Problem trying to extract words from string in PHP

Question

I'm trying to extract all words from a string into an array, but i am having some problems with spaces ( ).

This is what I do:

//Clean data to text only
$data = strip_tags($data);
$data = htmlentities($data, ENT_QUOTES, 'UTF-8');
$data = html_entity_decode($data, ENT_QUOTES, 'UTF-8');
$data = htmlspecialchars_decode($data);
$data = mb_strtolower($data, 'UTF-8');

//Clean up text from special chrs I don't want as words
$data = str_replace(',', '', $data);
$data = str_replace('.', '', $data);
$data = str_replace(':', '', $data);
$data = str_replace(';', '', $data);
$data = str_replace('*', '', $data);
$data = str_replace('?', '', $data);
$data = str_replace('!', '', $data);
$data = str_replace('-', ' ', $data);
$data = str_replace("\n", ' ', $data);
$data = str_replace("\r", ' ', $data);
$data = str_replace("\t", ' ', $data);
$data = str_replace("\0", ' ', $data);
$data = str_replace("\x0B", ' ', $data);
$data = str_replace(" ", ' ', $data);

//Clean up duplicated spaces
do {
   $data = str_replace('  ', ' ', $data);
} while(strpos($data, '  ') !== false);

//Make array
$clean_data = explode(' ', $data);

echo "<pre>";
var_dump($clean_data);
echo "</pre>";

This outputs:

array(58) {
  [0]=>
  string(5) " "
  [1]=>
  string(5) " "
  [2]=>
  string(11) "anläggning"
  [3]=>
  string(3) "med"
  [4]=>
  string(3) "den"
  [5]=>
  string(10) "erfarenhet"
  [6]=>
  string(3) "som"
}

If i check source for output i see that the first 2 array values is  .
No matter how I try, I can't remove this from the string. Any ideas?

UPDATE:
After some tweaking with code i manage to get following output:

array(56) {
  [0]=>
  string(1) "�" //Notice change. Instead of string length 5 it now says 1. But still its garbage.
  [1]=>
  string(1) "�"
  [2]=>
  string(11) "anläggning"
  [3]=>
  string(3) "med"
  [4]=>
  string(3) "den"
  [5]=>
  string(10) "erfarenhet"
  [6]=>
  string(3) "som"
  [7]=>
  string(5) "finns"
  [8]=>
  string(4) "inom"

Thanks!

ANSWER (for lazy people):

Even thou this is a slightly different approach to the problem, and it never really answers why I had the problems I had above (like leftover   and other extra weird spaces), I like it and it is a lot better than my original code.

Thanks to all who contributed to this!

//Clean data to text only
$data = strip_tags($data);
$data = html_entity_decode($data, ENT_QUOTES, 'UTF-8');
$data = htmlspecialchars_decode($data);
$data = mb_strtolower($data, 'UTF-8');

//Clean up text from special chrs
$data = str_replace(array("-"), ' ', $data);    

$clean_data = str_word_count($data, 1, 'äöå');

echo "<pre>";
var_dump($clean_data);
echo "</pre>";

Answer 1

Ok, the only thing you would have to do is to replace   with a space as you already do (only if the string really still contains   check @Andy E's answer to make sure that that your data does not contain any HTML entities.):

$data = str_replace(" ", ' ', $data);

Then you can use str_word_count to get the words:

$words = str_word_count($data, 1, 'äöåÄÖÅ');

P.S.: What is the sense of calling htmlentities first and then revert it again in with html_entity_decode anyway?

Update: Example:

$str = '      anläggning med den      erfahrenhet som åååÅ ÅÅ';
print_r(str_word_count($str, 1, 'äöåÄÖÅ'));

prints

Array
(
    [0] => anläggning
    [1] => med
    [2] => den
    [3] => erfahrenhet
    [4] => som
    [5] => åååÅ
    [6] => ÅÅ
)

Reading documentation helps :)

Answer 2

Is it possible you're "double encoding" any existing   parts of the string? You call htmlentities on the string before html_entity_decode, so any existing   characters would become  . You can prevent htmlentities from double encoding by providing false as the fourth parameter.

$data = htmlentities($data, ENT_QUOTES, 'UTF-8', false);
$data = html_entity_decode($data, ENT_QUOTES, 'UTF-8');

Also, bare in mind that you can pass an array for matches in str_replace:

$data = str_replace(array(',','.',':',';','*','?','!','-'), '', $data);

Answer 3

Instead of:

14x str_replace

do {
   $data = str_replace('  ', ' ', $data);
} while(strpos($data, '  ') !== false);

do:

$data = preg_replace('/[.*,:;?!]/', '', $data);
$data = preg_replace('/(?:\xC2\xA0|\s{2,}|-)/', ' ', $data);

Whereas 0xC2A0 is the non-breaking space ( ) and \s is any white-space character covering the repeated str_replace calls.

Answer 4

print_r( explode(" ", $data));

Update

define("WORD_COUNT_MASK", "/\p{L}[\p{L}\p{Mn}\p{Pd}'\x{2019}]*/u");

function str_word_count_utf8($str)
{
     preg_match_all(WORD_COUNT_MASK, $str, $matches);
     print_r( $matches);
}
str_word_count_utf8( $str);

Answer 5

$data = '&nbsp; cesadasdsadas <br /> &nbsp; dsadsadas';
$data = preg_replace('/&nbsp;/', ' ', $data);
var_dump($data);

Answer 6

maybe you should try this : http://php.net/manual/en/function.str-word-count.php

I've made something close to your goal recently :

    $words = array_unique(str_word_count($CONTENT." ".$TITLE, 1));
    sort($words);
    $words = addslashes (implode(" ", array_values($words)));

Bye.