5 ms·

Regular Expressions in CoffeeScript are Awesome

45 points by elijahmanor 15y ago

gmac 15y ago
Perl, Ruby and some other languages support a similar 'extended' RegEx syntax which allows comments and disregards whitespace (in Ruby, you use the 'x' flag on a Regexp literal). For those using plain JavaScript -- or wanting to do a one-off conversion -- I wrote a simple JS function that converts an extended RegEx to a plain one: http://blog.mackerron.com/2010/08/08/extended-multi-line-js-regexps/ http://blog.mackerron.com/2010/08/08/extended-multi-line-js-...
- elijahmanor 15y ago
  Yeah, my guess is that CoffeeScript took direction from those languages for this feature. Nice function to convert annotated string to a regex. I wonder if this is being considered in the next versions of JavaScript? Or better yet... native annotation? I like the CoffeeScript approach since that logic is done at compile time and not when it is being executed.
peter_l_downs 15y ago
Python, too: http://docs.python.org/library/re.html#re.VERBOSE http://docs.python.org/library/re.html#re.VERBOSE a = re.compile(r"""\d + # the integral part \. # the decimal point \d * # some fractional digits""", re.X) b = re.compile(r"\d+\.\d*")
- elijahmanor 15y ago
  Ohh nice, maybe there is where CoffeeScript got inspiration from? The syntax looks very similar.
  - berntb 15y ago
    Afaik Python, like most open source languages, use the PCRE lib, Perl Compatible Regular Expressions. Google it. PCRE isn't as good as Perl's regexps, but not bad. (I've helped people with PCRE problems that worked in Perl.) Edit: I googled myself. :-) The Python lib isn't pcre (anymore), it is something home rolled. The answer to the parent question about inspiration is: Most everyone has copied the Perl syntax (except, irritatingly, Emacs). Edit 2: herge -- Perl is ~ a superset of awk. Use that. (Check -n and -e flags in perldoc perlrun.)
    - herge 15y ago
      Vim and awk use the old syntax, and that irritates me to no end.
samarudge 15y ago
You shouldn't validate emails by regular expressions since, technically (at least according to spec) "()<>[]:;@,\\\"!#$%&'*+-/=?^_`{}| ~ ? ^_`{}|~."@example.org is a perfectly valid email address [1] [1] http://en.wikipedia.org/wiki/Email_address#Valid_email_addresses http://en.wikipedia.org/wiki/Email_address#Valid_email_addre...
- ColMustard 15y ago
  More like you shouldn't validate emails by using crappy regular expressions? AFAIK there are regular expressions that cover all valid email addresses (and some invalid, but better than not recognizing one that is valid). Although it gets pretty nasty: http://www.diablotin.com/librairie/autres/mre/chBB.html http://www.diablotin.com/librairie/autres/mre/chBB.html
  - samarudge 15y ago
    But since nearly every character is valid in an email address before the @ sign (within ""), it's much easier to validate `*@thisisavaliddomain.com` and just try sending an email. The way we do it, if the user enters an email that matches our reasonably comprehensive regexp we consider it valid, if it doesn't match we send the user an email and get them to verify it with a link, that way most users don't have to go through an annoying click-and-confirm email validation but we can still catch a few edge cases of people mistyping
- cynoclast 15y ago
  You shouldn't try to validate an email address by any means other than sending mail to it. If it works, it's valid.
phuff 15y ago
In JS you can comment regular expressions, too... :) var foo = new RegExp( '\\d\+' + // First digit '-' + // Delimiter '\\d\+' // Second digit ); foo.test("1234-5432");
- elijahmanor 15y ago
  True, but then you have the overhead of string concatenation... although in the large scheme of things that is probably very small, unless you are doing it in a huge loop, but then you'd cache that RegExp anyway.
  - natrius 15y ago
    Concatenation of string literals can be compiled away.
mike-cardwell 15y ago
His example, slightly modified for Perl: my $emailPattern = qr{ ^ #begin of line ([a-z0-9_.-]+) #one or more letters, numbers, _ . or - @ #followed by an @ sign ([\da-z.-]+) #then one or more letters, numbers, _ . or - \. #followed by a period ([a-z.]{2,6}) #followed by 2 to 6 letters or periods $ }xi; #end of line and ignore case if( 'john.smith@gmail.com' =~ $emailPattern ){ print "E-mail is valid\n"; } else { print "E-mail is invalid\n"; } EDIT: A much better way of doing it though: use Mail::Sendmail; if( 'john.smith@gmail.com' =~ /$Mail::Sendmail::address_rx/ ){ print "E-mail is valid\n"; } else { print "E-mail is invalid\n"; }
- elijahmanor 15y ago
  Nice! I like it ;)
- draegtun 15y ago
  And for just email address checking then even better IMHO is: use Email::Valid; if (Email::Valid->address('john.smith@gmail.com')) { print "E-mail is valid\n"; } else { print "Email is invalid\n" } ref: https://metacpan.org/module/Email::Valid https://metacpan.org/module/Email::Valid
shtylman 15y ago
I personally don't find this example very helpful in showing why I would do this. The regex is simple enough and to me the comments are the equivalent of "return here", needless and distracting.
- elijahmanor 15y ago
  The example was to show how you could do it, not necessarily that it was needed for that case. However, you might be surprised how many developers aren't familiar with that example regex.
jashkenas 15y ago
For a little dollop of recursivity in your morning coffee, here's the bit where the CoffeeScript compiler uses these extended regular expressions to lex, among other things, extended regular expressions... https://github.com/jashkenas/coffee-script/blob/master/src/lexer.coffee#L595-647 https://github.com/jashkenas/coffee-script/blob/master/src/l...
joshuahedlund 15y ago
I know the email validation was just an example, but since we're on the subject, if you've got an email validation that includes a maximum number of characters for the TLD you might want to update that before the new TLDs start rolling in later this year or next.
jriddycuz 15y ago
"Let's face it, regular expressions aren't for everyone." Wait, since when aren't they? I would have thought basic regex skills were a baseline shibboleth of a programmer. I even know several non-technical people that are proficient.
- Zancarius 15y ago
  Your comment reminded me of Jamie Zawinski's tongue-in-cheek quotation "Some people, when confronted with a problem, think 'I know, I'll use regular expressions.' Now they have two problems." (http://regex.info/blog/2006-09-15/247 http://regex.info/blog/2006-09-15/247) Jestful, off-the-wall comments aside, I love regex. It takes away a great deal of the drudgery associated with processing volumes and volumes of text. I can't imagine not having it. (Aside: Now that you mention it, I can think of a couple people who aren't programmers who use regex pretty frequently because it makes their life a little easier even if it isn't highly advanced, super-dense regex.)
SeoxyS 15y ago
That has nothing to do CoffeeScript, most regex parser implementation support this via a flag. I've seen it done in PHP, Ruby, Perl, Objective-C. For languages whose regular expression engine don't support this, it's easy enough to achieve it via string concatenation.
- RyanMcGreal 15y ago
  Python has it, too.
mrpollo 15y ago
is there any performance penalty?
- tikhonj 15y ago
  I'm pretty sure that these expressions are simplified to normal JavaScript expressions at compile time, so there is no performance penalty. Even if they were compiled at runtime, the performance penalty would probably be negligible unless you were randomly compiling the expression in the middle of an inner loop or something equally silly.
buddydvd 15y ago
What about trailing spaces? They need to be escaped by parenthesis?

