Regex Python Remove everything before the first letter in a string

Question

Another question from REGEX that none of the answers I found follows. I have a dataframe that some strings start erroneously with things that are not Letters, for example :

t = ['. Subordinam-se ao regime', '1º O acesso à informação']

The result I want is

'Subordinam-se ao regime', 'O acesso à informação'

I’m trying to re.sub(r'[[:alpha:]]+', '', t) unsuccessfully, what I’m doing wrong?

Answer 1

In his example, t is a list, so it’s no use passing it directly to re.sub. You have to pass one string at a time.

Therefore, one option is to use:

import re

t = ['. Subordinam-se ao regime',  '1º O acesso à informação']
for s in t:
    print(re.sub('^[^a-zA-Z]+', '', s))

regex uses the marker ^ which indicates the start of the string. Then [^a-zA-Z]+ (one or more occurrences of any character which nay be a letter).

That is, we replace anything that is not letter and is at the beginning of the string. The output will be:

Subordinam-se ao regime
O acesso à informação

---

If you have accented letters, you can switch to:

import re
from unicodedata import normalize

t = ['. Subordinam-se ao regime',  '1º O acesso à informação', ' - É blabla etc']
for s in t:
    print(re.sub('^[^a-zA-Z]+', '', normalize('NFD', s)))

I use normalize to convert the string to NFD. To understand what normalization is, read here, here and here, but basically letters like É are broken down into two: the letter E without accent and the accent itself. Thus, we also consider accented characters.

---

The problem is that when normalizing you change the original content.

Of course another alternative is to include all accented characters in regex, something like:

re.sub('^[^a-zA-ZáéíóúÁÉÍÓÚ]+', '', s)

I just put the letters in a high-pitched accent, but then just add the rest inside the brackets.

---

And it can also be done without regex:

from string import ascii_lowercase

# retorna a string a partir da primeira letra
def apos_primeira_letra(s):
    # coloque aqui todas as letras válidas
    letras_validas = ascii_lowercase + 'çáéíóúãõâô'
    for i, c in enumerate(s):
        if c.lower() in letras_validas:
            return s[i:] # retorna da letra em diante
    return s # se não encontrou nenhuma letra, retorna a própria string

t = ['. Subordinam-se ao regime',  '1º O acesso à informação', ' - É blabla etc']
for s in t:
    print(apos_primeira_letra(s))

The idea is to have all valid letters in letras_validas (I put only the tiny ones to facilitate, so when checking you use lower() for each character being checked).

Initially I had thought to use isalpha() to check if it is letter, but this method returns True for the character º, then I thought it best to fix the letters that I will consider valid.

Thus, the function apos_primeira_letra goes through the string, and if you find a letter, it returns everything from there. If you don’t find any, it returns the same string without modification.