Copyright © 2013 Splunk Inc.
Regex is Fun
David Clawson
SplunkYoda
2
Regular Expressions
•
“A regular expression is a pattern which specifies a set of
strings of characters; it is said to match certain strings.”
—Ken Thompson
•
QED Text Editor written by Ken in the 1970s
•
Invented in the 1940s
•
Help celebrate it’s 70th Year
3
Types of Regular Expressions
4
How is Regex used in Python?
Python “re”
Python's built-in "re" module provides excellent support for regular expressions, with a
modern and complete regex flavor.
The only significant features missing from Python's regex syntax are atomic grouping,
possessive quantifiers, and Unicode properties.
Using Regular Expressions in Python
The first thing to do is to import the regexp module into your script with “import re”.
5
How is Regex used in Python?
Call re.search(regex,
•
•
subject) to apply a regex pattern to a subject string.
The function returns None if the matching attempt fails, and a Match object otherwise.
The Match object stores details about the part of the string matched by the regular expression
pattern.
Since None evaluates to False, you can easily use re.search() in an if statement.
6
How is Regex used in Python?
Do not confuse re.search() with re.match().
•
Both functions do exactly the same, with the important distinction that re.search() will attempt
the pattern throughout the string, until it finds a match. re.match() on the other hand, only
attempts the pattern at the very start of the string.
7
How is Regex used in Python?
To get all matches from a string, call re.findall(regex, subject).
This will return an array of all non-overlapping regex matches in the string. "Non-overlapping"
means that the string is searched through from left to right, and the next match attempt starts
beyond the previous match.
If the regex contains one or more capturing groups, re.findall() returns an array of tuples,
with each tuple containing text matched by all the capturing groups.
The overall regex match is not included in the tuple, unless you place the entire regex
inside a capturing group.
8
How is Regex used in Python?
More efficient than re.findall() is re.finditer(regex, subject).
It returns an iterator that enables you to loop over the regex matches in the subject string: for m in
re.finditer(regex, subject). The for-loop variable m is a Match object with the details of the current
match.
9
How is Regex used in Splunk?
Field extraction
| rex field=_raw “%UC_CALLMANAGER-(?<Severity>\d+)-EndPointUnregistered:
Configure Line Breaking
LINE_BREAKER = [\r\n]+
Filtering and Routing Data to Queues
REGEX =(?m)^EventCode=(592|593)
Many more…….
10
Regex Testing Tools
• RegExr http://gskinner.com/RegExr/
• Reggy
http://reggyapp.com/
• RegexPal http://regexpal.com/
• Regex Buddy http://www.regexbuddy.com/
• Lars Olav Torvik http://regex.larsolavtorvik.com/
• Rubular http://rubular.com/
11
Regex Reference Texts
• http://www.regular-expressions.info/reference.html - from the
creators of RegexBuddy
• Introducing Regular Expressions by Michael Fitzgerald
• Mastering Regular Expressions by Jeffrey Friedl
• Regular Expressions Cookbook by Jan Goyvaerts
• Regular Expressions Pocket Reference by Tony Stubblebine
12
Basic Concepts of Regular Expressions
Because Knowing
leads to Doing
14
Simple Pattern Matching

Matching String Literals

Matching Digits and Non-Digits

Matching Word and Non-Word Characters

Matching Whitespace

Matching Any Character
15
Matching String Literals
Sample Apache Log
10.23.10.11 www.iamcool.com 10.100.0.11 - - [06/Dec/2012:14:39:03 -0800] "GET
/Facelift/answers/swelling HTTP/1.1" 301 20 14932 "-" "Mozilla/5.0 (compatible; Googlebot/2.1;
+http://www.google.com/bot.html)”
Literal String Match of the first ip address would be:
10.23.10.11
16
Matching Digits and Non-Digits
\d or \D or [0-9]
\d - match digit
\D – match non-digit (matches whitespace, punctuation and other characters not used in words)
[0-9] - match any number (called a character class)
[^0-9] – match any non-number
17
Matching Words and Non-Words
\w or \W
\w – match any word character and is essentially the same as the character class [a-zA-Z0-9]
\W – match any non-word character
18
Matching Whitespace
\s or \S
\s – match whitespace (Spaces, Tabs, Line Feeds and Carriage Returns)
\S – match any character that is not whitespace. Same as [^\s]
19
Character shorthands for whitespace
Character Shorthand
Description
\f
Form Feed
\h
Horizontal Whitespace
\H
Not Horizontal Whitespace
\n
Newline
\r
Carriage Return
\t
Horizontal Tab
\v
Vertical Tab (whitespace)
\V
Not vertical whitespace
20
Matching Any Character
Dot (.)
Matches any character but line ending characters
\b – matches a word boundary without consuming any characters
21
Boundaries and Alternation

