exit(1);
}
-$es = new Elasticsearch($GLOBALS['phinde']['elasticsearch']);
+function removeTags($doc, $tag) {
+ $elems = array();
+ foreach ($doc->getElementsbyTagName($tag) as $elem) {
+ $elems[] = $elem;
+ }
+ foreach ($elems as $elem) {
+ $elem->parentNode->removeChild($elem);
+ }
+}
$url = $argv[1];
+
+$req = new \HTTP_Request2($url);
+$req->setConfig('follow_redirects', true);
+$req->setConfig('connect_timeout', 5);
+$req->setConfig('timeout', 10);
+$req->setConfig('ssl_verify_peer', false);
+//FIXME: size limit
+
+$es = new Elasticsearch($GLOBALS['phinde']['elasticsearch']);
$existingDoc = $es->get($url);
if ($existingDoc && $existingDoc->status == 'indexed') {
- echo "URL already indexed: $url\n";
- exit(0);
+ $nMoDate = strtotime($existingDoc->modate);
+ $refreshtime = $GLOBALS['phinde']['refreshtime'];
+ if (time() - $nMoDate < $refreshtime) {
+ echo "URL already indexed less than $refreshtime seconds ago: $url\n";
+ exit(0);
+ }
+
+ $req->setHeader('If-Modified-Since: ' . date('r', $nMoDate));
}
//FIXME: sourcetitle, sourcelink
-//FIXME: enable redirects
-//FIXME: enable ssl
-$req = new \HTTP_Request2($url);
-$req->setConfig('connect_timeout', 5);
-$req->setConfig('timeout', 10);
$res = $req->send();
//FIXME: try-catch
-//FIXME: delete if 401 gone or 404 when updating
-if ($res->getStatus() !== 200) {
+if ($res->getStatus() === 304) {
+ //not modified since last time
+ //FIXME: store "last try" time
+ exit(0);
+} else if ($res->getStatus() !== 200) {
+ //FIXME: delete if 401 gone or 404 when updating
echo "Response code is not 200 but " . $res->getStatus() . ", stopping\n";
//FIXME: update status
exit(3);
//FIXME: update index only if changed since last index time
//FIXME: extract base url from html
-//FIXME: use final URL after redirects
+//FIXME: check if effective url needs updating
+$url = $res->getEffectiveUrl();
$base = new \Net_URL2($url);
$indexDoc = new \stdClass();
$doc = new \DOMDocument();
//@ to hide parse warning messages in invalid html
@$doc->loadHTML($res->getBody());
-$sx = simplexml_import_dom($doc);
+$dx = new \DOMXPath($doc);
+
+$xbase = $dx->evaluate('/html/head/base[@href]')->item(0);
+if ($xbase) {
+ $base = $base->resolve(
+ $xbase->attributes->getNamedItem('href')->textContent
+ );
+}
+
+$meta = $dx->evaluate('/html/head/meta[@name="robots" and @content]')
+ ->item(0);
+if ($meta) {
+ $robots = $meta->attributes->getNamedItem('content')->textContent;
+ foreach (explode(',', $robots) as $value) {
+ if (trim($value) == 'noindex') {
+ echo "URL does not want to be indexed: $url\n";
+ exit(0);
+ }
+ }
+}
+
+//remove script tags
+removeTags($doc, 'script');
+removeTags($doc, 'style');
+removeTags($doc, 'nav');
+
+//default content: <body>
+$xpContext = $doc->getElementsByTagName('body')->item(0);
+//FIXME: follow meta refresh, no body
+// example: https://www.gnu.org/software/coreutils/
+
+//use microformats content if it exists
+$xpElems = $dx->query(
+ "//*[contains(concat(' ', normalize-space(@class), ' '), ' e-content ')]"
+);
+if ($xpElems->length) {
+ $xpContext = $xpElems->item(0);
+} else if ($doc->getElementById('content')) {
+ //if there is an element with ID "content", we'll use this
+ $xpContext = $doc->getElementById('content');
+}
$indexDoc->url = $url;
+$indexDoc->schemalessUrl = Helper::noSchema($url);
$indexDoc->type = 'html';
$indexDoc->subtype = '';
$indexDoc->mimetype = $mimetype;
$indexDoc->author = new \stdClass();
-$arSxElems = $sx->xpath('/html/head/meta[@name="author"]');
-if (count($arSxElems)) {
- $indexDoc->author->name = trim($arSxElems[0]['content']);
