diff options
| -rw-r--r-- | README.md | 3 | ||||
| -rw-r--r-- | config.py | 90 | ||||
| -rw-r--r-- | gtkui.py | 113 | ||||
| -rw-r--r-- | hasher.py | 37 | ||||
| -rwxr-xr-x | kaylee.py | 206 | ||||
| -rw-r--r-- | kayleevc/__init__.py | 0 | ||||
| -rw-r--r-- | kayleevc/gui.py (renamed from gtktrayui.py) | 104 | ||||
| -rw-r--r-- | kayleevc/kaylee.py | 207 | ||||
| -rw-r--r-- | kayleevc/numbers.py (renamed from numberparser.py) | 3 | ||||
| -rw-r--r-- | kayleevc/recognizer.py (renamed from recognizer.py) | 0 | ||||
| -rw-r--r-- | kayleevc/util.py | 202 | ||||
| -rw-r--r-- | languageupdater.py | 80 |
12 files changed, 517 insertions, 528 deletions
diff --git a/README.md b/README.md index 7d59f01..5715733 100644 --- a/README.md +++ b/README.md @@ -7,7 +7,8 @@ but adds a lot of features that go beyond the original purpose of Blather. ## Requirements -1. pocketsphinx +1. Python 3 (tested with 3.5, may work with older versions) +1. pocketsphinx 5prealpha 2. gstreamer-1.0 (and what ever plugin has pocketsphinx support) 3. gstreamer-1.0 base plugins (required for ALSA) 4. python-gobject (required for GStreamer and the GTK-based UI) diff --git a/config.py b/config.py deleted file mode 100644 index 30f8df0..0000000 --- a/config.py +++ /dev/null @@ -1,90 +0,0 @@ -# This is part of Kaylee -# -- this code is licensed GPLv3 -# Copyright 2015-2016 Clayton G. Hobbs -# Portions Copyright 2013 Jezra - -import json -import os -from argparse import ArgumentParser, Namespace - -from gi.repository import GLib - -class Config: - """Keep track of the configuration of Kaylee""" - # Name of the program, for later use - program_name = "kaylee" - - # Directories - conf_dir = os.path.join(GLib.get_user_config_dir(), program_name) - cache_dir = os.path.join(GLib.get_user_cache_dir(), program_name) - data_dir = os.path.join(GLib.get_user_data_dir(), program_name) - - # Configuration files - command_file = os.path.join(conf_dir, "commands.conf") - opt_file = os.path.join(conf_dir, "options.json") - - # Cache files - history_file = os.path.join(cache_dir, program_name + "history") - hash_file = os.path.join(cache_dir, "hash.json") - - # Data files - strings_file = os.path.join(data_dir, "sentences.corpus") - lang_file = os.path.join(data_dir, 'lm') - dic_file = os.path.join(data_dir, 'dic') - - def __init__(self): - # Ensure necessary directories exist - self._make_dir(self.conf_dir) - self._make_dir(self.cache_dir) - self._make_dir(self.data_dir) - - # Set up the argument parser - self.parser = ArgumentParser() - self.parser.add_argument("-i", "--interface", type=str, - dest="interface", action='store', - help="Interface to use (if any). 'g' for GTK or 'gt' for GTK" + - " system tray icon") - - self.parser.add_argument("-c", "--continuous", - action="store_true", dest="continuous", default=False, - help="Start interface with 'continuous' listen enabled") - - self.parser.add_argument("-p", "--pass-words", - action="store_true", dest="pass_words", default=False, - help="Pass the recognized words as arguments to the shell" + - " command") - - self.parser.add_argument("-H", "--history", type=int, - action="store", dest="history", - help="Number of commands to store in history file") - - self.parser.add_argument("-m", "--microphone", type=int, - action="store", dest="microphone", default=None, - help="Audio input card to use (if other than system default)") - - self.parser.add_argument("--valid-sentence-command", type=str, - dest="valid_sentence_command", action='store', - help="Command to run when a valid sentence is detected") - - self.parser.add_argument("--invalid-sentence-command", type=str, - dest="invalid_sentence_command", action='store', - help="Command to run when an invalid sentence is detected") - - # Read the configuration file - self._read_options_file() - - # Parse command-line arguments, overriding config file as appropriate - self.parser.parse_args(namespace=self.options) - - def _make_dir(self, directory): - if not os.path.exists(directory): - os.makedirs(directory) - - def _read_options_file(self): - try: - with open(self.opt_file, 'r') as f: - self.options = json.load(f) - self.options = Namespace(**self.options) - except