/**
* {@inheritDoc}
*/
public function build(AbstractSource $source)
{
$sourceId = $source->sourceId();
$tags = is_array($source->data()->get('tags')) ? $source->data()->get('tags') : array();
$categories = is_array($source->data()->get('categories')) ? $source->data()->get('categories') : null;
$record = array('objectID' => md5($sourceId), 'title' => $source->data()->get('title'), 'content' => strip_tags($source->content()), 'link' => $source->permalink()->relativeUrlPath(), 'date' => $source->data()->get('calculated_date'));
// add em if we have em
if ($categories != NULL) {
$record["categories"] = $categories;
}
if ($tags != NULL) {
$record["tags"] = $tags;
}
return $record;
}
/**
* @param AbstractSource $source
* @return array
*/
private function parseSource(AbstractSource $source)
{
$document = array('objectID' => sha1($source->sourceId()), 'title' => $source->data()->get('title'), 'body' => strip_tags($source->content()), 'url' => rtrim($source->permalink()->relativeUrlPath(), '/') . '/', 'date' => $source->data()->get('calculated_date'));
$tags = is_array($source->data()->get('tags')) ? $source->data()->get('tags') : array();
if ($tags) {
$document['tags'] = $tags;
}
return $document;
}
/**
* @param AbstractSource $source
* @return bool
*/
private function shouldBeConverted(AbstractSource $source)
{
// File based whitelist has precedence
if (!empty($this->files)) {
foreach ($this->files as $fileName) {
if ($source->relativePathname() === $fileName) {
return true;
}
}
return false;
}
foreach ($this->extensions as $extension) {
if (fnmatch("*.{$extension}", $source->filename())) {
return true;
}
}
return false;
}
/**
* Initialize source
*
* @param bool $hasChanged Has the file changed?
*/
protected function init($hasChanged = null)
{
parent::init($hasChanged);
$originalData = $this->data;
if ($this->isRaw) {
$this->useFileReference = true;
$this->data = new Data();
} else {
$internetMediaType = $this->analyzer->detectFromFilename($this->file);
if ($internetMediaType && ('text' === $internetMediaType->getType() || $this->applicationXmlType->equals($internetMediaType))) {
// Only text files can be processed by Sculpin and since we
// have to read them here we are going to ensure that we use
// the content we read here instead of having someone else
// read the file again later.
$this->useFileReference = false;
// Additionally, any text file is a candidate for formatting.
$this->canBeFormatted = true;
$content = file_get_contents($this->file);
if (preg_match('/^\\s*(?:---[\\s]*[\\r\\n]+)(.*?)(?:---[\\s]*[\\r\\n]+)(.*?)$/s', $content, $matches)) {
$this->content = $matches[2];
if (preg_match('/^(\\s*[-]+\\s*|\\s*)$/', $matches[1])) {
// There is nothing useful in the YAML front matter.
$this->data = new Data();
} else {
// There may be YAML frontmatter
try {
$builder = new YamlDataBuilder($matches[1]);
$this->data = $builder->build();
} catch (\InvalidArgumentException $e) {
// Likely not actually YAML front matter available,
// treat the entire file as pure content.
echo ' ! ' . $this->sourceId() . ' ' . $e->getMessage() . ' !' . PHP_EOL;
$this->content = $content;
$this->data = new Data();
}
}
} else {
$this->content = $content;
$this->data = new Data();
$this->canBeFormatted = false;
}
} else {
$this->useFileReference = true;
$this->data = new Data();
}
}
if ($this->data->get('date')) {
if (!is_numeric($this->data->get('date'))) {
$this->data->set('date', strtotime($this->data->get('date')));
}
$this->data->set('calculated_date', $this->data->get('date'));
}
if ($originalData) {
$this->data->import($originalData, false);
}
}
/**
* @param AbstractSource $source
* @return array
*/
private function parseSource(AbstractSource $source)
{
$tags = is_array($source->data()->get('tags')) ? $source->data()->get('tags') : array();
$document = array('title' => $source->data()->get('title'), 'body' => strip_tags($source->content()), 'tags' => implode(', ', $tags), 'url' => rtrim($source->permalink()->relativeUrlPath(), '/') . '/');
return $document;
}