+$arXpElems = $dx->query('/html/head/meta[@name="author" and @content]');
+if ($arXpElems->length) {
+ $indexDoc->author->name = trim(
+ $arXpElems->item(0)->attributes->getNamedItem('content')->textContent
+ );
}
-$arSxElems = $sx->xpath('/html/head/link[@rel="author"]');
-if (count($arSxElems)) {
- $indexDoc->author->url = (string) $base->resolve($arSxElems[0]['href']);
+$arXpElems = $dx->query('/html/head/link[@rel="author" and @href]');
+if ($arXpElems->length) {
+ $indexDoc->author->url = trim(
+ $base->resolve(
+ $arXpElems->item(0)->attributes->getNamedItem('href')->textContent
+ )
+ );
+}
+
+
+$arXpElems = $dx->query('/html/head/title');
+if ($arXpElems->length) {
+ $indexDoc->title = trim(
+ $arXpElems->item(0)->textContent
+ );
}
-$indexDoc->title = (string) $sx->head->title;
foreach (array('h1', 'h2', 'h3', 'h4', 'h5', 'h6') as $headlinetype) {
$indexDoc->$headlinetype = array();
- //FIXME: limit to h-entry children
- foreach ($sx->xpath('//' . $headlinetype) as $xheadline) {
+ foreach ($xpContext->getElementsByTagName($headlinetype) as $xheadline) {
array_push(
$indexDoc->$headlinetype,
- trim(dom_import_simplexml($xheadline)->textContent)
+ trim($xheadline->textContent)
);
}
}
-//FIXME: limit to h-entry e-content
+//FIXME: split paragraphs
//FIXME: insert space after br
-//FIXME: remove javascript
$indexDoc->text = array();
-foreach ($doc->getElementsByTagName('body') as $body) {
- $indexDoc->text[] = trim(
- str_replace(
- array("\r\n", "\n", "\r", ' '),
- ' ',
- $body->textContent
- )
- );
-}
+$indexDoc->text[] = trim(
+ str_replace(
+ array("\r\n", "\n", "\r", ' '),
+ ' ',
+ $xpContext->textContent
+ )
+);
//tags
$tags = array();
-foreach ($sx->xpath('/html/head/meta[@name="keywords"]') as $xkeywords) {
- foreach (explode(',', $xkeywords['content']) as $keyword) {
+foreach ($dx->query('/html/head/meta[@name="keywords" and @content]') as $xkeywords) {
+ $keywords = $xkeywords->attributes->getNamedItem('content')->textContent;
+ foreach (explode(',', $keywords) as $keyword) {
$tags[trim($keyword)] = true;
}
}
$indexDoc->tags = array_keys($tags);
//dates
-$arSxdates = $sx->xpath('/html/head/meta[@name="DC.date.created"]');
-if (count($arSxdates)) {
- $indexDoc->crdate = date('c', strtotime((string) $arSxdates[0]['content']));
+$arXpdates = $dx->query('/html/head/meta[@name="DC.date.created" and @content]');
+if ($arXpdates->length) {
+ $indexDoc->crdate = date(
+ 'c',
+ strtotime(
+ $arXpdates->item(0)->attributes->getNamedItem('content')->textContent
+ )
+ );
}
//FIXME: keep creation date from database, or use modified date if we
// do not have it there
-$arSxdates = $sx->xpath('/html/head/meta[@name="DC.date.modified"]');
-if (count($arSxdates)) {
- $indexDoc->modate = date('c', strtotime((string) $arSxdates[0]['content']));
+$arXpdates = $dx->query('/html/head/meta[@name="DC.date.modified" and @content]');
+if ($arXpdates->length) {
+ $indexDoc->modate = date(
+ 'c',
+ strtotime(
+ $arXpdates->item(0)->attributes->getNamedItem('content')->textContent
+ )
+ );
} else {
$lm = $res->getHeader('last-modified');
if ($lm !== null) {
//language
//there may be "en-US" and "de-DE"
-$indexDoc->language = strtolower(substr((string) $sx['lang'], 0, 2));
+$xlang = $doc->documentElement->attributes->getNamedItem('lang');
+if ($xlang) {
+ $indexDoc->language = strtolower(substr($xlang->textContent, 0, 2));
+}
//FIXME: fallback, autodetection
//FIXME: check noindex
-
-//var_dump($indexDoc);
+//var_dump($indexDoc);die();
$indexDoc->status = 'indexed';
git.cweiske.de / phinde.git / blobdiff