FileNotFoundError: - # Make an empty options namespace - self.options = Namespace() diff --git a/gtkui.py b/gtkui.py deleted file mode 100644 index 1268ccf..0000000 --- a/gtkui.py +++ /dev/null @@ -1,113 +0,0 @@ -# This is part of Kaylee -# -- this code is licensed GPLv3 -# Copyright 2015-2016 Clayton G. Hobbs -# Portions Copyright 2013 Jezra - -import sys -import gi -from gi.repository import GObject -# Gtk -gi.require_version('Gtk', '3.0') -from gi.repository import Gtk, Gdk - -class UI(GObject.GObject): - __gsignals__ = { - 'command' : (GObject.SIGNAL_RUN_LAST, GObject.TYPE_NONE, (GObject.TYPE_STRING,)) - } - - def __init__(self, args, continuous): - GObject.GObject.__init__(self) - self.continuous = continuous - # Make a window - self.window = Gtk.Window(Gtk.WindowType.TOPLEVEL) - self.window.connect("delete_event", self.delete_event) - # Give the window a name - self.window.set_title("Kaylee") - self.window.set_resizable(False) - - layout = Gtk.VBox() - self.window.add(layout) - # Make a listen/stop button - self.lsbutton = Gtk.Button("Listen") - layout.add(self.lsbutton) - # Make a continuous button - self.ccheckbox = Gtk.CheckButton("Continuous Listen") - layout.add(self.ccheckbox) - - # Connect the buttons - self.lsbutton.connect("clicked", self.lsbutton_clicked) - self.ccheckbox.connect("clicked", self.ccheckbox_clicked) - - # Add a label to the UI to display the last command - self.label = Gtk.Label() - layout.add(self.label) - - # Create an accellerator group for this window - accel = Gtk.AccelGroup() - # Add the ctrl+q to quit - accel.connect(Gdk.keyval_from_name('q'), Gdk.ModifierType.CONTROL_MASK, - Gtk.AccelFlags.VISIBLE, self.accel_quit) - # Lock the group - accel.lock() - # Add the group to the window - self.window.add_accel_group(accel) - - def ccheckbox_clicked(self, widget): - checked = self.ccheckbox.get_active() - self.lsbutton.set_sensitive(not checked) - if checked: - self.lsbutton_stopped() - self.emit('command', "continuous_listen") - self.set_icon_active() - else: - self.emit('command', "continuous_stop") - self.set_icon_inactive() - - def lsbutton_stopped(self): - self.lsbutton.set_label("Listen") - - def lsbutton_clicked(self, button): - val = self.lsbutton.get_label() - if val == "Listen": - self.emit("command", "listen") - self.lsbutton.set_label("Stop") - # Clear the label - self.label.set_text("") - self.set_icon_active() - else: - self.lsbutton_stopped() - self.emit("command", "stop") - self.set_icon_inactive() - - def run(self): - # Set the default icon - self.set_icon_inactive() - self.window.show_all() - if self.continuous: - self.set_icon_active() - self.ccheckbox.set_active(True) - - def accel_quit(self, accel_group, acceleratable, keyval, modifier): - self.emit("command", "quit") - - def delete_event(self, x, y): - self.emit("command", "quit") - - def finished(self, text): - # If the continuous isn't pressed - if not self.ccheckbox.get_active(): - self.lsbutton_stopped() - self.set_icon_inactive() - self.label.set_text(text) - - def set_icon_active_asset(self, i): - self.icon_active = i - - def set_icon_inactive_asset(self, i): - self.icon_inactive = i - - def set_icon_active(self): - Gtk.Window.set_default_icon_from_file(self.icon_active) - - def set_icon_inactive(self): - Gtk.Window.set_default_icon_from_file(self.icon_inactive) diff --git a/hasher.py b/hasher.py deleted file mode 100644 index 4aebd51..0000000 --- a/hasher.py +++ /dev/null @@ -1,37 +0,0 @@ -# This is part of Kaylee -# -- this code is licensed GPLv3 -# Copyright 2015-2016 Clayton G. Hobbs -# Portions Copyright 2013 Jezra - -import json -import hashlib - -class Hasher: - """Keep track of hashes for Kaylee""" - - def __init__(self, config): - self.config = config - try: - with open(self.config.hash_file, 'r') as f: - self.hashes = json.load(f) - except IOError: - # No stored hash - self.hashes = {} - - def __getitem__(self, hashname): - try: - return self.hashes[hashname] - except (KeyError, TypeError): - return None - - def __setitem__(self, hashname, value): - self.hashes[hashname] = value - - def get_hash_object(self): - """Returns an object to compute a new hash""" - return hashlib.sha256() - - def store(self): - """Store the current hashes into a the hash file""" - with open(self.config.hash_file, 'w') as f: - json.dump(self.hashes, f) diff --git a/kaylee.py b/kaylee.py index 298a2c6..082d6b6 100755 --- a/kaylee.py +++ b/kaylee.py @@ -1,212 +1,10 @@ #!