WIPhash-collision

Signed-off-by: Florian Pritz <bluewind@xinu.at>

3 files changed, 91 insertions, 48 deletions

diff --git a/application/controllers/file.php b/application/controllers/file.php
index 2617d4840..25fbd27e3 100644
--- a/application/controllers/file.php
+++ b/application/controllers/file.php
@@ -90,7 +90,7 @@ class File extends MY_Controller {
 
 		$etag = "";
 		foreach ($files as $filedata) {
-			$etag = sha1($etag.$filedata["hash"]);
+			$etag = sha1($etag.$filedata["data_id"]);
 		}
 
 		// handle some common "lexers" here
@@ -128,7 +128,7 @@ class File extends MY_Controller {
 			handle_etag($etag);
 
 			$filedata = $files[0];
-			$filepath = $this->mfile->file($filedata["hash"]);
+			$filepath = $this->mfile->file($filedata["data_id"]);
 			$this->ddownload->serveFile($filepath, $filedata["filename"], "text/plain");
 			exit();
 		}
@@ -136,7 +136,7 @@ class File extends MY_Controller {
 		$this->load->library("output_cache");
 
 		foreach ($files as $key => $filedata) {
-			$file = $this->mfile->file($filedata['hash']);
+			$file = $this->mfile->file($filedata['data_id']);
 
 			// autodetect the lexer for highlighting if the URL contains a / after the ID (/ID/)
 			// /ID/lexer disables autodetection
@@ -272,9 +272,9 @@ class File extends MY_Controller {
 	{
 		// highlight the file and cache the result, fall back to plain text if $lexer fails
 		foreach (array($lexer, "text") as $lexer) {
-			$highlit = cache_function($filedata['hash'].'_'.$lexer, 100,
+			$highlit = cache_function($filedata['data_id'].'_'.$lexer, 100,
 									  function() use ($filedata, $lexer, $is_multipaste) {
-				$file = $this->mfile->file($filedata['hash']);
+				$file = $this->mfile->file($filedata['data_id']);
 				if ($lexer == "rmd") {
 					ob_start();
 
@@ -323,7 +323,7 @@ class File extends MY_Controller {
 	private function _tooltip_for_image($filedata)
 	{
 		$filesize = format_bytes($filedata["filesize"]);
-		list($width, $height) = getimagesize($this->mfile->file($filedata["hash"]));
+		list($width, $height) = getimagesize($this->mfile->file($filedata["data_id"]));
 		$upload_date = date("r", $filedata["date"]);
 
 		$tooltip  = "${filedata["id"]} - $filesize<br>";
@@ -405,7 +405,7 @@ class File extends MY_Controller {
 						$filename = $filedata["id"]."-".$filedata["filename"];
 					}
 					assert(!isset($seen[$filename]));
-					$a->addFile($this->mfile->file($filedata["hash"]), $filename);
+					$a->addFile($this->mfile->file($filedata["data_id"]), $filename);
 					$seen[$filename] = true;
 				}
 				$archive->gzip_compress();
@@ -457,7 +457,7 @@ class File extends MY_Controller {
 
 				if (count($ids) == 1) {
 					$filedata = $this->mfile->get_filedata($id);
-					$file = $this->mfile->file($filedata['hash']);
+					$file = $this->mfile->file($filedata['data_id']);
 					$type = $filedata['mimetype'];
 					$lexer = $this->mfile->should_highlight($type);
 
@@ -531,7 +531,7 @@ class File extends MY_Controller {
 			$filedata = $this->mfile->get_filedata($repaste_id);
 
 			if ($filedata !== false && $this->mfile->can_highlight($filedata["mimetype"])) {
-				$this->data["textarea_content"] = file_get_contents($this->mfile->file($filedata["hash"]));
+				$this->data["textarea_content"] = file_get_contents($this->mfile->file($filedata["data_id"]));
 			}
 		}
 
@@ -568,11 +568,11 @@ class File extends MY_Controller {
 			show_error("Failed to get file data");
 		}
 
