mirror of
https://github.com/Ahwxorg/librey/
synced 2024-12-06 19:16:48 +01:00
Implement brave search scraper
This commit is contained in:
+48
-2
@@ -1,7 +1,23 @@
|
||||
<?php
|
||||
class BraveSearchRequest extends EngineRequest {
|
||||
public function get_request_url() {
|
||||
// TODO build url for brave search
|
||||
$query_encoded = str_replace("%22", "\"", urlencode($this->query));
|
||||
|
||||
$results_language = $this->opts->language;
|
||||
$number_of_results = $this->opts->number_of_results;
|
||||
|
||||
$url = "https://search.brave.com/search?q=$query_encoded&nfpr=1&start=$this->page";
|
||||
|
||||
if (3 > strlen($results_language) && 0 < strlen($results_language)) {
|
||||
$url .= "&lr=lang_$results_language";
|
||||
$url .= "&hl=$results_language";
|
||||
}
|
||||
|
||||
if (3 > strlen($number_of_results) && 0 < strlen($number_of_results))
|
||||
$url .= "&num=$number_of_results";
|
||||
|
||||
if (isset($_COOKIE["safe_search"]))
|
||||
$url .= "&safe=medium";
|
||||
|
||||
return $url;
|
||||
}
|
||||
@@ -13,8 +29,38 @@
|
||||
if (!$xpath)
|
||||
return $results;
|
||||
|
||||
// TODO implement brave search scraper
|
||||
foreach($xpath->query("//div[@id='results']//div[contains(@class, 'snippet')]") as $result) {
|
||||
$url = $xpath->evaluate(".//a[contains(@class, 'h')]//@href", $result)[0];
|
||||
|
||||
if ($url == null)
|
||||
continue;
|
||||
|
||||
$url = $url->textContent;
|
||||
|
||||
if (!empty($results) && array_key_exists("url", $results) && end($results)["url"] == $url->textContent)
|
||||
continue;
|
||||
|
||||
$title = $xpath->evaluate(".//a[contains(@class, 'h')]//div[contains(@class, 'url')]", $result)[0];
|
||||
|
||||
if ($title == null)
|
||||
continue;
|
||||
$title = $title->textContent;
|
||||
|
||||
$description = $xpath->evaluate(".//div[contains(@class, 'snippet-content')]//div[contains(@class, 'snippet-description')]", $result)[0]->textContent;
|
||||
|
||||
array_push($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 $results;
|
||||
}
|
||||
|
||||
|
||||
@@ -46,7 +46,7 @@
|
||||
if ($url == null)
|
||||
continue;
|
||||
|
||||
if (!empty($results) && array_key_exists("url", $results) && end($results)["url"] == $url->textContent)
|
||||
if (!empty($results) && array_key_exists("url", $results) && end($results)["url"] == $url->textContent)
|
||||
continue;
|
||||
|
||||
$url = $url->textContent;
|
||||
|
||||
@@ -14,7 +14,6 @@
|
||||
check_ddg_bang($this->query, $opts);
|
||||
|
||||
if ($this->engine == "google") {
|
||||
|
||||
require "engines/text/google.php";
|
||||
$this->engine_request = new GoogleRequest($opts, $mh);
|
||||
}
|
||||
@@ -24,6 +23,11 @@
|
||||
$this->engine_request = new DuckDuckGoRequest($opts, $mh);
|
||||
}
|
||||
|
||||
if ($this->engine == "brave") {
|
||||
require "engines/text/brave.php";
|
||||
$this->engine_request = new BraveSearchRequest($opts, $mh);
|
||||
}
|
||||
|
||||
if (has_cooldown($this->engine, $this->opts->cooldowns) && !has_cached_results($this->engine_request->url)) {
|
||||
// TODO dont add it in the first place
|
||||
curl_multi_remove_handle($mh, $this->engine_request->ch);
|
||||
|
||||
Reference in New Issue
Block a user