text2digits 0.1.2

A small library to convert text numbers to digits in a string

Installation

pip3 install text2digits

Usage

Python 3 only!

from text2digits import text2digits
t2d = text2digits.Text2Digits()
t2d.convert("twenty ten and twenty one")
> 2010 and 21

It can handle a variety of phrases. Spoken/Informal and formal language:

"A random string" -> 'A random string'
"I am thirty six years old with a child that is four. I would like to get him four cars!" -> 'I am 36 years old with a child that is 4. I would like to get him 4 cars!'
"I was born in twenty ten" -> 'I was born in 2010'
"I was born in nineteen sixty four" -> 'I was born in 1964'
"I am the fourth cousin" -> 'I am the 4 cousin'
"I am twenty nine" -> 'I am 29'
"it was twenty ten and was negative thirty seven degrees" -> 'it was 2010 and was negative 37 degrees'
"thirty twenty one" -> '3021'
"one thousand six hundred sixty six" -> '1666'
"one thousand and six hundred and sixty six" -> '1666'
"sixteen sixty six" -> '1666'
"eleven hundred twelve" -> '1112'
"Sixteen and seven" -> '16 and 7'
"twenty ten and twenty one" -> '2010 and 21'
"I was born in nineteen ninety two and am twenty six years old!" -> 'I was born in 1992 and am 26 years old!'
"three forty five" -> '345'

I find this useful if using Alexa/Lex to convert audio to text and have to convert the text to digits.

Known Limitations

• Negative numbers: the word "negative" is preserved as-is rather than being converted to a unary minus (e.g. "negative five" → "negative 5", not "-5").
• Ordinal-to-digit conversion is lossy by default: ordinal suffixes are dropped unless add_ordinal_ending=True is passed (e.g. "third" → "3", not "3rd").

What Is Supported

• Decimal literals adjacent to scale words: "2.5 thousand" → "2500", "1.2345 hundred" → "123.45"
• Indian number system scales: lakh, crore, arab, kharab
• Ordinals: first, second, …, twentieth, hundredth, thousandth, …
• Spelling correction via similarity_threshold parameter
• Year-style concatenation: "twenty ten" → "2010"

Acknowledgements

I have heavily used code from the SO answers from here: https://stackoverflow.com/questions/493174/is-there-a-way-to-convert-number-words-to-integers and improved upon them