error.php: WIP error_handler implementation

inc/config.php

@@ -943,6 +943,10 @@
     // Location of thumbnail to use for deleted images.
     $config['image_deleted'] = 'static/deleted.png';
 
+    // When a thumbnailed image is going to be the same (in dimension), just copy the entire file and use
+    // that as a thumbnail instead of resizing/redrawing.
+    $config['minimum_copy_resize'] = false;
+
     // Maximum image upload size in bytes.
     $config['max_filesize'] = 10 * 1024 * 1024; // 10MB
     // Maximum image dimensions.
@@ -981,6 +985,15 @@
     // Set this to true if you're using Linux and you can execute `md5sum` binary.
     $config['gnu_md5'] = false;
 
+    // Use Tesseract OCR to retrieve text from images, so you can use it as a spamfilter.
+    $config['tesseract_ocr'] = false;
+
+    // Tesseract parameters
+    $config['tesseract_params'] = '';
+
+    // Tesseract preprocess command
+    $config['tesseract_preprocess_command'] = 'convert -monochrome %s -';
+
     // Number of posts in a "View Last X Posts" page
     $config['noko50_count'] = 50;
     // Number of posts a thread needs before it gets a "View Last X Posts" page.

inc/error.php

@@ -1,9 +1,222 @@
 <?php
+namespace Vichan;
 