45 points by elijahmanor 15y ago

gmac 15y ago

Perl, Ruby and some other languages support a similar 'extended' RegEx syntax which allows comments and disregards whitespace (in Ruby, you use the 'x' flag on a Regexp literal). For those using plain JavaScript -- or wanting to do a one-off conversion -- I wrote a simple JS function that converts an extended RegEx to a plain one: http://blog.mackerron.com/2010/08/08/extended-multi-line-js-regexps/ http://blog.mackerron.com/2010/08/08/extended-multi-line-js-...

elijahmanor 15y ago
Yeah, my guess is that CoffeeScript took direction from those languages for this feature. Nice function to convert annotated string to a regex. I wonder if this is being considered in the next versions of JavaScript? Or better yet... native annotation? I like the CoffeeScript approach since that logic is done at compile time and not when it is being executed.

peter_l_downs 15y ago

Python, too: http://docs.python.org/library/re.html#re.VERBOSE http://docs.python.org/library/re.html#re.VERBOSE a = re.compile(r"""\d + # the integral part \. # the decimal point \d * # some fractional digits""", re.X) b = re.compile(r"\d+\.\d*")

elijahmanor 15y ago
Ohh nice, maybe there is where CoffeeScript got inspiration from? The syntax looks very similar.
- berntb 15y ago
  Afaik Python, like most open source languages, use the PCRE lib, Perl Compatible Regular Expressions. Google it. PCRE isn't as good as Perl's regexps, but not bad. (I've helped people with PCRE problems that worked in Perl.) Edit: I googled myself. :-) The Python lib isn't pcre (anymore), it is something home rolled. The answer to the parent question about inspiration is: Most everyone has copied the Perl syntax (except, irritatingly, Emacs). Edit 2: herge -- Perl is ~ a superset of awk. Use that. (Check -n and -e flags in perldoc perlrun.)
  - herge 15y ago
    Vim and awk use the old syntax, and that irritates me to no end.

samarudge 15y ago

You shouldn't validate emails by regular expressions since, technically (at least according to spec) "()<>[]:;@,\\\"!#$%&'*+-/=?^_`{}| ~ ? ^_`{}|~."@example.org is a perfectly valid email address [1] [1] http://en.wikipedia.org/wiki/Email_address#Valid_email_addresses http://en.wikipedia.org/wiki/Email_address#Valid_email_addre...

