ReNamer:Regular Expressions

From den4b Wiki
Revision as of 16:41, 27 November 2012 by Den4b (talk | contribs) (final cleanup and bumped up section heading levels)
Jump to navigation Jump to search

Introduction

Regular Expressions (RegEx) allow you to use precise search conditions, such as:

  • Your search string must be located at the beginning (or at the end) of a line,
  • The specified text must occur n times, etc.

RegEx expressions use $ . ^ { [ ( | ) * + ? \ characters (called metacharacters) in various combinations to specify all these conditions.The search engine interprets these metacharacters, rather than finding a literal match for them.

The RegEx search expression is called a "RegEx pattern", because a single expression can match a large number of actual text that has the specified pattern. For example, the RegEx pattern b?t matches with bat, bet, bit, bot and but, etc.

Remember that RegEx strings are case-sensitive (The words cat, CAT, cAt, Cat, caT, cAT, CAt and CaT are not equivalent).

So our RegEx pattern b?t also matches with bAt, bEt, bIt, bOt and bUt, etc.

Also, note that even the digits (0-9) are "numeric characters" for RegEx.

So our RegEx pattern b?t also matches with b5t, b7t, etc.

In this section, the RegEx expressions (patterns) are shown in bold orange. The target strings (which are compared with the RegEx expression for a possible match) are shown in bold black. A part of the target text is color-coded to provide a clue as to why a certain part matches (green color), or does not match (red color)

Simple (literal) matches

When the search string does not contain any metacharacters, the RegEx engine works like "normal" search. (it tries to find an exact copy of the search string.) (This is also known as "literal match").

If you want to find a literal match for a metacharacter, put a backslash \ before it. (The \ character is called "escape character", because it lets the metacharacter escape from its special duty, and lets it act as a normal character. Its combination with a metacharacter is called "escape sequence").

For example, metacharacter ^ matches the beginning of string, but \^ matches the character ^.

Note that the RegEx pattern \\ matches the character \.

RegEx pattern Matches Remarks
foobar
foobar
 This RegEx pattern does not contain any metacharacters; so all characters are matched literally.
\^FooBarPtr
^FooBarPtr
 The \^ escape sequence searches for the character ^ literally .

Escape sequences

We already saw one use of escape sequence (above).

Specific escape sequences are interpreted as special conditions, as listed below.

RegEx pattern matches
\xnn
 Character represented by the hex code nn
\x{nnnn}
 two bytes char with hex code nnnn (unicode)
\t
 tab (HT/TAB), same as \x09 (Hex 09)
\n
 new line (NL), same as \x0a (Hex 0a)
\r
 carriage return (CR), same as \x0d (Hex 0d)
\f
 form feed (FF), same as \x0c (Hex 0c)
foo\x20bar
 matches foo bar (note the space in the middle), but does not match foobar
\tfoobar
 matches foobar preceded by a tab (the tab is needed for the match)

Note that the tab, new line, carriage return, and form feed are known as "white spaces". But RegEx can distinguish between them. This allows you to make high-precision searches.

Character Classes

A character class is a list of characters in square brackets [], which will match any one (and only one-) character from the list.

Note that:

  • The characters are not separated with a comma or a space.
  • If you repeat any character in the list, it is considered only once (duplicates are ignored).
  • A hyphen - is used to indicate range of characters.
RegEx Pattern Remarks
[abdef]
 Matches d, e, or f (only one character), but no other characters
[c-m]
 Matches any one (and only one) of the small alphabetical characters, from c to m
[G-J]
 Matches any one (and only one) of the capital alphabetical characters from G to J
[a-zA-Z]
 Matches any one (and only one) of the alphabetical characters (capital or small)
[5-8]
 Matches any one (and only one) of numerical characters from 5 to 8
[\n-\x0D]

Matches any one (and only one) of #10, #11, #12 or #13
(Note the use of escape sequence inside a class)

There are some special conditions:

  • If you do not want any of the characters in the specified class, then place ^ at the very beginning of the list (RegEx interprets that as "none of the characters listed in this class").
  • If you want [ or ] itself to be a member of a class, put it at the start or end of the list, or create a escape sequence (by putting \ before it).
RegEx Pattern Remarks
[-az]
 matches a, z, and -
(since – is put at the beginning, the escape sequence is not needed)
[a\-z]
 matches a, z, and -
(since – is not at the beginning/end, the escape sequence is needed)
[^0-9]
 matches any non-digit character
[]-a]
 matches any character from ] to a.
(since ] is at the beginning, the escape sequence is not needed)
foob[aeiou]r
 Matches with foobar, foober, etc. but not foobbr, foobcr, etc.
foob[^aeiou]r
 Matches with foobbr, foobcr etc. but not foobar, foober, etc.

