PCRE with variable data

silent

I am extremely rusty using regex...
The data is basically this :

(Some question here -- may contain numbers)
a. An answer
b. another answer

and so on... the problem I am having is that the number of answers varies...
any ideas on a regex that parse this?

sneakyimp

assuming the newlines are there and stuff....hmmmmm

if the data is really consistent in structure, try this...i had to use two patterns.

<?

$haystack = "1. (Some question here -- may contain numbers)
a. An answer
b. another answer
c. a third answer

2. another question
a. answer here
b. another answer
c. a third answer";

// this pattern matches a question plus answers
$pattern = "#(\d+\.[^\n]+\n)([a-z]\.[^\n^\r]+\n*)+#is";

if (preg_match_all($pattern, $haystack, $matches)) {
  echo 'questions found:' . +sizeof($matches[0]) . '<br>';

  //$matches[1] has the questions
  $questions = array();
  foreach($matches[1] as $question) {
    $questions[] = $question;
  }

  $answer_pattern = "#^[a-z]\.[^\n]+#mis";
  $answers = array();
  foreach($matches[0] as $key => $match0) {
    // we need to extract the questions from each match
    if (preg_match_all($answer_pattern, $match0, $answer_matches)) {
      foreach($answer_matches[0] as $akey => $answer) {
        $answers[$key][$akey]= $answer;
      }
    }
  }
} else {
  echo 'no match';
}


$num = sizeof($questions);
for($i=0; $i<$num; $i++) {
  echo $questions[$i] . "<br>";
  foreach($answers[$i] as $answer) {
    echo $answer . "<br>";
  }
  echo "</p>";
}
?>

silent

I really like the method you are using, but it's only returning the first answer for each question...

output:

questions found:2
1. (Some question here -- may contain numbers) 
a. An answer 


2. another question 
a. answer here

any ideas?

sneakyimp

it works for me:

http://adfotos.com/demo/chump2.php

silent

That's weird... I think my PCRE lib is jacked up... i don't think it was compiled with support for the # operator...

I'll have to check... but thanks for the help!

Weedpacket

silent wrote:
i don't think it was compiled with support for the # operator...

It's not an operator. It's just the symbol sneakyimp picked to mark the start and end of the regexp. Usually it's /, but there's no law that says it has to be.

I'd be more likely to start with something like

$q_and_a = preg_split("/\n(?=\d)/", $haystack);

Assuming that each answer and each question is on a single line. (That really would be easier, otherwise it could get hairy trying to tell where one answer leaves off and the next begins.

I'd prefer to split on a double-linebreak (and in PHP5 that wouldn't even need regular expressions), and if you know what line-ending convention (Unix vs. Windows) this thing will be running under that's what I'd go for.

In fact, I'd probably go for that anyway:

$linebreak="
";

preg_split("/$linebreak$linebreak/", $haystack);