diff --git a/src/configs/Createdb.php b/src/configs/Createdb.php index 12c6e63af..262a953c5 100755 --- a/src/configs/Createdb.php +++ b/src/configs/Createdb.php @@ -1039,7 +1039,7 @@ $scrapers = [ "contains(@href, '/sites/default/files') or " . "contains(@content, 'Drupal')]|" . "//*[contains(@class, 'drupalorg-site-main')]", - "//*[contains(@id, 'main')]", + "//*[@id='main']", "//*[contains(@class, 'd-animations-settings')]\n" . "//*[contains(@class,'element-invisible')]\n" . "//*[contains(@class,'visually-hidden')]\n" . @@ -1053,7 +1053,7 @@ $scrapers = [ "//thead\n", ""], ["MEDIAWIKI", "//meta[contains(@content, 'MediaWiki')]", - "//main|//*[contains(@id, 'content')]", + "//main|//*[@id='content']", "//*[contains(@class, 'nmbox')]\n" . "//*[contains(@class, 'hatnote')]\n" . "//*[contains(@class, 'infobox')]\n" . @@ -1108,20 +1108,20 @@ $scrapers = [ "string-length(@id) >4]|//div[contains(@class, 'homepagewrapper')]|" . "//div[contains(@class, 'homepage-row')]/parent::*|" . "//div[contains(@class, 'archive-grid')]/parent::*", - "//*[contains(@id, 'entry-comments')]\n" . + "//*[@id='entry-comments']\n" . "//*[contains(@class, 'sharedaddy')]\n" . "//*[contains(@class, 'blog-subscribe')]\n" . - "//*[contains(@id, 'entry-side')]\n" . + "//*[@id='entry-side']\n" . "//*[contains(@class, 'signup')]\n" . "//*[contains(@class, 'module-donate')]", ""], ["YIOOP", "//meta[@name='generator' and contains(@content,'Yioop')]", "//div[contains(@class, 'body-container')]", - "//*[contains(@id, 'message')]\n" . - "//*[contains(@id, 'help')]\n" . + "//*[@id='message']\n" . + "//*[@id='help']\n" . "//*[contains(@class, 'query-statistics')]\n" . - "//*[contains(@id, 'suggest')]\n" . - "//*[contains(@id, 'spell')]", + "//*[@id='suggest']\n" . + "//*[@id='spell']", ""], ]; foreach ($scrapers as $scraper) { diff --git a/src/data/public_default.db b/src/data/public_default.db index 1bedb8e6d..17cd0bef2 100644 Binary files a/src/data/public_default.db and b/src/data/public_default.db differ