Merge pull request #17 from gunsodo/feat/number-to-lao

Implement `num_to_laoword`

laonlp/util/__init__.py

@@ -17,11 +17,13 @@
 __all__ = [
     "lao_digit_to_arabic_digit",
     "arabic_digit_to_lao_digit",
+    "num_to_laoword",
     "remove_tone_mark",
 ]
 from laonlp.util.digitconv import (
     lao_digit_to_arabic_digit,
     arabic_digit_to_lao_digit,
+    num_to_laoword,
 )
 from laonlp.util.lao import (
     remove_tone_mark

laonlp/util/digitconv.py

@@ -28,6 +28,11 @@
     "ເກົ້າ",
     "ສູນ"
 ]
+_places = [
+    "", "ສິບ", "ຮ້ອຍ", "ພັນ", "ຫມື່ນ", "ແສນ", "ລ້ານ", "ຕື້",
+]
+_exceptions = {"ຫມື່ນ": "ສິບ", "ນຶ່ງສິບ": "ສິບ", "ສອງສິບນຶ່ງ": "ຊາວເອັດ", "ສອງສິບ": "ຊາວ", "ສິບນຶ່ງ": "ສິບເອັດ"}
+
 _dict_lao_arabic = dict(zip(list(NUMBERS), list(_arabic_numerals)))
 _dict_arabic_lao = dict(zip(list(_arabic_numerals), list(NUMBERS)))
 _lao_arabic_table = str.maketrans(_dict_lao_arabic)
@@ -56,9 +61,50 @@ def arabic_digit_to_lao_digit(text: str) -> str:
     return text.translate(_arabic_lao_table)
 
 
-def number2lao(numbers: int):
+def num_to_laoword(number: int):
     """
-    Numbers to Lao pronunciation
+    Number to Lao word
+
+    :param number int: Integer to be converted
+    :return: returns a string of Lao word representation of the integer
+    :rtype: str
     """
-    # TODO
-    return ""
+    output = ""
+    prefix = ""
+
+    if number is None:
+        return ""
+
+    if number == 0:
+        return _pronunciation[-1]
+
+    sign = number < 0
+    number = str(abs(number))
+
+    # Special case > 1e9
+    if len(number) >= 10:
+        prefix = num_to_laoword(int(number[:-9])) + _places[-1]
+        number = number[-9:]
+
+    prev_value = ""
+
+    for place, value in enumerate(list(number[::-1])):
+        if place % 6 == 0 and place > 0:
+            output = _places[6] + output
+
+        if value != "0":
+            output = _pronunciation[int(value) - 1] + _places[place % 6] + output
+
+        # Special place exception
+        if place % 6 == 3 and prev_value == "0":
+            output = _places[3] + output
+
+        prev_value = value
+
+    for search, replac in _exceptions.items():
+        output = output.replace(search, replac)
+
+    if sign:
+        output = "ລົບ" + output
+
+    return prefix + output

setup.py

@@ -23,8 +23,9 @@
     requirements = [i.strip() for i in f.readlines()]
 
 extras = {
-    "word_vector": ["gensim", "huggingface-hub"],
+    "word_vector": ["scipy<1.11.0", "gensim", "huggingface-hub"],
     "full": [
+        "scipy<1.11.0",
         "gensim",
         "huggingface-hub"
     ]

tests/test_util.py

@@ -1,24 +1,39 @@
 # -*- coding: utf-8 -*-
 
 import unittest
-from laonlp.util import *
+from laonlp.util import (
+    lao_digit_to_arabic_digit,
+    arabic_digit_to_lao_digit,
+    remove_tone_mark,
+    num_to_laoword,
+)
 
 
 class TestTagPackage(unittest.TestCase):
     def test_lao_digit_to_arabic_digit(self):
         self.assertEqual(
             lao_digit_to_arabic_digit("໑໒໓໔໕໖໗໘໙໐"),
-            '1234567890'
+            "1234567890"
         )
 
     def test_arabic_digit_to_lao_digit(self):
         self.assertEqual(
-            arabic_digit_to_lao_digit('1234567890'),
+            arabic_digit_to_lao_digit("1234567890"),
             "໑໒໓໔໕໖໗໘໙໐"
         )
 
     def test_remove_tone_mark(self):
         self.assertEqual(
             remove_tone_mark("ຜູ້"),
-            'ຜູ'
+            "ຜູ"
         )
+
+    def test_num_to_laoword(self):
+        self.assertEqual(num_to_laoword(None), "")
+        self.assertEqual(num_to_laoword(0), "ສູນ")
+        self.assertEqual(num_to_laoword(112), "ນຶ່ງຮ້ອຍສິບສອງ")
+        self.assertEqual(num_to_laoword(-273), "ລົບສອງຮ້ອຍເຈັດສິບສາມ")
+        self.assertEqual(num_to_laoword(12101), "ສິບສອງພັນນຶ່ງຮ້ອຍນຶ່ງ")
+        self.assertEqual(num_to_laoword(20000), "ຊາວພັນ")
+        self.assertEqual(num_to_laoword(987654321), "ເກົ້າຮ້ອຍແປດສິບເຈັດລ້ານຫົກແສນຫ້າສິບສີ່ພັນສາມຮ້ອຍຊາວເອັດ")
+        self.assertEqual(num_to_laoword(11987654321), "ສິບເອັດຕື້ເກົ້າຮ້ອຍແປດສິບເຈັດລ້ານຫົກແສນຫ້າສິບສີ່ພັນສາມຮ້ອຍຊາວເອັດ")