Catholicism Computes » Substrings in Ruby

Substrings in Ruby

June 3, 2007 at 12:33

Filed under: Computing — Pistos @ 12:33

Most Ruby programmers know how to get substrings out of strings using the ways described in the Ruby core documentation for the String class. For reference, I quote:

   a = "hello there"
   a[1]                   #=> 101
   a[1,3]                 #=> "ell"
   a[1..3]                #=> "ell"
   a[-3,2]                #=> "er"
   a[-4..-2]              #=> "her"
   a[12..-1]              #=> nil
   a[-2..-4]              #=> ""
   a[/[aeiou](.)\1/]      #=> "ell"
   a[/[aeiou](.)\1/, 0]   #=> "ell"
   a[/[aeiou](.)\1/, 1]   #=> "l"
   a[/[aeiou](.)\1/, 2]   #=> nil
   a["lo"]                #=> "lo"
   a["bye"]               #=> nil

I almost always use the string[ /regexp/, 1 ] method, myself. However, all of the above only let you extract one substring. We can do multi-variable assignment in Ruby:

a, b, c = 1, 2, 3
  # a == 1; b == 2; c == 3
a, b = b, a
  # a == 2; b == 1

So why can’t we do multi-variable substring extraction? Ah, but we can! As described in the MatchData documentation:

multi-variable substring extraction

all, first, second = *(
    /(\w+) +(\w+)/.match "Here is a sentence."
)
  # all == "Here is"
  # first == "Here"
  # second == "is"

Get the same result the other way around, if this ordering makes more sense to you:

all, first, second = *(
    "Here is a sentence.".match /(\w+) +(\w+)/
)
  # all == "Here is"
  # first == "Here"
  # second == "is"

Now, what we really need is a way to do it without having to give a container for the whole match (“all” in the above code), preferably something with more concise syntax. Wouldn’t this be neat:

# This is not valid Ruby code!
first, second =
  "Here is a sentence."[ /(\w+) +(\w+)/ ]

No related posts.

Comments (7) Permalink TrackBack

7 Comments »

manveru says:

June 4, 2007 at 05:35

first, second = ‘this is a’.scan(/w+/)[0,2]

["this", "is"]
Pistos says:

June 4, 2007 at 09:20

manveru:

Your backslash got swallowed up by my Markdown plugin:
```
'this is a'.scan(/\w+/)[0,2]
```
But anyway: Thanks for pointing out this technique. However, I’m not sure it would let us grab arbitrary regexp groups, such as:
```
weekday, month, monthday, year =
"Monday, June 4, 2007"[
  /^(\w+), (\w+) (\d+), (\d+)/
]
```
manveru says:

June 4, 2007 at 11:11

Yeah, i noticed that but didn’t want to double-post.
To parse something like this i would use:
```
  string = 'Monday, June 4, 2007'
  format = '%A, %B %d, %Y'
  date = Date.strptime(string, format)
  puts date
  # => 2007-06-04
```
But i get your point, groups would be quite handy at times instead of:
```
  'Monday, June 4, 2007'.match(/^(\\w+), (\\w+) (\\d+), (\\d+)/).captures
  # ["Monday", "June", "4", "2007"]
```
This time i even previewed the post
One thing i see though is that markdown doesn’t work like the docs say, 4 spaces indentation should result in a code-block.
Pistos says:

June 4, 2007 at 11:38

Hrm:
```
  test code
  right here.
```
Seems to work for me? The way I’m getting the highlighted code blocks is actually via the syntax highlight Wordpress plugin. Add code blocks on my blog with <pre lang=”ruby”>.
manveru says:

June 4, 2007 at 12:16

Yes, but it still tries to remove the \
Pistos says:

June 4, 2007 at 12:53

Oh, I see. Well… I guess we need to use <pre> and preview.
Pistos says:

June 10, 2009 at 08:07

Ruby 1.9 has named captures. With this we can name capture groups in a regexp, and then local variables are assigned the captured strings.
```
s = "a123 bxyz"
/a(?<my_a>\S+) +b(?<my_b>\S+)/ =~ s
my_a #=> "123"
my_b #=> "xyz"
```