-function error_handler($errno, $errstr, $errfile, $errline) {
-	if (error_reporting() & $errno) {
-		$config['debug'] = true;
-		error("$errstr in $errfile at line $errline");
+class ErrorHandler {
+	private static array|false $error_recursion = false;
+
+	private bool $include_details;
+	/**
+	 * @var callable
+	 */
+	private mixed $logger_fn;
+
+
+	private static function isCli(): bool {
+		return defined('STDIN') || \PHP_SAPI === 'cli' || (\stristr(\PHP_SAPI, 'cgi') && \getenv('TERM'))
+			|| (empty($_SERVER['REMOTE_ADDR']) && !isset($_SERVER['HTTP_USER_AGENT']) && \count($_SERVER['argv']) > 0);
+	}
+
+	private static function errnoToLogLevel(int $errno): int {
+		switch ($errno) {
+			default:
+			case \E_ERROR:
+			case \E_PARSE:
+			case \E_CORE_ERROR:
+			case \E_COMPILE_ERROR:
+			case \E_RECOVERABLE_ERROR:
+				return \LOG_EMERG;
+			case \E_USER_ERROR:
+				return \LOG_ERR;
+			case \E_WARNING:
+			case \E_CORE_WARNING:
+			case \E_COMPILE_WARNING:
+			case \E_USER_WARNING:
+				return \LOG_WARNING;
+			case \E_NOTICE:
+			case \E_DEPRECATED:
+			case \E_USER_NOTICE:
+			case \E_USER_DEPRECATED:
+				return \LOG_NOTICE;
+		}
+	}
+
+	public static function handle_error(
+		int $level,
+		string $message,
+		callable $logger,
+		bool $display_details,
+		string $file,
+		int $line,
+		string $trace_str,
+		array $trace
+	): void {
+		global $board, $mod, $config, $db_error;
+
+		$newline = self::isCli() ? '\n' : '<br>';
+
+		if (self::$error_recursion !== false) {
+			list($o_message, $o_file, $o_line, $o_trace) = self::$error_recursion;
+			if ($display_details) {
+				echo("Error recursion detected with $message at $file:$line. Backtrace:$newline$trace_str{$newline}Original error: $o_message at $o_file:$o_line. Original backtrace:$newline$o_trace");
+			} else {
+				echo("Error recursion detected with $message{$newline}Original error: $o_message");
+			}
+			exit;
+		}
+
+		self::$error_recursion = [ $message, $file, $line, $trace_str ];
+
+		// Message for the system, be it cli or error logging facility.
+		$sys_msg = "Error: $message at $file:$line.\nBacktrace: $trace_str\n";
+
+		if (self::isCli()) {
+			// Running from CLI. Always display the details.
+			echo($sys_msg);
+			exit;
+		}
+
+		// Log the error if we aren't running from cli.
+		$logger($level, $sys_msg);
+
+		// Recycled code.
+		$debug_stuff = [];
+		if ($config['debug']) {
+			if (isset($db_error)) {
+				$debug_stuff = \array_combine([ 'SQLSTATE', 'Error code', 'Error message' ], $db_error);
+			}
+			$debug_stuff['backtrace'] = $trace;
+
+			// Remove the password from SQL errors.
+			$pw = $config['db']['password'];
+			$debug_callback = function($item) use (&$debug_callback, $pw) {
+				if (\is_array($item)) {
+					$item = \array_filter($item, $debug_callback);
+				}
+				return $item !== $pw || !$pw;
+			};
+			$debug_stuff = \array_map($debug_callback, $debug_stuff);
+		}
+
+		\header("{$_SERVER['SERVER_PROTOCOL']} 500 Internal Server Error");
+
+		// Error via json.
+		if (isset($_POST['json_response'])) {
+			\header('Content-Type: text/json; charset=utf-8');
+			$data = [ 'error' => $message ];
+			if (!empty($config) && $config['debug'] && !empty($debug_stuff)) {
+				$data['debug'] = $debug_stuff;
+			}
+			echo(\json_encode($data));
+			exit;
+		}
+
+		// Pretty print.
+		die(Element('page.html', [
+			'config' => $config,
+			'title' => _('Error'),
+			'subtitle' => _('An error has occured.'),
+			'body' => Element('error.html', [
+				'config' => $config,
+				'message' => $message,
+				'mod' => $mod,
+				'board' => isset($board) ? $board : false,
+				'debug' => \str_replace("\n", '&#10;', utf8tohtml(print_r($debug_stuff, true)))
+			])
+		]));
+	}
+
+	private function default_logger(int $level, string $message) {
+		global $config;
+
+		if (isset($config['syslog']) && $config['syslog']) {
+			_syslog($level, $message);
+		} else {
+			\error_log($message);
+		}
+	}
+
+	public function __construct() {
+		$this->include_details = self::isCli();
+		$this->logger_fn = [ __CLASS__, 'default_logger' ];
+	}
+
+	public function setIncludeDetails(bool $include_details) {
+		$this->include_details = $include_details;
+	}
+
+	public function setLogger(?callable $logger_fn) {
+		$this->logger_fn = $logger_fn !== null ? $logger_fn : [ __CLASS__, 'default_logger' ];
+	}
+
+	public function installGlobally() {
+		$logger_fn = $this->logger_fn;
+		$include_details = $this->include_details || self::isCli();
+
+		\set_error_handler(function($errno, $errstr, $errfile, $errline) use ($logger_fn, $include_details) {
+			$errno = \error_reporting() & $errno;
+			if ($errno !== 0) {
+				$level = self::errnoToLogLevel($errno);
+
+				// https://stackoverflow.com/a/35930682
+				$e = new \Exception();
+				$trace_str = $e->getTraceAsString();
+				$trace = $e->getTrace();
+				self::handle_error($level, $errstr, $logger_fn, $include_details, $errfile, $errline, $trace_str, $trace);
+			}
+			return false;
+		});
+
+		\set_exception_handler(function($exception) use ($logger_fn, $include_details) {
+			$message = $exception->getMessage();
+			$file = $exception->getFile();
+			$line = $exception->getLine();
+			$trace_str = $exception->getTraceAsString();
+			$trace = $exception->getTrace();
+
+			self::handle_error(\LOG_EMERG, $message, $logger_fn, $include_details, $file, $line, $trace_str, $trace);
+		});
+
+		\register_shutdown_function(function() use ($logger_fn, $include_details) {
+			$error = \error_get_last();
+			if ($error !== null) {
+				$level = self::errnoToLogLevel($error['type']);
+				$message = $error['message'];
+				$file = $error['file'];
+				$line = $error['line'];
+
+				self::handle_error($level, $message, $logger_fn, $include_details, $file, $line, '<UNAVAILABLE>', []);
+			}
+		});
+	}
+}
+
+/**
+ * Install a fancy error handler with the given logger.
+ *
+ * @param ?callable $logger Called when there is to log something. Use null to use the minimal default implementation.
+ *                          The callable is invoke with two parameters, the first being the log level (integer), as
+ *                          listed by the syslog constants, the second is the log message (string).
+ *                          Consider that $config may not have been initialized yet.
+ */
+function install_error_handler(bool $basic, ?callable $logger_fn): void {
+	global $config;
+
+	$logger_fn ??= function($level, $message) use ($config) {
+		if (isset($config['syslog']) && $config['syslog']) {
+			_syslog($level, $message);
+		} else {
+			\error_log($message);
+		}
+	};
+
+	if (!$basic) {
+		\set_error_handler(function($errno, $errstr, $errfile, $errline) use ($config) {
+			if (\error_reporting() & $errno) {
+				$config['debug'] = true;
+				\error("$errstr in $errfile at line $errline");
+			}
+			return false;
+		});
 	}
 	return false;
 }
@@ -87,7 +300,7 @@ function error($message, $priority = true, $debug_stuff = false) {
 			'message' => $message,
 			'mod' => $mod,
 			'board' => isset($board) ? $board : false,
-			'debug' => str_replace("\n", '&#10;', utf8tohtml(print_r($debug_stuff, true)))
+			'debug' => \str_replace("\n", '&#10;', utf8tohtml(\print_r($debug_stuff, true)))
 		))
 	]));
 }