-		$cache_key = $filedata['hash'].'_thumb_'.$thumb_size;
+		$cache_key = $filedata['data_id'].'_thumb_'.$thumb_size;
 
 		$thumb = cache_function($cache_key, 100, function() use ($filedata, $thumb_size){
 			$CI =& get_instance();
-			$img = new libraries\Image($this->mfile->file($filedata["hash"]));
+			$img = new libraries\Image($this->mfile->file($filedata["data_id"]));
 			$img->makeThumb($thumb_size, $thumb_size);
 			$thumb = $img->get(IMAGETYPE_JPEG);
 
@@ -596,8 +596,9 @@ class File extends MY_Controller {
 		$user = $this->muser->get_userid();
 
 		$query = $this->db
-			->select('id, filename, mimetype, date, hash, filesize')
+			->select('files.id, filename, mimetype, date, hash, hash_collision_counter, filesize')
 			->from('files')
+			->join('file_storage', 'file_storage.id = files.file_storage_id')
 			->where('user', $user)
 			->where_in('mimetype', array('image/jpeg', 'image/png', 'image/gif'))
 			->order_by('date', 'desc')
@@ -609,7 +610,7 @@ class File extends MY_Controller {
 				continue;
 			}
 			$query[$key]["tooltip"] = $this->_tooltip_for_image($item);
-			$query[$key]["orientation"] = libraries\Image::get_exif_orientation($this->mfile->file($item["hash"]));
+			$query[$key]["orientation"] = libraries\Image::get_exif_orientation($this->mfile->file($item["data_id"]));
 		}
 
 		$this->data["items"] = $query;
@@ -630,7 +631,7 @@ class File extends MY_Controller {
 
 		// key: database field name; value: display name
 		$fields = array(
-			"id" => "ID",
+			"files.id" => "ID",
 			"filename" => "Filename",
 			"mimetype" => "Mimetype",
 			"date" => "Date",
@@ -647,14 +648,16 @@ class File extends MY_Controller {
 
 		$items = $this->db->select(implode(',', array_keys($fields)))
 			->from('files')
+			->join('file_storage', 'file_storage.id = files.file_storage_id')
 			->where('user', $user)
 			->get()->result_array();
 
 		$query = $this->db->query("
-			SELECT m.url_id id, sum(f.filesize) filesize, m.date, '' hash, '' mimetype, concat(count(*), ' file(s)') filename
+			SELECT m.url_id id, sum(fs.filesize) filesize, m.date, '' hash, '' mimetype, concat(count(*), ' file(s)') filename
 			FROM multipaste m
 			JOIN multipaste_file_map mfm ON m.multipaste_id = mfm.multipaste_id
 			JOIN files f ON f.id = mfm.file_url_id
+			JOIN file_storage fs ON f.file_storage_id = fs.id
 			WHERE m.user_id = ?
 			GROUP BY m.url_id
 			", array($user))->result_array();
@@ -688,8 +691,9 @@ class File extends MY_Controller {
 		$total_size = $this->db->query("
 			SELECT sum(filesize) sum
 			FROM (
-				SELECT DISTINCT hash, filesize
+				SELECT DISTINCT hash, hash_collision_counter,  filesize
 				FROM files
+				JOIN file_storage ON file_storage.id = files.file_storage_id
 				WHERE user = ?
 			) sub
 			", array($user))->row_array();
@@ -950,6 +954,9 @@ class File extends MY_Controller {
 		}
 
 		foreach ($files as $key => $file) {
+			// FIXME: use refactored code from api-rework branch here
+			// FIXME: remove hash_collision_counter from db and simply use the
+			// ID (so we can atomically insert and the get the last_insert_id
 			$limits = $this->muser->get_upload_id_limits();
 			$id = $this->mfile->new_id($limits[0], $limits[1]);
 			$hash = md5_file($file['tmp_name']);
@@ -1078,7 +1085,7 @@ class File extends MY_Controller {
 
 		$small_upload_size = $this->config->item('small_upload_size');
 
