Peppy BOM fix

README.md

@@ -17,7 +17,10 @@ Pull requests are welcome! I doubt this project would get any traffic, so feel f
  - [x] Replace all backgrounds with 1x1 black images
  - [x] Remove storyboard specific files
  - [x] Remove background videos
- - [x] Remove junk files (skins, hitsounds, leftovers from mapping etc.)
+ - [x] Remove beatmap skins
+ - [x] Remove beatmap keyed hitsounds
+ - [x] Remove beatmap default hitsounds
+ - [x] Remove junk files (mapping garbage, unused storyboard, random files, etc.)
  - [ ] Export maps to .osz
  - [ ] Delete maps based on star difficulty
  - [ ] Blacklist: the maps that should not be touched by this program

libraries/optimizer.php

@@ -1,6 +1,44 @@
 <?php
 class optimizer
 {
+	public static function is_default_hitsound(string $path) : bool
+	{
+		$string = strtolower(basename($path));
+		$pattern = '(taiko-)?(drum|normal|soft|nightcore)-((hit|slider)(normal|clap|finish|whistle|tick|slide))\d*\.([wW][aA][vV]|[mM][pP]3|[oO][gG][gG])';
+		return mb_ereg_match($pattern, $string);
+	}
+	
+	public static function is_skinnable_sound(string $path) : bool
+	{
+		$string = strtolower(basename($path));
+		// take your time to appreciate this extremely performance penalty inducing line, then look for a skin image one too
+		$pattern = '((heartbeat|seeya|welcome|key-(confirm|delete|movement|press)|back-button-(click|hover)|check-(on|off)|click-(close|short|short-confirm)|(menu|pause)(click|back|hit|-(back|charts|direct|edit|exit|freeplay|multiplayer|options|play|continue|retry|loop))?(-(click|hover))?|select-(expand|difficulty)|shutter|sliderbar|whoosh|match-(confirm|join|leave|(not)?ready|start)|metronomelow|count(\d+s)?|gos|readys|comboburst|combobreak|failsound|section(pass|fail)|applause|spinner(spin|bonus))|(taiko-)?(drum|normal|soft|nightcore)-((hit|slider)(normal|clap|finish|whistle|tick|slide)))-?\d*\.([wW][aA][vV]|[mM][pP]3|[oO][gG][gG])';
+		return mb_ereg_match($pattern, $string);
+	}
+	
+	public static function is_skinnable_image(string $path) : bool
+	{
+		$string = strtolower(basename($path));
+		// take a look at the sound version too
+		$pattern = '(mania-|taiko-?)?((key|note)(\d+|[sS])[dDhHlLtT]?|stage-(right|left|bottom|light|hint)|warningarrow|pippidon(clear|fail|idle|kiai)|bigcircle|drum-(inner|outer)|roll-(middle|end)|bar-(left|right)(-glow)?|mode-(osu|taiko|fruits|mania)(-med|-small)?|menu-(osu|background|snow|back|button-background)|button-(left|middle|right)|cursor(middle|trail|-(smoke|ripple))?|welcome_text|playfield|selection-(selectoptions(-over)?|mod-(novideo|autoplay|cinema|doubletime|easy|fadein|flashlight|halftime|hardrock|hidden|key(\d+|coop)|mirror|nightcore|nofail|perfect|random|relax2?|spunout|suddendeath|target|freemodallowed|touchdevice)|(mod[es]|random|options)(-over)?|tab)|options-offset-tick|play-(skip|unranked|warningarrow)|arrow-(pause|warning)|masking-border|multi-skipped|section-(fail|pass)|count|go|ready|hit|inputoverlay-(background|key)|pause-(overlay|back|continue|replay|retry)|scorebar-(bg|colou?r|ki|kidanger2?|marker)|score(entry)?(-(percent|x))?|ranking-([xXsSaAbBcCdD][hH]?|back|accuracy|graph|maxcombo|panel|perfect|title|replay|retry|winner)(-small)?|fail-background|volume-bg|comboburst(-(fruits|mania))?|default|(approach|hit)circle(select)?|followpoint|lighting[lLnN]?|slider((start|end)circle|(score)?point|followcircle|b(-(nd|spec))?|-(fail|flower-group))?|reversearrow|spinner-(approachcircle|rpm|clear|spin|background|circle|metre|osu|glow|bottom|top|middle2?|warning)|particle|fruit-(catcher-(idle|fail|kiai)|ryuuta|pear|grapes|apple|orange|bananas|drop)|star|coin|fps(-fps)?)(-?overlay)?(-?(\d+|comma|dot)[kg]?)*(@2x)?\.([pP][nN][gG]|[jJ][pP][eE]?[gG])(-effect)?';
+		return mb_ereg_match($pattern, $string);
+	}
+	
+	public static function is_skinnable_other(string $path) : bool
+	{
+		$string = strtolower(basename($path));
+		if (strtolower($string) == "skin.ini") return true;
+		if (mb_stripos($string, "k.ini") !== false) return true; // legacy mania ini
+		return false;
+	}
+	
+	public static function is_skinnable(string $path) : bool
+	{
+		return self::is_skinnable_sound($path) ||
+			self::is_skinnable_image($path) ||
+			self::is_skinnable_other($path);
+	}
+	
 	public static function blacken_image(string $file) : void
 	{
 		$black_png = "./resources/black.png";
@@ -27,7 +65,7 @@ class optimizer
 	{
 		foreach ($library->get_videos() as $file)
 		{
-			unlink($file);
+			if (file_exists($file)) unlink($file);
 		}
 	}
 	
