Merge pull request #4389 from pixelfed/staging

Autospam Advanced Detection

CHANGELOG.md

@@ -12,6 +12,7 @@
 
 ### Added
 - Added store remote media on S3 config setting, disabled by default ([51768083](https://github.com/pixelfed/pixelfed/commit/51768083))
+- Added Autospam Advanced Detection ([132a58de](https://github.com/pixelfed/pixelfed/commit/132a58de))
 
 ### Updates
 - Update admin dashboard, fix search and dropdown menu ([dac0d083](https://github.com/pixelfed/pixelfed/commit/dac0d083))

app/Http/Controllers/Admin/AdminAutospamController.php

@@ -0,0 +1,255 @@
+<?php
+
+namespace App\Http\Controllers\Admin;
+
+use DB, Cache;
+use App\{
+	AccountInterstitial,
+	DiscoverCategory,
+	DiscoverCategoryHashtag,
+	Hashtag,
+	Media,
+	Profile,
+	Status,
+	StatusHashtag,
+	User
+};
+use App\Models\ConfigCache;
+use App\Models\AutospamCustomTokens;
+use App\Services\AccountService;
+use App\Services\ConfigCacheService;
+use App\Services\StatusService;
+use Carbon\Carbon;
+use Illuminate\Http\Request;
+use Illuminate\Validation\Rule;
+use League\ISO3166\ISO3166;
+use Illuminate\Support\Str;
+use Illuminate\Support\Facades\Storage;
+use Illuminate\Support\Facades\Validator;
+use Illuminate\Support\Facades\Http;
+use App\Http\Controllers\PixelfedDirectoryController;
+use \DateInterval;
+use \DatePeriod;
+use App\Http\Resources\AdminSpamReport;
+use App\Util\Lexer\Classifier;
+use App\Jobs\AutospamPipeline\AutospamPretrainPipeline;
+use App\Jobs\AutospamPipeline\AutospamPretrainNonSpamPipeline;
+use App\Jobs\AutospamPipeline\AutospamUpdateCachedDataPipeline;
+use Illuminate\Support\Facades\URL;
+use App\Services\AutospamService;
+
+trait AdminAutospamController
+{
+	public function autospamHome(Request $request)
+	{
+		return view('admin.autospam.home');
+	}
+
+	public function getAutospamConfigApi(Request $request)
+	{
+		$open = Cache::remember('admin-dash:reports:spam-count', 3600, function() {
+			return AccountInterstitial::whereType('post.autospam')->whereNull('appeal_handled_at')->count();
+		});
+
+		$closed = Cache::remember('admin-dash:reports:spam-count-closed', 3600, function() {
+			return AccountInterstitial::whereType('post.autospam')->whereNotNull('appeal_handled_at')->count();
+		});
+
+		$thisWeek = Cache::remember('admin-dash:reports:spam-count-stats-this-week ', 86400, function() {
+			$sr = config('database.default') == 'pgsql' ? "to_char(created_at, 'MM-YYYY')" : "DATE_FORMAT(created_at, '%m-%Y')";
+			$gb = config('database.default') == 'pgsql' ? [DB::raw($sr)] : DB::raw($sr);
+			$s = AccountInterstitial::select(
+					DB::raw('count(id) as count'),
+					DB::raw($sr . " as month_year")
+				)
+				->where('created_at', '>=', now()->subWeeks(52))
+				->groupBy($gb)
+				->get()
+				->map(function($s) {
+					$dt = now()->parse('01-' . $s->month_year);
+					return [
+						'id' => $dt->format('Ym'),
+						'x' => $dt->format('M Y'),
+						'y' => $s->count
+					];
+				})
+				->sortBy('id')
+				->values()
+				->toArray();
+			return $s;
+		});
+
+		$files = [
+			'spam' => [
+				'exists' => Storage::exists(AutospamService::MODEL_SPAM_PATH),
+				'size' => 0
+			],
+			'ham' => [
+				'exists' => Storage::exists(AutospamService::MODEL_HAM_PATH),
+				'size' => 0
+			],
+			'combined' => [
+				'exists' => Storage::exists(AutospamService::MODEL_FILE_PATH),
+				'size' => 0
+			]
+		];
+
+		if($files['spam']['exists']) {
+			$files['spam']['size'] = Storage::size(AutospamService::MODEL_SPAM_PATH);
+		}
+
+		if($files['ham']['exists']) {
+			$files['ham']['size'] = Storage::size(AutospamService::MODEL_HAM_PATH);
+		}
+
+		if($files['combined']['exists']) {
+			$files['combined']['size'] = Storage::size(AutospamService::MODEL_FILE_PATH);
+		}
+
+		return [
+			'autospam_enabled' => (bool) config_cache('pixelfed.bouncer.enabled') ?? false,
+			'nlp_enabled' => (bool) AutospamService::active(),
+			'files' => $files,
+			'open' => $open,
+			'closed' => $closed,
+			'graph' => collect($thisWeek)->map(fn($s) => $s['y'])->values(),
+			'graphLabels' => collect($thisWeek)->map(fn($s) => $s['x'])->values()
+		];
+	}
+
+    public function getAutospamReportsClosedApi(Request $request)
+    {
+		$appeals = AdminSpamReport::collection(
