added baidu, the best search engine

This commit is contained in:
lolcat 2025-07-27 21:46:03 -04:00
parent 0bdd5e73df
commit f43feff0aa
3 changed files with 2248 additions and 3 deletions

View File

@ -949,6 +949,7 @@ class frontend{
"crowdview" => "Crowdview", "crowdview" => "Crowdview",
"mwmbl" => "Mwmbl", "mwmbl" => "Mwmbl",
"mojeek" => "Mojeek", "mojeek" => "Mojeek",
"baidu" => "Baidu",
"solofield" => "Solofield", "solofield" => "Solofield",
"marginalia" => "Marginalia", "marginalia" => "Marginalia",
"wiby" => "wiby", "wiby" => "wiby",
@ -969,6 +970,7 @@ class frontend{
"startpage" => "Startpage", "startpage" => "Startpage",
"qwant" => "Qwant", "qwant" => "Qwant",
"yep" => "Yep", "yep" => "Yep",
"baidu" => "Baidu",
"solofield" => "Solofield", "solofield" => "Solofield",
"pinterest" => "Pinterest", "pinterest" => "Pinterest",
"flickr" => "Flickr", "flickr" => "Flickr",
@ -993,6 +995,7 @@ class frontend{
"google" => "Google", "google" => "Google",
"startpage" => "Startpage", "startpage" => "Startpage",
"qwant" => "Qwant", "qwant" => "Qwant",
"baidu" => "Baidu",
"solofield" => "Solofield" "solofield" => "Solofield"
] ]
]; ];
@ -1008,7 +1011,8 @@ class frontend{
"startpage" => "Startpage", "startpage" => "Startpage",
"qwant" => "Qwant", "qwant" => "Qwant",
"yep" => "Yep", "yep" => "Yep",
"mojeek" => "Mojeek" "mojeek" => "Mojeek",
"baidu" => "Baidu"
] ]
]; ];
break; break;

View File

@ -240,12 +240,13 @@ class fuckhtml{
public function getElementsByFuzzyAttributeValue(string $name, string $value, $collection = null){ public function getElementsByFuzzyAttributeValue(string $name, string $value, $collection = null){
$elems = $this->getElementsByAttributeName($name, $collection); $elems = $this->getElementsByAttributeName($name, $collection);
$value = $value =
explode( explode(
" ", " ",
trim( trim(
preg_replace( preg_replace(
'/ +/', '/\s+/',
" ", " ",
$value $value
) )
@ -258,7 +259,18 @@ class fuckhtml{
foreach($elem["attributes"] as $attrib_name => $attrib_value){ foreach($elem["attributes"] as $attrib_name => $attrib_value){
$attrib_value = explode(" ", $attrib_value); $attrib_value =
explode(
" ",
trim(
preg_replace(
'/\s+/',
" ",
$attrib_value
)
)
);
$ac = count($attrib_value); $ac = count($attrib_value);
$nc = count($value); $nc = count($value);
$cr = 0; $cr = 0;

2229
scraper/baidu.php Normal file

File diff suppressed because it is too large Load Diff