cnoan 1.1.2

Convert Chinese numerals and Arabic numerals.

Chinese numerals and Arabic numerals conversion

There should be emojis here (●'◡'●)

Project Description and Description English Version 工程说明与描述中文版本

♠ cnoan is a toolkit to quickly convert Chinese numbers and Arabic numbers! in
cn refers to Chinese numerals
an refers specifically to Arabic numerals
o stands for reciprocal

♥ The mutual in the interchange is difficult to translate /(ㄒoㄒ)/~~ If the first letter of mutual is used, then the name of this project will feel like 🐎 people (spread bad speech 🔪, close the small black house❎); Roughly, it means mutual rotation. The two are connected through c, which can only reflect the meaning of one-way 👉; Therefore, o is used to connect in the middle, which can reflect the concept of ·mutual· ·ring· to a certain extent;

♦ This project is based on the ideas and guidance of cn2an to update the functions of the problems encountered or existing; Welcome to star and follow, everyone to maintain and improve together;

♣ Hey, it's great ★,°:.☆(￣▽￣)/$:.°★ .

Directory Structure

   Catalog Name                 Function and Description                         What's New
|---an2cn.py         Convert Arabic numerals to Chinese numerals               Newly defined class names
|---base.py          This is the base class of the project, which contains the base class of ConvertBase None
|---cn2an.py         Convert Chinese numbers to Arabic numbers Newly defined class names
|---config.yaml      The configuration of the project, mainly the definition of the matching rules Add the abnormal field
|---setup.py         Project packaging, packaging and publishing Add my information

|---translate.py     Convert the content of the sentence to determine the conversion,     use the abnormal field
                     With parameters Modified regular expression
|---Auxiliary        Definition of basic functions in utils.py                                    None
|---requirement.txt  The package required by the project                                          None

---

Project features

basic function

1.1 Chinese numbers => Arabic numbers

• Support Chinese numbers => Arabic numbers;
• Support Uppercase Chinese numbers => Arabic numbers;
• Support Chinese numbers and Arabic numbers => Arabic numbers;

1.2 Arabic numbers => Chinese numbers

• Support Arabic numbers => Chinese numbers;
• Support Arabic numbers => Uppercase Chinese numbers;
• Support Arabic numerals => Uppercase RMB;

1.3 Sentence Transformation

• Support Chinese numbers => Arabic numbers;

  • support date;
  • support score;
  • support percent;
  • support Celsius;

• Support Arabic numbers => Chinese numbers;

  • support date;
  • support score;
  • support percent;
  • support Celsius;

1.4 Others

• support decimal;
• support negative numbers;
• Support for HTTP API.

Feature updates and fixes

• 🎈 Redefine the field position to be translated (translated) (●'◡'●) The original project (transform+cn2en) will have the following situations

  Seven up and eight down --> 7 up and eight down
  2 people --> 2 people
  

  In fact, in practical applications, we do not want it to be converted. Therefore, the premise of conversion is redefined in this project

  'raw':
      self.cn_pattern = f" negative trillion]+"
  

  'Now':
        self.cn_pattern = f"negative?-?positive?\+?([01234567890][\s\t]*[1000000000 trillion]+)(dot [01234567890]+)?"
  

  Of course, I can't guarantee that this rule can help you solve the corresponding business needs. Therefore, you can redefine it in self.cn_pattern of translate.

• 🎈 Introduced isolation conversion and regression of abnormal words o(￣▽￣)ブ <<<<<<< HEAD When the above redefinition criteria are defined, situations such as Wanning, In case, and seven up and eight down can be avoided, but it is still necessary to sigh the vastness and profoundness of Chinese

When the above redefinition criteria are defined, situations such as Wanning, In case, and seven up and eight down can be avoided, but it is still necessary to sigh the vastness and profoundness of Chinese

307cd2b822ddce624506eb5f7d53b04b123f8bad

'E.g':
    'One Five Ten'
...

If the word is thrown in directly, the following results will be obtained:

:return: 'One five 10'

This doesn't work. In this project, I classify this similar content as abnormal words, refer to abnormal_words in config

'Thinking':
# encoder
  masks = ['1510', ''] # list[str, str, ...] define abnormal words
  inputs = str('XXXxxx')
  mask_contents = {}
  for index, item in enumerate(masks):
      if item in inputs:
          mask = f'_MASK_{index}_'
          mask_contents[mask] = item
          inputs = inputs.replace(item, mask)
# decoder
  for contents in list(mask_contents.keys()):
      if contents in output:
          output = output.replace(contents, mask_contents[contents])

• 🎈 Modify a point in the original project There is a situation in the original engineering quantity: when liang, gan, yi, etc. appear in the text, and are not the content to be converted, when the following demo segment is executed,

  inputs = str('XXXXxxx')
  inputs = inputs.replace("twenty", "twenty").replace("half", "0.5").replace("two", "2")

These words will be converted in advance, so this project will do the corresponding conversion after regularization judgment.