js/post-menu.js

@@ -104,10 +104,8 @@ function buildMenu(e) {
 
 function addButton(post) {
 	var $ele = $(post);
-	// Use unicode code with ascii variant selector
-	// https://stackoverflow.com/questions/37906969/how-to-prevent-ios-from-converting-ascii-into-emoji
 	$ele.find('input.delete').after(
-		$('<a>', {href: '#', class: 'post-btn', title: 'Post menu'}).text('\u{25B6}\u{fe0e}')
+		$('<a>', {href: '#', class: 'post-btn', title: 'Post menu'}).text('►')
 	);
 }
 

post.php

@@ -1402,13 +1402,13 @@ function handle_post(Context $ctx)
                     $file['thumbwidth'] = $size[0];
                     $file['thumbheight'] = $size[1];
                 } elseif (
-                    (($config['strip_exif'] && isset($file['exif_stripped']) && $file['exif_stripped']) || !$config['strip_exif']) &&
+                    $config['minimum_copy_resize'] &&
                     $image->size->width <= $config['thumb_width'] &&
                     $image->size->height <= $config['thumb_height'] &&
                     $file['extension'] == ($config['thumb_ext'] ? $config['thumb_ext'] : $file['extension'])
                 ) {
                     // Copy, because there's nothing to resize
-                    copy($file['tmp_name'], $file['thumb']);
+                    coopy($file['tmp_name'], $file['thumb']);
 
                     $file['thumbwidth'] = $image->size->width;
                     $file['thumbheight'] = $image->size->height;
@@ -1551,6 +1551,35 @@ function handle_post(Context $ctx)
                 }
             }
 
+            if ($config['tesseract_ocr'] && $file['thumb'] != 'file') {
+                // Let's OCR it!
+                $fname = $file['tmp_name'];
+
+                if ($file['height'] > 500 || $file['width'] > 500) {
+                    $fname = $file['thumb'];
+                }
+
+                if ($fname == 'spoiler') {
+                    // We don't have that much CPU time, do we?
+                } else {
+                    $tmpname = __DIR__ . "/tmp/tesseract/" . rand(0, 10000000);
+
+                    // Preprocess command is an ImageMagick b/w quantization
+                    $error = shell_exec_error(sprintf($config['tesseract_preprocess_command'], escapeshellarg($fname)) . " | " .
+                        'tesseract stdin ' . escapeshellarg($tmpname) . ' ' . $config['tesseract_params']);
+                    $tmpname .= ".txt";
+
+                    $value = @file_get_contents($tmpname);
+                    @unlink($tmpname);
+
+                    if ($value && trim($value)) {
+                        // This one has an effect, that the body is appended to a post body. So you can write a correct
+                        // spamfilter.
+                        $post['body_nomarkup'] .= "<tinyboard ocr image $key>" . htmlspecialchars($value) . "</tinyboard>";
+                    }
+                }
+            }
+
             if (!isset($dont_copy_file) || !$dont_copy_file) {
                 if (isset($file['file_tmp'])) {
                     if (!@rename($file['tmp_name'], $file['file'])) {
@@ -1598,6 +1627,11 @@ function handle_post(Context $ctx)
         }
     }
 
+    // Do filters again if OCRing
+    if ($config['tesseract_ocr'] && !hasPermission($config['mod']['bypass_filters'], $board['uri']) && !$dropped_post) {
+        do_filters($ctx, $post);
+    }
+
     if (!hasPermission($config['mod']['postunoriginal'], $board['uri']) && $config['robot_enable'] && checkRobot($post['body_nomarkup']) && !$dropped_post) {
         undoImage($post);
         if ($config['robot_mute']) {

stylesheets/style.css

@@ -380,7 +380,6 @@ form table tr td div.center {
 
 .file {
   float: left;
-  min-width: 100px;
 }
 
 .file:not(.multifile) .post-image {
@@ -391,10 +390,6 @@ form table tr td div.center {
   float: none;
 }
 
-.file.multifile {
-  margin: 0 10px 0 0;
-}
-
 .file.multifile > p {
   width: 0px;
   min-width: 100%;

templates/themes/categories/frames.html

@@ -96,7 +96,6 @@
               grid-column: 1;
               grid-row: 3;
               width: 100%;
-              word-break: break-all;
             }
             
             .modlog {

templates/themes/categories/news.html

@@ -11,7 +11,7 @@
     .home-description {
       margin: 20px auto 0 auto;
       text-align: center;
-      max-width: 700px;
+      max-width: 700px;"
     }
   </style>
 	{{ boardlist.top }}