/usr/bin/env python3 - # This is part of Kaylee # -- this code is licensed GPLv3 # Copyright 2015-2016 Clayton G. Hobbs # Portions Copyright 2013 Jezra -import sys -import signal -import os.path -import subprocess -from gi.repository import GObject, GLib - -from recognizer import Recognizer -from config import Config -from languageupdater import LanguageUpdater -from numberparser import NumberParser -from hasher import Hasher - - -class Kaylee: - - def __init__(self): - self.ui = None - self.options = {} - ui_continuous_listen = False - self.continuous_listen = False - - # Load configuration - self.config = Config() - self.options = vars(self.config.options) - self.commands = self.options['commands'] - - # Create number parser for later use - self.number_parser = NumberParser() - - # Create a hasher - self.hasher = Hasher(self.config) - - # Create the strings file - self.update_strings_file_if_changed() - - if self.options['interface']: - if self.options['interface'] == "g": - from gtkui import UI - elif self.options['interface'] == "gt": - from gtktrayui import UI - else: - print("no GUI defined") - sys.exit() - - self.ui = UI(self.options, self.options['continuous']) - self.ui.connect("command", self.process_command) - # Can we load the icon resource? - icon = self.load_resource("icon.png") - if icon: - self.ui.set_icon_active_asset(icon) - # Can we load the icon_inactive resource? - icon_inactive = self.load_resource("icon_inactive.png") - if icon_inactive: - self.ui.set_icon_inactive_asset(icon_inactive) - - if self.options['history']: - self.history = [] - - # Update the language if necessary - self.language_updater = LanguageUpdater(self.config) - self.language_updater.update_language_if_changed() - - # Create the recognizer - self.recognizer = Recognizer(self.config) - self.recognizer.connect('finished', self.recognizer_finished) +from kayleevc import kaylee - def update_voice_commands_if_changed(self): - """Use hashes to test if the voice commands have changed""" - stored_hash = self.hasher['voice_commands'] - - # Calculate the hash the voice commands have right now - hasher = self.hasher.get_hash_object() - for voice_cmd in self.commands.keys(): - hasher.update(voice_cmd.encode('utf-8')) - # Add a separator to avoid odd behavior - hasher.update('\n'.encode('utf-8')) - new_hash = hasher.hexdigest() - - if new_hash != stored_hash: - self.create_strings_file() - self.hasher['voice_commands'] = new_hash - self.hasher.store() - - def create_strings_file(self): - # Open the strings file - with open(self.config.strings_file, 'w') as strings: - # Add command words to the corpus - for voice_cmd in sorted(self.commands.keys()): - strings.write(voice_cmd.strip().replace('%d', '') + "\n") - # Add number words to the corpus - for word in self.number_parser.number_words: - strings.write(word + "\n") - - def log_history(self, text): - if self.options['history']: - self.history.append(text) - if len(self.history) > self.options['history']: - # Pop off the first item - self.history.pop(0) - - # Open and truncate the history file - with open(self.config.history_file, 'w') as hfile: - for line in self.history: - hfile.write(line + '\n') - - def run_command(self, cmd): - """Print the command, then run it""" - print(cmd) - subprocess.call(cmd, shell=True) - - def recognizer_finished(self, recognizer, text): - t = text.lower() - numt, nums = self.number_parser.parse_all_numbers(t) - # Is there a matching command? - if t in self.commands: - # Run the valid_sentence_command if it's set - if self.options['valid_sentence_command']: - subprocess.call(self.options['valid_sentence_command'], - shell=True) - cmd = self.commands[t] - # Should we be passing words? - if self.options['pass_words']: - cmd += " " + t - self.run_command(cmd) - self.log_history(text) - elif numt in self.commands: - # Run the valid_sentence_command if it's