.wait
+end
+
 macro testrefuser2uid( path )
   {% if flag?(:windows) %}
     File.exists?( {{path}} ) && ( File.read( {{ path }} ) =~ /^[0-9]+$/ )
@@ -281,7 +327,82 @@ end
 
 lastvoice = Array(String).new
 
-# obs-websocket
+# Put tts stuff into the same thread so each playback blocks the next
+spawn do
+  loop do
+    begin
+      while t2stuple = t2sipc.receive
+      voice, text = [ *t2stuple ]
+        if ( match = voice.match( /^Microsoft-([A-Za-z]+)/ ) )
+          if ( match[1] =~ /Lili|Mary|Mike|Sam|Anna/ )
+            msttsvoice="Microsoft #{match[1]}"
+          else
+            msttsvoice="Microsoft #{match[1]} Desktop"
+          end
+          p = Process.new(
+              "powershell.exe",
+              [ "-Command", "
+                Add-Type -AssemblyName System.Speech;
+                $speak = New-Object System.Speech.Synthesis.SpeechSynthesizer;
+                $speak.SelectVoice(\"#{msttsvoice}\");
+                $speak.Speak($Input);
+                $speak.Finalize;
+              "],
+              input: Process::Redirect::Pipe, output: STDOUT
+            )
+          p.input.puts text
+          p.input.close
+          p.wait
+        elsif gcloud && ( match = voice.match( /^([a-zA-Z]{2,3}-[a-zA-Z]{2})/ ) ) # Google cloud voice
+          request = Hash( String, Hash( String, String ) ){
+            "input" => { "text" => text },
+            "audioConfig" => { "audioEncoding" => "MP3" },
+            "voice" => {
+              "name" => voice,
+              "languageCode" => match[1],
+            },
+          }
+          body = request.to_json
+          ssl_context = OpenSSL::SSL::Context::Client.new
+          ssl_context.verify_mode = OpenSSL::SSL::VerifyMode::NONE
+
+          headers = HTTP::Headers.new
+          headers["Content-Type"] = "application/json; charset=utf-8"
+
+          response = HTTP::Client.exec( "POST", "https://texttospeech.googleapis.com/v1/text:synthesize?key=#{settings["gcloud_token"]}", headers, body, tls: ssl_context )
+
+          response.body
+
+          filepath="#{settings["tempdir"]}#{Time.utc.to_unix}.mp3"
+          json=JSON.parse(response.body)
+          File.write( filepath, Base64.decode_string( json["audioContent"].as_s ) )
+          playmp3file( filepath )
+          File.delete( filepath )
+        elsif aws # AWS polly voices
+          filepath="#{settings["tempdir"]}#{Time.utc.to_unix}.mp3"
+          p = Process.new(
+            "aws.exe", [
+              "polly", "synthesize-speech",
+              "--output-format", "mp3",
+              "--voice-id", voice,
+              "--text", text,
+              filepath
+            ], output: STDOUT, error: STDERR
+          )
+          p.wait
+          playmp3file( filepath )
+          File.delete( filepath )
+        else # unknown
+          STDERR.puts "Voice not recognized or available."
+        end
+      end
+    rescue ex
+      puts ex
+    end
+  end
+end
+
+# obs-websocket thread
 spawn do
   loop do
     begin
@@ -408,7 +529,7 @@ spawn do
       obs_pubsub.run
     rescue ex : Socket::ConnectError
       # these are a bit noisy
-      pp ex
+      #pp ex
       obs_pubsub && obs_pubsub.close
     rescue ex : IO::Error
       pp ex
@@ -424,14 +545,14 @@ spawn do
   end
 end
 
-# IRC
+# main thread: IRC
 loop do
   begin
 
-   bot = Twitch::IRC::Client.new( nick: settings["chat_user"], token: "oauth:" + settings["access_token"], log_mode: true )
+    bot = Twitch::IRC::Client.new( nick: settings["chat_user"], token: "oauth:" + settings["access_token"], log_mode: true )
     bot.tags = [ "membership", "tags", "commands" ]
 
-    # Outgoing
+    # Outgoing IRC message thread
     # "Most IRC servers limit messages to 512 bytes in length, including the trailing CR-LF characters."
     # PRIVMSG #channel message\r\n
     spawn do
@@ -450,7 +571,7 @@ loop do
         snifflast = Time.utc.to_unix
         t2smsg( settings, chatuser + " sniff null" )
       end
-      if ( t2sreturn = t2s( settings, userdir, chatuser, message.params[1] ) )
+      if ( t2sreturn = t2s( t2sipc, settings, userdir, chatuser, message.params[1] ) )
         lastvoice.insert( 0, t2sreturn )
         lastvoice = lastvoice[0..4]
       end
@@ -470,7 +591,7 @@ loop do
           end
         else
           if match[2]?
-              voicesub = match[2].downcase
+            voicesub = match[2].downcase
             if voicesub =~ /^(disabled|null|disable|none)$/
               if File.exists?( userdir + "/voicesub" )
                 File.delete( userdir + "/voicesub" )
@@ -601,7 +722,7 @@ loop do
         #sleep 0.5
         # I should probably set this up to add the source, and then delete the source at the end of playback.
         request["visible"] = true
-        ircipc.send( { "##{settings["channel"]}", "Playing media-#{cmd}" } )
+        #ircipc.send( { "##{settings["channel"]}", "Playing media-#{cmd}" } )
         obsipc.send( request.to_json )
       elsif ( ( cmd =~ /^(status|title)$/ ) && ( mod || own ) )
         begin
@@ -793,6 +914,10 @@ loop do
     # loop to reconnect
   rescue ex
     pp ex
+    {% if flag?(:windows) %}
+      puts "press enter to end program"
+      gets
+    {% end %}
     exit 1
   end
 end