Matching the Beginning and End of Line

List of Regex Special Character

Alternation and Regex Options

Subpatterns

Capturing and Named Groups

Character Classes

Negated Character Classes
22
Matching Beginning and End of Line
^ OR $
^ - matches the beginning of a line
$ - matches the end of a line
23
List of Regex Special Characters
.^*+?|(){}[]\-
. -matches any character
^ -matches beginning of the line
* -matches zero or more
+ -matches one or more
? –matches one or more
| -used for alternation (choice of patterns to match)
() –used for grouping
{} –used as a quantifier
[] –used with character classes
\ -used to make a character literal or as a special regex character
- -hyphen is used in a character class range
24
Alternation and Options
| OR ?
| -gives choice of alternate patterns to match, ie: (THE|The|the)
(?i) – Case insensitive
(?J) –allow duplicate names
(?m) –match on duplicate lines
(?s) –match on a single line
(?U) –match lazy
(?X) –Ignore whitespace, comments
(?-…) –Unset or turn off options
25
Subpatterns
Group(s) within a group
(THE|The|the) -has three subpatterns
(tT)h(e|eir) –matches the, The, their, Their
26
Capturing and Named Groups
()
(?<name>…) OR (?P<name>…)
 Store their content in memory
(it is) (time to eat)
$1
$2
(?<Severity>\d)
 Splunk creates a field of Severity from this named group
27
Character Classes
[]
[aeiou] –only matches the characters inside of the brackets
[0-9] –matches a range of characters, using a hyphen
[a-zA-Z0-9] –matches all alphanumeric characters
28
Negated Character Classes
[^…]
*** Super important – especially for Splunk field extractions ***
[^aeiou] –matches all consonants and NOT vowels
[^\s] – match everything that is not a space
29
Quantifiers

Greedy, Lazy, Possessive

Matching a certain number of times
30
Greedy, Lazy, Possessive
* + ?
* - match zero of more times
.* -will match all of the characters in the subject text (want to avoid this)
+ -match one or more
\d+ -match all of the digits until there aren’t any more - greedy
? –match 0 or 1 of the preceeding token.
colou?r –matches either color or colour
31
Matching a Certain Number of Times
{}
\d{3} -matches 3 digits only
\d{1,3} –matches range of 1 to 3 digits
\d{1,} -same as \d+
\d{0,} -same as \d*
\d{0,1} -same as \d?
32
Any Thoughts, Ideas, Feedback?
splunkyoda@splunk.com
33
Optimized Regular Expressions
Because fast is elegant!
Optimize Regular Expressions
Good
Better
(whiskey)
(?:whiskey)
Capture groups add unnecessary overhead and impact overall
performance use them only when necessary.
Optimize Regular Expressions
Good
Better
splunk|splash
spl(?:unk|ash)
Try to “factor” on the left, when you can, while exposing required
text. Less alternation is better.
Optimize Regular Expressions
Good
Better
(?:aussie$|gypsie$)
(?:aus|gyp)sie$
Try to “factor” on the right when input text is close to end of the
line. Most regex engines will anchor at end of line when “$” is
present.
Optimize Regular Expressions
Good
Better
0{3,7}
0000{0,4}
Typically exposing required or literal text makes the engine
execute the regex faster
Optimize Regular Expressions
Good
Better
(.)*
.*
Useless parenthesis add unnecessary overhead. As above, use
them only when necessary.
Optimize Regular Expressions
Good
Better
matty[:]
matty:
The character class/set (indicated by []) will add unnecessary
overhead when not needed.
Optimize Regular Expressions
Good
Better
^genti|^collar
^(?:genti|collar)
Anchoring the regex at the beginning of the line will result in
improved performance with most regex engines.
Optimize Regular Expressions
Good
Better
delaney$|connery$
(delaney|connery)$
I said, anchor the regex!
Optimize Regular Expressions
Good
Better
^src.*:
^src[^:]*:
Using a negated character class/set instead of lazy/greedy
quantifiers will typically result in faster regexes. Lazy/greedy
quantifiers will make the regex engines backtrack which
ultimately impacts overall performance.
Optimize Regular Expressions
Good
Better
bride|brian
bri(?:de|an)
Full alternation is more expensive than partial alternation. Also,
in this case the regex engine will alternate only AFTER ‘bri’ has
been matched.
Optimize Regular Expressions
Good
Better
(?:edu|com|net|…)
(?:com|edu|net|…)
Leading the engine to a match by placing the most popular match
first may result in faster execution in some engines.
Optimize Regular Expressions
Good
Better
^.*(answer)
^.{42}(answer)
Specifying an exact position inside the string and leading the
engine to a match, will help improve performance drastically
compared to using a simple greedy/lazy quantifier.
Optimize Regular Expressions
Good
Better
.*?a
^.*a
If ‘a’ is near the end of the input string will match faster as less
backtracking will be required.
Optimize Regular Expressions
Good
Better
.*a
^.*?a
If ‘a’ is near the beginning of the input string the regex engine
will match faster.
Optimize Regular Expressions
Good
Better
:[^:]*:
:[^:]*+:
Ex. in ‘ :destination’ the second regex fails faster.
Optimize Regular Expressions
Good
Better
:[^:]*:
:(?>[^:]*):
Same as above, using different notation. Explanation:
Atomic grouping or possessive quantifiers instruct the regex
engine not to keep the states captured by * or + therefore
preventing it from unsuccessfully backtracking and in turn failing
faster.
Python for the Masses