-		$query = $this->db->select('hash, id, user')
+		$query = $this->db->select('hash, hash_collision_counter, id, user')
 			->from('files')
 			->where('date <', $oldest_time)
 			->or_where('('.$this->db->_protect_identifiers('user').' = 0 AND '
@@ -1086,7 +1093,7 @@ class File extends MY_Controller {
 			->get()->result_array();
 
 		foreach($query as $row) {
-			$file = $this->mfile->file($row['hash']);
+			$file = $this->mfile->file($row['data_id']);
 			if (!file_exists($file)) {
 				$this->mfile->delete_id($row["id"]);
 				continue;
@@ -1095,10 +1102,10 @@ class File extends MY_Controller {
 			if ($row["user"] == 0 || filesize($file) > $small_upload_size) {
 				if (filemtime($file) < $oldest_time) {
 					unlink($file);
-					$this->mfile->delete_hash($row["hash"]);
+					$this->mfile->delete_data_id($row["data_id"]);
 				} else {
 					$this->mfile->delete_id($row["id"]);
-					if ($this->mfile->stale_hash($row["hash"])) {
+					if ($this->mfile->stale_data_id($row["data_id"])) {
 						unlink($file);
 					}
 				}
@@ -1166,9 +1173,9 @@ class File extends MY_Controller {
 			return;
 		}
 
-		$hash = $file_data["hash"];
-		$this->mfile->delete_hash($hash);
-		echo "removed hash \"$hash\"\n";
+		$data_id = $file_data["data_id"];
+		$this->mfile->delete_data_id($data_id);
+		echo "removed data_id \"$data_id\"\n";
 	}
 
 	function update_file_metadata()
@@ -1187,9 +1194,9 @@ class File extends MY_Controller {
 				->get()->result_array();
 
 			foreach ($query as $key => $item) {
-				$hash = $item["hash"];
-				$filesize = intval(filesize($this->mfile->file($hash)));
-				$mimetype = $this->mfile->mimetype($this->mfile->file($hash));
+				$data_id = $item["data_id"];
+				$filesize = intval(filesize($this->mfile->file($data_id)));
+				$mimetype = $this->mfile->mimetype($this->mfile->file($data_id));
 
 				$this->db->where('hash', $hash)
 					->set(array(
diff --git a/application/helpers/filebin_helper.php b/application/helpers/filebin_helper.php
index e5637ce94..acec213a9 100644
--- a/application/helpers/filebin_helper.php
+++ b/application/helpers/filebin_helper.php
@@ -283,4 +283,31 @@ function cache_function($key, $ttl, $function)
 	return $content;
 }
 
+function files_are_equal($a, $b)
+{
+	$chunk_size = 8*1024;
+
+	// Check if filesize is different
+	if (filesize($a) !== filesize($b)) {
+		return false;
+	}
+
+	// Check if content is different
+	$ah = fopen($a, 'rb');
+	$bh = fopen($b, 'rb');
+
+	$result = true;
+	while (!feof($ah) && !feof($bh)) {
+		if (fread($ah, $chunk_size) !== fread($bh, $chunk_size)) {
+			$result = false;
+			break;
+		}
+	}
+
+	fclose($ah);
+	fclose($bh);
+
+	return $result;
+}
+
 # vim: set noet:
diff --git a/application/models/mfile.php b/application/models/mfile.php
index eee2c4e5b..d6851a6ef 100644
--- a/application/models/mfile.php
+++ b/application/models/mfile.php
@@ -70,27 +70,30 @@ class Mfile extends CI_Model {
 	function get_filedata($id)
 	{
 		$query = $this->db
-			->select('id, hash, filename, mimetype, date, user, filesize')
+			->select('files.id, hash, hash_collision_counter, filename, mimetype, date, user, filesize')
 			->from('files')
-			->where('id', $id)
+			->join('file_storage', 'file_storage.id = files.file_storage_id')
+			->where('files.id', $id)
 			->limit(1)
 			->get();
 
 		if ($query->num_rows() > 0) {
-			return $query->row_array();
+			$data = $query->row_array();
+			$data["data_id"] = $data["hash"]."-".$data["hash_collision_counter"];
+			return $data;
 		} else {
 			return false;
 		}
 	}
 