+			AccountInterstitial::orderBy('id', 'desc')
+			->whereType('post.autospam')
+			->whereIsSpam(true)
+			->whereNotNull('appeal_handled_at')
+			->cursorPaginate(6)
+			->withQueryString()
+		);
+
+		return $appeals;
+    }
+
+    public function postAutospamTrainSpamApi(Request $request)
+    {
+    	$aiCount = AccountInterstitial::whereItemType('App\Status')
+        	->whereIsSpam(true)
+        	->count();
+        abort_if($aiCount < 100, 422, 'You don\'t have enough data to pre-train against.');
+
+        $existing = Cache::get('pf:admin:autospam:pretrain:recent');
+        abort_if($existing, 422, 'You\'ve already run this recently, please wait 30 minutes before pre-training again');
+        AutospamPretrainPipeline::dispatch();
+        Cache::put('pf:admin:autospam:pretrain:recent', 1, 1440);
+
+        return [
+        	'msg' => 'Success!'
+        ];
+    }
+
+    public function postAutospamTrainNonSpamSearchApi(Request $request)
+    {
+    	$this->validate($request, [
+    		'q' => 'required|string|min:1'
+    	]);
+
+    	$q = $request->input('q');
+
+    	$res = Profile::whereNull(['status', 'domain'])
+    		->where('username', 'like', '%' . $q . '%')
+    		->orderByDesc('followers_count')
+    		->take(10)
+    		->get()
+    		->map(function($p) {
+    			$acct = AccountService::get($p->id, true);
+    			return [
+    				'id' => (string) $p->id,
+    				'avatar' => $acct['avatar'],
+    				'username' => $p->username
+    			];
+    		})
+    		->values();
+    	return $res;
+    }
+
+    public function postAutospamTrainNonSpamSubmitApi(Request $request)
+    {
+    	$this->validate($request, [
+    		'accounts' => 'required|array|min:1|max:10'
+    	]);
+
+    	$accts = $request->input('accounts');
+
+    	$accounts = Profile::whereNull(['domain', 'status'])->find(collect($accts)->map(function($a) { return $a['id'];}));
+
+    	abort_if(!$accounts || !$accounts->count(), 422, 'One or more of the selected accounts are not valid');
+
+    	AutospamPretrainNonSpamPipeline::dispatch($accounts);
+    	return $accounts;
+    }
+
+    public function getAutospamCustomTokensApi(Request $request)
+    {
+		return AutospamCustomTokens::latest()->cursorPaginate(6);
+    }
+
+    public function saveNewAutospamCustomTokensApi(Request $request)
+    {
+    	$this->validate($request, [
+    		'token' => 'required|unique:autospam_custom_tokens,token',
+    	]);
+
+    	$ct = new AutospamCustomTokens;
+    	$ct->token = $request->input('token');
+    	$ct->weight = $request->input('weight');
+    	$ct->category = $request->input('category') === 'spam' ? 'spam' : 'ham';
+    	$ct->note = $request->input('note');
+    	$ct->active = $request->input('active');
+    	$ct->save();
+
+    	AutospamUpdateCachedDataPipeline::dispatch();
+    	return $ct;
+    }
+
+    public function updateAutospamCustomTokensApi(Request $request)
+    {
+    	$this->validate($request, [
+    		'id' => 'required',
+    		'token' => 'required',
+    		'category' => 'required|in:spam,ham',
+    		'active' => 'required|boolean'
+    	]);
+
+    	$ct = AutospamCustomTokens::findOrFail($request->input('id'));
+    	$ct->weight = $request->input('weight');
+    	$ct->category = $request->input('category');
+    	$ct->note = $request->input('note');
+    	$ct->active = $request->input('active');
+    	$ct->save();
+
+    	AutospamUpdateCachedDataPipeline::dispatch();
+
+    	return $ct;
+    }
+
+    public function exportAutospamCustomTokensApi(Request $request)
+    {
+    	abort_if(!Storage::exists(AutospamService::MODEL_SPAM_PATH), 422, 'Autospam Dataset does not exist, please train spam before attempting to export');
+    	return Storage::download(AutospamService::MODEL_SPAM_PATH);
+    }
+
+    public function enableAutospamApi(Request $request)
+    {
+    	ConfigCacheService::put('autospam.nlp.enabled', true);
+    	Cache::forget(AutospamService::CHCKD_CACHE_KEY);
+    	return ['msg' => 'Success'];
+    }
+
+    public function disableAutospamApi(Request $request)
+    {
+    	ConfigCacheService::put('autospam.nlp.enabled', false);
+    	Cache::forget(AutospamService::CHCKD_CACHE_KEY);
+    	return ['msg' => 'Success'];
+    }
+}