@@ -35,12 +73,12 @@ class optimizer
 	{
 		foreach ($library->get_storyboards() as $file)
 		{
-			unlink($file);
+			if (file_exists($file)) unlink($file);
 		}
 		
 		foreach ($library->get_osb_files() as $file)
 		{
-			unlink($file);
+			if (file_exists($file)) unlink($file);
 		}
 		
 		$empty = self::build_empty_dir_list($library);
@@ -53,30 +91,40 @@ class optimizer
 	private static function build_removand_sublist(string $folder) : array
 	{
 		$queue = array();
+		$lowercase = array();
 		
 		foreach (glob(utils::globsafe($folder) . "/*") as $file)
 		{
 			if (is_dir($file))
 			{
-				$queue = array_merge($queue, self::build_removand_sublist($file)); // recursion
+				list($new_queue, $new_lowercase) = self::build_removand_sublist($file); // recursion
+				
+				$lowercase = array_merge($lowercase, $new_lowercase);
+				$queue = array_merge($queue, $new_queue);
 			}
 			else
 			{
-				$queue[] = strtolower($file);
+				$queue[] = $file;
+				$lowercase[] = strtolower($file);
 			}
 		}
 		
-		return $queue;
+		return array($queue, $lowercase);
 	}
 	
 	public static function build_removand_list(osu_library $library) : array
 	{
 		$queue = array();
+		$lowercase = array();
+		
 		foreach ($library->get_folders() as $folder)
 		{
-			$queue = array_merge($queue, self::build_removand_sublist($folder));
+			list($new_queue, $new_lowercase) = self::build_removand_sublist($folder);
+			$queue = array_merge($queue, $new_queue);
+			$lowercase = array_merge($lowercase, $new_lowercase);
 		}
-		return $queue;
+		
+		return array($queue, $lowercase);
 	}
 	
 	public static function build_empty_dir_sublist(string $folder) : array
@@ -114,44 +162,89 @@ class optimizer
 	
 	public static function build_excluded_list(osu_library $library) : array
 	{
-		$bg = $library->get_backgrounds();
+		$background_files = $library->get_backgrounds();
-		$vid = $library->get_videos();
+		$video_files = $library->get_videos();
-		$sb = $library->get_storyboards();
+		$storyboard_files = $library->get_storyboards();
-		$a = $library->get_audiofiles();
+		$hitsound_files = $library->get_hitsounds();
-		$sbf = $library->get_osb_files();
+		$audio_files = $library->get_audiofiles();
-		$osf = $library->get_osu_files();
+		$osb_files = $library->get_osb_files();
-		$excluded = array_merge($bg, $vid, $sb, $a, $sbf, $osf);
+		$osu_files = $library->get_osu_files();
 		
-		$lowercase = array();
+		$essential_excluded = array_merge($background_files, $video_files, $audio_files);
+		$physical_excluded = array_merge($osb_files, $osu_files);
+		$other_excluded = array_merge($storyboard_files, $hitsound_files);
+		$excluded = array_merge($essential_excluded, $physical_excluded, $other_excluded);
+		
+		$lowercase_excluded = array();
 		foreach ($excluded as $key => $value)
 		{
-			$lowercase[$key] = strtolower($value);
+			$lowercase_excluded[$key] = strtolower($value);
 		}
 		
