discover.php

<?php
error_reporting(E_ALL|E_STRICT);

$input = "david@davidrecordon.com";
$input = "https://davidrecordon.com";
$input = "www.davidrecordon.com";

$parsed = parse_input($input);
$after  = normalize_input($parsed);
echo $input . ' -> ' . $after . "\n";

$curl = curl_init();
if (!$curl) {
  die("No curl!\n");
}
curl_setopt($curl, CURLOPT_RETURNTRANSFER, true);

$user_endpoint = discover_endpoint($curl, $parsed['scheme'], $parsed['domain'], $after);
echo "  endpoint: $user_endpoint\n";


function parse_input($input) {
  $url = @parse_url($input);
  if (!$url) {
    die("That isn't a valid identifier!\n");
  }
  if (isset($url['user']) || isset($url['pass'])) {
    die("We don't support username:password syntax.\n");
  }
  if (isset($url['fragment'])) {
    die("We don't support URL fragments.\n");
  }
  if (isset($url['query'])) {
    die("Query parameters in OpenIDs are a bad idea!\n");
  }

  $identifier = array(
    'scheme' => NULL,
    'user'   => NULL,
    'domain' => NULL,
    'port'   => NULL,
    'path'   => NULL,
  );

  // email?
  if (strstr($url['path'], '@')) {
    list($user, $domain) = preg_split('/@/', $url['path']);

    $identifier['scheme'] = 'acct';
    $identifier['user']   = $user;
    $identifier['domain'] = $domain;

  } else { // some form of URL
    if (isset($url['host'])) {
      $identifier['scheme'] = strtolower($url['scheme']);
      $identifier['domain'] = $url['host'];
      $identifier['port'] = isset($url['port']) ? $url['port'] : NULL;
      $identifier['path'] = isset($url['path']) ? $url['path'] : NULL;

    } elseif (isset($url['path'])) { // user didn't include a scheme
      if (strstr($url['path'], '/')) {
        list($domain, $path) = preg_split('/\//', $url['path']);
      } else {
        $domain = $url['path'];
        $path = NULL;
      }

      $identifier['scheme'] = 'http';
      $identifier['domain'] = $domain;
      $identifier['port'] = isset($url['port']) ? $url['port'] : NULL;
      $identifier['path'] = $path ? '/' . $path : NULL;
    }
  }

  // make sure we got something
  if (!$identifier['domain']) {
    die("Not sure what you gave us: $input\n");
  }

  // make sure we ended up with a valid scheme
  if (!in_array($identifier['scheme'], array('http', 'https', 'acct'))) {
    die("Invalid URI scheme\n");
  }

  return $identifier;
}

// turn what we figured out the user entered into a canonicalized string
function normalize_input($parsed) {
  $identifier = '';
  if ($parsed['scheme'] == 'acct') {
    $identifier = 'acct:' . $parsed['user'] . '@' . $parsed['domain'];
  } else {
    $identifier = $parsed['scheme'] . '://' . $parsed['domain'];
    if ($parsed['port']) {
      $identifier .= ':' . $parsed['port'];
    }
  
     $identifier .= $parsed['path'];
  }

  return $identifier;
}

function discover_endpoint($curl, $scheme, $domain, $identifier) {
  // first try SSL
  curl_setopt($curl, CURLOPT_URL, 'https://' . $domain . '/.well-known/host-meta.json');
  $host_meta = curl_exec($curl);

  // don't fallback to HTTP if the user explicitly said HTTPS
  if (!$host_meta && $scheme != 'https') {
    curl_setopt($curl, CURLOPT_URL, 'http://' . $domain . '/.well-known/host-meta.json');
    $host_meta = curl_exec($curl);
  }
  if (!$host_meta || !isset($host_meta['lrdd'])) {
    die("Couldn't find a valid host meta file for your domain: $domain\n");
  }
  $host_meta = json_decode($host_meta, true);

  // Host-priority is the default if it isn't set
  if ($host_meta['lrdd']['priority'] == 'Resource-priority') {
    if ($lrdd = _fetch_lrdd($curl, $host_meta, $identifier)) {
      if (isset($lrdd['openid'])) {
        return $lrdd['openid']['href'];
      }
    }

    if ($endpoint = _fetch_identifier($curl, $identifier)) {
      return $endpoint;
    }
  
  } else {
    if ($endpoint = _fetch_identifier($curl, $identifier)) {
      return $endpoint;
    }

    if ($lrdd = _fetch_lrdd($curl, $host_meta, $identifier)) {
      if (isset($lrdd['openid'])) {
        return $lrdd['openid']['href'];
      }
    }
  }

  return false;
}

function _fetch_lrdd($curl, $host_meta, $identifier) {
  if (!isset($host_meta['lrdd']['template'])) {
    die("No LRDD template.\n");
  }
  $url = str_replace('{uri}', $identifier, $host_meta['lrdd']['template']);
  curl_setopt($curl, CURLOPT_URL, $url);
  $json = curl_exec($curl);
  $lrdd = json_decode($json, true);

  if ($lrdd['lrdd']['subject'] != $identifier) {
    die("LRDD subject doesn't match...but it should!\n");
  }

  return $lrdd;
}

function _fetch_identifier($curl, $identifier) {
  // can't fetch emails
  if (preg_match('/^acct:/', $identifier)) {
    return false;
  }

  curl_setopt($curl, CURLOPT_URL, $identifier);
  $html = curl_exec($curl);  
  if (!$html) {
    die("Couldn't fetch the identifier: $identifier\n");
  }
  
  // this now needs to parse the dom and look for a link tag in the head
  // with a rel of openid. then return the href value or false

  return false;
}

?>