num2words library - Luxembourgish Number and Text Normalization

This repository contains an enhanced version of num2words with comprehensive Luxembourgish language support, including a powerful text normalizer for converting numbers, dates, times, and other numerical expressions to Luxembourgish text.

⚠️ Important: This Luxembourgish-enhanced version is only available from this GitHub repository and is not available on PyPI.

Installation

Since this Luxembourgish-enhanced version is not available on PyPI, you must install it directly from this GitHub repository:

git clone https://github.com/PeterGilles/num2words.git
cd num2words
pip install -e .

Or install directly from GitHub:

pip install git+https://github.com/PeterGilles/num2words.git

Luxembourgish Language Support

This version includes comprehensive Luxembourgish language support with the following features:

Language Code: lb (Luxembourgish)

Basic Usage

Supported to options:

• cardinal (default): Convert to cardinal numbers (e.g., 42 → zweeavéierzeg)
• ordinal: Convert to ordinal numbers (e.g., 42 → zweeavéierzegsten)
• ordinal_num: Convert to ordinal numbers as numerals (e.g., 42 → 42.)
• year: Convert to year form (e.g., 2023 → zweedausenddräianzwanzeg)
• currency: Convert to currency form (e.g., 1.50 EUR → eent Euro an fofzeg Cent)

Command line with Luxembourgish:

$ num2words 10001 -l lb
dausendeent
$ num2words 24,120.10 -l lb
véieranzwanzegdausendhonnertzwanzeg Komma zéng
$ num2words 24,120.35 -l lb
véieranzwanzegdausendhonnertzwanzeg Komma fënnefandrësseg
$ num2words 24.120,35 -l lb
véieranzwanzegdausendhonnertzwanzeg Komma fënnefandrësseg
$ num2words 1,234.56 -l lb
dausendzweehonnertvéierandrësseg Komma sechsafofzeg
$ num2words 1.234,56 -l lb
dausendzweehonnertvéierandrësseg Komma sechsafofzeg

Note: Both English-style (comma as thousands, dot as decimal) and European-style (dot as thousands, comma as decimal) number formats are supported for all decimal and large number conversions.

In Python code:

>>> from num2words import num2words
>>> num2words(42, lang='lb')
'zweeavéierzeg'
>>> num2words(42, lang='lb', to='ordinal')
'zweeavéierzegsten'
>>> num2words(2023, lang='lb', to='year')
'zweedausenddräianzwanzeg'

Luxembourgish Text Normalizer

The repository includes a comprehensive text normalizer (luxembourgish_normalizer.py) that converts various numerical expressions in Luxembourgish text to their word forms.

Installation and Usage:

python luxembourgish_normalizer.py input_file.txt
# or
echo "Your text here" | python luxembourgish_normalizer.py

Supported Normalization Types

The Luxembourgish normalizer script (luxembourgish_normalizer.py) recognizes and converts the following types of expressions:

1. Numbers and Large Numbers - Recognizes: Standalone numbers, numbers with spaces or dots as thousand separators, numbers with decimals (dot or comma) - Converts: All numbers to their full Luxembourgish word form - Examples:

   • 42 → zweeavéierzeg
   • 1 234 → dausendzweehonnertvéierandrësseg
   • 40 000 → véierzegdausend
   • 24,120.35 → véieranzwanzegdausendhonnertzwanzeg Komma fënnefandrësseg
   • 24.120,35 → véieranzwanzegdausendhonnertzwanzeg Komma fënnefandrësseg
   • 1.234,56 → dausendzweehonnertvéierandrësseg Komma sechsafofzeg

2. Dates and Ordinals - Recognizes: Full dates, numeric dates, ordinals with dot - Converts: Dates to Luxembourgish with correct ordinal and month, ordinals to correct form before nouns - Examples:

   • 30. Abrëll 2010 → drëssegsten Abrëll zweedausendzéng
   • 22.3. → zweeanzwanzegste Mäerz
   • 9. Plaz → néngte Plaz
   • den 1. Juni → den éischte Juni
   • de 4. Juni → de véierte Juni

3. Times - Recognizes: Times in HH:MM, HHhMM, or HH.MM format - Converts: To [hour] Auer [minute] in Luxembourgish, only for valid times (hours 0–24, minutes 0–59) - Examples:

   • 10:34 → zéng Auer véierandrësseg
   • 10h34 → zéng Auer véierandrësseg
   • 17.40 → siwwenzéng Auer véierzeg
   • 8:15 → aacht Auer fofzéng

4. Match Results - Recognizes: Sports scores in X:Y or X-Y format - Converts: To [score1] zu [score2] using correct Luxembourgish number words - Examples:

   • 1:1 → eent zu eent
   • 2:0 → zwee zu null
   • 71:56 → eenasiwwenzeg zu sechsafofzeg
   • 3-2 → dräi zu zwee

5. Percentages - Recognizes: Numbers followed by % (with or without space) - Converts: To [number] Prozent with correct decimal handling - Examples:

   • 25% → fënnefanzwanzeg Prozent
   • 93,9% → dräiannonzeg Komma néng Prozent
   • 50 % → fofzeg Prozent
   • 1,25% → eent Komma fënnefanzwanzeg Prozent

6. Units and Measurements - Recognizes: Numbers with units (temperature, volume, weight, distance, time) - Converts: To full Luxembourgish with correct unit and number form, including feminine forms - Examples:

   • 90° → nonzeg Grad
   • 50 kg → fofzeg Kilogramm
   • 100 ml → honnert Milliliter
   • 60 km → sechzeg Kilometer
   • 2 Stonnen → zwou Stonnen
   • 1ml → ee Milliliter
   • 500 gr → fënnefhonnert Gramm

7. Phone Numbers - Recognizes: Lines containing Telefon, Tel, or Phone with digit groups - Converts: Each digit to word, always including null for zeros and leading zeros - Examples:

   • Tel: 08 123 456 → Tel: null aacht eent zwee dräi véier fënnef sechs
   • Telefon: 352 123 456 → Telefon: dräi fënnef zwee eent zwee dräi véier fënnef sechs

8. Abbreviations - Recognizes: 2+ consecutive uppercase letters - Converts: To Luxembourgish letter pronunciation, or as a word if in custom dictionary - Examples:

   • VW → FAUWEE
   • CSV → ZEEÄSSFAU
   • FIFA → FIFA (custom dictionary word)
   • NATO → NATO (custom dictionary word)

9. Years with Suffixes - Recognizes: Decades like 1970er, 80er - Converts: To full Luxembourgish decade form - Examples:

   • 1970er → nonzénghonnertsiwwenzeger
   • 1980er → nonzénghonnertachtzeger
   • 80er → achtzeger
   • an den 1970er Joren → an den nonzénghonnertsiwwenzeger Joren

10. Currency - Recognizes: Numbers with currency codes or symbols (EUR, €, USD, $, GBP, £, etc.) - Converts: To full Luxembourgish currency form with correct grammar - Examples:

    • 1,50 EUR → een Euro a fofzeg Cent
    • 1,50€ → een Euro a fofzeg Cent
    • 1.50 EUR → een Euro a fofzeg Cent
    • 2,25 USD → zwee Dollar a fënnefanzwanzeg Cent
    • 1,01 GBP → ee Pond an ee Penny
    • huet 1,50 EUR kritt → huet een Euro a fofzeg Cent kritt

Each normalization type is context-aware and applies Luxembourgish grammar and phonological rules for natural, correct output. The normalizer handles both European-style (comma as decimal) and English-style (dot as decimal) number formats.