set - if self.options['valid_sentence_command']: - subprocess.call(self.options['valid_sentence_command'], - shell=True) - cmd = self.commands[numt] - cmd = cmd.format(*nums) - # Should we be passing words? - if self.options['pass_words']: - cmd += " " + t - self.run_command(cmd) - self.log_history(text) - else: - # Run the invalid_sentence_command if it's set - if self.options['invalid_sentence_command']: - subprocess.call(self.options['invalid_sentence_command'], - shell=True) - print("no matching command {0}".format(t)) - # If there is a UI and we are not continuous listen - if self.ui: - if not self.continuous_listen: - # Stop listening - self.recognizer.pause() - # Let the UI know that there is a finish - self.ui.finished(t) - - def run(self): - if self.ui: - self.ui.run() - else: - self.recognizer.listen() - - def quit(self): - sys.exit() - - def process_command(self, UI, command): - print(command) - if command == "listen": - self.recognizer.listen() - elif command == "stop": - self.recognizer.pause() - elif command == "continuous_listen": - self.continuous_listen = True - self.recognizer.listen() - elif command == "continuous_stop": - self.continuous_listen = False - self.recognizer.pause() - elif command == "quit": - self.quit() - - def load_resource(self, string): - local_data = os.path.join(os.path.dirname(__file__), 'data') - paths = ["/usr/share/kaylee/", "/usr/local/share/kaylee", local_data] - for path in paths: - resource = os.path.join(path, string) - if os.path.exists(resource): - return resource - # If we get this far, no resource was found - return False - - -if __name__ == "__main__": - # Make our kaylee object - kaylee = Kaylee() - # Init gobject threads - GObject.threads_init() - # We want a main loop - main_loop = GObject.MainLoop() - # Handle sigint - signal.signal(signal.SIGINT, signal.SIG_DFL) - # Run the kaylee +if __name__ == '__main__': kaylee.run() - # Start the main loop - try: - main_loop.run() - except: - print("time to quit") - main_loop.quit() - sys.exit() - diff --git a/kayleevc/__init__.py b/kayleevc/__init__.py new file mode 100644 index 0000000..e69de29 --- /dev/null +++ b/kayleevc/__init__.py diff --git a/gtktrayui.py b/kayleevc/gui.py index f595f26..3369e33 100644 --- a/gtktrayui.py +++ b/kayleevc/gui.py @@ -10,7 +10,7 @@ from gi.repository import GObject gi.require_version('Gtk', '3.0') from gi.repository import Gtk, Gdk -class UI(GObject.GObject): +class GTKTrayInterface(GObject.GObject): __gsignals__ = { 'command' : (GObject.SIGNAL_RUN_LAST, GObject.TYPE_NONE, (GObject.TYPE_STRING,)) } @@ -105,3 +105,105 @@ class UI(GObject.GObject): def set_icon_inactive(self): self.statusicon.set_from_file(self.icon_inactive) + +class GTKInterface(GObject.GObject): + __gsignals__ = { + 'command' : (GObject.SIGNAL_RUN_LAST, GObject.TYPE_NONE, (GObject.TYPE_STRING,)) + } + + def __init__(self, args, continuous): + GObject.GObject.__init__(self) + self.continuous = continuous + # Make a window + self.window = Gtk.Window(Gtk.WindowType.TOPLEVEL) + self.window.connect("delete_event", self.delete_event) + # Give the window a name + self.window.set_title("Kaylee") + self.window.set_resizable(False) + + layout = Gtk.VBox() + self.window.add(layout) + # Make a listen/stop button + self.lsbutton = Gtk.Button("Listen") + layout.add(self.lsbutton) + # Make a continuous button + self.ccheckbox = Gtk.CheckButton("Continuous Listen") + layout.add(self.ccheckbox) + + # Connect the buttons + self.lsbutton.connect("clicked", self.lsbutton_clicked) + self.ccheckbox.connect("clicked", self.ccheckbox_clicked) + + # Add a label to the UI to display the last command + self.label = Gtk.Label() + layout.add(self.label) + + # Create an accellerator group for this window + accel = Gtk.AccelGroup() + # Add the ctrl+q to quit + accel.connect(Gdk.keyval_from_name('q'), Gdk.ModifierType.CONTROL_MASK, + Gtk.AccelFlags.VISIBLE, self.accel_quit) + # Lock the group + accel.lock() + # Add the group to the window + self.window.add_accel_group(accel) + + def ccheckbox_clicked(self, widget): + checked = self.ccheckbox.get_active() + self.lsbutton.set_sensitive(not checked) + if checked: + self.lsbutton_stopped() + self.emit('command', "continuous_listen") + self.set_icon_active() + else: + self.emit('command', "continuous_stop") + self.set_icon_inactive() + + def lsbutton_stopped(self): + self.lsbutton.set_label("Listen") + + def lsbutton_clicked(self, button): + val = self.lsbutton.get_label() + if val == "Listen": + self.emit("command", "listen") + self.lsbutton.set_label("Stop") + # Clear the label + self.label.set_text("") + self.set_icon_active() + else: + self.lsbutton_stopped() + self.emit("command", "stop") + self.set_icon_inactive() + + def run(self): + # Set the default icon + self.set_icon_inactive() + self.window.show_all() + if self.continuous: + self.set_icon_active() + self.ccheckbox.set_active(True) + + def accel_quit(self, accel_group, acceleratable, keyval, modifier): + self.emit("command", "quit") + + def delete_event(self, x, y): + self.emit("command", "quit") + + def finished(self, text): + # If the continuous isn't pressed + if not self.ccheckbox.get_active(): + self.lsbutton_stopped() + self.set_icon_inactive() + self.label.set_text(text) + + def set_icon_active_asset(self, i): + self.icon_active = i + + def set_icon_inactive_asset(self, i): + self.icon_inactive = i + + def set_icon_active(self): + Gtk.Window.set_default_icon_from_file(self.icon_active) + + def set_icon_inactive(self): + Gtk.Window.set_default_icon_from_file(self.icon_inactive) diff --git a/kayleevc/kaylee.py b/kayleevc/kaylee.py new file mode 100644 index 0000000..a99f9d4 --- /dev/null +++ b/kayleevc/kaylee.py @@ -0,0 +1,207 @@ +# This is part of Kaylee +# -- this code is licensed GPLv3 +# Copyright 2015-2016 Clayton G. Hobbs +# Portions Copyright 2013 Jezra + +import sys +import signal +import os.path +import subprocess +from gi.repository import GObject, GLib + +from kayleevc.recognizer import Recognizer +from kayleevc.util import * +from kayleevc.numbers import NumberParser + + +class Kaylee: + + def __init__(self): + self.ui = None + self.options = {} + ui_continuous_listen = False + self.continuous_listen = False + + # Load configuration + self.config = Config() + self.options = vars(self.config.options) + self.commands = self.options['commands'] + + # Create number parser for later use + self.number_parser = NumberParser() + + # Create a hasher + self.hasher = Hasher(self.config) + + # Create the strings file + self.update_voice_commands_if_changed() + + if self.options['interface']: + if self.options['interface'] == "g": + from kayleevc.gui import GTKInterface as UI + elif self.options['interface'] == "gt": + from kayleevc.gui import GTKTrayInterface as UI + else: + print("no GUI defined") + sys.exit() + + self.ui = UI(self.options, self.options['continuous']) + self.ui.connect("command", self.process_command) + # Can we load the icon resource? + icon = self.load_resource("icon.png") + if icon: + self.ui.set_icon_active_asset(icon) + # Can we load the icon_inactive resource? + icon_inactive = self.load_resource("icon_inactive.png") + if icon_inactive: + self.ui.set_icon_inactive_asset(icon_inactive) + + if self.options['history']: + self.history = [] + + # Update the language if necessary + self.language_updater = LanguageUpdater(self.config) + self.language_updater.update_language_if_changed() + + # Create the recognizer + self.recognizer = Recognizer(self.config) + self.recognizer.connect('finished', self.recognizer_finished) + + def update_voice_commands_if_changed(self): + """Use hashes to test if the voice commands have changed""" + stored_hash = self.hasher['voice_commands'] + + # Calculate the hash the voice commands have right now + hasher = self.hasher.get_hash_object() + for voice_cmd in self.commands.keys(): + hasher.update(voice_cmd.encode('utf-8')) + # Add a separator to avoid odd behavior + hasher.update('\n'.encode('utf-8')) + new_hash = hasher.hexdigest() + + if new_hash != stored_hash: + self.create_strings_file() + self.hasher['voice_commands'] = new_hash + self.hasher.store() + + def create_strings_file(self): + # Open the strings file + with open(self.config.strings_file, 'w') as strings: + # Add command words to the corpus + for voice_cmd in