Wsrtjtyk

In a string (in reality it's much bigger):

s = """

BeginA

Qwerty

Element 11 35

EndA



BeginB

Element 12 38

...

Element 198 38

EndB



BeginA

Element 81132 38

SomethingElse

EndA



BeginB

Element 12 39

Element 198 38

EndB

"""

how to replace every Element <anythinghere> 38 which is inside a BeginB...EndB block (and only those!) by Element ABC?

I was trying with:

s = re.sub(r'Element .* 38', 'Element ABC', s)

but this doesn't detect if it's in a BeginB...EndB block. How to do this?

Use two expressions:

block = re.compile(r'BeginB[sS]+?EndB')

element = re.compile(r'Element.*?b38b')



def repl(match):

    return element.sub('Element ABC', match.group(0))



nstring = block.sub(repl, string)

print(nstring)

This yields

BeginA

Qwerty

Element 11 35

EndA



BeginB

Element ABC

...

Element ABC

EndB



BeginA

Element 81132 38

SomethingElse

EndA



BeginB

Element 12 39

Element ABC

EndB

See a demo on ideone.com.

Without re.compile (just to get the idea):

def repl(match):

    return re.sub(r'Element.*?b38b', 'Element ABC', match.group(0))



print re.sub(r'BeginB[sS]+?EndB', repl, s)

The important idea here is the fact that re.sub's second parameter can be a function.

You could very well do it without a function but you'd need the newer regex module which supports G and K:

rx = re.compile(r'''

    (?:G(?!A)|BeginB)

    (?:(?!EndB)[sS])+?K

    Element.+?b38b''', re.VERBOSE)



string = rx.sub('Element ABC', string)

print(string)

See another demo for this one on regex101.com as well.

Try the following:

r'(?s)(?<=BeginB)s+Elements+(d+)s+d+.*?(?=EndB)'

You can test it here.

For your example, I would echo @Jan's answer and use two separate regular expressions:

import re



restrict = re.compile(r'(?s)(?<=BeginB).*?(?=EndB)')

pattern = re.compile(r'Elements+(d+)s+38')



def repl(block):



    return pattern.sub('Element ABC', block.group(0))



out = restrict.sub(repl, s)