gistfile1.php

<?php

// Tell mbstring to always use UTF-32, which is what the API uses.
mb_internal_encoding("UTF-32");

// These come from the API
$text = $object->text;

$pieces = [];  // the pieces of the new text we're building
$offset = 0;  // the amount of adjustment positive or negative we've seen so far
$text_index = 0;  // current pos in text

// Load all entities into one array and sort by pos.
// If you're not displaying, and just care about links, replace this
// section with $entities = $object->entities->links;
$entities = [];

foreach($object->entities->hashtags as $entity) {
	$entity->type = 'hashtag';
	$entities[] = $entity;
}
foreach($object->entities->links as $entity) {
	$entity->type = 'link';
	$entities[] = $entity;
}
foreach($object->entities->mentions as $entity) {
	$entity->type = 'mention';
	$entities[] = $entity;
}
usort($entities, function($a, $b) {
	return $b->pos - $a->pos;
});
 

foreach($entities as $entity) {
	$entity_start = $entity->pos
	$entity_end = $entity->pos + $entity->len
	// this key not always present, default to entity_end
	if(isset($entity->amended_len)) {
		$entity_amended_end = $entity->pos + $entity->amended_len
	} else {
		$entity_amended_end = $entity_end;
	}
 
	if ($text_index != $entity_start) {
		// catch anything before the start of the entity
		$pieces[] = mb_substr($text, $text_index, $entity_start - $text_index);
	}
 
	// object are looped by reference, so adjust in place
	$entity->pos += $offset;
 
	if($entity_end != $entity_amended_end) {
		$offset += $entity_end - $entity_amended_end;
	}

	$pieces[] = mb_substr($text, $entity->pos, $entity->len);

	$text_index = $entity_amended_end;
}
 
// catch anything that comes after the last entity
$pieces[] = mb_substr($text, $text_index);

$new_text = implode('', $pieces);