Merge pull request #100 from astahlman/master

Platform independent audio output. Stop hardcoding /home/jasper.

9 files changed, 112 insertions, 25 deletions

diff --git a/boot/boot.py b/boot/boot.py
index 456dd12..12b60eb 100755
--- a/boot/boot.py
+++ b/boot/boot.py
@@ -3,12 +3,16 @@
 import os
 import json
 import urllib2
+import sys
 
 import vocabcompiler
+import traceback
 
-def say(phrase, OPTIONS = " -vdefault+m3 -p 40 -s 160 --stdout > ../static/audio/say.wav"):
-    os.system("espeak " + json.dumps(phrase) + OPTIONS)
-    os.system("aplay -D hw:1,0 ../static/audio/say.wav")
+lib_path = os.path.abspath('../client')
+sys.path.append(lib_path)
+
+import speaker as speak
+speaker = speak.newSpeaker()
 
 def configure():
     try:
@@ -19,17 +23,17 @@ def configure():
         vocabcompiler.compile("../client/sentences.txt", "../client/dictionary.dic", "../client/languagemodel.lm")
 
         print "STARTING CLIENT PROGRAM"
-        os.system("/home/pi/jasper/client/start.sh &")
+        os.system("$JASPER_HOME/jasper/client/start.sh &")
 
     except:
-
         print "COULD NOT CONNECT TO NETWORK"
-        say("Hello, I could not connect to a network. Please read the documentation to configure your Raspberry Pi.")
+        traceback.print_exc()
+        speaker.say("Hello, I could not connect to a network. Please read the documentation to configure your Raspberry Pi.")
 
 if __name__ == "__main__":
     print "==========STARTING JASPER CLIENT=========="
     print "=========================================="
     print "COPYRIGHT 2013 SHUBHRO SAHA, CHARLIE MARSH"
     print "=========================================="
-    say("Hello.... I am Jasper... Please wait one moment.")
+    speaker.say("Hello.... I am Jasper... Please wait one moment.")
     configure()
diff --git a/boot/boot.sh b/boot/boot.sh
index 5ea1b3d..e712c0d 100755
--- a/boot/boot.sh
+++ b/boot/boot.sh
@@ -1,6 +1,17 @@
-cd /home/pi/jasper/boot/
+#!/bin/bash
+if [[ -z "$JASPER_HOME" ]]; then
+    if [[ -d "/home/pi" ]]; then
+        JASPER_HOME="/home/pi"
+        export JASPER_HOME;
+    else
+        echo "Error: \$JASPER_HOME is not set."
+        exit 0;
+    fi
+fi
+
+cd $JASPER_HOME/jasper/boot
 LD_LIBRARY_PATH="/usr/local/lib"
 export LD_LIBRARY_PATH
 PATH=$PATH:/usr/local/lib/
 export PATH
-python boot.py &
\ No newline at end of file
+python boot.py &
diff --git a/boot/test.py b/boot/test.py
index c6bd3ff..af5f982 100644
--- a/boot/test.py
+++ b/boot/test.py
@@ -1,4 +1,8 @@
 import os
+
+if os.environ.get('JASPER_HOME') is None:
+    os.environ['JASPER_HOME'] = '/home/pi'
+
 import sys
 import unittest
 from mock import patch
@@ -12,6 +16,10 @@ sys.path.append(mod_path)
 
 import g2p
 
+class UnorderedList(list):
+    def __eq__(self, other):
+        return sorted(self) == sorted(other)
+    
 
 class TestVocabCompiler(unittest.TestCase):
 
@@ -33,7 +41,7 @@ class TestVocabCompiler(unittest.TestCase):
 
                 # 'words' is appended with ['MUSIC', 'SPOTIFY']
                 # so must be > 2 to have received WORDS from modules
-                translateWords.assert_called_once_with(words)
+                translateWords.assert_called_once_with(UnorderedList(words))
                 self.assertTrue(text2lm.called)
         os.remove(sentences)
         os.remove(dictionary)
diff --git a/client/g2p.py b/client/g2p.py
index 06fef14..6d72bfe 100644
--- a/client/g2p.py
+++ b/client/g2p.py
@@ -4,6 +4,7 @@ import re
 
