/**
* Gets the list of prioritised extractor classes
*
* @return array
*/
protected static function get_extractor_classes()
{
// Check cache
if (self::$sorted_extractor_classes) {
return self::$sorted_extractor_classes;
}
// Generate the sorted list of extractors on demand.
$classes = ClassInfo::subclassesFor("FileTextExtractor");
array_shift($classes);
$classPriorities = array();
foreach ($classes as $class) {
$classPriorities[$class] = Config::inst()->get($class, 'priority');
}
arsort($classPriorities);
// Save classes
$sortedClasses = array_keys($classPriorities);
return self::$sorted_extractor_classes = $sortedClasses;
}
/**
* Tries to parse the file contents if a FileTextExtractor class exists to handle the file type, and returns the text.
* The value is also cached into the File record itself.
*
* @param $forceParse If false, the file content is only parsed on demand. If true, the content parsing is forced, bypassing the
* cached version
* @return String
*/
function extractFileAsText($forceParse = false)
{
if (!$forceParse && $this->owner->FileContentCache) {
return $this->owner->FileContentCache;
}
// Determine which extractor can process this file.
$extractor = FileTextExtractor::for_file($this->owner->FullPath);
if (!$extractor) {
return null;
}
$text = $extractor->getContent($this->owner->FullPath);
if (!$text) {
return null;
}
$this->owner->FileContentCache = $text;
$this->owner->write();
return $text;
}
/**
* Tries to parse the file contents if a FileTextExtractor class exists to handle the file type, and returns the text.
* The value is also cached into the File record itself.
*
* @param boolean $disableCache If false, the file content is only parsed on demand.
* If true, the content parsing is forced, bypassing the cached version
* @return string
*/
public function extractFileAsText($disableCache = false)
{
if (!$disableCache) {
$text = $this->getTextCache()->load($this->owner);
if ($text) {
return $text;
}
}
// Determine which extractor can process this file.
$extractor = FileTextExtractor::for_file($this->owner->FullPath);
if (!$extractor) {
return null;
}
$text = $extractor->getContent($this->owner->FullPath);
if (!$text) {
return null;
}
$this->getTextCache()->save($this->owner, $text);
return $text;
}
/**
* @param String $path
* @return FileTextExtractor
*/
static function for_file($path)
{
$extension = pathinfo($path, PATHINFO_EXTENSION);
if (!self::$sorted_extractor_classes) {
// Generate the sorted list of extractors on demand.
$classes = ClassInfo::subclassesFor("FileTextExtractor");
array_shift($classes);
$sortedClasses = array();
foreach ($classes as $class) {
$sortedClasses[$class] = Config::inst()->get($class, 'priority');
}
arsort($sortedClasses);
self::$sorted_extractor_classes = $sortedClasses;
}
foreach (self::$sorted_extractor_classes as $className => $priority) {
$formatter = new $className();
$matched = array_filter($formatter->supportedExtensions(), function ($compare) use($extension) {
return strtolower($compare) == strtolower($extension);
});
if ($matched) {
return $formatter;
}
}
}
