php iconv translit for removing accents: not working as excepted?

Question

consider this simple code:

echo iconv(\'UTF-8\', \'ASCII//TRANSLIT\', \'è\');

it prints

 `e

instead of ju

Answer 1

It happen with me with pure iconv without php. The Trick was to set LANG environment value to en_US.UTF-8 (it was hu_HU.UTF-8 before, in my case). After it worked as expected.

Answer 2

I'm tempted to say "nothing", although this is a little outside my expertise. PHP's iconv() is notorious, and the inspiration for many workarounds, including

• dropping to the system's iconv utility (Unix & Linux)
• crafting a lookup table
• replacing all accented characters with an ASCII equivalent as kind of a preprocessing stage
• setting LC_COLLATE (which doesn't seem to work for everyone)
• use htmlentities() instead of iconv()

Read the comments for iconv() documentation for more inspiration. (Or commiseration. Too close to call.)

Answer 3

It seems the standard way to handle this is with a "removing accents" function which you can find in library's like flourish or CMS's like Wordpress. Iconv seems to be unable to translate accents (and rightly so) since this isn't a good idea for anything other than URL slugs.

Answer 4

I have this standard function to return valid url strings without the invalid url characters. The magic seems to be in the line after the //remove unwanted characters comment.

This is taken from the Symfony framework documentation: http://www.symfony-project.org/jobeet/1_4/Doctrine/en/08 which in turn is taken from http://php.vrana.cz/vytvoreni-pratelskeho-url.php but i don't speak Czech ;-)

function slugify($text)
{
  // replace non letter or digits by -
  $text = preg_replace('#[^\\pL\d]+#u', '-', $text);

  // trim
  $text = trim($text, '-');

  // transliterate
  if (function_exists('iconv'))
  {
    $text = iconv('utf-8', 'us-ascii//TRANSLIT', $text);
  }

  // lowercase
  $text = strtolower($text);

  // remove unwanted characters
  $text = preg_replace('#[^-\w]+#', '', $text);

  if (empty($text))
  {
    return 'n-a';
  }

  return $text;
}

echo slugify('é'); // --> "e"

Answer 5

When doing transliteration, you have to make sure that your LC_COLLATE is properly set, otherwise the default POSIX will be used.

Look at http://uk3.php.net/manual/en/function.setlocale.php

Answer 6

cf @tchrist, with INTL php extension

http://fr2.php.net/manual/en/book.intl.php

preg_replace('/\pM*/u','',normalizer_normalize( $mystring, Normalizer::FORM_D));

eéèêëiîïoöôuùûüaâäÅ Ἥ ŐǟǠ ǺƶƈƉųŪŧȬƀ␢ĦŁȽŦ ƀǖ becomes

eeeeeiiiooouuuuaaaA Η OaA AƶƈƉuUŧOƀ␢ĦŁȽŦ ƀu

---

As tchrist emphasises, not all unicode characters are considered decomposable:

extract from Unicode charts:

U0080.pdf

00CF Ï LATIN CAPITAL LETTER I WITH DIAERESIS

≡ 0049 I 0308 ¨

NB this symbol « ≡ » indicate an available decomposition

00D0 Ð LATIN CAPITAL LETTER ETH

→ 00F0 ð latin small letter eth

→ 0110 Đ latin capital letter d with stroke

→ 0189 Ɖ latin capital letter african d

no decomposition available, IMHO strangely (we could consider ASCII letter D as an acceptable equivalent).

U0100.pdf

0110 Đ LATIN CAPITAL LETTER D WITH STROKE

→ 00D0 Ð latin capital letter eth

→ 0111 đ latin small letter d with stroke

→ 0189 Ɖ latin capital letter african d

even stranger: this one is identified as LATIN CAPITAL LETTER D (with stroke), but not decomposable as such! Perhaps a cooler solution should be to get the unicode description of each char, and compare it with the description of each ascii char (and replace accordingly). Anyone? ;-]

cf http://unicode.org/Public/UNIDATA/UnicodeData.txt