ColMustard 15y ago
More like you shouldn't validate emails by using crappy regular expressions? AFAIK there are regular expressions that cover all valid email addresses (and some invalid, but better than not recognizing one that is valid). Although it gets pretty nasty: http://www.diablotin.com/librairie/autres/mre/chBB.html http://www.diablotin.com/librairie/autres/mre/chBB.html
- samarudge 15y ago
  But since nearly every character is valid in an email address before the @ sign (within ""), it's much easier to validate `*@thisisavaliddomain.com` and just try sending an email. The way we do it, if the user enters an email that matches our reasonably comprehensive regexp we consider it valid, if it doesn't match we send the user an email and get them to verify it with a link, that way most users don't have to go through an annoying click-and-confirm email validation but we can still catch a few edge cases of people mistyping
cynoclast 15y ago
You shouldn't try to validate an email address by any means other than sending mail to it. If it works, it's valid.

phuff 15y ago

In JS you can comment regular expressions, too... :) var foo = new RegExp( '\\d\+' + // First digit '-' + // Delimiter '\\d\+' // Second digit ); foo.test("1234-5432");

elijahmanor 15y ago
True, but then you have the overhead of string concatenation... although in the large scheme of things that is probably very small, unless you are doing it in a huge loop, but then you'd cache that RegExp anyway.
- natrius 15y ago
  Concatenation of string literals can be compiled away.

mike-cardwell 15y ago

His example, slightly modified for Perl: my $emailPattern = qr{ ^ #begin of line ([a-z0-9_.-]+) #one or more letters, numbers, _ . or - @ #followed by an @ sign ([\da-z.-]+) #then one or more letters, numbers, _ . or - \. #followed by a period ([a-z.]{2,6}) #followed by 2 to 6 letters or periods $ }xi; #end of line and ignore case if( 'john.smith@gmail.com' =~ $emailPattern ){ print "E-mail is valid\n"; } else { print "E-mail is invalid\n"; } EDIT: A much better way of doing it though: use Mail::Sendmail; if( 'john.smith@gmail.com' =~ /$Mail::Sendmail::address_rx/ ){ print "E-mail is valid\n"; } else { print "E-mail is invalid\n"; }

elijahmanor 15y ago
Nice! I like it ;)
draegtun 15y ago
And for just email address checking then even better IMHO is: use Email::Valid; if (Email::Valid->address('john.smith@gmail.com')) { print "E-mail is valid\n"; } else { print "Email is invalid\n" } ref: https://metacpan.org/module/Email::Valid https://metacpan.org/module/Email::Valid

shtylman 15y ago

I personally don't find this example very helpful in showing why I would do this. The regex is simple enough and to me the comments are the equivalent of "return here", needless and distracting.

elijahmanor 15y ago
The example was to show how you could do it, not necessarily that it was needed for that case. However, you might be surprised how many developers aren't familiar with that example regex.

jashkenas 15y ago

For a little dollop of recursivity in your morning coffee, here's the bit where the CoffeeScript compiler uses these extended regular expressions to lex, among other things, extended regular expressions... https://github.com/jashkenas/coffee-script/blob/master/src/lexer.coffee#L595-647 https://github.com/jashkenas/coffee-script/blob/master/src/l...

joshuahedlund 15y ago

I know the email validation was just an example, but since we're on the subject, if you've got an email validation that includes a maximum number of characters for the TLD you might want to update that before the new TLDs start rolling in later this year or next.

jriddycuz 15y ago

"Let's face it, regular expressions aren't for everyone." Wait, since when aren't they? I would have thought basic regex skills were a baseline shibboleth of a programmer. I even know several non-technical people that are proficient.

Zancarius 15y ago
Your comment reminded me of Jamie Zawinski's tongue-in-cheek quotation "Some people, when confronted with a problem, think 'I know, I'll use regular expressions.' Now they have two problems." (http://regex.info/blog/2006-09-15/247 http://regex.info/blog/2006-09-15/247) Jestful, off-the-wall comments aside, I love regex. It takes away a great deal of the drudgery associated with processing volumes and volumes of text. I can't imagine not having it. (Aside: Now that you mention it, I can think of a couple people who aren't programmers who use regex pretty frequently because it makes their life a little easier even if it isn't highly advanced, super-dense regex.)

SeoxyS 15y ago

That has nothing to do CoffeeScript, most regex parser implementation support this via a flag. I've seen it done in PHP, Ruby, Perl, Objective-C. For languages whose regular expression engine don't support this, it's easy enough to achieve it via string concatenation.

RyanMcGreal 15y ago
Python has it, too.

mrpollo 15y ago

is there any performance penalty?

tikhonj 15y ago
I'm pretty sure that these expressions are simplified to normal JavaScript expressions at compile time, so there is no performance penalty. Even if they were compiled at runtime, the performance penalty would probably be negligible unless you were randomly compiling the expression in the middle of an inner loop or something equally silly.

buddydvd 15y ago

What about trailing spaces? They need to be escaped by parenthesis?