Add wunsen

docker_requirements.txt

@@ -30,3 +30,5 @@ OSKut==1.3
 nlpo3==1.2.2
 thai-nner==0.3
 spacy==2.3.*
+wunsen==0.0.1
+khanaa==0.0.6

docs/api/transliterate.rst

@@ -11,6 +11,7 @@ Modules
 .. autofunction:: transliterate
 .. autofunction:: pronunciate
 .. autofunction:: puan
+.. automodule:: pythainlp.transliterate.wunsen.WunsenTransliterate
 
 Romanize Engines
 ----------------

pythainlp/transliterate/wunsen.py

@@ -0,0 +1,81 @@
+# -*- coding: utf-8 -*-
+"""
+Transliterating Japanese/Korean/Vietnamese romanization text to Thai text
+By Wunsen
+
+:See Also:
+    * `GitHub \
+        <https://github.com/cakimpei/wunsen>`_
+"""
+from wunsen import ThapSap
+
+
+class WunsenTransliterate:
+    """
+    Transliterating Japanese/Korean/Vietnamese romanization text to Thai text
+    by Wunsen
+
+    :See Also:
+        * `GitHub \
+            <https://github.com/cakimpei/wunsen>`_
+    """
+    def __init__(self) -> None:
+        self.thap_value = None
+        self.lang = None
+        self.jp_input = None
+
+    def transliterate(self, text: str, lang: str, jp_input: str = None):
+        """
+        Use Wunsen for transliteration
+
+        :param str text: text wants transliterated to Thai text.
+        :param str lang: source language
+        :param str jp_input: japanese input method (for japanese only)
+
+        :return: Thai text
+        :rtype: str
+
+        :Options for lang:
+            * *jp* - Japanese (from Hepburn romanization)
+            * *ko* - Korean (from Revised Romanization)
+            * *vi* - Vietnamese (Latin script)
+        :Options for jp_input:
+            * *Hepburn-no diacritic* - Hepburn-no diacritic (without macron)
+
+        :Example:
+        ::
+            from pythainlp.transliterate.wunsen import WunsenTransliterate
+
+            wt = WunsenTransliterate()
+
+            wt.transliterate("ohayō", lang="jp")
+            # output: 'โอฮาโย'
+
+            wt.transliterate(
+                "ohayou",
+                lang="jp",
+                jp_input="Hepburn-no diacritic"
+            )
+            # output: 'โอฮาโย'
+
+            wt.transliterate("annyeonghaseyo", lang="ko")
+            # output: 'อันนย็องฮาเซโย'
+
+            wt.transliterate("xin chào", lang="vi")
+            # output: 'ซีน จ่าว'
+        """
+        if self.lang != lang or self.jp_input != jp_input:
+            if lang == "jp":
+                if jp_input is None:
+                    self.thap_value = ThapSap("ja")
+                else:
+                    self.thap_value = ThapSap("ja", input=jp_input)
+                self.jp_input = jp_input
+            elif lang == "ko" or lang == "vi":
+                self.jp_input = None
+                self.thap_value = ThapSap(lang)
+            else:
+                raise NotImplementedError(
+                    "The %s language is not implemented." % lang
+                )
+        return self.thap_value.thap(text)

setup.py

@@ -55,6 +55,7 @@
         "torch>=1.0.0",
         "transformers>=4.6.0",
     ],
+    "wunsen": ["wunsen>=0.0.1"],
     "textaugment": [
         "bpemb",
         "gensim>=4.0.0"
@@ -107,7 +108,8 @@
         "oskut>=1.3",
         "nlpo3>=1.2.2",
         "onnxruntime>=1.10.0",
-        "thai_nner"
+        "thai_nner",
+        "wunsen>=0.0.1"
     ],
 }
 

tests/test_transliterate.py

@@ -6,6 +6,7 @@
 from pythainlp.transliterate import romanize, transliterate, pronunciate, puan
 from pythainlp.transliterate.ipa import trans_list, xsampa_list
 from pythainlp.transliterate.thai2rom import ThaiTransliterator
+from pythainlp.transliterate.wunsen import WunsenTransliterate
 from pythainlp.corpus import remove
 
 _BASIC_TESTS = {
@@ -156,6 +157,31 @@ def test_transliterate_iso11940(self):
             "p̣hās̛̄āịthy"
         )
 
+    def test_transliterate_wunsen(self):
+        wt = WunsenTransliterate()
+        self.assertEqual(
+            wt.transliterate("ohayō", lang="jp"),
+            'โอฮาโย'
+        )
+        self.assertEqual(
+            wt.transliterate(
+                "ohayou",
+                lang="jp",
+                jp_input="Hepburn-no diacritic"
+            ),
+            'โอฮาโย'
+        )
+        self.assertEqual(
+            wt.transliterate("annyeonghaseyo", lang="ko"),
+            'อันนย็องฮาเซโย'
+        )
+        self.assertEqual(
+            wt.transliterate("xin chào", lang="vi"),
+            'ซีน จ่าว'
+        )
+        with self.assertRaises(NotImplementedError):
+            wt.transliterate("xin chào", lang="vii")
+
     def test_pronunciate(self):
         self.assertEqual(pronunciate(""), "")
         remove("thai_w2p")