mirror of
https://git.bakhai.co.in/FbIN/LibreY.git
synced 2025-11-06 04:31:30 +05:30
747e827fef
fix(google): Use updated Google search endpoint via unixfox's research for SearXNG (b531406ce3/engines/text/google.php) Add initial Bing support Note, Bing obfuscates anchor links but it was trivial to determine. (f9f51c6b74) Update bing.php, decode URL Added urldecode to base64uri-decoded URL to enable proper parsing by get_base_url. (97c085cf95) Ignore Bing's relative links Updated parser for Bing links to ignore links that don't fit "results", such as relative links. Only non-obfuscated links and de-obfuscated absolute links will correctly make it through the parser. (7f12ad2950) https://github.com/Ahwxorg/LibreY/pull/218 https://github.com/Ahwxorg/LibreY/pull/215
92 lines
No EOL
3.3 KiB
PHP
92 lines
No EOL
3.3 KiB
PHP
<?php
|
|
class BingSearchRequest extends EngineRequest {
|
|
public function get_request_url() {
|
|
$query_encoded = str_replace("%22", "\"", urlencode($this->query));
|
|
|
|
$results_language = $this->opts->language;
|
|
$number_of_results = $this->opts->number_of_results;
|
|
|
|
// TODO figure out how to not autocorrect
|
|
$url = "https://www.bing.com/search?q=$query_encoded&first=" . ((10 * $this->page) + 1);
|
|
|
|
// TODO language setting
|
|
if (!is_null($results_language))
|
|
$url .= "&lang=$results_language";
|
|
|
|
return $url;
|
|
}
|
|
|
|
public function parse_results($response) {
|
|
$results = array();
|
|
$xpath = get_xpath($response);
|
|
|
|
if (!$xpath)
|
|
return $results;
|
|
|
|
foreach($xpath->query("//ol[@id='b_results']//li") as $result) {
|
|
$href_url = $xpath->evaluate(".//h2//a//@href", $result)[0];
|
|
|
|
if ($href_url == null)
|
|
continue;
|
|
|
|
$possible_url = $href_url->textContent;
|
|
|
|
$possible_url_query = parse_url($possible_url, PHP_URL_QUERY);
|
|
|
|
if ($possible_url_query == false)
|
|
continue;
|
|
|
|
parse_str($possible_url_query, $possible_url);
|
|
|
|
if (!array_key_exists('u', $possible_url))
|
|
continue;
|
|
|
|
$possible_url = $possible_url['u'];
|
|
|
|
if (str_starts_with($possible_url, "a1aHR0c"))
|
|
{
|
|
// First two characters are irrelevant, strip for later
|
|
$possible_url = substr($possible_url, 2);
|
|
}
|
|
if (str_starts_with($possible_url, "aHR0c"))
|
|
{
|
|
// Base64 "coded", extract and decode
|
|
$possible_url = str_replace('-', '+', $possible_url);
|
|
$possible_url = str_replace('_', '/', $possible_url);
|
|
$url = urldecode(base64_decode($possible_url, true));
|
|
} else
|
|
$url = $possible_url;
|
|
|
|
if (str_starts_with($url, "a1"))
|
|
continue; // It's probably a Bing-relative link such as for video, skip it.
|
|
|
|
if (!empty($results) && array_key_exists("url", $results) && end($results)["url"] == $url->textContent)
|
|
continue;
|
|
|
|
$title = $xpath->evaluate(".//h2//a", $result)[0];
|
|
|
|
if ($title == null)
|
|
continue;
|
|
|
|
$title = $title->textContent;
|
|
|
|
$description = ($xpath->evaluate(".//div[contains(@class, 'b_caption')]//p", $result)[0] ?? null) ?->textContent ?? '';
|
|
|
|
array_push($results,
|
|
array (
|
|
"title" => htmlspecialchars($title),
|
|
"url" => htmlspecialchars($url),
|
|
// base_url is to be removed in the future, see #47
|
|
"base_url" => htmlspecialchars(get_base_url($url)),
|
|
"description" => $description == null ?
|
|
TEXTS["result_no_description"] :
|
|
htmlspecialchars($description)
|
|
)
|
|
);
|
|
|
|
}
|
|
return $results;
|
|
}
|
|
|
|
}
|
|
?>
|