Pattern Matching - Extraction

• pulling substrings that match a pattern our of a string
• String contain groups of patterns

Extraction Examples

• extract email address from a customer message
• extract links or content from a html tag
• look for valid textual input from a survey
• brand name information from a product description

Extraction Groups

• a group is defined by enclosing a pattern in parenthesis
• a pattern con contain several groups
• Groups are referred to by the order they were defined (starting at 1 )
• Group zero refers to the entire string

Grouping Characters

(...)
# starts and ends a group of regex patterns
(?P<name>...)
# starts and ends a named group name of regex patterns

match_object.group(group1, group2,...)
# return a string of the match specified by group
# group can be either an integer index or a string
# if no arguments are passed, defaults to group zero

Match Object Methods

match_object.group()
# return a tuple of all the subgroups of the match

Match Groups Example 1

import re
pattern = '(\w*)ware'
        # prefix ware
for word in ['software','hardware','dinnerware']:
    #3 words we want to extract from
    match = re.search(pattern, word)
    if match:
        print(match.group(1), match.groups())
        # oprint the first group [match.group(1)] and the entire word [match.groups()]
# OUTPUT
# soft software
# hard hardware
# dinner dinnerware

Match Groups Example 2

import re
pattern = '([A-z ]+)\s?(\d+)'
        # first group letter, second of numbers, in between optional white space character
for word in ['S100','Porsch911','dinnerware', 'Cyperdyne 101']:
    #products we want to extract from
    match = re.search(pattern, word)
    if match:
        print(f'Brand: {match.group(1)}\tModel: {match.groups(2)}')
        # match.group(1) refers to the part with the letters 
# OUTPUT
# Brand: S Model: 100
# Brand: Porsche Model: 911
# Brand: Cyperdyne Model: 101

Named Groups

• instead of number we can also use names of groups
• group() accepts the string name as an argument
• Defining a group anme consists of adding the following string after the opening parenthesis: ?P

Named Match Groups Example

import re
pattern = '(?P<brand>[A-z ]+)\s?(?P<model>\d+)'
for word in ['S100','Porsch911','dinnerware', 'Cyperdyne 101']:
    match = re.search(pattern, word)
    if match:
        print(f'Brand: {match.group("brand")}\tModel: {match.groups("model")}')
# OUTPUT
# Brand: S Model: 100
# Brand: Porsche Model: 911
# Brand: Cyperdyne Model: 101