 TEMP_FILENAME = "g2ptemp"
 PHONE_MATCH = re.compile(r'<s> (.*) </s>')
+PHONETISAURUS_PATH = os.environ['JASPER_HOME'] + "/phonetisaurus"
 
 
 def parseLine(line):
@@ -16,7 +17,7 @@ def parseOutput(output):
 
 def translateWord(word):
     out = subprocess.check_output(['phonetisaurus-g2p', '--model=%s' %
-                                  os.path.expanduser("~/phonetisaurus/g014b2b.fst"), '--input=%s' % word])
+                                  PHONETISAURUS_PATH + "/g014b2b.fst", '--input=%s' % word])
     return parseLine(out)
 
 
@@ -34,8 +35,8 @@ def translateWords(words):
 
 
 def translateFile(input_filename, output_filename=None):
-    out = subprocess.check_output(['phonetisaurus-g2p', '--model=%s' % os.path.expanduser(
-        "~/phonetisaurus/g014b2b.fst"), '--input=%s' % input_filename, '--words', '--isfile'])
+    out = subprocess.check_output(['phonetisaurus-g2p', '--model=%s' % 
+        PHONETISAURUS_PATH + "/g014b2b.fst", '--input=%s' % input_filename, '--words', '--isfile'])
     out = parseOutput(out)
 
     if output_filename:
@@ -51,5 +52,5 @@ def translateFile(input_filename, output_filename=None):
 
 if __name__ == "__main__":
 
-    translateFile(os.path.expanduser("~/phonetisaurus/sentences.txt"),
-                  os.path.expanduser("~/phonetisaurus/dictionary.dic"))
+    translateFile(PHONETISAURUS_PATH + "/phonetisaurus/sentences.txt",
+                  PHONETISAURUS_PATH + "/phonetisaurus/dictionary.dic")
diff --git a/client/main.py b/client/main.py
index 0b29c75..6ba3645 100644
--- a/client/main.py
+++ b/client/main.py
@@ -1,5 +1,6 @@
 import yaml
 import sys
+import speaker
 from conversation import Conversation
 
 
@@ -20,7 +21,7 @@ if __name__ == "__main__":
 
     profile = yaml.safe_load(open("profile.yml", "r"))
 
-    mic = Mic("languagemodel.lm", "dictionary.dic",
+    mic = Mic(speaker.newSpeaker(), "languagemodel.lm", "dictionary.dic",
               "languagemodel_persona.lm", "dictionary_persona.dic")
 
     addendum = ""
diff --git a/client/mic.py b/client/mic.py
index d746ba1..f4b70df 100644
--- a/client/mic.py
+++ b/client/mic.py
@@ -22,17 +22,18 @@ class Mic:
     speechRec = None
     speechRec_persona = None
 
-    def __init__(self, lmd, dictd, lmd_persona, dictd_persona, lmd_music=None, dictd_music=None):
+    def __init__(self, speaker, lmd, dictd, lmd_persona, dictd_persona, lmd_music=None, dictd_music=None):
         """
             Initiates the pocketsphinx instance.
 
             Arguments:
+            speaker -- handles platform-independent audio output
             lmd -- filename of the full language model
             dictd -- filename of the full dictionary (.dic)
             lmd_persona -- filename of the 'Persona' language model (containing, e.g., 'Jasper')
             dictd_persona -- filename of the 'Persona' dictionary (.dic)
         """
-
+        self.speaker = speaker
         hmdir = "/usr/local/share/pocketsphinx/model/hmm/en_US/hub4wsj_sc_8k"
 
         if lmd_music and dictd_music:
@@ -237,7 +238,7 @@ class Mic:
         if THRESHOLD == None:
             THRESHOLD = self.fetchThreshold()
 
-        os.system("aplay -D hw:1,0 ../static/audio/beep_hi.wav")
+        self.speaker.play("../static/audio/beep_hi.wav")
 