app/Http/Controllers/AdminController.php

@@ -21,6 +21,7 @@ use Carbon\Carbon;
 use Illuminate\Http\Request;
 use Illuminate\Support\Facades\Redis;
 use App\Http\Controllers\Admin\{
+	AdminAutospamController,
 	AdminDirectoryController,
 	AdminDiscoverController,
 	AdminHashtagsController,
@@ -43,6 +44,7 @@ use App\Models\CustomEmoji;
 class AdminController extends Controller
 {
 	use AdminReportController,
+	AdminAutospamController,
 	AdminDirectoryController,
 	AdminDiscoverController,
 	AdminHashtagsController,

app/Jobs/AutospamPipeline/AutospamPretrainNonSpamPipeline.php

@@ -0,0 +1,58 @@
+<?php
+
+namespace App\Jobs\AutospamPipeline;
+
+use Illuminate\Bus\Queueable;
+use Illuminate\Contracts\Queue\ShouldBeUnique;
+use Illuminate\Contracts\Queue\ShouldQueue;
+use Illuminate\Foundation\Bus\Dispatchable;
+use Illuminate\Queue\InteractsWithQueue;
+use Illuminate\Queue\SerializesModels;
+use App\Util\Lexer\Classifier;
+use App\AccountInterstitial;
+use App\Profile;
+use App\Status;
+use Illuminate\Support\Facades\Storage;
+use App\Services\AutospamService;
+
+class AutospamPretrainNonSpamPipeline implements ShouldQueue
+{
+	use Dispatchable, InteractsWithQueue, Queueable, SerializesModels;
+
+	public $classifier;
+	public $accounts;
+
+	/**
+	 * Create a new job instance.
+	 */
+	public function __construct($accounts)
+	{
+		$this->accounts = $accounts;
+		$this->classifier = new Classifier();
+	}
+
+	/**
+	 * Execute the job.
+	 */
+	public function handle(): void
+	{
+		$classifier = $this->classifier;
+		$accounts = $this->accounts;
+
+        foreach($accounts as $acct) {
+        	Status::whereNotNull('caption')
+        		->whereScope('public')
+        		->whereProfileId($acct->id)
+        		->inRandomOrder()
+        		->take(400)
+        		->pluck('caption')
+        		->each(function($c) use ($classifier) {
+        			$classifier->learn($c, 'ham');
+        		});
+        }
+
+		Storage::put(AutospamService::MODEL_HAM_PATH, $classifier->export());
+
+        AutospamUpdateCachedDataPipeline::dispatch()->delay(5);
+	}
+}

app/Jobs/AutospamPipeline/AutospamPretrainPipeline.php

@@ -0,0 +1,63 @@
+<?php
+
+namespace App\Jobs\AutospamPipeline;
+
+use Illuminate\Bus\Queueable;
+use Illuminate\Contracts\Queue\ShouldBeUnique;
+use Illuminate\Contracts\Queue\ShouldQueue;
+use Illuminate\Foundation\Bus\Dispatchable;
+use Illuminate\Queue\InteractsWithQueue;
+use Illuminate\Queue\SerializesModels;
+use App\Util\Lexer\Classifier;
+use App\AccountInterstitial;
+use App\Status;
+use Illuminate\Support\Facades\Storage;
+use App\Services\AutospamService;
+
+class AutospamPretrainPipeline implements ShouldQueue
+{
+    use Dispatchable, InteractsWithQueue, Queueable, SerializesModels;
+
+    public $classifier;
+
+    /**
+     * Create a new job instance.
+     */
+    public function __construct()
+    {
+        $this->classifier = new Classifier();
+    }
+
+    /**
+     * Execute the job.
+     */
+    public function handle(): void
+    {
+        $classifier = $this->classifier;
+
+        $aiCount = AccountInterstitial::whereItemType('App\Status')
+        	->whereIsSpam(true)
+        	->count();
+
+		if($aiCount < 100) {
+			return;
+		}
+
+        AccountInterstitial::whereItemType('App\Status')
+        	->whereIsSpam(true)
+        	->inRandomOrder()
+        	->take(config('autospam.nlp.spam_sample_limit'))
+        	->pluck('item_id')
+        	->each(function ($ai) use($classifier) {
+        		$status = Status::whereNotNull('caption')->find($ai);
+        		if(!$status) {
+        			return;
+        		}
+        		$classifier->learn($status->caption, 'spam');
+        	});
+
+        Storage::put(AutospamService::MODEL_SPAM_PATH, $classifier->export());
+
+        AutospamUpdateCachedDataPipeline::dispatch()->delay(5);
+    }
+}