sorted(self.commands.keys()): + strings.write(voice_cmd.strip().replace('%d', '') + "\n") + # Add number words to the corpus + for word in self.number_parser.number_words: + strings.write(word + "\n") + + def log_history(self, text): + if self.options['history']: + self.history.append(text) + if len(self.history) > self.options['history']: + # Pop off the first item + self.history.pop(0) + + # Open and truncate the history file + with open(self.config.history_file, 'w') as hfile: + for line in self.history: + hfile.write(line + '\n') + + def run_command(self, cmd): + """Print the command, then run it""" + print(cmd) + subprocess.call(cmd, shell=True) + + def recognizer_finished(self, recognizer, text): + t = text.lower() + numt, nums = self.number_parser.parse_all_numbers(t) + # Is there a matching command? + if t in self.commands: + # Run the valid_sentence_command if it's set + if self.options['valid_sentence_command']: + subprocess.call(self.options['valid_sentence_command'], + shell=True) + cmd = self.commands[t] + # Should we be passing words? + if self.options['pass_words']: + cmd += " " + t + self.run_command(cmd) + self.log_history(text) + elif numt in self.commands: + # Run the valid_sentence_command if it's set + if self.options['valid_sentence_command']: + subprocess.call(self.options['valid_sentence_command'], + shell=True) + cmd = self.commands[numt] + cmd = cmd.format(*nums) + # Should we be passing words? + if self.options['pass_words']: + cmd += " " + t + self.run_command(cmd) + self.log_history(text) + else: + # Run the invalid_sentence_command if it's set + if self.options['invalid_sentence_command']: + subprocess.call(self.options['invalid_sentence_command'], + shell=True) + print("no matching command {0}".format(t)) + # If there is a UI and we are not continuous listen + if self.ui: + if not self.continuous_listen: + # Stop listening + self.recognizer.pause() + # Let the UI know that there is a finish + self.ui.finished(t) + + def run(self): + if self.ui: + self.ui.run() + else: + self.recognizer.listen() + + def quit(self): + sys.exit() + + def process_command(self, UI, command): + print(command) + if command == "listen": + self.recognizer.listen() + elif command == "stop": + self.recognizer.pause() + elif command == "continuous_listen": + self.continuous_listen = True + self.recognizer.listen() + elif command == "continuous_stop": + self.continuous_listen = False + self.recognizer.pause() + elif command == "quit": + self.quit() + + def load_resource(self, string): + local_data = os.path.join(os.path.dirname(__file__), '..', 'data') + paths = ["/usr/share/kaylee/", "/usr/local/share/kaylee", local_data] + for path in paths: + resource = os.path.join(path, string) + if os.path.exists(resource): + return resource + # If we get this far, no resource was found + return False + + +def run(): + # Make our kaylee object + kaylee = Kaylee() + # Init gobject threads + GObject.threads_init() + # We want a main loop + main_loop = GObject.MainLoop() + # Handle sigint + signal.signal(signal.SIGINT, signal.SIG_DFL) + # Run the kaylee + kaylee.run() + # Start the main loop + try: + main_loop.run() + except: + print("time to quit") + main_loop.quit() + sys.exit() diff --git a/numberparser.py b/kayleevc/numbers.py index fec07f2..6d41b63 100644 --- a/numberparser.py +++ b/kayleevc/numbers.py @@ -1,10 +1,9 @@ #!/usr/bin/env python -# numberparser.py - Translate words to decimal - # This is part of Kaylee # -- this code is licensed GPLv3 # Copyright 2015-2016 Clayton G. Hobbs # Portions Copyright 2013 Jezra + import re # Define the mappings from words to numbers diff --git a/recognizer.py b/kayleevc/recognizer.py index b54c055..b54c055 100644 --- a/recognizer.py +++ b/kayleevc/recognizer.py diff --git a/kayleevc/util.py b/kayleevc/util.py new file mode 100644 index 0000000..5c93b7f --- /dev/null +++ b/kayleevc/util.py @@ -0,0 +1,202 @@ +# This is part of Kaylee +# -- this code is licensed GPLv3 +# Copyright 2015-2016 Clayton G. Hobbs +# Portions Copyright 2013 Jezra + +import re +import json +import hashlib +import os +from argparse import ArgumentParser, Namespace + +import requests + +from gi.repository import GLib + +class Config: + """Keep track of the configuration of Kaylee""" + # Name of the program, for later use + program_name = "kaylee" + + # Directories + conf_dir = os.path.join(GLib.get_user_config_dir(), program_name) + cache_dir = os.path.join(GLib.get_user_cache_dir(), program_name) + data_dir = os.path.join(GLib.get_user_data_dir(), program_name) + + # Configuration files + command_file = os.path.join(conf_dir, "commands.conf") + opt_file = os.path.join(conf_dir, "options.json") + + # Cache files + history_file = os.path.join(cache_dir, program_name + "history") + hash_file = os.path.join(cache_dir, "hash.json") + + # Data files + strings_file = os.path.join(data_dir, "sentences.corpus") + lang_file = os.path.join(data_dir, 'lm') + dic_file = os.path.join(data_dir, 'dic') + + def __init__(self): + # Ensure necessary directories exist + self._make_dir(self.conf_dir) + self._make_dir(self.cache_dir) + self._make_dir(self.data_dir) + + # Set up the argument parser + self.parser = ArgumentParser() + self.parser.add_argument("-i", "--interface", type=str, + dest="interface", action='store', + help="Interface to use (if any). 'g' for GTK or 'gt' for GTK" + + " system tray icon") + + self.parser.add_argument("-c", "--continuous", + action="store_true", dest="continuous", default=False, + help="Start interface with 'continuous' listen enabled") + + self.parser.add_argument("-p", "--pass-words", + action="store_true", dest="pass_words", default=False, + help="Pass the recognized words as arguments to the shell" + + " command") + + self.parser.add_argument("-H", "--history", type=int, + action="store", dest="history", + help="Number of commands to store in history file") + + self.parser.add_argument("-m", "--microphone", type=int, + action="store", dest="microphone", default=None, + help="Audio input card to use (if other than system default)") + + self.parser.add_argument("--valid-sentence-command", type=str, + dest="valid_sentence_command", action='store', + help="Command to run when a valid sentence is detected") + + self.parser.add_argument("--invalid-sentence-command", type=str, + dest="invalid_sentence_command", action='store', + help="Command to run when an invalid sentence is detected") + + # Read the configuration file + self._read_options_file() + + # Parse command-line arguments, overriding config file as appropriate + self.parser.parse_args(namespace=self.options) + + def _make_dir(self, directory): + if not os.path.exists(directory): + os.makedirs(directory) + + def _read_options_file(self): + try: + with open(self.opt_file, 'r') as f: + self.options = json.load(f) + self.options = Namespace(**self.options) + except FileNotFoundError: + # Make an empty options namespace + self.options = Namespace() + +class Hasher: + """Keep track of hashes for Kaylee""" + + def __init__(self, config): + self.config = config + try: + with open(self.config.hash_file, 'r') as f: + self.hashes = json.load(f) + except IOError: + # No stored hash + self.hashes = {} + + def __getitem__(self, hashname): + try: + return self.hashes[hashname] + except (KeyError, TypeError): + return None + + def __setitem__(self, hashname, value): + self.hashes[hashname] = value + + def get_hash_object(self): + """Returns an object to compute a new hash""" + return hashlib.sha256() + + def store(self): + """Store the current hashes into a the hash file""" + with open(self.config.hash_file, 'w') as f: + json.dump(self.hashes, f) + +class LanguageUpdater: + """ + Handles updating the language using the online lmtool. + + This class provides methods to check if the corpus has changed, and to + update the language to match the new corpus using the lmtool. This allows + us to automatically update the language if the corpus has changed, saving + the user from having to do this manually. + """ + + def __init__(self, config): + self.config = config + self.hasher = Hasher(config) + + def update_language_if_changed(self): + """Test if the language has changed, and if it has, update it""" + if self.language_has_changed(): + self.update_language() + self.save_language_hash() + + def language_has_changed(self): + """Use hashes to test if the