Merge pull request #139 from PyThaiNLP/g2p

Add pythainlp.g2p.ipa

Author: wannaphong

pythainlp/corpus/wordnet.py

@@ -3,7 +3,6 @@
 WordNet
 """
 import nltk
-from nltk.corpus import wordnet
 
 try:
     nltk.data.find("corpora/omw")
@@ -15,6 +14,7 @@
 except LookupError:
     nltk.download("wordnet")
 
+from nltk.corpus import wordnet
 
 def synsets(word, pos=None, lang="tha"):
     return wordnet.synsets(lemma=word, pos=pos, lang=lang)

pythainlp/g2p/__init__.py

@@ -0,0 +1,21 @@
+# -*- coding: utf-8 -*-
+try:
+    import epitran
+except ImportError:
+    from pythainlp.tools import install_package
+
+    install_package("epitran")
+    try:
+        import epitran
+    except ImportError:
+        raise ImportError("ImportError: Try 'pip install epitran'")
+epi = epitran.Epitran('tha-Thai')
+class ipa:
+	def __init__(self,text=""):
+		self.text=text
+	def str(self):
+		return epi.transliterate(self.text)
+	def list(self):
+		return epi.trans_list(self.text)
+	def xsampa_list(self):
+		return epi.xsampa_list(self.text)

tests/__init__.py

@@ -30,9 +30,9 @@
 from pythainlp.tag import pos_tag, pos_tag_sents
 from pythainlp.tokenize import etcc, isthai, syllable_tokenize, tcc, word_tokenize
 from pythainlp.util import listtext_num2num, normalize
+from pythainlp.g2p import ipa
 from pythainlp.text import Text
 
-
 class TestUM(unittest.TestCase):
     """
     ทดสอบการทำงาน
@@ -261,6 +261,12 @@ def test_ner(self):
     def test_Text(self):
         self.assertIsNotNone(Text("ทดสอบภาษาไทย"))
 
+    def test_ipa(self):
+        t=ipa("คน")
+        self.assertEqual(t.str(),'kʰon')
+        self.assertIsNotNone(t.list())
+        self.assertIsNotNone(t.xsampa_list())
+
 
 if __name__ == "__main__":
     unittest.main()