-		return $lowercase;
+		$peppy_excluded_lowercase = array();
+		foreach ($lowercase_excluded as $value)
+		{
+			$extension = pathinfo($value, PATHINFO_EXTENSION) ?? "";
+			
+			$directory = pathinfo($value, PATHINFO_DIRNAME) ?? "";
+			$directory = !empty($directory) ? $directory. "/" : ""; // append slash if set
+			
+			$filename = pathinfo($value, PATHINFO_FILENAME);
+			
+			if (mb_ereg_match("([jJ][pP][eE]?[gG]|[pP][nN][gG])", $extension))
+			{
+				$image_extensions = [ "png", "jpg", "jpeg", strtolower($extension) ];
+				$image_extensions = array_unique($image_extensions);
+				foreach ($image_extensions as $image_extension)
+				{
+					$peppy_excluded_lowercase[] = $directory . $filename . "." . $image_extension;
+				}
+			}
+			else if (mb_ereg_match("([wW][aA][vV]|[mM][pP]3|[oO][gG][gG])", $extension))
+			{
+				$sound_extensions = [ "wav", "mp3", "ogg", strtolower($extension) ];
+				$sound_extensions = array_unique($sound_extensions);
+				foreach ($sound_extensions as $sound_extension)
+				{
+					$peppy_excluded_lowercase[] = $directory . $filename . "." . $sound_extension;
+				}
+			}
+			else
+			{
+				$peppy_excluded_lowercase[] = $value;
+			}
+		}
+		
+		return $peppy_excluded_lowercase;
 	}
 	
 	public static function remove_other(osu_library $library) : void
 	{
-		// $time_start = microtime(true);
+		// get the path and lowercase paths
+		list($removand, $removand_lower) = self::build_removand_list($library);
 		
-		$removand = self::build_removand_list($library);
+		// get the lowercase exclusions
 		$exclusions = self::build_excluded_list($library);
-		$final = array_diff($removand, $exclusions);
+		
+		// subtract the exclusions (lowercase because osu! is case-insensitive)
+		$check = array_diff($removand_lower, $exclusions);
+		
+		// take the original paths based on the keys from the subtraction above
+		// (this is only important for case-sensitive file systems like unix)
+		$final = array_intersect_key($removand, $check);
 		
 		foreach ($final as $file)
 		{
-			unlink($file);
+			if (self::is_skinnable($file)) continue; // ignore default hitsounds
+			// if (file_exists($file)) unlink($file);
+			echo $file . "<br />";
 		}
+		exit(0);
 		
 		$empty = self::build_empty_dir_list($library);
 		foreach ($empty as $folder)
 		{
 			rmdir($folder);
 		}
-		
-		// $time_end = microtime(true);
-		// $time = $time_end - $time_start;
-		// echo " in {$time} seconds.";
 	}
 }

libraries/osu_library.php

@@ -273,7 +273,8 @@ class osu_library
 				if (!empty($beatmap["background"]))
 				{
 					$path = $beatmapset["path"] . "/" . $beatmap["background"];
-					if (file_exists($path)) $backgrounds[] = $path;
+					// if (file_exists($path))
+					$backgrounds[] = $path;
 				}
 			}
 		}
@@ -295,7 +296,8 @@ class osu_library
 				if (!empty($beatmap["video"]))
 				{
 					$path = $beatmapset["path"] . "/" . $beatmap["video"];
-					if (file_exists($path)) $videos[] = $path;
+					// if (file_exists($path))
+					$videos[] = $path;
 				}
 			}
 		}
@@ -305,52 +307,71 @@ class osu_library
 		return $videos;
 	}
 	
-	public function get_osu_files() : array
+	public function get_parsed_files() : array
 	{
 		$db = $this->get_library();
 		
-		$osu_files = array();
+		$files = array();
 		foreach ($db as $beatmapset)
 		{
 			foreach($beatmapset["difficulties"] as $beatmap)
 			{
-				if (!empty($beatmap["format"]) && $beatmap["format"] != "storyboard")
+				if (!empty($beatmap["path"]) && !empty($beatmap["format"]))
 				{
-					if (file_exists($beatmap["path"]))
+					$files[] = $beatmap;
-					{
-						$osu_files[] = $beatmap["path"];
-					}
 				}
 			}
 		}
 		
-		$osu_files = array_unique($osu_files);
+		return $files;
+	}
+	
+	public function get_osu_files() : array
+	{
+		$files = $this->get_parsed_files();
+		
+		$osu_files = array();
+		foreach ($files as $file)
+		{
+			if (mb_strpos($beatmap["format"], "osu file format " !== false))
+			{
+				$osu_files[] = $beatmap["path"];
+			}
+		}
 		
 		return $osu_files;
 	}
 	
 	public function get_osb_files() : array
 	{
-		$db = $this->get_library();
+		$files = $this->get_parsed_files();
 		
-		$storyboard_files = array();
+		$osb_files = array();
-		foreach ($db as $beatmapset)
+		foreach ($files as $file)
 		{
-			foreach($beatmapset["difficulties"] as $beatmap)
+			if ($beatmap["format"] == "storyboard")
 			{
-				if (!empty($beatmap["format"]) && $beatmap["format"] == "storyboard")
+				$osb_files[] = $beatmap["path"];
-				{
-					if (file_exists($beatmap["path"]))
-					{
-						$storyboard_files[] = $beatmap["path"];
-					}
-				}
 			}
 		}
 		
