ltakens · January 11, 2023 21:22
diff --git a/python_email_address_finder.py b/python_email_address_finder.py
 """Here are some keywords one might use to find this email regex:

 - Python etract email addresses using regular expression
 - Python extract email from text
 - How to extract email addresses from a block of text using Python
 """
 import doctest
 import re


 EMAIL_REGEX = re.compile(
    r'''(                                       # start capture full address
    ([a-z0-9\.\+-_]|\(dot\)|\[dot\])+           # e.g. 'john_91[dot]goodman'
    (@|\(at\)|\[at\])                           # '@' or '(at)' or '[at]'
    [\w-]+                                      # e.g. 'a-subdomain'
    ((\(dot\)|\[dot\]|\.)([a-z-_]+)){1,4}       # e.g. 'domain.com.au'
    )''',                                       # end capture full address
    re.VERBOSE | re.IGNORECASE
 )


 def find_all_email_addresses(text: str) -> list:
    """Return all email addresses from text.

    >>> find_all_email_addresses(
    ...     '''Hello [email protected],
    ...     Are you still at john+goodman.is.awesome(at)his-domain.com.au?
    ...     Or have you changed your email format to
    ...     john[at]his(dot)domain.com[dot]au?'''
    ... )
    ['[email protected]', '[email protected]', '[email protected]']
    """
    results = []
    for addr, *_ in re.findall(EMAIL_REGEX, text):
        addr = re.sub('\(at\)|\[at\]', '@', addr, flags=re.IGNORECASE)
        addr = re.sub('\(dot\)|\[dot\]', '.', addr, flags=re.IGNORECASE)
        results.append(addr)

    return results


 if __name__ == '__main__':
    doctest.testmod()
	"""Here are some keywords one might use to find this email regex:

	- Python etract email addresses using regular expression
	- Python extract email from text
	- How to extract email addresses from a block of text using Python
	"""
	import doctest
	import re


	EMAIL_REGEX = re.compile(
	r'''( # start capture full address
	([a-z0-9\.\+-_]\|\(dot\)\|\[dot\])+ # e.g. 'john_91[dot]goodman'
	(@\|\(at\)\|\[at\]) # '@' or '(at)' or '[at]'
	[\w-]+ # e.g. 'a-subdomain'
	((\(dot\)\|\[dot\]\|\.)([a-z-_]+)){1,4} # e.g. 'domain.com.au'
	)''', # end capture full address
	re.VERBOSE \| re.IGNORECASE
	)


	def find_all_email_addresses(text: str) -> list:
	"""Return all email addresses from text.

	>>> find_all_email_addresses(
	... '''Hello [email protected],
	... Are you still at john+goodman.is.awesome(at)his-domain.com.au?
	... Or have you changed your email format to
	... john[at]his(dot)domain.com[dot]au?'''
	... )
	['[email protected]', '[email protected]', '[email protected]']
	"""
	results = []
	for addr, *_ in re.findall(EMAIL_REGEX, text):
	addr = re.sub('\(at\)\|\[at\]', '@', addr, flags=re.IGNORECASE)
	addr = re.sub('\(dot\)\|\[dot\]', '.', addr, flags=re.IGNORECASE)
	results.append(addr)

	return results


	if __name__ == '__main__':
	doctest.testmod()
No results found