- refactored scraper filter function

- added ability to specify default scrapers in environment variables
2026-05-15 23:44:01 -05:00
parent b13dceaeb5
commit 9c2921c7fa
3 changed files with 133 additions and 151 deletions
@@ -0,0 +1,115 @@
+<?php
+
+const DEFAULT_SCRAPERS = [
+    "web" => config::SCRAPER_WEB,
+    "images" => config::SCRAPER_IMAGES,
+    "news" => config::SCRAPER_NEWS,
+    "videos" => config::SCRAPER_VIDEOS,
+    "music" => config::SCRAPER_MUSIC,
+];
+
+const AVAILABLE_SCRAPERS = [
+    "web" => [
+        "ddg" => "DuckDuckGo",
+        //"yahoo" => "Yahoo!",
+        "brave" => "Brave",
+        "yandex" => "Yandex",
+        "google" => "Google",
+        "google_api" => "Google API",
+        "google_cse" => "Google CSE",
+        "yahoo_japan" => "Yahoo! JAPAN",
+        "startpage" => "Startpage",
+        "qwant" => "Qwant",
+        "ghostery" => "Ghostery",
+        "yep" => "Yep",
+        "mwmbl" => "Mwmbl",
+        "mojeek" => "Mojeek",
+        "baidu" => "Baidu",
+        "coccoc" => "Cốc Cốc",
+        "solofield" => "Solofield",
+        "marginalia" => "Marginalia",
+        "wiby" => "wiby"
+    ],
+    "images" => [
+        "ddg" => "DuckDuckGo",
+        "yandex" => "Yandex",
+        "brave" => "Brave",
+        "google" => "Google",
+        "google_api" => "Google API",
+        "google_cse" => "Google CSE",
+        "yahoo_japan" => "Yahoo! JAPAN",
+        "startpage" => "Startpage",
+        "qwant" => "Qwant",
+        "baidu" => "Baidu",
+        "solofield" => "Solofield",
+        "pinterest" => "Pinterest",
+        "cara" => "Cara",
+        "flickr" => "Flickr",
+        "pexels" => "Pexels",
+        "pixabay" => "Pixabay",
+        "unsplash" => "Unsplash",
+        "fivehpx" => "500px",
+        "vsco" => "VSCO",
+        "imgur" => "Imgur",
+        "ftm" => "FindThatMeme"
+    ],
+
+    "videos" => [
+        "yt" => "YouTube",
+        //"archiveorg" => "Archive.org",
+        "vimeo" => "Vimeo",
+        //"odysee" => "Odysee",
+        "sepiasearch" => "Sepia Search",
+        //"fb" => "Facebook videos",
+        "ddg" => "DuckDuckGo",
+        "brave" => "Brave",
+        "yandex" => "Yandex",
+        "google" => "Google",
+        "yahoo_japan" => "Yahoo! JAPAN",
+        "startpage" => "Startpage",
+        "qwant" => "Qwant",
+        "baidu" => "Baidu",
+        "coccoc" => "Cốc Cốc",
+        "solofield" => "Solofield"
+    ],
+
+    "news" => [
+        "ddg" => "DuckDuckGo",
+        "brave" => "Brave",
+        "google" => "Google",
+        "yahoo_japan" => "Yahoo! JAPAN",
+        "startpage" => "Startpage",
+        "qwant" => "Qwant",
+        "mojeek" => "Mojeek",
+        "baidu" => "Baidu"
+    ],
+
+    "music" => [
+        "sc" => "SoundCloud",
+        "swisscows" => "Swisscows (SoundCloud)",
+        //"spotify" => "Spotify"
+    ],
+
+    "booru" => [
+        "safebooru" => "Safebooru",
+        "konachan" => "Konachan",
+        "tbib" => "The Big Imageboard",
+        "gelbooru" => "Gelbooru",
+        "yandere" => "Yande.re",
+        "sankakucomplex" => "SankakuComplex",
+        "soybooru" => "SoyBooru",
+    ],
+];
+
+function get_scraper(string $page, string $selected): string
+{
+    $available = AVAILABLE_SCRAPERS[$page];
+    if (!$available) {
+        return "__NONEXISTENT__";
+    }
+    if ($available[$selected]) {
+        return $selected;
+    }
+    $default = DEFAULT_SCRAPERS[$page];
+    return $available[$default] ? $default : array_key_first($available);
+}