-		$storyboard_files = array_unique($storyboard_files);
+		return $osb_files;
+	}
 	
-		return $storyboard_files;
+	public function get_broken_files() : array
+	{
+		$files = $this->get_parsed_files();
+		
+		$broken_files = array();
+		foreach ($files as $file)
+		{
+			if ($beatmap["format"] == "unknown")
+			{
+				$broken_files[] = $beatmap["path"];
+			}
+		}
+		
+		return $broken_files;
 	}
 	
 	public function get_storyboards() : array
@@ -365,7 +386,8 @@ class osu_library
 				foreach ($beatmap["storyboard"] ?? array() as $storyelement)
 				{
 					$path = $beatmapset["path"] . "/" . $storyelement;
-					if (file_exists($path)) $storyboards[] = $path;
+					// if (file_exists($path)) 
+					$storyboards[] = $path;
 				}
 			}
 		}
@@ -375,6 +397,29 @@ class osu_library
 		return $storyboards;
 	}
 	
+	public function get_hitsounds() : array
+	{
+		$db = $this->get_library();
+		
+		$hitsounds = array();
+		foreach ($db as $beatmapset)
+		{
+			foreach($beatmapset["difficulties"] as $beatmap)
+			{
+				foreach ($beatmap["hitsounds"] ?? array() as $hitsound)
+				{
+					$path = $beatmapset["path"] . "/" . $hitsound;
+					// if (file_exists($path))
+					$hitsounds[] = $path;
+				}
+			}
+		}
+		
+		$hitsounds = array_unique($hitsounds);
+		
+		return $hitsounds;
+	}
+	
 	public function get_audiofiles() : array
 	{
 		$db = $this->get_library();
@@ -387,7 +432,8 @@ class osu_library
 				if (!empty($beatmap["General"]["AudioFilename"]))
 				{
 					$path = $beatmapset["path"] . "/" . $beatmap["General"]["AudioFilename"];
-					if (file_exists($path)) $audiofiles[] = $path;
+					// if (file_exists($path)) 
+					$audiofiles[] = $path;
 				}
 			}
 		}
@@ -396,4 +442,58 @@ class osu_library
 		
 		return $audiofiles;
 	}
+	
+	public function get_missing_files() : array
+	{
+		$db = $this->get_library();
+		
+		$missing = array();
+		foreach ($db as $beatmapset)
+		{
+			foreach($beatmapset["difficulties"] as $beatmap)
+			{
+				if (!empty($beatmap["General"]["AudioFilename"]))
+				{
+					$path = $beatmapset["path"] . "/" . $beatmap["General"]["AudioFilename"];
+					if (!file_exists($path)) $missing[] = $path;
+				}
+				
+				if (!empty($beatmap["background"]))
+				{
+					$path = $beatmapset["path"] . "/" . $beatmap["background"];
+					if (!file_exists($path)) $missing[] = $path;
+				}
+				
+				if (!empty($beatmap["video"]))
+				{
+					$path = $beatmapset["path"] . "/" . $beatmap["video"];
+					if (!file_exists($path)) $missing[] = $path;
+				}
+				
+				foreach ($beatmap["storyboard"] ?? array() as $storyelement)
+				{
+					$path = $beatmapset["path"] . "/" . $storyelement;
+					if (!file_exists($path)) $missing[] = $path;
+				}
+				
+				foreach ($beatmap["hitsounds"] ?? array() as $hitsound)
+				{
+					$path = $beatmapset["path"] . "/" . $hitsound;
+					if (!file_exists($path)) $missing[] = $path;
+				}
+				
+				if (!empty($beatmap["format"]))
+				{
+					if (!file_exists($beatmap["path"]))
+					{
+						$missing[] = $beatmap["path"];
+					}
+				}
+			}
+		}
+		
+		$missing = array_unique($missing);
+		
+		return $missing;
+	}
 }