language has changed""" + self.stored_hash = self.hasher['language'] + + # Calculate the hash the language file has right now + hasher = self.hasher.get_hash_object() + with open(self.config.strings_file, 'rb') as sfile: + buf = sfile.read() + hasher.update(buf) + self.new_hash = hasher.hexdigest() + + return self.new_hash != self.stored_hash + + def update_language(self): + """Update the language using the online lmtool""" + print('Updating language using online lmtool') + + host = 'http://www.speech.cs.cmu.edu' + url = host + '/cgi-bin/tools/lmtool/run' + + # Submit the corpus to the lmtool + response_text = "" + with open(self.config.strings_file, 'rb') as corpus: + files = {'corpus': corpus} + values = {'formtype': 'simple'} + + r = requests.post(url, files=files, data=values) + response_text = r.text + + # Parse response to get URLs of the files we need + path_re = r'.*<title>Index of (.*?)</title>.*' + number_re = r'.*TAR([0-9]*?)\.tgz.*' + for line in response_text.split('\n'): + # If we found the directory, keep it and don't break + if re.search(path_re, line): + path = host + re.sub(path_re, r'\1', line) + # If we found the number, keep it and break + elif re.search(number_re, line): + number = re.sub(number_re, r'\1', line) + break + + lm_url = path + '/' + number + '.lm' + dic_url = path + '/' + number + '.dic' + + self._download_file(lm_url, self.config.lang_file) + self._download_file(dic_url, self.config.dic_file) + + def save_language_hash(self): + self.hasher['language'] = self.new_hash + self.hasher.store() + + def _download_file(self, url, path): + r = requests.get(url, stream=True) + if r.status_code == 200: + with open(path, 'wb') as f: + for chunk in r: + f.write(chunk) diff --git a/languageupdater.py b/languageupdater.py deleted file mode 100644 index 3f36d06..0000000 --- a/languageupdater.py +++ /dev/null @@ -1,80 +0,0 @@ -# This is part of Kaylee -# -- this code is licensed GPLv3 -# Copyright 2015-2016 Clayton G. Hobbs -# Portions Copyright 2013 Jezra - -import re - -import requests - -from hasher import Hasher - -class LanguageUpdater: - - def __init__(self, config): - self.config = config - self.hasher = Hasher(config) - - def update_language_if_changed(self): - """Test if the language has changed, and if it has, update it""" - if self.language_has_changed(): - self.update_language() - self.save_language_hash() - - def language_has_changed(self): - """Use hashes to test if the language has changed""" - self.stored_hash = self.hasher['language'] - - # Calculate the hash the language file has right now - hasher = self.hasher.get_hash_object() - with open(self.config.strings_file, 'rb') as sfile: - buf = sfile.read() - hasher.update(buf) - self.new_hash = hasher.hexdigest() - - return self.new_hash != self.stored_hash - - def update_language(self): - """Update the language using the online lmtool""" - print('Updating language using online lmtool') - - host = 'http://www.speech.cs.cmu.edu' - url = host + '/cgi-bin/tools/lmtool/run' - - # Submit the corpus to the lmtool - response_text = "" - with open(self.config.strings_file, 'rb') as corpus: - files = {'corpus': corpus} - values = {'formtype': 'simple'} - - r = requests.post(url, files=files, data=values) - response_text = r.text - - # Parse response to get URLs of the files we need - path_re = r'.*<title>Index of (.*?)</title>.*' - number_re = r'.*TAR([0-9]*?)\.tgz.*' - for line in response_text.split('\n'): - # If we found the directory, keep it and don't break - if re.search(path_re, line): - path = host + re.sub(path_re, r'\1', line) - # If we found the number, keep it and break - elif re.search(number_re, line): - number = re.sub(number_re, r'\1', line) - break - - lm_url = path + '/' + number + '.lm' - dic_url = path + '/' + number + '.dic' - - self._download_file(lm_url, self.config.lang_file) - self._download_file(dic_url, self.config.dic_file) - - def save_language_hash(self): - self.hasher['language'] = self.new_hash - self.hasher.store() - - def _download_file(self, url, path): - r = requests.get(url, stream=True) - if r.status_code == 200: - with open(path, 'wb') as f: - for chunk in r: - f.write(chunk) |