app/Jobs/AutospamPipeline/AutospamUpdateCachedDataPipeline.php

@@ -0,0 +1,79 @@
+<?php
+
+namespace App\Jobs\AutospamPipeline;
+
+use Illuminate\Bus\Queueable;
+use Illuminate\Contracts\Queue\ShouldBeUnique;
+use Illuminate\Contracts\Queue\ShouldQueue;
+use Illuminate\Foundation\Bus\Dispatchable;
+use Illuminate\Queue\InteractsWithQueue;
+use Illuminate\Queue\SerializesModels;
+use App\Models\AutospamCustomTokens;
+use Illuminate\Support\Facades\Storage;
+use App\Services\AutospamService;
+use Cache;
+
+class AutospamUpdateCachedDataPipeline implements ShouldQueue
+{
+	use Dispatchable, InteractsWithQueue, Queueable, SerializesModels;
+
+	/**
+	 * Create a new job instance.
+	 */
+	public function __construct()
+	{
+	}
+
+    /**
+     * Execute the job.
+     */
+    public function handle(): void
+    {
+        $spam = json_decode(Storage::get(AutospamService::MODEL_SPAM_PATH), true);
+    	$newSpam = AutospamCustomTokens::whereCategory('spam')->get();
+    	foreach($newSpam as $ns) {
+    		$key = strtolower($ns->token);
+    		if(isset($spam['words']['spam'][$key])) {
+    			$spam['words']['spam'][$key] = $spam['words']['spam'][$key] + $ns->weight;
+    		} else {
+    			$spam['words']['spam'][$key] = $ns->weight;
+    		}
+    	}
+    	$newSpamCount = count($spam['words']['spam']);
+    	$spam['documents']['spam'] = $newSpamCount;
+    	arsort($spam['words']['spam']);
+    	Storage::put(AutospamService::MODEL_SPAM_PATH, json_encode($spam, JSON_UNESCAPED_SLASHES|JSON_PRETTY_PRINT));
+
+    	$ham = json_decode(Storage::get(AutospamService::MODEL_HAM_PATH), true);
+    	$newHam = AutospamCustomTokens::whereCategory('ham')->get();
+    	foreach($newHam as $ns) {
+    		$key = strtolower($ns->token);
+    		if(isset($spam['words']['ham'][$key])) {
+    			$ham['words']['ham'][$key] = $ham['words']['ham'][$key] + $ns->weight;
+    		} else {
+    			$ham['words']['ham'][$key] = $ns->weight;
+    		}
+    	}
+
+    	$newHamCount = count($ham['words']['ham']);
+    	$ham['documents']['ham'] = $newHamCount;
+    	arsort($ham['words']['ham']);
+
+    	Storage::put(AutospamService::MODEL_HAM_PATH, json_encode($ham, JSON_UNESCAPED_SLASHES|JSON_PRETTY_PRINT));
+
+    	$combined = [
+    		'documents' => [
+    			'spam' => $newSpamCount,
+    			'ham' => $newHamCount,
+    		],
+    		'words' => [
+    			'spam' => $spam['words']['spam'],
+    			'ham' => $ham['words']['ham']
+    		]
+    	];
+
+    	Storage::put(AutospamService::MODEL_FILE_PATH, json_encode($combined, JSON_PRETTY_PRINT,JSON_UNESCAPED_SLASHES));
+    	Cache::forget(AutospamService::MODEL_CACHE_KEY);
+    	Cache::forget(AutospamService::CHCKD_CACHE_KEY);
+    }
+}

app/Models/AutospamCustomTokens.php

@@ -0,0 +1,11 @@
+<?php
+
+namespace App\Models;
+
+use Illuminate\Database\Eloquent\Factories\HasFactory;
+use Illuminate\Database\Eloquent\Model;
+
+class AutospamCustomTokens extends Model
+{
+    use HasFactory;
+}

app/Services/AutospamService.php

@@ -0,0 +1,78 @@
+<?php
+
+namespace App\Services;
+
+use Illuminate\Support\Facades\Cache;
+use Illuminate\Support\Facades\Storage;
+use App\Util\Lexer\Classifier;
+
+class AutospamService
+{
+	const CHCKD_CACHE_KEY = 'pf:services:autospam:nlp:checked';
+	const MODEL_CACHE_KEY = 'pf:services:autospam:nlp:model-cache';
+	const MODEL_FILE_PATH = 'nlp/active-training-data.json';
+	const MODEL_SPAM_PATH = 'nlp/spam.json';
+	const MODEL_HAM_PATH  = 'nlp/ham.json';
+
+	public static function check($text)
+	{
+		if(!$text || strlen($text) == 0) {
+			false;
+		}
+		if(!self::active()) {
+			return null;
+		}
+		$model = self::getCachedModel();
+		$classifier = new Classifier;
+		$classifier->import($model['documents'], $model['words']);
+		return $classifier->most($text) === 'spam';
+	}
+
+	public static function eligible()
+	{
+		return Cache::remember(self::CHCKD_CACHE_KEY, 86400, function() {
+			if(!config_cache('pixelfed.bouncer.enabled') || !config('autospam.enabled')) {
+				return false;
+			}
+
+			if(!Storage::exists(self::MODEL_SPAM_PATH)) {
+				return false;
+			}
+
+			if(!Storage::exists(self::MODEL_HAM_PATH)) {
+				return false;
+			}
+
+			if(!Storage::exists(self::MODEL_FILE_PATH)) {
+				return false;
+			} else {
+				if(Storage::size(self::MODEL_FILE_PATH) < 1000) {
+					return false;
+				}
+			}
+
+			return true;
+		});
+	}
+
+	public static function active()
+	{
+		return config_cache('autospam.nlp.enabled') && self::eligible();
+	}
+
+	public static function getCachedModel()
+	{
+		if(!self::active()) {
+			return null;
+		}
+
+		return Cache::remember(self::MODEL_CACHE_KEY, 86400, function() {
+			$res = Storage::get(self::MODEL_FILE_PATH);
+			if(!$res || empty($res)) {
+				return null;
+			}
+
+			return json_decode($res, true);
+		});
+	}
+}

app/Services/ConfigCacheService.php

@@ -69,7 +69,9 @@ class ConfigCacheService
 				'instance.landing.show_directory',
 				'instance.landing.show_explore',
 				'instance.admin.pid',
-				'instance.banner.blurhash'
+				'instance.banner.blurhash',
+
+				'autospam.nlp.enabled',
 				// 'system.user_mode'
 			];
 

