Hyphenation

eng_to_ipa/transcribe.py

@@ -92,12 +92,26 @@ def get_cmu(tokens_in, db_type="sql"):
     """query the SQL database for the words and return the phonemes in the order of user_in"""
     result = fetch_words(tokens_in, db_type)
     ordered = []
+    space = ' '
+
     for word in tokens_in:
         this_word = [[i[1] for i in result if i[0] == word]][0]
         if this_word:
             ordered.append(this_word[0])
         else:
-            ordered.append(["__IGNORE__" + word])
+            if( word.find('-') != -1 ):
+                # we couldn't transliterate a hyphenated word - try word parts
+                tmpresult = get_cmu( word.split('-'), db_type=db_type )
+                if not tmpresult[0][0].startswith('__IGNORE__') and not tmpresult[1][0].startswith('__IGNORE__'):
+                    this_word = []
+                    for this_word1 in tmpresult[0]:
+                        for this_word2 in tmpresult[1]:
+                            this_word.append(this_word1 + space + this_word2)
+                    ordered.append(this_word)
+                else:
+                    ordered.append(["__IGNORE__" + word])
+            else:
+                ordered.append(["__IGNORE__" + word])
     return ordered
 
 

tests/test_transcribe.py

@@ -0,0 +1,17 @@
+# -*- coding: utf-8 -*-
+
+# USAGE:
+# PYTHONPATH=".." python test_transcribe.py 
+
+from eng_to_ipa import transcribe
+import transcribe_fixtures
+import sys
+
+words0 = "on-demand".split()
+
+class TestConversion_default(transcribe_fixtures.BaseConversion):
+    @classmethod
+    def setUpClass(self):
+        self.words0 = words0
+        self.cmu0   = [['aa1 n d ih0 m ae1 n d', 'ao1 n d ih0 m ae1 n d']]
+        self.lang   = None

tests/transcribe_fixtures.py

@@ -0,0 +1,14 @@
+# -*- coding: utf-8 -*-
+
+# USAGE:
+# PYTHONPATH=".." python test_transcribe.py 
+
+from eng_to_ipa import transcribe
+import unittest
+
+class BaseConversion(unittest.TestCase):
+    """Simple unit testing for the transcribe function(s)."""
+
+    def test_get_cmu_on_demand(self):
+        res0 = transcribe.get_cmu(self.words0, db_type='sql')
+        self.assertEqual(res0, self.cmu0)