Clock goes back to the speaking-mode SL #4079

This PATCH is just a workaround to the real problem:

     Clock is using gst 0.11 but we have installed the 'espeak' plugin
     for Gst 1.0 because we need that version for the shell
     itself. So, Clock does not find the proper plugin and tries to
     use the 'espeak' command directily but it fails as well because a
     'portaudio' lib issue.

This PATCH makes 'espeak' to create a WAV file in a temporary file and
then plays it with the command 'playwave'. It's the same idea used in
Speak, with the difference than Speak uses gstreamer to play the wav
file.

This workaround will not be necessary when Clock uses Gtk3 and Gst
1.0.

Signed-off-by: Manuel Kaufmann <humitos@gmail.com>

1 files changed, 18 insertions, 1 deletions

diff --git a/speaker.py b/speaker.py
index f33dbad..1963747 100755
--- a/speaker.py
+++ b/speaker.py
@@ -14,6 +14,7 @@ Controls the espeak program available on the OLPC XO laptop.
 
 import sys
 import os
+import tempfile
 
 from gettext import gettext as _
 
@@ -51,12 +52,28 @@ class Speaker:
     # http://espeak.sourceforge.net/languages.html to see if your language is supported.
     VOICE = _("en")
 
+    # Temporary file to save the 'espeak -w' output and then play it
+    # with 'playwave' command
+    WAV_TEMP_FILE = tempfile.mktemp()
+
+    # 'espeak' command to generate the wav file to be played with 'playwave'
+    ESPEAK_COMMAND = 'espeak -p%s -s%s -g%s -v%s "%s" -w %s'
+
+    # 'playwave' command that plays the wav generated by 'espeak'
+    PLAYWAVE_COMMAND = 'playwave %s' % WAV_TEMP_FILE
 
     def speak(self, text):
         """Speaks aloud the given text.
         """
         text = text.replace("\"", "\\\"")
-        child = os.popen("espeak -p%s -s%s -g%s -v%s \"%s\"" % (Speaker.PITCH, Speaker.SPEED, Speaker.WORD_GAP, Speaker.VOICE, text))
+        child = os.popen(Speaker.ESPEAK_COMMAND % (
+                Speaker.PITCH, Speaker.SPEED, Speaker.WORD_GAP, Speaker.VOICE,
+                text, Speaker.WAV_TEMP_FILE))
+        data = child.read()
+        child.close()
+
+        # This is a workaround for SL #4079
+        child = os.popen(Speaker.PLAYWAVE_COMMAND)
         data = child.read()
         child.close()