app/Util/Lexer/Classifier.php

@@ -0,0 +1,178 @@
+<?php
+
+namespace App\Util\Lexer;
+
+use Brick\Math\BigDecimal;
+use Illuminate\Support\Collection;
+use Illuminate\Support\Str;
+
+class Classifier
+{
+	/**
+	 * @var ?callable(string): array<int, string>
+	 */
+	private $tokenizer;
+
+	/**
+	 * @var array<string, array<string, int>>
+	 */
+	private array $words = [];
+
+	/**
+	 * @var array<string, int>
+	 */
+	private array $documents = [];
+
+	private bool $uneven = false;
+
+	/**
+	 * @param callable(string): array<int, string> $tokenizer
+	 */
+	public function setTokenizer(callable $tokenizer): void
+	{
+		$this->tokenizer = $tokenizer;
+	}
+
+	/**
+	 * @return Collection<int, string>
+	 */
+	public function tokenize(string $string): Collection
+	{
+		if ($this->tokenizer) {
+			/** @var array<int, string> */
+			$tokens = call_user_func($this->tokenizer, $string);
+
+			return collect($tokens);
+		}
+
+		return Str::of($string)
+			->lower()
+			->matchAll('/[[:alpha:]]+/u');
+	}
+
+	/**
+	 * @return $this
+	 */
+	public function learn(string $statement, string $type): self
+	{
+		foreach ($this->tokenize($statement) as $word) {
+			$this->incrementWord($type, $word);
+		}
+
+		$this->incrementType($type);
+
+		return $this;
+	}
+
+	/**
+	 * @return Collection<string, string>
+	 */
+	public function guess(string $statement): Collection
+	{
+		$words = $this->tokenize($statement);
+
+		return collect($this->documents)
+			->map(function ($count, string $type) use ($words) {
+				$likelihood = $this->pTotal($type);
+
+				foreach ($words as $word) {
+					$likelihood *= $this->p($word, $type);
+				}
+
+				return (string) BigDecimal::of($likelihood);
+			})
+			->sortDesc();
+	}
+
+	public function most(string $statement): string
+	{
+		/** @var string */
+		return $this->guess($statement)->keys()->first();
+	}
+
+	/**
+	 * @return self
+	 */
+	public function uneven(bool $enabled = false): self
+	{
+		$this->uneven = $enabled;
+
+		return $this;
+	}
+
+	/**
+	 * Increment the document count for the type
+	 */
+	private function incrementType(string $type): void
+	{
+		if (! isset($this->documents[$type])) {
+			$this->documents[$type] = 0;
+		}
+
+		$this->documents[$type]++;
+	}
+
+	/**
+	 * Increment the word count for the given type
+	 */
+	private function incrementWord(string $type, string $word): void
+	{
+		$ignored = config('autospam.ignored_tokens');
+		if(!$ignored) {
+			$ignored = ['the', 'a', 'of', 'and'];
+		} else {
+			$ignored = explode(',', $ignored);
+		}
+		if ($type == 'spam' && in_array($word, $ignored)) {
+			return;
+		}
+		if (! isset($this->words[$type][$word])) {
+			$this->words[$type][$word] = 0;
+		}
+
+		$this->words[$type][$word]++;
+	}
+
+	/**
+	 * @return float|int
+	 */
+	private function p(string $word, string $type)
+	{
+		$count = $this->words[$type][$word] ?? 0;
+
+		return ($count + 1) / (array_sum($this->words[$type]) + 1);
+	}
+
+	/**
+	 * @return float|int
+	 */
+	private function pTotal(string $type)
+	{
+		return $this->uneven
+			? ($this->documents[$type] + 1) / (array_sum($this->documents) + 1)
+			: 1;
+	}
+
+	public function export()
+	{
+		$words = $this->words;
+		$words = collect($words)
+			->map(function($w) {
+				arsort($w);
+				return $w;
+			})
+			->all();
+		return json_encode([
+			'_ns' => 'https://pixelfed.org/ns/nlp',
+			'_v' => '1.0',
+			'documents' => $this->documents,
+			'words' => $words
+		], JSON_PRETTY_PRINT|JSON_UNESCAPED_SLASHES);
+	}
+
+	public function import($documents, $words)
+	{
+		$this->documents = $documents;
+		$this->words = $words;
+	}
+}

app/Util/Sentiment/Bouncer.php

@@ -10,6 +10,7 @@ use App\Services\NotificationService;
 use App\Services\StatusService;
 use App\Jobs\ReportPipeline\AutospamNotifyAdminViaEmail;
 use App\Notification;
+use App\Services\AutospamService;
 
 class Bouncer {
 
@@ -65,6 +66,12 @@ class Bouncer {
 			return (new self)->handle($status);
 		}
 
+		if(AutospamService::active()) {
+			if(AutospamService::check($status->caption)) {
+				return (new self)->handle($status);
+			}
+		}
+
 		$recentKey = 'pf:bouncer_v0:recent_by_pid:' . $status->profile_id;
 		$recentTtl = now()->addHours(28);
 

config/autospam.php

@@ -0,0 +1,37 @@
+<?php
+
+return [
+
+	/*
+    |--------------------------------------------------------------------------
+    | Enable Autospam
+    |--------------------------------------------------------------------------
+    |
+    | Autospam uses NLP and other techniques to detect and mitigate potential
+    | spam posts from users on your server.
+    | We recommend enabling this when you have open registrations, regardless
+    | of how many users you have.
+    |
+    */
+
+    'enabled' => env('PF_BOUNCER_ENABLED', false),
+
+
+    /*
+    |--------------------------------------------------------------------------
+    | Ignored Tokens
+    |--------------------------------------------------------------------------
+    |
+    | Ignored tokens are for commonly used words that may impact the detection.
+    | These tokens should be lowercase and not contain spaces or non alpha-
+    | numerical characters and be in comma-separated format.
+    |
+    */
+
+    'ignored_tokens' => env('PF_AUTOSPAM_IGNORED_TOKENS', 'the,a,of,and'),
+
+    'nlp' => [
+    	'enabled' => false,
+    	'spam_sample_limit' => env('PF_AUTOSPAM_NLP_SPAM_SAMPLE_LIMIT', 200),
+    ]
+];

