diff options
| author | lolcat <will@lolcat.ca> | 2026-07-05 01:57:56 -0400 |
|---|---|---|
| committer | Pinapelz <yukais@pinapelz.com> | 2026-07-06 21:32:18 -0700 |
| commit | d69c3170795b02ec7aca7317a7fe64cd76a9c76b (patch) | |
| tree | 1306faf65450a29af0e0e50c43543f00c432238b | |
| parent | d075661645705819d85f6ae7cfd7549194c6a348 (diff) | |
i overengineered the fuck out of that baidu scraper jesus
| -rw-r--r-- | scraper/baidu.php | 38 |
1 files changed, 26 insertions, 12 deletions
diff --git a/scraper/baidu.php b/scraper/baidu.php index efb14ca..b81c224 100644 --- a/scraper/baidu.php +++ b/scraper/baidu.php @@ -666,14 +666,24 @@ class baidu{ // // Don't parse as a search result if it's a card // - $card = + $tpl_probe = $this->fuckhtml - ->getElementsByClassName( - "cosc-card", - $div + ->getElementsByAttributeValue( + "tpl", + "www_index", + [$datafield] ); - if(count($card) !== 0){ + if( + count($tpl_probe) === 0 && + count( + $this->fuckhtml + ->getElementsByClassName( + "cosc-card", + $div + ) + ) !== 0 + ){ // // Parse chinese youtube shorts @@ -870,10 +880,13 @@ class baidu{ // class:FYB_RD -> News garbage, IGNORE $result = - $this->fuckhtml - ->getElementsByClassName( - "result", - [$datafield] + array_merge( + $tpl_probe, + $this->fuckhtml + ->getElementsByClassName( + "result", + [$datafield] + ) ); if(count($result) !== 0){ @@ -885,7 +898,7 @@ class baidu{ $title = $this->fuckhtml ->getElementsByClassName( - "sc-link", + "cosc-title-a", "a" ); @@ -905,8 +918,9 @@ class baidu{ $description = $this->fuckhtml - ->getElementsByClassName( - "c-color", + ->getElementsByAttributeValue( + "data-sanssr-cmpt", + "card/www-summary-1", $div ); |