Merge branch 'dev' of https://github.com/PyThaiNLP/pythainlp into dev

Author: wannaphong

docs/api/util.rst

@@ -32,6 +32,7 @@ Modules
 .. autofunction:: remove_zw
 .. autofunction:: reorder_vowels
 .. autofunction:: text_to_arabic_digit
+.. autofunction:: text_to_num
 .. autofunction:: text_to_thai_digit
 .. autofunction:: thai_strftime
 .. autofunction:: thai_to_eng
@@ -40,5 +41,6 @@ Modules
 .. autofunction:: thaiword_to_num
 .. autofunction:: thaiword_to_time
 .. autofunction:: time_to_thaiword
+.. autofunction:: words_to_num
 .. autoclass:: Trie
    :members:

pythainlp/corpus/sentenceseg_crfcut.model

@@ -41,6 +41,8 @@
     "thaiword_to_num",
     "thaiword_to_time",
     "time_to_thaiword",
+    "text_to_num",
+    "words_to_num",
 ]
 
 from pythainlp.util.collate import collate
@@ -85,4 +87,4 @@
 from pythainlp.util.thaiwordcheck import is_native_thai
 from pythainlp.util.time import thai_time, thaiword_to_time, time_to_thaiword
 from pythainlp.util.trie import Trie, dict_trie
-from pythainlp.util.wordtonum import thaiword_to_num
+from pythainlp.util.wordtonum import thaiword_to_num, text_to_num, words_to_num

pythainlp/util/__init__.py

@@ -6,8 +6,10 @@
 https://colab.research.google.com/drive/148WNIeclf0kOU6QxKd6pcfwpSs8l-VKD#scrollTo=EuVDd0nNuI8Q
 """
 import re
+from typing import List
 
 from pythainlp.tokenize import Tokenizer
+from pythainlp.corpus import thai_words
 
 _ptn_digits = r"(|หนึ่ง|เอ็ด|สอง|ยี่|สาม|สี่|ห้า|หก|เจ็ด|แปด|เก้า)"
 _ptn_six_figures = (
@@ -45,6 +47,23 @@
 _tokenizer = Tokenizer(custom_dict=_valid_tokens)
 
 
+def _check_is_thainum(word: str):
+    for j in list(_digits.keys()):
+        if j in word:
+            return (True, 'num')
+    for j in ["สิบ", "ร้อย", "พัน", "หมื่น", "แสน", "ล้าน", "จุด", "ลบ"]:
+        if j in word:
+            return (True, 'unit')
+    return (False, None)
+
+
+_dict_words = [i for i in list(thai_words()) if not _check_is_thainum(i)[0]]
+_dict_words += list(_digits.keys())
+_dict_words += ["สิบ", "ร้อย", "พัน", "หมื่น", "แสน", "ล้าน", "จุด"]
+
+_tokenizer_thaiwords = Tokenizer(_dict_words)
+
+
 def thaiword_to_num(word: str) -> int:
     """
     Converts the spelled-out numerals in Thai scripts into an actual integer.
@@ -102,3 +121,94 @@ def thaiword_to_num(word: str) -> int:
         accumulated = -accumulated
 
