Add options to break up the captcha image storage with hash-digit subdirectories to avoid trawling through a giant directory on every hit

FancyCaptcha.php

@@ -29,6 +29,9 @@ if ( defined( 'MEDIAWIKI' ) ) {
 global $wgCaptchaDirectory;
 $wgCaptchaDirectory = "$wgUploadDirectory/captcha"; // bad default :D
 
+global $wgCaptchaDirectoryLevels;
+$wgCaptchaDirectoryLevels = 0; // To break into subdirectories
+
 global $wgCaptchaSecret;
 $wgCaptchaSecret = "CHANGE_THIS_SECRET!";
 
@@ -110,9 +113,48 @@ class FancyCaptcha extends SimpleCaptcha {
 	 * @return mixed tuple of (salt key, text hash) or false if no image to find
 	 */
 	function pickImage() {
-		global $wgCaptchaDirectory;
-		$n = mt_rand( 0, $this->countFiles( $wgCaptchaDirectory ) );
-		$dir = opendir( $wgCaptchaDirectory );
+		global $wgCaptchaDirectory, $wgCaptchaDirectoryLevels;
+		return $this->pickImageDir(
+			$wgCaptchaDirectory,
+			$wgCaptchaDirectoryLevels );
+	}
+	
+	function pickImageDir( $directory, $levels ) {
+		if( $levels ) {
+			$dirs = array();
+			
+			// Check which subdirs are actually present...
+			$dir = opendir( $directory );
+			while( false !== ($entry = readdir( $dir ) ) ) {
+				if( ctype_xdigit( $entry ) && strlen( $entry ) == 1 ) {
+					$dirs[] = $entry;
+				}
+			}
+			closedir( $dir );
+			
+			$place = mt_rand( 0, count( $dirs ) - 1 );
+			// In case all dirs are not filled,
+			// cycle through next digits...
+			for( $j = 0; $j < count( $dirs ); $j++ ) {
+				$char = $dirs[($place + $j) % count( $dirs )];
+				$return = $this->pickImageDir( "$directory/$char", $levels - 1 );
+				if( $return ) {
+					return $return;
+				}
+			}
+			// Didn't find any images in this directory... empty?
+			return false;
+		} else {
+			return $this->pickImageFromDir( $directory );
+		}
+	}
+	
+	function pickImageFromDir( $directory ) {
+		if( !is_dir( $directory ) ) {
+			return false;
+		}
+		$n = mt_rand( 0, $this->countFiles( $directory ) );
+		$dir = opendir( $directory );
 
 		$count = 0;
 
@@ -121,7 +163,7 @@ class FancyCaptcha extends SimpleCaptcha {
 		while( false !== $entry ) {
 			$entry = readdir( $dir );
 			if( preg_match( '/^image_([0-9a-f]+)_([0-9a-f]+)\\.png$/', $entry, $matches ) ) {
-				$size = getimagesize( "$wgCaptchaDirectory/$entry" );
+				$size = getimagesize( "$directory/$entry" );
 				$pick = array(
 					'salt' => $matches[1],
 					'hash' => $matches[2],
@@ -156,7 +198,6 @@ class FancyCaptcha extends SimpleCaptcha {
 
 	function showImage() {
 		global $wgOut, $wgRequest;
-		global $wgCaptchaDirectory;
 
 		$wgOut->disable();
 
@@ -172,7 +213,7 @@ class FancyCaptcha extends SimpleCaptcha {
 
 			$salt = $info['salt'];
 			$hash = $info['hash'];
-			$file = $wgCaptchaDirectory . DIRECTORY_SEPARATOR . "image_{$salt}_{$hash}.png";
+			$file = $this->imagePath( $salt, $hash );
 
 			if( file_exists( $file ) ) {
 				header( 'Content-type: image/png' );
@@ -184,6 +225,18 @@ class FancyCaptcha extends SimpleCaptcha {
 		return false;
 	}
 	
+	function imagePath( $salt, $hash ) {
+		global $wgCaptchaDirectory, $wgCaptchaDirectoryLevels;
+		$file = $wgCaptchaDirectory;
+		$file .= DIRECTORY_SEPARATOR;
+		for( $i = 0; $i < $wgCaptchaDirectoryLevels; $i++ ) {
+			$file .= $hash{$i};
+			$file .= DIRECTORY_SEPARATOR;
+		}
+		$file .= "image_{$salt}_{$hash}.png";
+		return $file;
+	}
+
 	/**
 	 * Show a message asking the user to enter a captcha on edit
 	 * The result will be treated as wiki text

captcha.py

@@ -21,6 +21,7 @@
 #
 # Further tweaks by Brion Vibber <brion@pobox.com>:
 # 2006-01-26: Add command-line options for the various parameters
+# 2007-02-19: Add --dirs param for hash subdirectory splits
 
 import random
 import Image
@@ -98,6 +99,22 @@ def gen_captcha(text, fontname, fontsize, file_name):
 	# save the image, in format determined from filename
 	im.save(file_name)
 
+def gen_subdir(basedir, hash, levels):
+	"""Generate a subdirectory path out of the first _levels_
+	characters of _hash_, and ensure the directories exist
+	under _basedir_."""
+	subdir = None
+	for i in range(0, levels):
+		char = hash[i]
+		if subdir:
+			subdir = os.path.join(subdir, char)
+		else:
+			subdir = char
+		fulldir = os.path.join(basedir, subdir)
+		if not os.path.exists(fulldir):
+			os.mkdir(fulldir)
+	return subdir
+	
 if __name__ == '__main__':
 	"""This grabs random words from the dictionary 'words' (one
 	word per line) and generates a captcha image for each one,
@@ -112,9 +129,10 @@ if __name__ == '__main__':
 	output = "."
 	count = 20
 	fill = 0
+	dirs = 0
 	verbose = False
 	
-	opts, args = getopt.getopt(sys.argv[1:], "", ["font=", "wordlist=", "key=", "output=", "count=", "fill=", "verbose"])
+	opts, args = getopt.getopt(sys.argv[1:], "", ["font=", "wordlist=", "key=", "output=", "count=", "fill=", "dirs=", "verbose"])
 	for o, a in opts:
 		if o == "--font":
 			font = a
@@ -128,6 +146,8 @@ if __name__ == '__main__':
 			count = int(a)
 		if o == "--fill":
 			fill = int(a)
+		if o == "--dirs":
+			dirs = int(a)
 		if o == "--verbose":
 			verbose = True
 	
@@ -149,6 +169,9 @@ if __name__ == '__main__':
 		# 64 bits of hash is plenty for this purpose
 		hash = md5.new(key+salt+word+key+salt).hexdigest()[:16]
 		filename = "image_%s_%s.png" % (salt, hash)
+		if dirs:
+			subdir = gen_subdir(output, hash, dirs)
+			filename = os.path.join(subdir, filename)
 		if verbose:
 			print filename
 		gen_captcha(word, font, 40, os.path.join(output, filename))