Add thai_text_braille

README.md

@@ -19,6 +19,18 @@ print(thai_word_braille("แมว"))
 
 This function support Thai and number only.
 
+**Thai text to braille**
+
+```python
+from thaibraille import thai_text_braille
+
+print(thai_text_braille("แมวกิน ปลา"))
+
+# output: ['⠣⠍⠺', '⠛⠃⠝', ' ', '⠯⠇⠡']
+```
+
+This function support Thai and number only.
+
 ## License
 
 

setup.py

@@ -9,7 +9,7 @@
 
 setup(
     name="thaibraille",
-    version="0.1dev0",
+    version="0.1dev1",
     description="Thai Braille for Natural Language Processing.",
     long_description=readme,
     long_description_content_type="text/markdown",

tests/test_text.py

@@ -8,3 +8,6 @@ def test_thai_word_braille(self):
         self.assertEqual(thai_word_braille("ลิ้น"), '⠇⠃⠲⠝')
         self.assertEqual(thai_word_braille("ว่าง"), '⠺⠔⠡⠻')
         self.assertEqual(thai_word_braille("แก้ม"), '⠣⠛⠲⠍')
+
+    def test_thai_text_braille(self):
+        self.assertEqual(thai_text_braille("แมวกิน   ปลา"), ['⠣⠍⠺', '⠛⠃⠝', '   ', '⠯⠇⠡'])

thaibraille/text/__init__.py

@@ -1,3 +1,3 @@
 # -*- coding: utf-8 -*-
-__all__ = ["thai_braille_mapping_dict", "Braille", "thai_word_braille"]
-from thaibraille.text.core import thai_braille_mapping_dict, Braille, thai_word_braille
+__all__ = ["thai_braille_mapping_dict", "Braille", "thai_word_braille", "thai_text_braille"]
+from thaibraille.text.core import thai_braille_mapping_dict, Braille, thai_word_braille, thai_text_braille

thaibraille/text/core.py

@@ -78,6 +78,7 @@
     "๋":['356'],
     "์":['456'],
     "ๆ":['3'],
+    " ":['-1']
 }
 
 dict_2 = {
@@ -93,7 +94,7 @@
 
 _v1=["เ-tอ", "เ-ีtย", "เ-ืtอ", "-ัtว", "เ-tา", "เ-tาะ"]
 
-char_trie = Trie(list(thai_braille_mapping_dict.keys())+_v1)
+char_trie = Trie(list(thai_braille_mapping_dict.keys())+_v1+[" "])
 
 
 _vowel_patterns =[i.replace("-","([ก-ฮ])").replace("t","([่้๊๋])")+",\\1"+i.replace("t","")+"\\2" for i in _v1]
@@ -106,15 +107,26 @@ def _replace_vowels(word: str) -> str:
 
     return word
 
-def thai_word_braille(word):
+def thai_word_braille(word: str) -> str:
     word = _replace_vowels(word)
     _temp = []
     for i in word_tokenize(word,custom_dict=char_trie, engine="mm"):
-       _temp.append(thai_braille_mapping_dict[i])
+       if i.isspace() and len(i)>1:
+        for k in list(i):
+            _temp.append(thai_braille_mapping_dict[k])
+       else:
+        _temp.append(thai_braille_mapping_dict[i])
     _b = Braille(_temp)
     return _b.tobraille()
 
 
+def thai_text_braille(text: str) -> list:
+    _list_braille = []
+    for word in word_tokenize(text):
+        _list_braille.append(thai_word_braille(word))
+    return _list_braille
+
+
 class Braille:
 	'''
     List Braille to Braille
@@ -133,6 +145,7 @@ def __init__(self,data):
 		else:
 			self.data = sorted(list(data)) # แปลงเป็น list พร้อมเรียงจากน้อยไปมาก
 		self.db = {
+            "-1":" ",
             '0':'⠀',
             '1':'⠁',
             '3':'⠂',