Removing Arabic Diacritics using Python

Question

I want to filter my text by removing Arabic diacritics using Python.

For example:

Context	Text
Before filtering	اللَّهمَّ اغْفِرْ لنَا ولوالدِينَا
After filtering	اللهم اغفر لنا ولوالدينا

I have found that this can be done using CAMeL Tools but I am not sure how.

Ackerley

Looks like this answer has what you are looking for? https://mcmap.net/q/40948/-what-is-the-best-way-to-remove-accents-normalize-in-a-python-unicode-string – Capable

Unfortunately, it does not work for Arabic – Ackerley

The trick is to ensure text is normalised to NFC, rather than NFD, then strip non-spacing marks (diacritics). – Rhythmandblues

''.join(c for c in unicodedata.normalize('NFC', text) if unicodedata.category(c) != 'Mn') – Rhythmandblues

Answer 1

You can use the library pyArabic like this:

import pyarabic.araby as araby

before_filter="اللَّهمَّ اغْفِرْ لنَا ولوالدِينَا"
after_filter = araby.strip_diacritics(before_filter)

print(after_filter)
# will print : اللهم اغفر لنا ولوالدينا

You can try different strip filters:

araby.strip_harakat(before_filter)  # 'اللّهمّ اغفر لنا ولوالدينا'
araby.strip_lastharaka(before_filter)  # 'اللَّهمَّ اغْفِرْ لنَا ولوالدِينَا'
araby.strip_shadda(before_filter)  # 'اللَهمَ اغْفِرْ لنَا ولوالدِينَا'
araby.strip_small(before_filter)  # 'اللَّهمَّ اغْفِرْ لنَا ولوالدِينَا'
araby.strip_tashkeel(before_filter)  # 'اللَّهمَّ اغْفِرْ لنَا ولوالدِينَا'
araby.strip_tatweel(before_filter)  # 'اللَّهمَّ اغْفِرْ لنَا ولوالدِينَا'

Tricycle

Answer 2

You really don't need to use any library for this, just plain regex:

import re
text = 'اللَّهمَّ اغْفِرْ لنَا ولوالدِينَا '    
output=re.sub(u'[\u064e\u064f\u0650\u0651\u0652\u064c\u064b\u064d\u0640\ufc62]','',text)
print(output)
#اللهم اغفر لنا ولوالدينا 

Cathexis

Answer 3

Oneliner:

text = 'text with Arabic Diacritics to be removed'    
text = ''.join([t for t in text if t not in ['ِ', 'ُ', 'ٓ', 'ٰ', 'ْ', 'ٌ', 'ٍ', 'ً', 'ّ', 'َ']])
print(text)

if you want the full list of Arabic Diacritics you can also get it from pyArabic, standalone example:

import unicodedata
try:
    unichr
except NameError:
    unichr = chr

text = 'اللَّهمَّ اغْفِرْ لنَا ولوالدِينَا '    
text = ''.join([t for t in text if t not in [unichr(x) for x in range(0x0600, 0x06ff) if unicodedata.category(unichr(x)) == "Mn"]])
print(text)

Endogen