Ruby正则表达式/(\\/.*?(?=\\/|$)){2}/

Question

My regexp behaves just like I want it to on http://regexr.com , but not like I want it in irb.

I'm trying to make a regular expression that will match the following:

A forward slash, 
then 2 * any number of random characters (i.e. `.*`), 
  up to but not including another / 
  OR the end of the string (whichever comes first)

I'm sorry as that was probably unclear, but it's my best attempt at an English translation.

Here's my current attempt and hopefully that will give you a better idea of what I'm trying to do:

/(\/.*?(?=\/|$)){2}/

The usage scenario is I want to be able to take a path like /foo/bar/baz/bin/bash and shorten it to the level I'm at in the filesystem, in this case the second level ( /foo/bar ). I'm trying to do this using the command path.scan(-regex-).shift .

Answer 1

The usage scenario is I want to be able to take a path like /foo/bar/baz/bin/bash and shorten it to the level I'm at in the filesystem, in this case the second level (/foo/bar)

Ruby already has a class for handling paths, Pathname . You can use Pathname#relative_path_from to do what you want.

require 'pathname'

path = Pathname.new("/foo/bar/baz/bin/bash")

# Normally you'd use Pathname.getwd
cwd  = Pathname.new("/foo/bar")

# baz/bin/bash
puts path.relative_path_from(cwd)

---

Regexes just invite problems, like assuming the path separator is / , not honoring escapes, and not dealing with extra / . For example, "//foo/bar//b\\\\/az/bin/bash" . // is particularly common in code which joins together directories using paths.join("/") or "#{dir}/#{file} .

For completeness, the general way you match a single piece of a path is this.

%r{^(/[^/]+)}

That's the beginning of the string, a / , then 1 or more characters which are not / . Using [^/]+ means you don't have to try and match an optional / or end of string, a very useful technique. Using %r{} means less leaning toothpicks .

But this is only applicable to a canonicalized path. It will fail on //foo//b\\\\/ar/ . You can try to fix up the regex to deal with that, or do your own canonicalization, but just use Pathname.