Pattern Matching With Regular Expressions¶

A common file processing requirement is to match strings within the file to a standard form, for example a file may contain list of names, numbers and email addresses. A email extraction would need to extract only those entries that matched which look like an email address.
Regular expressions, commonly called regexes, are ideally suited for this task and although they can become very complex it is also possible to perform many tasks with some relatively simple expressions.
At their simplest, a regular expression is simply a string of characters and this string would then match with only that exact string, e.g.

string in file: 'email'
regex: 'email' # This is a regular expression but albeit not a very
               # useful one as only matches with one word!

In reality regexes are used to search for a string that “has the form” of the regular expression, as in the above email example. For this to be possible we need to define some syntax that lets us specify things such as ‘a number is in a range’, ‘a letter is one of a set’, ‘a certain number of characters’ etc. For this to work some characters are considered special and when used in conjunction with each other they let the user specify the correct search criteria.
Here we will examine a few special characters, for a complete reference see http://www.regular-expressions.info/reference.html or search for “regular expression reference” online.

Special Characters¶

An asterisk * specifies that the character preceding it can appear zero or more times, e.g,

regex: 'a*b'
test: 'b'         # Matches as there are no occurrences of 'a'
test: 'ab'        # Matches as there is a single 'a'
test: 'aaaaaaaab' # Matches as there are multiple occurrences of 'a'
test: 'aaaabab'   # Matches as there is an occurrence of a string of
                  # a's followed by a b

A range of characters, or a “character class” is defined using square brackets [], e.g.

regex: '[a-z]'
test: 'm' # Matches as it is a lower case letter
test: 'M' # Fails as it is an upper case letter
test: '4' # Fails as it is a number

Several ranges can be specified such that they are all checked, e.g.

regex: '[a-z,A-Z,0-9]'
test: 'm'  # Matches!
test: 'M'  # Matches!
test: '4'  # Matches!
test: 'mm' #Fails as there are two characters

Combining ranges and the asterisk allows us to specify any number of alphanumeric characters!, e.g.

regex: '[a-z,A-Z,0-9]*'
test: 'mm'    # Matches
test: 'a0123' # Matches

To specify an exact number of characters use braces {}, e.g.

regex: 'a{2}'
test: 'abab'  # Fails as there is not two consecutive a's in the string
test: 'aaaab' # Matches

For more complicated regular expressions it is not obvious whether you have written the expression correctly so it can be useful to check that it matches as you expect. For such tests there are online tools available such as the regex tester at http://www.regular-expressions.info/javascriptexample.html. Simply type in your regex and a test string and it will tell you a match can be found within your string.

Regular Expressions in Python¶

Python contains a regular expression module, called re that allows strings to be tested against regular expressions with a few lines of code. Reference: http://docs.python.org/2/library/re.html
The compile function also takes another optional argument controlling the matching process, all of which are documented at the above location. Here we pass the RE.IGNORECASE option meaning that a case-insensitive match is performed.
Example:

import re

def checkForMatch(checker, test):
    if checker.match(test) != None:
        print('String matches!')
    else:
        print('String does not contain a match')
# End of function definition

checker = re.compile('[a-z]')
checkForMatch(checker, 'a')
checkForMatch(checker, '9')

checker = re.compile('[a-z]', re.IGNORECASE)
checkForMatch(checker, 'a')
checkForMatch(checker, 'A')

Gives the output:

String matches!
String does not contain a match
String matches!
String matches!

Below we provide a more complex example of using regular expressions and a place where they would actually be used in a practical sense. The scenarios concern parsing a file with multiple lines of the form

Running 13 tests.............OK!

where the line has to start with the word ‘Running’ and end with the word ‘OK!’ or the test is considered a failure.

Regular expressions make parsing such a file a relatively simple matter once the regular expression is known. Here is the full example:

import re

filetestsRun = 'testResults.log'
f = open(filetestsRun,'r')
reTestCount = re.compile("Running\\s*(\\d+)\\s*test", re.IGNORECASE)
reCrashCount = re.compile("OK!")
reFailCount = re.compile("Failed\\s*(\\d+)\\s*of\\s*(\\d+)\\s*tests", re.IGNORECASE)
testCount = 0
failCount = 0
testsPass = True
for line in f.readlines():
    m=reTestCount.search(line)
    if m:
        testCount += int(m.group(1))
        m=reCrashCount.search(line)
        if not m:
            failCount += 1
            testsPass = False
    m=reFailCount.match(line)
    if m:
        # Need to decrement failCount because crashCount will
        # have incremented it above
        failCount -= 1
        failCount += int(m.group(1))
        testsPass = False

f.close()

print("Tests Passed: {}".format(testsPass))
print("Tests Failed: {}".format(failCount))
print("Total Tests: {}".format(testCount))

The loop keeps track of test crashes and failures by using regular expressions to match the required text within each line of the file

Category:Tested Examples