Regex to compare strings with Umlaut and non-Umlaut variations

Question

Can anyone help me with a javascript regular expression that I can use to compare strings that are the same, taking into acccount their non-Umlaut-ed versions.

for example, in German the word Grüße can also be written Gruesse. These two strings are to be considered identical. The mappings (ignoring casings for the moment) are:

• ä = ae
• ü = ue
• ö = oe
• ß = ss

As there are not many "couplets" to consider I could do a replace for each variation, but I'm wondering if there is a more elegant way, especially as this use case might need to be extended in future to include e.g. Scandanavian characters...

Answer 1

something like

tr = {"ä":"ae", "ü":"ue", "ö":"oe", "ß":"ss" }

replaceUmlauts = function(s) {
    return s.replace(/[äöüß]/g, function($0) { return tr[$0] })
}

compare = function(a, b) {
    return replaceUmlauts(a) == replaceUmlauts(b)
}

alert(compare("grüße", "gruesse"))

you can easily extends this by adding more entries to "tr"

not quite elegant, but works

Answer 2

In addition to stereofrogs answer:

tr = {"\u00e4":"ae", "\u00fc":"ue", "\u00f6":"oe", "\u00df":"ss" }

ersetzeUmlauts = function(s) {
    return s.replace(/[\u00e4|\u00fc|\u00f6|\u00df]/g, function($0) { return tr[$0] })
}

I was dealing with Umlauts in an Aptana/Eclipse script and the normal characters ('ä' etc.) didn't do the trick for me.

Answer 3

I have another way : ( purpose : sorting arrays )

function umlaut(str) {
 return str
  .replace(/Â|À|Å|Ã/g, "A")
  .replace(/â|à|å|ã/g, "a")
  .replace(/Ä/g, "AE")
  .replace(/ä/g, "ae")
  .replace(/Ç/g, "C")
  .replace(/ç/g, "c")
  .replace(/É|Ê|È|Ë/g, "E")
  .replace(/é|ê|è|ë/g, "e")
  .replace(/Ó|Ô|Ò|Õ|Ø/g, "O")
  .replace(/ó|ô|ò|õ/g, "o")
  .replace(/Ö/g, "OE")
  .replace(/ö/g, "oe")
  .replace(/Š/g, "S")
  .replace(/š/g, "s")
  .replace(/ß/g, "ss")
  .replace(/Ú|Û|Ù/g, "U")
  .replace(/ú|û|ù/g, "u")
  .replace(/Ü/g, "UE")
  .replace(/ü/g, "ue")
  .replace(/Ý|Ÿ/g, "Y")
  .replace(/ý|ÿ/g, "y")
  .replace(/Ž/g, "Z")
  .replace(/ž/, "z"); 
}

Answer 4

Regular expressions aren't quite powerful enough to do this properly, though you could hack it into almost working with them.

What you want is called Unicode Normalization. A Normalized string is one converted to a common form so you can compare them. You tagged your post "javascript", however, Javascript doesn't have a built in standard library to do this, and I am not aware of one offhand. Most server-side languages do have one, though. For example, the Normalizer Class in PHP. Python and Perl have equivalents, as do Microsoft stuff, I'm sure.

Check out the wikipedia article on Unicode Equivalence for more information.

Answer 5

You can use pipe as an or in a group for each matching like this (ä|ae).

Answer 6

One way is to process your regexp 'input' so that it replaces for example 'ä' with (ae|ä)' - not hardcode the mappings to your regexps. I am completely ignorant to javascript (ok, i know document.write() but that's about it) - but here is the same in pseudo code;

instead of doing

regexp_match("Grüße|Gruesse",somestring)

You should do something like:

mappings = (("ä","ae"),("ö","oe"),("ü","ue"))
def my_regexp_match(regexp,input) {
    for key,value in mappings {
         new_regexp = regexp.replace(key,"("+key+"|"+value+")")
    }
    regexp_match(new_regexp,input)
}
my_regexp_match("Grüße",somestring)

Sorry for being so "pythonic" - I do not know if you have re.compile() -like structure in javascript, but if you do - you should do the for -loop when compiling the matcher, not in my_regexp_match()