Merge pull request #2009 from MycroftAI/bugfix/voc-match

Fix behavior of MycroftSkills.voc_match()

mycroft/skills/core.py

@@ -26,6 +26,7 @@ from datetime import datetime, timedelta
 
 import abc
 import re
+from itertools import chain
 from adapt.intent import Intent, IntentBuilder
 from os.path import join, abspath, dirname, basename, exists
 from threading import Event, Timer
@@ -41,7 +42,8 @@ from mycroft.messagebus.message import Message
 from mycroft.metrics import report_metric, report_timing, Stopwatch
 from mycroft.skills.settings import SkillSettings
 from mycroft.skills.skill_data import (load_vocabulary, load_regex, to_alnum,
-                                       munge_regex, munge_intent_parser)
+                                       munge_regex, munge_intent_parser,
+                                       read_vocab_file)
 from mycroft.util import camel_case_split, resolve_resource_file
 from mycroft.util.log import LOG
 
@@ -716,15 +718,15 @@ class MycroftSkill:
             if not voc or not exists(voc):
                 raise FileNotFoundError(
                         'Could not find {}.voc file'.format(voc_filename))
-
-            with open(voc) as f:
-                self.voc_match_cache[cache_key] = f.read().splitlines()
-
-        # Check for match
-        if utt and any(i.strip() in utt
-                       for i in self.voc_match_cache[cache_key]):
-            return True
-        return False
+            # load vocab and flatten into a simple list
+            vocab = list(chain(*read_vocab_file(voc)))
+            self.voc_match_cache[cache_key] = vocab
+        if utt:
+            # Check for matches against complete words
+            return any([re.match(r'.*\b' + i + r'\b.*', utt)
+                        for i in self.voc_match_cache[cache_key]])
+        else:
+            return False
 
     def report_metric(self, name, data):
         """ Report a skill metric to the Mycroft servers

mycroft/skills/skill_data.py

@@ -25,6 +25,28 @@ from mycroft.messagebus.message import Message
 from mycroft.util.format import expand_options
 
 
+def read_vocab_file(path):
+    """ Read voc file.
+
+        This reads a .voc file, stripping out empty lines comments and expand
+        parentheses. It retruns each line as a list of all expanded
+        alternatives.
+
+        Arguments:
+            path (str): path to vocab file.
+
+        Returns:
+            List of Lists of strings.
+    """
+    vocab = []
+    with open(path, 'r', encoding='utf8') as voc_file:
+        for line in voc_file.readlines():
+            if line.startswith('#') or line.strip() == '':
+                continue
+            vocab.append(expand_options(line))
+    return vocab
+
+
 def load_vocab_from_file(path, vocab_type, bus):
     """Load Mycroft vocabulary from file
     The vocab is sent to the intent handler using the message bus
@@ -36,19 +58,15 @@ def load_vocab_from_file(path, vocab_type, bus):
         skill_id(str):  skill id
     """
     if path.endswith('.voc'):
-        with open(path, 'r', encoding='utf8') as voc_file:
-            for line in voc_file.readlines():
-                if line.startswith("#"):
-                    continue
-                parts = expand_options(line)
-                entity = parts[0]
+        for parts in read_vocab_file(path):
+            entity = parts[0]
+            bus.emit(Message("register_vocab", {
+                'start': entity, 'end': vocab_type
+            }))
+            for alias in parts[1:]:
                 bus.emit(Message("register_vocab", {
-                    'start': entity, 'end': vocab_type
+                    'start': alias, 'end': vocab_type, 'alias_of': entity
                 }))
-                for alias in parts[1:]:
-                    bus.emit(Message("register_vocab", {
-                        'start': alias, 'end': vocab_type, 'alias_of': entity
-                    }))
 
 
 def load_regex_from_file(path, bus, skill_id):

test/unittests/skills/locale/en-us/turn_off2_test.voc

@@ -0,0 +1 @@
+(turn off|switch off)

test/unittests/skills/locale/en-us/turn_off_test.voc

@@ -0,0 +1,3 @@
+turn off
+switch off
+

test/unittests/skills/test_core.py

@@ -484,6 +484,28 @@ class MycroftSkillTest(unittest.TestCase):
             # handler
             self.assertTrue('A:sched_handler1' not in [e[0] for e in s.events])
 
+    def test_voc_match(self):
+        s = SimpleSkill1()
+        s.root_dir = abspath(dirname(__file__))
+
+        self.assertTrue(s.voc_match("turn off the lights", "turn_off_test"))
+        self.assertTrue(s.voc_match("would you please turn off the lights",
+                                    "turn_off_test"))
+        self.assertFalse(s.voc_match("return office", "turn_off_test"))
+        self.assertTrue(s.voc_match("switch off the lights", "turn_off_test"))
+        self.assertFalse(s.voc_match("", "turn_off_test"))
+        self.assertFalse(s.voc_match("switch", "turn_off_test"))
+        self.assertFalse(s.voc_match("My hovercraft is full of eels",
+                                     "turn_off_test"))
+
+        self.assertTrue(s.voc_match("turn off the lights", "turn_off2_test"))
+        self.assertFalse(s.voc_match("return office", "turn_off2_test"))
+        self.assertTrue(s.voc_match("switch off the lights", "turn_off2_test"))
+        self.assertFalse(s.voc_match("", "turn_off_test"))
+        self.assertFalse(s.voc_match("switch", "turn_off_test"))
+        self.assertFalse(s.voc_match("My hovercraft is full of eels",
+                                     "turn_off_test"))
+
 
 class _TestSkill(MycroftSkill):
     def __init__(self):