libraries/osu_old_parser.php

@@ -1,263 +0,0 @@
-<?php
-class osu_old_parser
-{
-	public static function scan_parse_osu_file(string $osu_file) : array
-	{
-		$time_start = microtime(true);
-		$file = file($osu_file, FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES);
-		// osu files can get big, but why not load the full thing :3
-		
-		if (stripos($file[0], "osu file format") === false) return [ "error..." ];
-		
-		// some files had "ZERO WIDTH NO-BREAK SPACE" characters...
-		$format = explode("osu file format ", $file[0])[1];
-		unset($file[0]); // no longer needed
-		
-		$osu = array("Format" => $format);
-		$current_section = "UnofficialComments";
-		$osu[$current_section] = array();
-		$storyboard = array();
-		foreach ($file as $key => $line)
-		{
-			
-			if (mb_strpos($line, "[") === 0 && mb_strpos($line, "]") === (strlen($line)-1))
-			{
-				$current_section = str_replace([ "[", "]" ], "", $line);
-				if (!isset($osu[$current_section]))
-				{
-					$osu[$current_section] = array();
-				}
-				
-				// peppy is retarded so i have to do this...
-				switch ($current_section) {
-					case "General":
-					case "Editor":
-						$section_type = "key-value pairs";
-						$delimiter = ": ";
-						break;
-					case "Metadata":
-					case "Difficulty":
-						$section_type = "key-value pairs";
-						$delimiter = ":"; // notice the missing space
-						break;
-					case "Colours":
-						$section_type = "key-value pairs";
-						$delimiter = " : "; // WHY WOULD YOU DO THIS IF YOU ALREADY HAVE TWO TYPES OF KEY-VALUE PAIRS???????????????????
-						break;
-					case "Events":
-					case "TimingPoints":
-					case "HitObjects":
-						$section_type = "lists"; // yes, listS because one list per line
-						$delimiter = ",";
-						break;
-					default:
-						$section_type = "unknown";
-				}
-				
-				continue;
-			}
-			
-			
-			// only parse the ones needed
-			switch ($current_section) {
-				case "General":
-				case "Metadata":
-				case "Difficulty":
-				case "Events":
-					$skip = false;
-					break;
-				default:
-					$skip = true;
-			}
-			
-			if ($skip) continue;
-			
-			
-			if (mb_strpos($line, "//") === 0) // there were commented files that broke my script
-			{
-			}
-			else if ($section_type == "key-value pairs")
-			{
-				$delimiter_position = mb_strpos($line, $delimiter);
-				
-				$value = mb_substr($line, $delimiter_position + strlen($delimiter_position));
-				$osu[$current_section][mb_substr($line, 0, $delimiter_position)] = $value;
-			}
-			else if ($section_type == "lists")
-			{
-				$list = explode($delimiter, $line);
-				
-				// group events by type and start time
-				if ($current_section == "Events")
-				{
-					if (mb_strpos($line, " ") === 0) continue; // skip storyboard details lines
-					
-					// event types: https://github.com/ppy/osu/blob/master/osu.Game/Beatmaps/Legacy/LegacyEventType.cs
-					$list[0] = str_replace(
-						[ "Background", "Video", "Break", "Colour", "Sprite", "Sample", "Animation" ],
-						[ "0",          "1",     "2",     "3",      "4",      "5",      "6" ],
-						$list[0]
-					);
-					
-					if ($list[0] == "5" || $list[0] == "4")
-					{
-						$storyboard[] = trim(str_replace("\\", "/", $list[3]), "\"");
-					}
-					
-					if ($list[0] == "6")
-					{
-						$story_base = pathinfo(trim(str_replace("\\", "/", $list[3]), "\""));
-						if (empty($story_base["extension"])) $ext = "";
-						else $ext = "." . $story_base["extension"];
-						if (empty($story_base["dirname"])) $dir = "";
-						else $dir = $story_base["dirname"] . "/";
-						
-						for ($i = 0; $i < intval($list[6]); $i++)
-						{
-							$storyboard[] = $dir . $story_base["filename"] . $i . $ext;
-						}
-					}
-					
-					if (!isset($osu[$current_section][$list[0]]))