database/migrations/2023_05_15_050604_create_autospam_custom_tokens_table.php

@@ -0,0 +1,34 @@
+<?php
+
+use Illuminate\Database\Migrations\Migration;
+use Illuminate\Database\Schema\Blueprint;
+use Illuminate\Support\Facades\Schema;
+
+return new class extends Migration
+{
+    /**
+     * Run the migrations.
+     */
+    public function up(): void
+    {
+        Schema::create('autospam_custom_tokens', function (Blueprint $table) {
+            $table->bigIncrements('id');
+            $table->string('token')->index();
+            $table->integer('weight')->default(1)->index();
+            $table->boolean('is_spam')->default(true)->index();
+            $table->text('note')->nullable();
+            $table->string('category')->nullable()->index();
+            $table->boolean('active')->default(false)->index();
+            $table->unique(['token', 'category']);
+            $table->timestamps();
+        });
+    }
+
+    /**
+     * Reverse the migrations.
+     */
+    public function down(): void
+    {
+        Schema::dropIfExists('autospam_custom_tokens');
+    }
+};

resources/views/admin/autospam/home.blade.php

@@ -0,0 +1,12 @@
+@extends('admin.partial.template-full')
+
+@section('section')
+</div>
+<admin-autospam />
+@endsection
+
+@push('scripts')
+<script type="text/javascript">
+    new Vue({ el: '#panel'});
+</script>
+@endpush

resources/views/admin/partial/sidenav.blade.php

@@ -15,6 +15,13 @@
 						</a>
 					</li>
 
+					<li class="nav-item">
+						<a class="nav-link {{request()->is('*autospam*')?'active':''}}" href="{{route('admin.autospam')}}">
+							<i class="ni ni-bold-right text-primary"></i>
+							<span class="nav-link-text">Autospam <span class="badge badge-primary ml-1">NEW</span></span>
+						</a>
+					</li>
+
 					<li class="nav-item">
 						<a class="nav-link {{request()->is('*messages*')?'active':''}}" href="{{route('admin.messages')}}">
 							<i class="ni ni-bold-right text-primary"></i>
@@ -36,13 +43,6 @@
 						</a>
 					</li>
 
-					<li class="nav-item">
-						<a class="nav-link {{request()->is('*stories*')?'active':''}}" href="{{route('admin.stories')}}">
-							<i class="ni ni-bold-right text-primary"></i>
-							<span class="nav-link-text">Stories <span class="badge badge-primary ml-1">NEW</span></span>
-						</a>
-					</li>
-
 					<li class="nav-item">
 						<a class="nav-link {{request()->is('*users*')?'active':''}}" href="{{route('admin.users')}}">
 							<i class="ni ni-bold-right text-primary"></i>
@@ -60,22 +60,7 @@
 
 				<hr class="my-3">
 
-				<ul class="navbar-nav mb-md-3">
+				<ul class="navbar-nav">
-
-                    <li class="nav-item">
-                        <a class="nav-link {{request()->is('*directory*')?'active':''}}" href="{{route('admin.directory')}}">
-                            <i class="ni ni-bold-right text-primary"></i>
-                            <span class="nav-link-text">Directory <span class="badge badge-primary ml-1">NEW</span></span>
-                        </a>
-                    </li>
-
-					<li class="nav-item">
-						<a class="nav-link {{request()->is('*apps*')?'active':''}}" href="{{route('admin.apps')}}">
-							<i class="ni ni-bold-right text-primary"></i>
-							<span class="nav-link-text">Apps</span>
-						</a>
-					</li>
-
 					<li class="nav-item">
 						<a class="nav-link {{request()->is('*custom-emoji*')?'active':''}}" href="{{route('admin.custom-emoji')}}">
 							<i class="ni ni-bold-right text-primary"></i>
@@ -83,13 +68,6 @@
 						</a>
 					</li>
 
-					<li class="nav-item">
-						<a class="nav-link {{request()->is('*diagnostics*')?'active':''}}" href="{{route('admin.diagnostics')}}">
-							<i class="ni ni-bold-right text-primary"></i>
-							<span class="nav-link-text">Diagnostics <span class="badge badge-primary ml-1">NEW</span></span>
-						</a>
-					</li>
-
 					<li class="nav-item">
 						<a class="nav-link {{request()->is('*hashtags*')?'active':''}}" href="{{route('admin.hashtags')}}">
 							<i class="ni ni-bold-right text-primary"></i>
@@ -97,13 +75,6 @@
 						</a>
 					</li>
 
-					<li class="nav-item">
-						<a class="nav-link" href="/horizon">
-							<i class="ni ni-bold-right text-primary"></i>
-							<span class="nav-link-text">Horizon</span>
-						</a>
-					</li>
-
 					<li class="nav-item">
 						<a class="nav-link {{request()->is('*instances*')?'active':''}}" href="{{route('admin.instances')}}">
 							<i class="ni ni-bold-right text-primary"></i>
@@ -119,16 +90,57 @@
 					</li>
 