-	// return the folder in which the file with $hash is stored
-	function folder($hash) {
-		return $this->config->item('upload_path').'/'.substr($hash, 0, 3);
+	// return the folder in which the file with $data_id is stored
+	function folder($data_id) {
+		return $this->config->item('upload_path').'/'.substr($data_id, 0, 3);
 	}
 
-	// Returns the full path to the file with $hash
-	function file($hash) {
-		return $this->folder($hash).'/'.$hash;
+	// Returns the full path to the file with $data_id
+	function file($data_id) {
+		return $this->folder($data_id).'/'.$data_id;
 	}
 
 	// Return mimetype of file
@@ -136,10 +139,10 @@ class Mfile extends CI_Model {
 		if (!$filedata) {
 			return false;
 		}
-		$file = $this->file($filedata['hash']);
+		$file = $this->file($filedata['data_id']);
 
 		if (!file_exists($file)) {
-			$this->delete_hash($filedata["hash"]);
+			$this->delete_data_id($filedata["data_id"]);
 			return false;
 		}
 
@@ -158,7 +161,7 @@ class Mfile extends CI_Model {
 			// if the file has been uploaded multiple times the mtime is the time
 			// of the last upload
 			if (filemtime($file) < $remove_before) {
-				$this->delete_hash($filedata["hash"]);
+				$this->delete_data_id($filedata["data_id"]);
 			} else {
 				$this->delete_id($id);
 			}
@@ -171,7 +174,7 @@ class Mfile extends CI_Model {
 	public function get_timeout($id)
 	{
 		$filedata = $this->get_filedata($id);
-		$file = $this->file($filedata["hash"]);
+		$file = $this->file($filedata["data_id"]);
 
 		if ($this->config->item("upload_max_age") == 0) {
 			return -1;
@@ -195,11 +198,14 @@ class Mfile extends CI_Model {
 		}
 	}
 
-	private function unused_file($hash)
+	private function unused_file($data_id)
 	{
-		$query = $this->db->select('id')
+		list ($hash, $hash_collision_counter) = explode("-", $data_id);
+		$query = $this->db->select('files.id')
 			->from('files')
+			->join('file_storage', 'file_storage.id = files.file_storage_id')
 			->where('hash', $hash)
+			->where('hash_collision_counter', $hash_collision_counter)
 			->limit(1)
 			->get();
 
@@ -238,12 +244,12 @@ class Mfile extends CI_Model {
 		}
 
 		if ($filedata !== false) {
-			assert(isset($filedata["hash"]));
-			if ($this->unused_file($filedata['hash'])) {
-				if (file_exists($this->file($filedata['hash']))) {
-					unlink($this->file($filedata['hash']));
+			assert(isset($filedata["data_id"]));
+			if ($this->unused_file($filedata['data_id'])) {
+				if (file_exists($this->file($filedata['data_id']))) {
+					unlink($this->file($filedata['data_id']));
 				}
-				$dir = $this->folder($filedata['hash']);
+				$dir = $this->folder($filedata['data_id']);
 				if (file_exists($dir)) {
 					if (count(scandir($dir)) == 2) {
 						rmdir($dir);
@@ -254,11 +260,14 @@ class Mfile extends CI_Model {
 		return true;
 	}
 
-	public function delete_hash($hash)
+	public function delete_data_id($data_id)
 	{
-		$ids = $this->db->select('id')
+		list ($hash, $hash_collision_counter) = explode("-", $data_id);
+		$ids = $this->db->select('files.id')
 			->from('files')
+			->join('file_storage', 'file_storage.id = files.file_storage_id')
 			->where('hash', $hash)
+			->where('hash_collision_counter', $hash_collision_counter)
 			->get()->result_array();
 
 		foreach ($ids as $entry) {