-					{
-						$osu[$current_section][$list[0]] = array();
-					}
-					
-					if (!isset($osu[$current_section][$list[0]][$list[1]]))
-					{
-						$osu[$current_section][$list[0]][$list[1]] = array();
-					}
-					
-					$osu[$current_section][$list[0]][$list[1]][] = $list;
-				}
-				else
-				{
-					$osu[$current_section][] = $list;
-				}
-			}
-			else
-			{
-				$osu[$current_section][] = $line; // just dump the unknown...
-			}
-		}
-		unset($file); // remove the memory leak
-		
-		
-		
-		// return $osu;
-		
-		$set_id = $osu["Metadata"]["BeatmapSetID"] ?? false;
-		if ($set_id === false)
-		{
-			$temp = explode(" ", basename(dirname($osu_file)))[0];
-			if (is_numeric($temp))
-			{
-				$set_id = $temp;
-			}
-			else
-			{
-				$set_id = "";
-			}
-		}
-		
-		$background = str_replace("\\", "/", trim($osu["Events"][0][0][0][2] ?? "", "\""));
-		$audio = str_replace("\\", "/", trim($osu["General"]["AudioFilename"] ?? "", "\""));
-		$video = str_replace("\\", "/", trim($osu["Events"][1][array_key_first($osu["Events"][1] ?? array())][0][2] ?? "", "\""));
-		$storyboard = array_unique($storyboard);
-		
-		$map_id = intval($osu["Metadata"]["BeatmapID"] ?? 0);
-		if ($map_id < 1) $map_id = "";
-		$return = array(
-			"format" => $osu["Format"] ?? "",
-			"title" => $osu["Metadata"]["Title"] ?? "",
-			"artist" => $osu["Metadata"]["Artist"] ?? "",
-			"mapper" => $osu["Metadata"]["Creator"] ?? "",
-			"difficulty" => $osu["Metadata"]["Version"] ?? "",
-			"tags" => $osu["Metadata"]["Tags"] ?? "",
-			"background" => $background,
-			"audio" => $audio,
-			"video" => $video,
-			"storyboard" => $storyboard,
-			"id" => $map_id,
-			"set_id" => $set_id,
-		);
-		
-		$time_end = microtime(true);
-		$time = $time_end - $time_start;
-		$return["process_time"] = $time;
-		$return["hash"] = hash_file("md5", $osu_file);
-		
-		return $return;
-	}
-	
-	public static function scan_parse_osb_file(string $osb_file) : array
-	{
-		$time_start = microtime(true);
-		$file = file($osb_file, FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES);
-		
-		$storyboard = array();
-		$current_section = "UnofficialComments";
-		foreach ($file as $key => $line)
-		{
-			if (mb_strpos($line, "[") === 0 && mb_strpos($line, "]") === (strlen($line)-1))
-			{
-				$current_section = str_replace([ "[", "]" ], "", $line);
-				continue;
-			}
-			
-			if (!($current_section == "Events")) continue; // skip rest
-			
-			$list = explode(",", $line);
-			
-			$list[0] = str_replace(
-				[ "Background", "Video", "Break", "Colour", "Sprite", "Sample", "Animation" ],
-				[ "0",          "1",     "2",     "3",      "4",      "5",      "6" ],
-				$list[0]
-			);
-			
-			if ($list[0] == "5" || $list[0] == "4")
-			{
-				$storyboard[] = trim(str_replace("\\", "/", $list[3]), "\"");
-			}
-			
-			if ($list[0] == "6")
-			{
-				$story_base = pathinfo(trim(str_replace("\\", "/", $list[3]), "\""));
-				if (empty($story_base["extension"])) $ext = "";
-				else $ext = "." . $story_base["extension"];
-				if (empty($story_base["dirname"])) $dir = "";
-				else $dir = $story_base["dirname"] . "/";
-				
-				for ($i = 0; $i < intval($list[6]); $i++)
-				{
-					$storyboard[] = $dir . $story_base["filename"] . $i . $ext;
-				}
-			}
-				
-			$temp = explode(" ", basename(dirname($osb_file)))[0];
-			if (is_numeric($temp))
-			{
-				$set_id = $temp;
-			}
-			else
-			{
-				$set_id = "";
-			}
-		}
-		
-		$storyboard = array_unique($storyboard);
-		
-		$return = array(
-			"format" => "storyboard",
-			"storyboard" => $storyboard,
-			"set_id" => $set_id,
-		);
-		
-		$time_end = microtime(true);
-		$time = $time_end - $time_start;
-		$return["process_time"] = $time;
-		$return["hash"] = hash_file("md5", $osb_file);
-		
-		return $return;
-	}
-}