 					<li class="nav-item">
-						<a class="nav-link {{request()->is('*site-news*')?'active':''}}" href="/i/admin/site-news">
+						<a class="nav-link {{request()->is('*profiles*')?'active':''}}" href="/i/admin/profiles">
 							<i class="ni ni-bold-right text-primary"></i>
-							<span class="nav-link-text">Newsroom</span>
+							<span class="nav-link-text">Profiles</span>
 						</a>
 					</li>
 
 					<li class="nav-item">
-						<a class="nav-link {{request()->is('*profiles*')?'active':''}}" href="/i/admin/profiles">
+						<a class="nav-link {{request()->is('*stories*')?'active':''}}" href="{{route('admin.stories')}}">
 							<i class="ni ni-bold-right text-primary"></i>
-							<span class="nav-link-text">Profiles</span>
+							<span class="nav-link-text">Stories <span class="badge badge-primary ml-1">NEW</span></span>
+						</a>
+					</li>
+				</ul>
+
+				<hr class="my-3">
+
+				<ul class="navbar-nav mb-md-3">
+
+					<li class="nav-item">
+						<a class="nav-link {{request()->is('*apps*')?'active':''}}" href="{{route('admin.apps')}}">
+							<i class="ni ni-bold-right text-primary"></i>
+							<span class="nav-link-text">Apps</span>
+						</a>
+					</li>
+
+					<li class="nav-item">
+						<a class="nav-link {{request()->is('*diagnostics*')?'active':''}}" href="{{route('admin.diagnostics')}}">
+							<i class="ni ni-bold-right text-primary"></i>
+							<span class="nav-link-text">Diagnostics <span class="badge badge-primary ml-1">NEW</span></span>
+						</a>
+					</li>
+
+                    <li class="nav-item">
+                        <a class="nav-link {{request()->is('*directory*')?'active':''}}" href="{{route('admin.directory')}}">
+                            <i class="ni ni-bold-right text-primary"></i>
+                            <span class="nav-link-text">Directory <span class="badge badge-primary ml-1">NEW</span></span>
+                        </a>
+                    </li>
+
+					<li class="nav-item">
+						<a class="nav-link" href="/horizon">
+							<i class="ni ni-bold-right text-primary"></i>
+							<span class="nav-link-text">Horizon</span>
+						</a>
+					</li>
+
+
+					<li class="nav-item">
+						<a class="nav-link {{request()->is('*site-news*')?'active':''}}" href="/i/admin/site-news">
+							<i class="ni ni-bold-right text-primary"></i>
+							<span class="nav-link-text">Newsroom</span>
 						</a>
 					</li>
 
@@ -152,7 +164,6 @@
 							<span class="nav-link-text">System</span>
 						</a>
 					</li>
-
 				</ul>
 			</div>
 		</div>