• 🎈 Do a little bit of detail We often define an unwritten rule in our daily tasks and writing papers: 10,000 has to be written in the style of 10,000, so this project is also 'forced' to join this rule 😔

After the above series of operations, the final effect is as follows:

from cnoan.translate import Translate
inputs = '这人坏滴很，王尼玛一五一十的收入为一万元, 而两人却告诉我是二千元'
mode = 'cn2an'
tans = Translate()
print(tans.convert("这人坏滴很，王尼玛一五一十的收入为一万元, 而两人却告诉我是二千元", "cn2an"))
# 这人坏滴很，王尼玛一五一十的收入为10,000元, 而两人却告诉我是2,000元

---

Project installation and usage

Install

• Method 1:

  pip install cnocn
  

• Method 2:

git clone https://github.com/zhuofalin/cnoan.git
cd cnoan
python setup.py install

• Method 3:

  git clone https://github.com/zhuofalin/cnoan.git
  copy cnoan to your project
  

Usage

# 在文件首部引入包
import cnoan

# 查看当前版本号
print(cnoan.__version__)
# 0.5.16

3.1 Chinese numbers => Arabic numerals

最大支持到 10**16，即 千万亿，最小支持到 10**-16。

import cnoan

# 在 strict 模式（默认）下，只有严格符合数字拼写的才可以进行转化
output = cnoan.cn2an("一百二十三")
# 或者
output = cnoan.cn2an("一百二十三", "strict")
# output:
# 123

# 在 normal 模式下，可以将 一二三 进行转化
output = cnoan.cn2an("一二三", "normal")
# output:
# 123

# 在 smart 模式下，可以将混合拼写的 1百23 进行转化
output = cnoan.cn2an("1百23", "smart")
# output:
# 123

# 以上三种模式均支持负数
output = cnoan.cn2an("负一百二十三", "strict")
# output:
# -123

# 以上三种模式均支持小数
output = cnoan.cn2an("一点二三", "strict")
# output:
# 1.23

3.2 Arabic numerals => Chinese numbers

最大支持到10**16，即千万亿，最小支持到 10**-16。

import cnoan

# 在 low 模式（默认）下，数字转化为小写的Chinese numbers
output = cnoan.an2cn("123")
# 或者
output = cnoan.an2cn("123", "low")
# output:
# 一百二十三

# 在 up 模式下，数字转化为大写的Chinese numbers
output = cnoan.an2cn("123", "up")
# output:
# 壹佰贰拾叁

# 在 rmb 模式下，数字转化为人民币专用的描述
output = cnoan.an2cn("123", "rmb")
# output:
# 壹佰贰拾叁元整

# 以上三种模式均支持负数
output = cnoan.an2cn("-123", "low")
# output:
# 负一百二十三

# 以上三种模式均支持小数
output = cnoan.an2cn("1.23", "low")
# output:
# 一点二三

3.3 句子转化

⚠️：Experimental feature that may cause undesired conversions.

import cnoan

# 在 cn2an 方法（默认）下，可以将句子中的Chinese numbers转成Arabic numerals
output = cnoan.translate("小王捡了一百块钱")
# 或者
output = cnoan.translate("小王捡了一百块钱", "cn2an")
# output:
# 小王捡了100块钱

# 在 an2cn 方法下，可以将句子中的Chinese numbers转成Arabic numerals
output = cnoan.translate("小王捡了100块钱", "an2cn")
# output:
# 小王捡了一百块钱


## 支持日期
output = cnoan.translate("小王的生日是二零零一年三月四日", "cn2an")
# output:
# 小王的生日是2001年3月4日

output = cnoan.translate("小王的生日是2001年3月4日", "an2cn")
# output:
# 小王的生日是二零零一年三月四日

## 支持分数
output = cnoan.translate("抛出去的硬币为正面的概率是二分之一", "cn2an")
# output:
# 抛出去的硬币为正面的概率是1/2

output = cnoan.translate("抛出去的硬币为正面的概率是1/2", "an2cn")
# output:
# 抛出去的硬币为正面的概率是二分之一

## 支持百分比
## 支持摄氏度

---

License

---

communicate

If you have any questions, you can communicate with me through [email] ([e-mail]), and I will reply as soon as possible.

---

Thanks

• Thunder Bouble: A lot of useful feedback, including some bugs and new features;
• Damon Yu: Added support for full-width numbers and full-width symbols.

Reference

• cn2an core code analysis
• How to publish your own package to pypi
• Small traps in Python
• Chinese characters to Arabic numerals
• Chinese Text Normalization for Speech Processing
• The Best Tool of Chinese Number to Digits
• Microsoft Recognizers Text Overview
• process: data preprocessing pipeline