libraries/osu_parser.php

@@ -31,6 +31,29 @@ class osu_parser
 		);
 	}
 	
+	// fix backslash and double quotes
+	public static function fix_filename(string $filename) : string
+	{
+		return trim(str_replace("\\", "/", $filename), "\"");
+	}
+	
+	public static function file_ver_peppy(string $path)// : array|bool // see you in php8
+	{
+		if (!file_exists($path)) return false;
+		$file = file_get_contents($path); // read normally
+		if (empty($file)) return false;
+		
+		$bom = "\xef\xbb\xbf"; // the BOM character
+		$file = str_replace($bom, "", $file); // removing BOM
+		$file = str_replace("\r\n", "\n", $file); // win CRLF to unix LF
+		$file = str_replace("\r", "\n", $file); // old mac CR to unix LF
+		$lines = explode("\n", $file); // split using LF
+		$lines = array_filter($lines); // remove empty lines
+		$lines = array_values($lines); // reindex array
+		
+		return $lines;
+	}
+	
 	public function parse_osu_file_format(string $path, bool $skip_cache = false)// : array|bool // see you again in php8
 	{
 		if (!file_exists($path)) return false;
@@ -43,28 +66,35 @@ class osu_parser
 		
 		$time_start = microtime(true); // measure parsing time
 		
-		$file = file($path, FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES);
+		$file = self::file_ver_peppy($path);
+		if ($file === false) return false;
 		
 		$parsed = array();
 		
-		// file format declaration
+		$osu_file_format_declaration = "osu file format ";
-		if (stripos($file[0], "osu file format ") !== false)
+		if (stripos($file[0], $osu_file_format_declaration) !== false)
 		{
-			// the ranker script had a bug where random "ZERO WIDTH NO-BREAK SPACE"
+			// random characters were at beginning of the .osu files...
-			// characters were at beginning of the .osu files
+			$format = $osu_file_format_declaration . explode($osu_file_format_declaration, $file[0])[1];
-			$parsed["format"] = explode("osu file format ", $file[0])[1];
 			unset($file[0]); // no longer needed
 		}
 		else if (pathinfo($path, PATHINFO_EXTENSION) == "osb")
 		{
-			$parsed["format"] = "storyboard";
+			$format = "storyboard";
 		}
+		else
+		{
+			$format = "unknown";
+		}
+		
+		$parsed["format"] = $format;
+		
 		
 		$current_section = false;
 		$section_type = false;
 		$delimiter = false;
 		$variables = [ "keys" => array(), "values" => array()]; // variables for osb files $key=value pairs
-		$needed_sections = [ "General", "Metadata", "Difficulty", "Variables", "Events" ];
+		$needed_sections = [ "General", "Metadata", "Difficulty", "Variables", "Events", "HitObjects" ];
 		foreach ($file as $key => $line)
 		{
 			if (mb_strpos($line, "[") === 0 && mb_strpos($line, "]") === (strlen($line)-1))
@@ -135,6 +165,33 @@ class osu_parser
 						}
 					}
 				}
+				else if ($current_section == "HitObjects") // saving the whole thing would take up too much space v2
+				{
+					$last = array_key_last($list);
+					if (!empty($list[$last]) && mb_strpos($list[$last], ":") !== false)
+					{
+						$hitSample = explode(":", $list[$last]);
+						$last_sample = array_key_last($hitSample);
+						if (!empty($hitSample[$last_sample]))
+						{
+							$filename = $hitSample[$last_sample];
+							$filename = self::fix_filename($filename);
+							
+							// some maps leave out the extensions
+							// (and some maps have .wav filess pointing to .ogg files........)
+							if (empty(pathinfo($filename, PATHINFO_EXTENSION)))
+							{
+								$filename = $filename . ".wav"; // just to signal it's an "audio file"
+							}
+							
+							// init empty array
+							if (!isset($parsed["hitsounds"])) $parsed["hitsounds"] = array();
+							
+							// add the element to the hitsounds
+							$parsed["hitsounds"][] = $filename;
+						}
+					}
+				}
 				else
 				{
 					// init empty array
@@ -151,8 +208,9 @@ class osu_parser
 		}
 		unset($file); // remove the memory leak
 		
-		// storyboards are overloaded with dupes (renumber to make json export to arrays)
+		// storyboards/hitsounds are overloaded with dupes (renumber to make json export to arrays)
 		if (!empty($parsed["storyboard"])) $parsed["storyboard"] = array_values(array_unique($parsed["storyboard"]));
+		if (!empty($parsed["hitsounds"])) $parsed["hitsounds"] = array_values(array_unique($parsed["hitsounds"]));
 		
 		$time_end = microtime(true);
 		$parsing_time = $time_end - $time_start;
@@ -193,8 +251,7 @@ class osu_parser
 			$source_file = str_replace($variables["keys"], $variables["values"], $source_file);
 		}
 		
-		// fix backslash and double quotes
+		if ($source_file !== false) $source_file = self::fix_filename($source_file);
-		if ($source_file !== false) $source_file = trim(str_replace("\\", "/", $source_file), "\"");
 		
 		// fix leading dot slash
 		if (mb_strpos($source_file, "./") === 0) $source_file = mb_substr($source_file, 2);
@@ -222,6 +279,23 @@ class osu_parser
 			$source_files = array($source_file); // pack the single-source resources into an array
 		}
 		
+		// artificially add a "good" extension
+		foreach ($source_files as $source_key => $source_file)
+		{
+			// some maps leave out the extensions...
+			if (empty(pathinfo($source_file, PATHINFO_EXTENSION)))
+			{
+				if ($event_type == 5)
+				{
+					$source_files[$source_key] = $source_file . ".wav"; // just to signal it's an "audio file"
+				}
+				else
+				{
+					$source_files[$source_key] = $source_file . ".png"; // just to signal it's an "image file"
+				}
+			}
+		}
+		
 		return array(self::reverse_event_type($event_type), $source_files);
 	}
 	

main.php

@@ -3,6 +3,7 @@
 // todo: repack osz file
 // todo: dupe checker
 // todo: stardiff deleter
+// todo: mode deleter
 