resources/views/admin/settings/sidebar.blade.php

@@ -1,31 +1,76 @@
-@section('menu')
+  <div class="col-12 col-md-3">
     <ul class="nav flex-column settings-nav">
-  <li class="nav-item pl-3">
+      <li class="nav-item pl-3 {{request()->is('settings/home')?'active':''}}">
-    <a class="nav-link text-muted {{request()->is('*settings') ? 'font-weight-bold':''}}" href="{{route('admin.settings')}}">Home</a>
+        <a class="nav-link font-weight-light  text-muted" href="{{route('settings')}}">Account</a>
       </li>
-{{--   <li class="nav-item pl-3">
+      {{-- <li class="nav-item pl-3 {{request()->is('settings/accessibility')?'active':''}}">
-    <a class="nav-link text-muted {{request()->is('*settings/backups') ? 'font-weight-bold':''}}" href="{{route('admin.settings.backups')}}">Backups</a>
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.accessibility')}}">Accessibility</a>
-  </li> --}}
-  {{-- <li class="nav-item pl-3">
-    <a class="nav-link text-muted {{request()->is('*settings/config') ? 'font-weight-bold':''}}" href="{{route('admin.settings.config')}}">Configuration</a>
-  </li> --}}
-  {{-- <li class="nav-item pl-3">
-    <a class="nav-link text-muted {{request()->is('*settings/customize') ? 'font-weight-bold':''}}" href="{{route('admin.settings.customize')}}">Customize</a>
-  </li> --}}
-  {{-- <li class="nav-item pl-3">
-    <a class="nav-link text-muted {{request()->is('*settings/features') ? 'font-weight-bold':''}}" href="{{route('admin.settings.features')}}">Features</a>
-  </li> --}}
-  {{-- <li class="nav-item pl-3">
-    <a class="nav-link text-muted {{request()->is('*settings/maintenance') ? 'font-weight-bold':''}}" href="{{route('admin.settings.maintenance')}}">Maintenance</a>
-  </li> --}}
-  <li class="nav-item pl-3">
-    <a class="nav-link text-muted {{request()->is('*settings/page*') ? 'font-weight-bold':''}}" href="{{route('admin.settings.pages')}}">Pages</a>
       </li>
-  {{-- <li class="nav-item pl-3">
+      <li class="nav-item pl-3 {{request()->is('settings/email')?'active':''}}">
-    <a class="nav-link text-muted {{request()->is('*settings/storage') ? 'font-weight-bold':''}}" href="{{route('admin.settings.storage')}}">Storage</a>
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.email')}}">Email</a>
       </li> --}}
-  <li class="nav-item pl-3">
+      @if(config('pixelfed.user_invites.enabled'))
-    <a class="nav-link text-muted {{request()->is('*settings/system') ? 'font-weight-bold':''}}" href="{{route('admin.settings.system')}}">System</a>
+      <li class="nav-item pl-3 {{request()->is('settings/invites*')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.invites')}}">Invites</a>
+      </li>
+      @endif
+      <li class="nav-item pl-3 {{request()->is('settings/notifications')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.notifications')}}">Notifications</a>
+      </li>
+      {{-- <li class="nav-item pl-3 {{request()->is('settings/password')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.password')}}">Password</a>
+      </li> --}}
+      <li class="nav-item pl-3 {{request()->is('settings/privacy*')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.privacy')}}">Privacy</a>
+      </li>
+      {{-- <li class="nav-item pl-3 {{request()->is('settings/relationships*')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.relationships')}}">Relationships</a>
+      </li>
+      <li class="nav-item pl-3 {{request()->is('settings/reports*')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.reports')}}">Reports</a>
+      </li> --}}
+
+      {{-- <li class="nav-item pl-3 {{request()->is('settings/safety*')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.security')}}">Safety</a>
+      </li> --}}
+      <li class="nav-item pl-3 {{request()->is('settings/security*')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.security')}}">Security</a>
+      </li>
+      {{-- <li class="nav-item pl-3 {{request()->is('settings/sponsor*')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.sponsor')}}">Sponsor</a>
+      </li> --}}
+      <li class="nav-item">
+        <hr>
+      </li>
+      {{-- <li class="nav-item pl-3 {{request()->is('*import*')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.import')}}">Import</a>
+      </li> --}}
+      <li class="nav-item pl-3 {{request()->is('settings/import*')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.import')}}">Import</a>
+      </li>
+      <li class="nav-item pl-3 {{request()->is('settings/data-export')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.dataexport')}}">Data Export</a>
+      </li>
+
+      @if(config('pixelfed.oauth_enabled') == true)
+      {{-- <li class="nav-item pl-3 {{request()->is('settings/applications')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.applications')}}">Applications</a>
+      </li> --}}
+      <li class="nav-item pl-3 {{request()->is('settings/developers')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.developers')}}">Developers</a>
+      </li>
+      @endif
+
+
+      <li class="nav-item pl-3 {{request()->is('settings/labs*')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.labs')}}">Labs</a>
+      </li>
+
+      <li class="nav-item pl-3 {{request()->is('settings/arch*')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.labs')}}">Archived Posts</a>
+      </li>
+      <li class="nav-item pl-3 {{request()->is('settings/moglod*')?'active':''}}">
+        <a class="nav-link font-weight-light text-muted" href="{{route('settings.labs')}}">Moderation Log</a>
       </li>
     </ul>
-@endsection
+  </div>

routes/web.php

@@ -94,6 +94,8 @@ Route::domain(config('pixelfed.domain.admin'))->prefix('i/admin')->group(functio
 
 	Route::get('directory/home', 'AdminController@directoryHome')->name('admin.directory');
 
+	Route::get('autospam/home', 'AdminController@autospamHome')->name('admin.autospam');
+
 	Route::prefix('api')->group(function() {
 		Route::get('stats', 'AdminController@getStats');
 		Route::get('accounts', 'AdminController@getAccounts');
@@ -129,6 +131,17 @@ Route::domain(config('pixelfed.domain.admin'))->prefix('i/admin')->group(functio
 		Route::get('reports/spam/all', 'AdminController@reportsApiSpamAll');
 		Route::get('reports/spam/get/{id}', 'AdminController@reportsApiSpamGet');
 		Route::post('reports/spam/handle', 'AdminController@reportsApiSpamHandle');
+		Route::post('autospam/config', 'AdminController@getAutospamConfigApi');
+		Route::post('autospam/reports/closed', 'AdminController@getAutospamReportsClosedApi');
+		Route::post('autospam/train', 'AdminController@postAutospamTrainSpamApi');
+		Route::post('autospam/search/non-spam', 'AdminController@postAutospamTrainNonSpamSearchApi');
+		Route::post('autospam/train/non-spam', 'AdminController@postAutospamTrainNonSpamSubmitApi');
+		Route::post('autospam/tokens/custom', 'AdminController@getAutospamCustomTokensApi');
+		Route::post('autospam/tokens/store', 'AdminController@saveNewAutospamCustomTokensApi');
+		Route::post('autospam/tokens/update', 'AdminController@updateAutospamCustomTokensApi');
+		Route::post('autospam/tokens/export', 'AdminController@exportAutospamCustomTokensApi');
+		Route::post('autospam/config/enable', 'AdminController@enableAutospamApi');
+		Route::post('autospam/config/disable', 'AdminController@disableAutospamApi');
 	});
 });
 

storage/app/.gitignore

@@ -2,5 +2,6 @@
 !backups/
 !public/
 !remcache/
+!nlp/
 !cities.json
 !.gitignore

storage/app/nlp/.gitignore

@@ -0,0 +1,2 @@
+*
+!.gitignore