PHP Regex Get the html tag

  // String a ser tratada
  $string = "<table>|<tr>[<td>#VALOR#</td>]</tr>|</table>"; 

  // Expressão regular 
  $regex  = "#\<table\>\|\<tr\>\[\<td\>(.*)\<\/td\>\]\<\/tr\>\|\<\/table\>#"; 

  // Extrai o conteudo
  preg_match_all($regex,$string,$retorno,PREG_PATTERN_ORDER);

  // Valor #VALOR#
  $valor = $retorno[1][0];

  // Exibi o valor
  echo $valor;

I will not go into all the explanations about Do not use REGEX for HTML again.

I think what you want is this:

~<table>.*?(<tr>.*?(<td>(.*?)</td>).*?</tr>).*?</table>~

Explanation

• <table> - literal should start the sentence.
• .*? - anything as little as possible until you fit the next sentence.
• <tr> - literal must have this sentence.
• .*? - anything as little as possible until you fit the next sentence.
• <td> - literal must have this sentence.
• .*? - anything as little as possible until you fit the next sentence. (its value will be here).

With this you have created 4 groups:

• 0 - The captured string itself.
• 1 - From <tr> ... </tr> .
• 2 - From <td> ... </td> .
• 3 - The value of <td> .

Addendum

• Use the modified s to dot capture \n (line break).

Example

REGEX101

You can not do what you're after with just three exact matches. You can not just capture WW from the WAW string in just one capturing group, even using non-capturing groups .

What gives to do, however, is as follows:

$string = "<table>|<tr>[<td>#VALOR#</td>]</tr>|</table>";

$regex = "#(<table>)\|(<tr>)\[<td>([^<]*)<\/td>\](<\/tr>)\|(<\/table>)#";

preg_match($regex, $string, $retorno);

$match1 = $retorno[1] . $retorno[5];
$match2 = $retorno[2] . $retorno[4];
$match3 = $retorno[3];

echo $match1 . "\n";
echo $match2 . "\n";
echo $match3 . "\n";

At the end, the variables $match1 , $match2 and $match3 will have the values <table></table> , <tr></tr> and #VALOR# , respectively, which is what you want.

And you can see the regex working in regex 101 .

Considerations:

The regex assumes that the only variable value in its string is "# VALUE #", which can assume any string that does not have the character < ;

• The regex does not handle blanks. If the string starts with < table> all the captures would fail.