Regular Expression - replace word except within a

Question

Writing a globalization module for a web application and I need a regexp to replace all instances of a word with another word (the translation) - except - words found within a URL/URI.

EDIT: I forgot to mention that I'm using Ruby, so I can't use 'Lookbehind'

Answer 1

• Split on URI regular expression; include the URI's in the result.
• For each piece:
  • if it is a URI, leave it alone
  • otherwise, do word replacement
• Join the pieces

Code:

# From RFC 3986 Appendix B, with these modifications:
#   o Spaces disallowed
#   o All groups non-matching, except for added outermost group
#   o Not anchored
#   o Scheme required
#   o Authority required
URI_REGEX = %r"((?:(?:[^ :/?#]+):)(?://(?:[^ /?#]*))(?:[^ ?#]*)(?:\?(?:[^ #]*))?(?:#(?:[^ ]*))?)"

def replace_except_uris(text, old, new)
  text.split(URI_REGEX).collect do |s|
    if s =~ URI_REGEX
      s
    else
      s.gsub(old, new)
    end
  end.join
end

text = <<END
stack http://www.stackoverflow.com stack
stack http://www.somewhere.come/stack?stack=stack#stack stack
END

puts replace_except_uris(text, /stack/, 'LINKED-LIST')

# => LINKED-LIST http://www.stackoverflow.com LINKED-LIST
# => LINKED-LIST http://www.somewhere.come/stack?stack=stack#stack LINKED-LIST

Answer 2

Have you tried splitting your text into words and iterating over the words? Then you can examine each word, determine if it's a URI, translate it if it isn't.

Answer 3

You can probaby use something like

(?<!://[^ ]*)\bfoo\b

But this probably isn't perfect, it just looks that the word doesn't appear in a single non-whitespace string of characters that don't have :// somewhere before the word.

PS Home:\> "foo foobar http://foo_bar/baz?gak=foobar baz foo" -replace '(?<!://[^ ]*)\bfoo\b', 'FOO'
FOO foobar http://foo_bar/baz?gak=foobar baz FOO