ServiceHelper.php

<?php
class ServiceHelper {
  
  // @text is the full tweet text
  // @shortCode is the code part of the shortlink, like nXXX0
  public function shortenForTwitter($text, $shortCode=FALSE, $shortDomain=FALSE) {

    $placeholders = array();
    $rawTweet = $text;

    // Catch anything that Twitter might think is a link and wrap in a t.co placeholder
    while(preg_match('|(https?://)?[a-z0-9\-]+\.[a-z]{2,10}/?([^ ]+)?|i', $rawTweet, $match, PREG_OFFSET_CAPTURE)) {
      // Count these entities as 20-characters.
      // Generate placeholder text, a sequence of characters unlikely to be used in body text
      $placeholder = ':::'.substr(sha1($match[0].rand(1000000,9999999)), 0, 17);
      $rawTweet = substr_replace($rawTweet, $placeholder, $match[0][1], strlen($match[0][0]));
      $placeholders[$placeholder] = $match[0][0];
      #echo "Raw: " . $rawTweet . "\n";
    }

    if($shortCode)
      $citationLinkTemplate = 'http://t.co/xxxxxxxx';
    else
      $citationLinkTemplate = '';

    $shortText = $this->shortenToLengthWithCitation($rawTweet, 140, $shortCode, $shortDomain, $citationLinkTemplate);

    // Re-expand the placeholders before sending to Twitter
    foreach($placeholders as $key=>$val) {
      $shortText = str_replace($key, $val, $shortText);
    }

    #echo "Tweet: " . $shortText . "\n";

    return $shortText;
  }

  public function shortenForAppDotNet($text, $shortCode=FALSE, $shortDomain=FALSE) {
    if($shortCode)
      $citationLinkTemplate = $shortDomain.'/'.$shortCode;
    else
      $citationLinkTemplate = '';

    $shortText = $this->shortenToLengthWithCitation($text, 256, $shortCode, $shortDomain, $citationLinkTemplate);
    return $shortText;
  }

  public $citationLinkPosition;
  public $citationLinkLength;
  public $citationLinkText;

  private function shortenToLengthWithCitation($text, $length, $shortCode=FALSE, $shortDomain=FALSE, $citationLinkTemplate='') {
    // Save 17 characters ' (aaron.pk xxxxx)' for the end
    $shortText = $text;
    $words = explode(' ', $shortText);
    $hashtags = array();
    $shortened = FALSE;

    // Shorten the text to fit in 140-n chars
    if($shortCode)
      $citation = ' (' . $citationLinkTemplate . ')';
    else
      $citation = '';

    while(strlen($shortText) > ($length-strlen($citation))) {
      // Preserve hashtags. Find the last word that is not a hashtag and remove it.
      $found = FALSE;
      for($i=count($words)-1; $i>=0 && $found==FALSE; $i--) {
        // Catch hashtags and save them
        if(!preg_match('/#[^ ]+/', $words[$i])) {
          array_splice($words, $i, 1);
          $found = TRUE;
        } else {
          $hashtags[] = $words[$i];
          array_splice($words, $i, 1);
        }
      }
      // Generate a sample shortText with the new arrays. This is not the final text that will be returned.
      $shortText = implode(' ', array_merge($words, $hashtags));
      $shortened = TRUE;
    }

    // If the text was shortened to fit the permalink, then re-shorten the tweet to include the linked permalink and ellipses
    if($shortened) {
      $shortText = $text;
      $words = explode(' ', $shortText);
      $hashtags = array();

      if($shortCode)
        $citation = '... '.$citationLinkTemplate;
      else
        $citation = '...';

      while(strlen($shortText) > ($length-strlen($citation))) {
        $found = FALSE;
        for($i=count($words)-1; $i>=0 && $found==FALSE; $i--) {
          if(!preg_match('/#[^ ]+/', $words[$i])) {
            array_splice($words, $i, 1);
            $found = TRUE;
          } else {
            $hashtags[] = $words[$i];
            array_splice($words, $i, 1);
          }
        }
        // Generate a sample shortText with the new arrays. This is not the final text that will be returned.
        $shortText = implode(' ', array_merge($words, $hashtags));
      }
    }

    // Put the words back together with the hashtags at the end
    if($shortened) {
      // Trim off punctuation. This actually makes the tweet possibly slightly under the limit.
      $shortText = trim(implode(' ', $words), ',.:;') . '... ' . implode(' ', $hashtags);
    }

    #echo "Short: $shortText\n";
    #echo "Length: " . strlen($shortText) . "\n";

    // Add the citation to the end
    if($shortCode) {
      if($shortened) {
        $this->citationLinkPosition = strlen(trim($shortText) . ' ');

        if(preg_match('/^http/', $citationLinkTemplate))
          $this->citationLinkText = 'http://' . $shortDomain . '/' . $shortCode;
        else
          $this->citationLinkText = $shortDomain . '/' . $shortCode;

        $this->citationLinkLength = strlen($this->citationLinkText);
        $shortText = trim($shortText) . ' ' . $this->citationLinkText;
      } else {
        $this->citationLinkPosition = strlen(trim($shortText) . ' (');

        if(preg_match('/^http/', $citationLinkTemplate))
          $this->citationLinkText = 'http://' . $shortDomain . '/' . $shortCode;
        else
          $this->citationLinkText = $shortDomain . '/' . $shortCode;

        $this->citationLinkLength = strlen($this->citationLinkText);
        $shortText = trim($shortText) . ' (' . $this->citationLinkText . ')';
      }
    }

    return $shortText;
  }

}


/** ******************************************** **/

$sample = array(
  'The real test. Posting from the xtendr iPhone app to my site, syndicating to Twitter, app.net and PuSH. #indieweb',
  'Random post with app.net short link and a full link http://social-igniter.com/blog/2012/08/why-we-support-app-dot-net and some text after, probably too long to post to Twitter as is',
  'Naturally the Twitter syndication failed, because it caught my 7-character "app.net" and turned it into a 20-character "t.co" URL....',
  'Testing my new posting API. This one should use link annotations. #indieweb',
  'I thought I was going to stop using Twitter because of their ToS changes. Turns out it\'s because of this: http://farm9.staticflickr.com/8434/7805846606_7cd9304bff_o.png',
  'This is the first good argument for app.net that I\'ve heard: http://social-igniter.com/blog/2012/08/why-we-support-app-dot-net http://news.ycombinator.com/item?id=4367965',
  'I had a dream last night where I was an XCode project. Every time the alarm went off, that was running "Build and Run." If I would fail to compile, I\'d have to fix the errors and try again. The next time the alarm went off, I\'d attempt to "run" again, and I kept failing to compile a few times when I kept hitting "snooze." The last thing I did before I actually woke up was fix a missing semicolon which finally allowed me to compile and then I was awake. #programming',
  'Just got in the world\'s dumbest bike accident on the waterfront. This adorable kid tried to give me a high five as I biked by, but I overcompensated with my front brake and flipped over my handlebars. I had no idea it was that easy to flip over.',
);

foreach($sample as $text) {

  echo $text . "\n";
  echo "Original Length: " . strlen($text) . "\n";

  $helper = new ServiceHelper();
  #$short = $helper->shortenForAppDotNet($text, 'nXXX0', 'aaron.pk');
  $short = $helper->shortenForTwitter($text, 'nXXX0', 'aaron.pk');
  echo $short . "\n";
  echo "Length: " . strlen($short) . "\n";
  echo "\n";
}