Passcode: wv=GWB
Regular expressions refers to a tiny formal language for describing and then acting on (searching, replacing, etc.) regular patterns in text. It is the basis for all sorts of machine parsing of text.
aka “RegEx” or “regex,” variously pronounced.
aka “GREP” after the venerable Unix tool (Global Regular Expression and Print). Technically, GREP is an application that uses regex. Since 2003 the OED has recognized ‘grep’ as both a noun and a verb.
“Wildcards” (as in MSWord) are a similar thing, but different; tends to be ad-hoc, as opposed to the formal and complete system that regex is.
Where will you find regex? Most importantly, in text editors. Sublime Text. BBEdit. Brackets. Whichever.
Regex is in lots of other software, including:
A “regular expression” represents a pattern in text, codified in a tiny formal language.
The simplest kind of pattern is a string – just a sequence of characters, like a word or a sentence, or a phone number. If we search for “foo” in a text, we will find instances of those three letters together.
But we can imagine other kinds of regular patterns that are more interesting and/or useful to us in our work: Like, all the things that are capitalized. Words that end in “ing.” URLs. XML tags. Things that should be tabular but aren’t. Phone numbers; dates; timecodes...
Regex gives us a handle on such things. It allows us to describe abstract patterns in text, and then grab hold of those things and do things with them – whether that’s just finding them or doing something more active, like changing them, or collecting them, or doing some other processing on them.
This little formal language is actually very simple. There are only about 9 things to remember.
(Chapter 8 in the BBEdit user manual is excellent – the best single reference I’ve seen.
As of April 2021, it’s at https://s3.amazonaws.com/BBSW-download/BBEdit_13.5.6_User_Manual.pdf)
A B C a b c
1 2 3
Horatio, orange, apple
\t (tab)
\d \w \s .
(any digit, any 'word' element, any space, and any character at all)
? * +
(optional; optional and repeatable; repeatable)
(0 or 1; 0 or more; 1 or more)
a|b|c
this|that
(apples|oranges)|vegetables
(Jan|Feb|Mar|Apr|May) (\d+)
[abcedfg]
[A-Za-z0-9]
^ $
(match at start of line; end of line)
\? \. \+
\[ \( \)
pattern+ (='greedy')
pattern+? (='non-greedy')
...very useful when matching pairs of quotation marks, or nested HTML tags.
\1 \2 \3
= 1st, 2nd, 3rd (etc) sets of matched parentheses in search string can be *replaced* individually.
There are some other pieces, but they’re specific enough that you can google them if you need them.
Because of the tangled history of computing, there are actually THREE different legacy ways to represent a line break.1 Most software is smart enough to be entirely agnostic about them, so this doesn’t present a problem 99% of the time.
But if we’re searching, we have to be very specific, so you need to know that line breaks come in these flavours:
\n (the Unix-derived LF "line feed" character)
\r (the old Apple CR "carriage return" character)
\r\n (microsoft's excellent compromise)
Bottom line: if you’re searching for line endings with \n and it doesn’t work, try \r. And vice-versa.
ALSO: Old text files – such as those from Project Gutenberg – split paragraphs into lines of 70-80 characters. More modern practice, assuming that software can ‘wrap’ lines automatically, will leave an entire paragraph to a single logical ‘line’ (that is, with an LF and/or CR at the end of the whole paragraph).
Again, being mindful of these distinctions will help you figure out how to make your regex patterns do what you want them to do.
https://regex101.com/ is a website that provides an interactive environment for building regexes and testing them.
Try your skill with these Regex Exercises… better still, try your skill at saying that five times fast.