         # prepare recording stream
         audio = pyaudio.PyAudio()
@@ -267,7 +268,7 @@ class Mic:
             if average < THRESHOLD * 0.8:
                 break
 
-        os.system("aplay -D hw:1,0 ../static/audio/beep_lo.wav")
+        self.speaker.play("../static/audio/beep_lo.wav")
 
         # save the audio data
         stream.stop_stream()
@@ -291,6 +292,4 @@ class Mic:
     def say(self, phrase, OPTIONS=" -vdefault+m3 -p 40 -s 160 --stdout > say.wav"):
         # alter phrase before speaking
         phrase = alteration.clean(phrase)
-
-        os.system("espeak " + json.dumps(phrase) + OPTIONS)
-        os.system("aplay -D hw:1,0 say.wav")
+        self.speaker.say(phrase)
diff --git a/client/speaker.py b/client/speaker.py
new file mode 100644
index 0000000..1b1255a
--- /dev/null
+++ b/client/speaker.py
@@ -0,0 +1,57 @@
+"""
+A Speaker handles audio output from Jasper to the user
+
+Speaker methods:
+    say - output 'phrase' as speech
+    play - play the audio in 'filename'
+    isAvailable - returns True if the platform supports this implementation
+"""
+import os
+import json
+
+class eSpeakSpeaker:
+    """
+    Uses the eSpeak speech synthesizer included in the Jasper disk image
+    """
+    @classmethod
+    def isAvailable(cls):
+        return os.system("which espeak") == 0
+
+    def say(self, phrase, OPTIONS=" -vdefault+m3 -p 40 -s 160 --stdout > say.wav"):
+        os.system("espeak " + json.dumps(phrase) + OPTIONS)
+        self.play("say.wav")
+
+    def play(self, filename):
+        os.system("aplay -D hw:1,0 " + filename)
+
+class saySpeaker:
+    """
+    Uses the OS X built-in 'say' command
+    """
+
+    @classmethod
+    def isAvailable(cls):
+        return os.system("which say") == 0
+
+    def shellquote(self, s):
+        return "'" + s.replace("'", "'\\''") + "'"
+
+    def say(self, phrase):
+        os.system("say " + self.shellquote(phrase))
+
+    def play(self, filename):
+        os.system("afplay " + filename)
+
+def newSpeaker():
+    """
+    Returns:
+        A speaker implementation available on the current platform
+
+    Raises:
+        ValueError if no speaker implementation is supported on this platform
+    """
+
+    for cls in [eSpeakSpeaker, saySpeaker]:
+        if cls.isAvailable():
+            return cls()
+    raise ValueError("Platform is not supported")
diff --git a/client/start.sh b/client/start.sh
index f3eea8b..cc33f9c 100755
--- a/client/start.sh
+++ b/client/start.sh
@@ -1,3 +1,3 @@
-cd /home/pi/jasper/client/
+cd $JASPER_HOME/jasper/client
 rm -rf ../old_client
 python main.py &
diff --git a/client/test.py b/client/test.py
index 80191ca..fb7d147 100644
--- a/client/test.py
+++ b/client/test.py
@@ -1,3 +1,8 @@
+import os
+
+if os.environ.get('JASPER_HOME') is None:
+    os.environ['JASPER_HOME'] = '/home/pi'
+
 import unittest
 import argparse
 from mock import patch
@@ -6,6 +11,7 @@ import yaml
 import test_mic
 import g2p
 import brain
+import speaker
 
 
 def activeInternet():
@@ -23,7 +29,7 @@ class TestMic(unittest.TestCase):
         self.time_clip = "../static/audio/time.wav"
 
         from mic import Mic
-        self.m = Mic("languagemodel.lm", "dictionary.dic",
+        self.m = Mic(speaker.newSpeaker(), "languagemodel.lm", "dictionary.dic",
                      "languagemodel_persona.lm", "dictionary_persona.dic")
 
     def testTranscribeJasper(self):