@@@ The "]-a" example would need a clarification as to what is the natural sequence of characters, and where is a superset of all possible characters described?

Predefined Classes

Some of the character classes are used so often that RegEx has predefined escape sequences to represent them.

RegEx Pattern Remarks
\w
 an alphanumeric character, including an underscore (_)
\W
 a non-alphanumeric character
\d
 a numeric character
\D
 a non-numeric character
\s
 any space (same as the [ \t\n\r\f] class)
\S
 a non space
.
 any character in line (the symbol is just a dot)

Notice that the capitalized letter is used to negate (for example, compare \w with \W)

Word/Text Boundaries

A word boundary (\b) is a spot between two characters that has a \w on one side of it and a \W on the other side of it (in either order), counting the imaginary characters off the beginning and end of the string as matching a \W.

RegEx Pattern Remarks
\b
 word boundary
\B
 not word boundary
\A
 start of text (^ is an alternative)
\Z
 end of text ($ is an alternative)

These markers are combined with the search string to specify where exactly you want the search string to be. For example, \bhis\b will search for a whole word his, but will ignore this, history or whistle.

Iterators (Quantifiers)

Iterators (quantifiers) are meta-characters that specify how many times the preceding expression has to repeat, A typical example is to find a 3-to-5 digit number.

RegEx newbies often place the iterators after the character that needs to repeat. Just remember that RegEx syntax is exact opposite of the usual English syntax. So, instead of "four dogs", we would have to say "dogs four", RegEx-style.

Iterators can be 'Greedy' or 'Non-Greedy'. Greedy means the expression grabs as much matching text as possible. In contrast, the non-greedy expression tries to match as little as possible.

For example,

  • when b+ (a greedy expression) is applied to string abbbbc, it returns bbbb,
  • but when b+? (a non-greedy expression) is applied to abbbbc, it returns only b.

Note that a ? attached to a greedy expression makes it non-greedy.

RegEx pattern Remarks Greedy? Remarks
*
 zero or more
Yes
 equivalent to {0,}
+
 one or more
Yes
 equivalent to {1,}
?
 zero or one equivalent to {0,1}
{n}
 exactly n times
Yes
{n,}
 at least n times
Yes
{n,m}
 at least n but not more than m times
Yes
*?
 zero or more
No
 equivalent to {0,}?
+?
 one or more
No
 equivalent to {1,}?
??
 zero or one
No
 equivalent to {0,1}?
{n}?
 exactly n times
No
{n,}?
 at least n times
No
{n,m}?
 at least n but not more than m times
No

Let us see some examples:

RegEx pattern Remarks
foob.*r
 matches foobar, foobalkjdflkj9r and foobr
foob.+r
 matches foobar, foobalkjdflkj9r but not foobr
foob.?r
 matches foobar, foobbr and foobr but not foobalkj9r
fooba{2}r
 matches foobaar
fooba{2,}r
 matches foobaar, foobaaar, foobaaaar etc. but not foobar
fooba{2,3}r
 matches foobaar, or foobaaar but not foobaaaar or foobar

Alternatives

A RegEx expression can have multiple alternative characters or subexpressions. The metacharacter | is used to separate the alternatives.

For example, fee|fie|foe will match with fee, fie, or foe in the target string.

It is difficult to understand where each alternative starts and ends. This is why it is a common practice to include alternatives in parentheses, to make it easier to understand.

For example, fee|fie|foe can be written as f(e|i|o)e, to make it easier to understand.

Alternatives are tried from left to right, so the first alternative found for which the entire expression matches, is the one that is chosen. For example, when matching foo|foot against barefoot, only the foo part will match, because that is the first alternative tried, and it successfully matches the target string. (This is important when you are capturing matched text using parentheses.)

RegEx Pattern Remarks
foo(bar|foo) matches foobar or foofoo

Also remember that alternatives cannot be used inside a character class (square brackets), because | is interpreted as a literal within []. That means [fee|fie|foe] is same as [feio|]. (The other characters are treated as duplicates, and ignored).

