Merge pull request #801 from PyThaiNLP/dev

Update 4.0 branch

pythainlp/__init__.py

@@ -17,7 +17,7 @@
 #
 # URL: <https://pythainlp.github.io/>
 # For license information, see LICENSE
-__version__ = "4.0.0"
+__version__ = "4.0.2"
 
 thai_consonants = "กขฃคฅฆงจฉชซฌญฎฏฐฑฒณดตถทธนบปผฝพฟภมยรลวศษสหฬอฮ"  # 44 chars
 

pythainlp/khavee/core.py

@@ -15,8 +15,7 @@
 from typing import List, Union
 from pythainlp.tokenize import subword_tokenize
 from pythainlp.util import sound_syllable
-
-
+from pythainlp.util import remove_tonemark
 class KhaveeVerifier:
     def __init__(self):
         """
@@ -185,10 +184,15 @@ def check_sara(self, word: str)-> str:
             sara.append('ออ') 
 
         # incase บ่ 
-        if 'บ่' in word:
+        if 'บ่' == word:
             sara = []
             sara.append('ออ')
-
+        if 'ํ' in word:
+            sara = []
+            sara.append('อำ')
+        if 'เ' in word and 'ื' in word and 'อ' in word:
+            sara = []
+            sara.append('เอือ')
         if sara == []:  
             return 'Cant find Sara in this word'
         else:
@@ -215,11 +219,8 @@ def check_marttra(self, word: str) -> str:
         """
         if word[-1] == 'ร' and word[-2] in ['ต','ท'] :
             word = word[:-1]
-        if '์' in word[-1]:
-            if 'ิ' in word[-2] or 'ุ' in word[-2]:
-                word = word[:-3]
-            else:
-                word = word[:-2]
+        word = self.handle_karun_sound_silence(word)
+        word = remove_tonemark(word)
         if 'ำ' in word or ('ํ' in word and 'า' in word) or 'ไ' in word or 'ใ' in word:
             return 'กา'
         elif word[-1] in ['า','ะ','ิ','ี','ุ','ู','อ'] or ('ี' in word and 'ย' in word[-1]) or ('ื' in word and 'อ' in word[-1]):
@@ -417,7 +418,6 @@ def check_klon(self, text: str,k_type: int=8) -> Union[List[str], str]:
     def check_aek_too(self, text: Union[List[str], str], dead_syllable_as_aek:bool = False) -> Union[List[bool], List[str], bool, str]:
         """
         Thai tonal word checker
-
         :param Union[List[str], str] text: Thai word or list of Thai words
         :param bool dead_syllable_as_aek: if True, dead syllable will be considered as aek
         :return: the check if the word is aek or too or False(not both) or list of the check if input is list
@@ -453,3 +453,22 @@ def check_aek_too(self, text: Union[List[str], str], dead_syllable_as_aek:bool =
             return 'aek'
         else:
             return False
+
+    def handle_karun_sound_silence(self, word: str) -> str:
+        """
+        Handle sound silence in Thai word using '์' character (Karun)
+        by stripping all the characters before the 'Karun' character that should be silenced
+
+        :param str text: Thai word
+        :return: Thai word with silence word stripped
+        :rtype: str
+        """
+        sound_silenced = True if word.endswith('์') else False
+        if not sound_silenced:
+            return word
+        thai_consonants = "กขฃคฅฆงจฉชซฌญฎฏฐฑฒณดตถทธนบปผฝพฟภมยรลวศษสหฬอฮ"
+        locate_silenced = word.rfind('์') - 1
+        can_silence_two = True if word[locate_silenced-2] in thai_consonants else False
+        cut_off = 2 if can_silence_two else 1
+        word = word[:locate_silenced + 1 - cut_off]
+        return word

pythainlp/khavee/example.py

@@ -16,12 +16,13 @@
 # True
 
 # การตรวจสอบคำสำผัสที่ผิด
-print('สรร ขวาน',kv.is_sumpus('สรร','ขวาน'))
+print('เพื่อน ล้วน',kv.is_sumpus('เพื่อน','ล้วน'))
 # False
 
 # การตรวจสอบคำ ครุ ลหุ 
 print('สรร',kv.check_karu_lahu('สรร'))
 #karu
+
 # การตรวจสอบคำ ครุ ลหุ 
 print('ชิชะ',kv.check_karu_lahu('ชิชะ'))
 # lahu
@@ -66,3 +67,5 @@
 # -> False, aek, too
 print(kv.check_aek_too(['เอง', 'เอ่ง', 'เอ้ง'])) # ใช้ List ได้เหมือนกัน
 # -> [False, 'aek', 'too']
+print(kv.check_aek_too(['ห๊ะ', 'เอ่ง', 'เอ้ง'], dead_syllable_as_aek=True)) # ใช้ List ได้เหมือนกัน และสามารถตั้งค่า ให้นับคำที่เสียงตายเป็นเอกได้ ตามการเช็คคฉันทลักษณ์กลอน
+# -> ['aek', 'aek', 'too']

setup.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 4.0.0
+current_version = 4.0.1
 commit = True
 tag = True
 parse = (?P<major>\d+)\.(?P<minor>\d+)\.(?P<patch>\d+)(\-(?P<release>[a-z]+)(?P<build>\d+))?

setup.py

@@ -143,7 +143,7 @@
 
 setup(
     name="pythainlp",
-    version="4.0.0",
+    version="4.0.2",
     description="Thai Natural Language Processing library",
     long_description=readme,
     long_description_content_type="text/markdown",