     return accumulated
+
+
+def _decimal_unit(words: list) -> float:
+    _num = 0.0
+    for i, v in enumerate(words):
+        _num += int(thaiword_to_num(v)) / (10**(i+1))
+    return _num
+
+
+def words_to_num(words: list) -> float:
+    """
+    Thai Words to float
+
+    :param str text: Thai words
+    :return: float of words
+    :rtype: float
+
+    :Example:
+    ::
+
+        from pythainlp.util import words_to_num
+
+        words_to_num(["ห้า", "สิบ", "จุด", "เก้า", "ห้า"])
+        # output: 50.95
+
+    """
+    num = 0
+    if "จุด" not in words:
+        num = thaiword_to_num(''.join(words))
+    else:
+        words_int = ''.join(words[:words.index("จุด")])
+        words_float = words[words.index("จุด") + 1:]
+        num = thaiword_to_num(words_int)
+        if num <= -1:
+            num -= _decimal_unit(words_float)
+        else:
+            num += _decimal_unit(words_float)
+
+    return num
+
+
+def text_to_num(text: str) -> List[str]:
+    """
+    Thai text to list thai word with floating point number
+
+    :param str text: Thai text with the spelled-out numerals
+    :return: list of thai words with float value of the input
+    :rtype: List[str]
+
+    :Example:
+    ::
+
+        from pythainlp.util import text_to_num
+
+        text_to_num("เก้าร้อยแปดสิบจุดเก้าห้าบาทนี่คือจำนวนทั้งหมด")
+        # output: ['980.95', 'บาท', 'นี่', 'คือ', 'จำนวน', 'ทั้งหมด']
+
+        text_to_num("สิบล้านสองหมื่นหนึ่งพันแปดร้อยแปดสิบเก้าบาท")
+        # output: ['10021889', 'บาท']
+
+    """
+    _temp = _tokenizer_thaiwords.word_tokenize(text)
+    thainum = []
+    last_index = -1
+    list_word_new = []
+    for i, word in enumerate(_temp):
+        if (
+            _check_is_thainum(word)[0]
+            and last_index+1 == i
+            and i+1 == len(_temp)
+        ):
+            thainum.append(word)
+            list_word_new.append(str(words_to_num(thainum)))
+        elif _check_is_thainum(word)[0] and last_index+1 == i:
+            thainum.append(word)
+            last_index = i
+        elif _check_is_thainum(word)[0]:
+            thainum.append(word)
+            last_index = i
+        elif (
+            not _check_is_thainum(word)[0]
+            and last_index+1 == i
+            and last_index != -1
+        ):
+            list_word_new.append(str(words_to_num(thainum)))
+            thainum = []
+            list_word_new.append(word)
+        else:
+            list_word_new.append(word)
+            last_index = -1
+    return list_word_new

pythainlp/util/wordtonum.py

@@ -45,6 +45,8 @@
     thai_to_eng,
     thaiword_to_num,
     thai_keyboard_dist,
+    text_to_num,
+    words_to_num,
 )
 
 
@@ -107,6 +109,40 @@ def test_number(self):
         with self.assertRaises(TypeError):
             thaiword_to_num(["หนึ่ง"])
 
+        self.assertEqual(words_to_num("ศูนย์"), 0)
+        self.assertEqual(words_to_num("แปด"), 8)
+        self.assertEqual(words_to_num("ยี่สิบ"), 20)
+        self.assertEqual(words_to_num("ร้อยสิบสอง"), 112)
+        self.assertEqual(words_to_num("ลบแปด"), -8)
+        self.assertEqual(words_to_num("ลบยี่สิบ"), -20)
+        self.assertEqual(words_to_num("ลบร้อยสิบสอง"), -112)
+        self.assertEqual(
+            words_to_num("หกล้านหกแสนหกหมื่นหกพันหกร้อยหกสิบหก"), 6666666
+        )
+        self.assertEqual(words_to_num("สองล้านสามแสนหกร้อยสิบสอง"), 2300612)
+        self.assertEqual(words_to_num("หนึ่งร้อยสิบล้าน"), 110000000)
+        self.assertEqual(
+            words_to_num("สิบห้าล้านล้านเจ็ดสิบสอง"), 15000000000072
+        )
+        self.assertEqual(words_to_num("หนึ่งล้านล้าน"), 1000000000000)
+        self.assertEqual(
+            words_to_num("สองแสนสี่หมื่นสามสิบล้านสี่พันล้าน"),
+            240030004000000000,
+        )
+        self.assertEqual(words_to_num("ร้อยสิบล้านแปดแสนห้าพัน"), 110805000)
+        self.assertEqual(words_to_num("ลบหนึ่ง"), -1)
+        text = "ลบหนึ่งร้อยล้านสี่แสนห้าพันยี่สิบเอ็ด"
+        self.assertEqual(num_to_thaiword(words_to_num(text)), text)
+        self.assertIsNotNone(
+            text_to_num("เก้าร้อยแปดสิบจุดเก้าห้าบาทนี่คือจำนวนทั้งหมด")
+        )
+        self.assertIsNotNone(
+            text_to_num("สิบล้านสองหมื่นหนึ่งพันแปดร้อยแปดสิบเก้าบาท")
+        )
+        self.assertIsNotNone(
+            text_to_num("สิบล้านสองหมื่นหนึ่งพันแปดร้อยแปดสิบเก้า")
+        )
+
         self.assertEqual(
             arabic_digit_to_thai_digit("ไทยแลนด์ 4.0"), "ไทยแลนด์ ๔.๐"
         )