 
 // imports
@@ -68,7 +69,7 @@ if (isset($_GET["purify"]))
 $start = file_get_contents("resources/start.html");
 $start = str_replace("{{ STYLE }}", file_get_contents("resources/style.css"), $start);
 echo $start;
-dump($lib, "lib");
+// dump($lib, "lib");
 echo '<a href="./?scan">[Scan]</a> ';
 echo '<a href="./?rescan">[Force rescan]</a> ';
 echo '<a href="./?blacken">[Blacken]</a> ';

resources/allhitsounds.txt

@@ -0,0 +1,109 @@
+check-off.wav
+check-on.wav
+click-close.wav
+click-short-confirm.wav
+menuback.wav
+menuhit.wav
+select-difficulty.wav
+select-expand.wav
+shutter.wav
+sliderbar.wav
+count.wav
+count1s.wav
+count2s.wav
+count3s.wav
+gos.wav
+readys.wav
+drum-hitclap.wav
+drum-hitfinish.wav
+drum-hitnormal.wav
+drum-hitwhistle.wav
+drum-sliderslide.wav
+drum-slidertick.wav
+drum-sliderwhistle.wav
+applause.wav
+combobreak.wav
+comboburst(-n).wav
+failsound.wav
+metronomelow.wav
+sectionfail.wav
+sectionpass.wav
+menu-background.jpg
+back-button-click.wav
+back-button-hover.wav
+menu-back-click.wav
+menu-back-hover.wav
+menu-charts-click.wav
+menu-charts-hover.wav
+menu-direct-click.wav
+menu-direct-hover.wav
+menu-edit-click.wav
+menu-edit-hover.wav
+menu-exit-click.wav
+menu-exit-hover.wav
+menu-freeplay-click.wav
+menu-freeplay-hover.wav
+menu-multiplayer-click.wav
+menu-multiplayer-hover.wav
+menu-options-click.wav
+menu-options-hover.wav
+menu-play-click.wav
+menu-play-hover.wav
+heartbeat.wav
+seeya.wav
+welcome.wav
+click-short.wav
+menuclick.wav
+key-confirm.wav
+key-delete.wav
+key-movement.wav
+key-press-1.wav
+key-press-2.wav
+key-press-3.wav
+key-press-4.wav
+match-confirm.wav
+match-join.wav
+match-leave.wav
+match-notready.wav
+match-ready.wav
+match-start.wav
+nightcore-clap.wav
+nightcore-finish.wav
+nightcore-hat.wav
+nightcore-kick.wav
+normal-hitclap.wav
+normal-hitfinish.wav
+normal-hitnormal.wav
+normal-hitwhistle.wav
+normal-sliderslide.wav
+normal-slidertick.wav
+normal-sliderwhistle.wav
+spinnerbonus.wav
+spinnerspin.wav
+taiko-drum-hitclap.wav
+taiko-drum-hitfinish.wav
+taiko-drum-hitnormal.wav
+taiko-drum-hitwhistle.wav
+taiko-normal-hitclap.wav
+taiko-normal-hitfinish.wav
+taiko-normal-hitnormal.wav
+taiko-normal-hitwhistle.wav
+taiko-soft-hitclap.wav
+taiko-soft-hitfinish.wav
+taiko-soft-hitnormal.wav
+taiko-soft-hitwhistle.wav
+pause-back-click.wav
+pause-back-hover.wav
+pause-continue-click.wav
+pause-continue-hover.wav
+pause-hover.wav
+pause-loop.wav
+pause-retry-click.wav
+pause-retry-hover.wav
+soft-hitclap.wav
+soft-hitfinish.wav
+soft-hitnormal.wav
+soft-hitwhistle.wav
+soft-sliderslide.wav
+soft-slidertick.wav
+soft-sliderwhistle.wav

resources/no-format-declaration.txt

@@ -0,0 +1,3 @@
+113273 HTT - Gohan wa Okazu/HTT - Gohan wa Okazu (YunoFanatic) [S.Star's 7K Another].osu
+360118 LamazeP - Koi no Program Hatsudou (feat Hatsune Miku)/LamazeP - Koi no Program Hatsudou (feat. Hatsune Miku) (Sonnyc) [Euny's Hard].osu
+661398 Baby's breath - Habataki no Birthday/Baby's breath - Habataki no Birthday (ShogunMoon) [Nao's Advanced].osu