/**
* All of the following names will be encoded to 'Émile Zola':
* '%C3%89mile_Zola', '%C3%A9mile_Zola', ' %C3%A9mile Zola ', ' %C3%A9mile _ Zola ', ' Émile _ Zola '
*
* TODO: maybe we should expect (require) the name to be normalized, e.g. with uppercase
* first letter and without duplicate spaces or spaces at start or end?
* Would make this method much simpler.
*
* @param $name encoded MediaWiki page name, e.g. '%C3%89mile_Zola'.
* Must not include the namespace (e.g. 'Template:').
*/
public static function wikiDecode($name)
{
PhpUtil::assertString($name, 'name');
// make first character uppercase
$name = StringUtil::mb_ucfirst(self::cleanSpace(urldecode($name)));
return $name;
}
/**
* @param $link MediaWiki link target
*/
public static function parse($link)
{
PhpUtil::assertString($link, 'link');
self::init();
$link = urldecode($link);
if (strpos($link, '#') !== false) {
throw new WikiParserException('Invalid title: "' . $link . '" (Contains #)');
}
$parts = explode(':', $link, 2);
if (count($parts) === 2) {
$prefix = mb_strtolower(WikiUtil::cleanSpace($parts[0]));
// TODO: handle interwiki links like [[:de:Foo]]
if (strlen($prefix) === 0) {
throw new WikiParserException('cannot handle link [' . $link . ']');
}
// TODO: handle special prefixes, e.g. [[q:Foo]] links to WikiQuotes
if (isset(self::$nsCodes[$prefix])) {
$code = self::$nsCodes[$prefix];
$name = StringUtil::mb_ucfirst(WikiUtil::cleanSpace($parts[1]));
return new WikiTitle($code, $name);
}
}
$name = StringUtil::mb_ucfirst(WikiUtil::cleanSpace($link));
return new WikiTitle(self::NS_MAIN, $name);
}
public function extract($node, $subjectUri, $pageContext)
{
$this->curl = curl_init();
$url = sprintf(LocalConfiguration::abstractPageUrlFormat, $this->language, $node->getRoot()->getTitle()->encoded());
curl_setopt($this->curl, CURLOPT_URL, $url);
curl_setopt($this->curl, CURLOPT_RETURNTRANSFER, true);
curl_setopt($this->curl, CURLOPT_BINARYTRANSFER, true);
curl_setopt($this->curl, CURLOPT_FAILONERROR, true);
$text = curl_exec($this->curl);
if (false === $text) {
//echo curl_error($this->curl);
// TODO
/*
('error', 'nr '.curl_errno($this->curl). "\n".curl_error($this->curl));
$this->log('error', $this->getExtractorID().' call to '.$url. ' failed');
$this->log('error', 'Please define AbstractExtractor.page_url_format correctly in your option file, e.g. dbpedia.ini.');
$this->log('error', 'nr '.curl_errno($this->curl). "\n".curl_error($this->curl));
// die($this->getExtractorID().' call to '.$url. ' failed');
echo($this->getExtractorID().' call to '.$url. ' failed');
return new ExtractionResult($pageID, $this->language, $this->getExtractorID());
*/
}
if (!empty($text)) {
$text = StringUtil::htmlDecode($text);
$shorttext = $this->short($text);
if (!empty($shorttext)) {
$quad = new RdfQuad($subjectUri, $this->ontology->getProperty(self::SHORT_PROPERTY), $shorttext, $node->getSourceUri());
$this->destination->addQuad($quad);
}
$quad = new RdfQuad($subjectUri, $this->ontology->getProperty(self::LONG_PROPERTY), $text, $node->getSourceUri());
$this->destination->addQuad($quad);
}
return true;
}
public function parse(Node $node)
{
$result = self::nodeToString($node);
//Clean text
$result = WikiUtil::removeWikiEmphasis($result);
$result = StringUtil::htmlDecode($result);
if (strlen($result) > 0) {
return $result;
}
}
public function generate($baseUri, $node)
{
if (isset($node)) {
//Retrieve text
$text = $this->nodeToText($node);
//Normalize text
$text = WikiUtil::removeWikiEmphasis($text);
$text = StringUtil::htmlDecode($text);
$text = preg_replace('/ +/', ' ', $text);
//remove duplicate spaces
$text = str_replace('(', ' ', $text);
$text = str_replace(')', ' ', $text);
$text = strip_tags($text);
$text = substr($text, 0, 50);
$text = trim($text);
$text = str_replace(' ', '_', $text);
$text = urlencode($text);
//Test if the base URI ends with a prefix of text
$baseLen = strlen($baseUri);
$textLen = strlen($text);
for ($i = $baseLen - 1; $i > 0 && $baseLen - $i < $textLen; $i--) {
if (substr_compare($baseUri, $text, $i, $textLen, true) === 0) {
$text = substr($text, $baseLen - $i);
break;
}
}
//Remove leading underscore
if (!empty($text) && $text[0] === '_') {
$text = substr($text, 1);
}
//Generate URI
$uri = $baseUri . '__' . $text;
} else {
$uri = $baseUri;
}
//Resolve collisions
if (!isset($this->uris[$uri])) {
//No collision
$this->uris[$uri] = 1;
} else {
//Collision found
$index = $this->uris[$uri];
$this->uris[$uri] = $index + 1;
$uri .= '__' . $index;
}
return $uri;
}
/**
* @param $baseDir must end with a directory separator (slash or backslash)
* @param $skipNames names (not paths) of files and directories to skip, e.g. '.svn'.
* If not given, all files and directories will be included.
* @param $paths array of strings, paths of files to use, relative to base dir,
* using forward slashes. If not given, all files and directories will be included.
*/
public function __construct($baseDir, $skipNames = null, $paths = null)
{
PhpUtil::assertString($baseDir, 'base dir');
$baseDir = str_replace('\\', '/', realpath($baseDir));
if (!is_dir($baseDir)) {
throw new \InvalidArgumentException('base dir must be an existing directory, but is ' . $baseDir);
}
// make sure that $baseDir ends with /
if (!StringUtil::endsWith($baseDir, '/')) {
$baseDir .= '/';
}
if ($skipNames !== null) {
PhpUtil::assertArray($skipNames, 'skip names');
} else {
$skipNames = array();
}
if ($paths !== null) {
PhpUtil::assertArray($paths, 'paths');
}
$this->baseDir = $baseDir;
$this->skipNames = $skipNames;
$this->paths = $paths;
}
private function templateMappingPath()
{
if ($this->parsedTitle->nsCode() !== LocalConfiguration::templateMappingNsCode) {
return false;
}
$path = $this->parsedTitle->encoded();
$suffix = LocalConfiguration::templateMappingSuffix;
if (!StringUtil::endsWith($path, $suffix)) {
return false;
}
return substr($path, 0, -strlen($suffix));
}
/**
* Hack that cuts off 'dbpedia/' from the start of the string
* or replaces the first '/' by a ':', so 'dbpedia/Person' becomes 'Person'
* and 'foaf/name' becomes 'foaf:name'.
* @param $name wiki-encoded page title
* @return $name without 'dbpedia/' prefix or with first '/' replaced by a ':'.
* @throws InvalidArgumentException if $name does not include a '/'
*/
public static function getName($name)
{
if (StringUtil::startsWith($name, 'dbpedia/')) {
return substr($name, 8);
} else {
$slash = strpos($name, '/');
if ($slash === false) {
throw new \InvalidArgumentException('missing namespace in page title ' . $name);
} else {
$name[$slash] = ':';
return $name;
}
}
}