Subexpressions

Parts of any RegEx pattern can be enclosed in brackets (), just like using brackets in a mathematics formula. Each part that is enclosed in brackets is called a "subexpression".

The brackets serve two main purposes:

  • Better readability, as in the mathematical formula a+(b+c).
  • Make a functional group, as in the mathematical formula a(b+c). This group is evaluated first.

Let us see some examples:

RegEx Pattern Remarks
(fee)|(fie)|(foe)
 Much better readability than the equivalent RegEx pattern fee|fie|foe.
(foobar){2,3}
 Matches with the entire enclosed string foobar repeated 2 or 3 times.

(i.e., matches with foobarfoobar or foobarfoobarfoobar)
(The iterator acts on the entire subexpression. Compare with the example below!)

foobar{2,3}
 Matches with fooba followed by the character r repeated 2 or 3 times.
(i.e., matches with foobarr or foobarrr) (The iterator acts only on the last character.)
foob([0-9]|a+)r matches only the character foob0r, foob1r, foobar, foobaar, foobaaaar, etc.
(The subexpression is evaluated first.)

Backreferences

You must have told (or heard-) jokes like this one:

"Two guys walk in a bar. The first guy says.... Then the second guy replies....".

Then you are already familiar with backreferences!

A "backreference" is a numbered reference to a previously mentioned thing.

RegEx also has backreferences. Let us understand how backreferences are defined in RegEx.

The RegEx engine tries to find text that matches the whole RegEx pattern. If a matching text is found, the RegEx engine identifies the matching text for each of the subexpressions in the pattern.

At this stage, the RegEx engine gives numbers to these matching parts:

  • The text that matches the entire RegEx expression takes the number '0'.
  • The text matching any subexpression is given a number based on the position of that subexpression inside the pattern. In other words, text matching the nth subexpression will take the number 'n'.

Now we use those numbers to refer to the entire pattern and/or subexpressions. (That is why these numbers are called "backreference".)

The backreference to the nth subexpression is written as \n.

The backreferences can be used to compose the RegEx pattern itself, as shown below:

(.)\1+
 matches aaaa and cc (any single character that is repeated twice or more)
(.+)\1+
 matches aaaa, cc, abababab, 123123

(a set of one or more characters, repeated twice or more)

(The character-sets are alternately colored blue and pink for easy identification. Observe how a RegEx pattern can match quite different text!)

Substitution of text using backreference

The backreferences are also used in find-and-replace operations, to re-assemble new text from old.

  • The expressions \1 through \9 serve as backreferences to the subexpressions found in the RegEx pattern. The expression \0 is used to represent the text that matches the whole RegEx pattern. These are used in the "find" part of the operation.
  • The expressions $1 through $9 represent the actual text that matches the respective subexpressions. These are used in the "replace" part of the operation.
  • The expressions $0 refers to the whole original name. Note: it is not necessary to enclosed them in round brackets () for this use, $0 is just there.

The replacement text is typically a combination of-

  • The text that matched the subexpressions, and
  • Some new text.

Note that the RegEx pattern may have some parts that are not enclosed in (). (In other words, it may have parts that are not subexpressions.) Such parts are not used in the replacement text.

Here are some "find-and-replace" examples:

Expression Replace Description
(.*) (.*)
$2, $1
 Switch two words around and put a comma after the resulting first word. Example: if input string is "John Smith", then output will be "Smith, John".

Notice that the replacement text also has additional literal text in the middle (comma and space).

\b(\d{2})-(\d{2})-(\d{4})\b
$3-$2-$1
 Find date sequences in dd-mm-yyyy format and reverse them into yyyy-mm-dd format.
(e.g. 25-10-2007 is converted to 2007-10-25).

Note: This is not a very robust example, because \d can represent any digit in range of 0-9. That means sequences like 99-99-9999 also will match this pattern, resulting in a problem. This in fact shows that you need to be careful with RegEx patterns!

\[.*?\]
 Remove the contents of the [...] (square brackets), and the brackets too.
(Replace with nothing means deleting.)

Limitations for binary data

One of the known limitation of RegEx engine when working with binary data is that the input string is not searched beyond the first occurrence of NULL character (\x00). This would not affect file names because there are simply no NULL characters in them, but may affect parsing of binary content of files when working in Pascal Script for example.

External links

Retrieved from "http://www.den4b.com/w/index.php?title=ReNamer:Regular_Expressions&oldid=2358"