Parse a list of attribute/values in PHP.

Question

Given a string with attribute/value pairs such as

attr1="some text" attr2 = "some other text" attr3= "some weird !@'#$\"=+ text"

the goal is to parse it and output an associative array, in this case:

array('attr1' => 'some text',
      'attr2' => 'some other text',
      'attr3' => 'some weird !@\'#$\"=+ text')

Note the inconsistent spacing around the equal signs, the escaped double quote in the input, and the escaped single quote in the output.

Answer 1

Try something like this:

$text = "attr1=\"some text\" attr2 = \"some other text\" attr3= \"some weird !@'#$\\\"=+ text\"";
echo $text;
preg_match_all('/(\S+)\s*=\s*"((?:\\\\.|[^\\"])*)"/', $text, $matches, PREG_SET_ORDER);
print_r($matches);

which produces:

attr1="some text" attr2 = "some other text" attr3= "some weird !@'#$\"=+ text"

Array
(
    [0] => Array
        (
            [0] => attr1="some text"
            [1] => attr1
            [2] => some text
        )

    [1] => Array
        (
            [0] => attr2 = "some other text"
            [1] => attr2
            [2] => some other text
        )

    [2] => Array
        (
            [0] => attr3= "some weird !@'#$\"=+ text"
            [1] => attr3
            [2] => some weird !@'#$\"=+ text
        )

)

And a short explanation:

(\S+)               // match one or more characters other than white space characters
                    // > and store it in group 1
\s*=\s*             // match a '=' surrounded by zero or more white space characters 
"                   // match a double quote
(                   // open group 2
  (?:\\\\.|[^\\"])* //   match zero or more sub strings that are either a backslash
                    //   > followed by any character, or any character other than a
                    //   > backslash
)                   // close group 2
"                   // match a double quote

Answer 2

EDIT: This regex fails if the value ends in a backslash like attr4="something\\"

I don't know PHP, but since the regex would be essentially the same in any language, this is how I did it in ActionScript:

var text:String = "attr1=\"some text\" attr2 = \"some other text\" attr3= \"some weird !@'#$\\\"=+ text\"";

var regex:RegExp = /\s*(\w+)\s*=\s*(?:"(.*?)(?<!\\)")\s*/g;

var result:Object;
while(result = regex.exec(text))
    trace(result[1] + " is " + result[2]);

And I got the following out put:

attr1 is some text
attr2 is some other text
attr3 is some weird !@'#$\"=+ text