public $id; // int(4) primary_key not_null
public $urlhash; // varchar(64) unique_key
public $url; // text
+ public $filehash; // varchar(64) indexed
public $mimetype; // varchar(50)
public $size; // int(4)
- public $title; // varchar(255)
+ public $title; // text()
public $date; // int(4)
public $protected; // int(4)
- public $filename; // varchar(255)
+ public $filename; // text()
public $width; // int(4)
public $height; // int(4)
public $modified; // timestamp() not_null default_CURRENT_TIMESTAMP
const URLHASH_ALG = 'sha256';
+ const FILEHASH_ALG = 'sha256';
public static function schemaDef()
{
return array(
'fields' => array(
'id' => array('type' => 'serial', 'not null' => true),
- 'urlhash' => array('type' => 'varchar', 'length' => 64, 'description' => 'sha256 of destination URL (url field)'),
- 'url' => array('type' => 'text', 'description' => 'destination URL after following redirections'),
+ 'urlhash' => array('type' => 'varchar', 'length' => 64, 'not null' => true, 'description' => 'sha256 of destination URL (url field)'),
+ 'url' => array('type' => 'text', 'description' => 'destination URL after following possible redirections'),
+ 'filehash' => array('type' => 'varchar', 'length' => 64, 'not null' => false, 'description' => 'sha256 of the file contents, only for locally stored files of course'),
'mimetype' => array('type' => 'varchar', 'length' => 50, 'description' => 'mime type of resource'),
'size' => array('type' => 'int', 'description' => 'size of resource when available'),
- 'title' => array('type' => 'varchar', 'length' => 255, 'description' => 'title of resource when available'),
+ 'title' => array('type' => 'text', 'description' => 'title of resource when available'),
'date' => array('type' => 'int', 'description' => 'date of resource according to http query'),
'protected' => array('type' => 'int', 'description' => 'true when URL is private (needs login)'),
- 'filename' => array('type' => 'varchar', 'length' => 255, 'description' => 'if a local file, name of the file'),
+ 'filename' => array('type' => 'text', 'description' => 'if file is stored locally (too) this is the filename'),
'width' => array('type' => 'int', 'description' => 'width in pixels, if it can be described as such and data is available'),
'height' => array('type' => 'int', 'description' => 'height in pixels, if it can be described as such and data is available'),
'unique keys' => array(
'file_urlhash_key' => array('urlhash'),
),
+ 'indexes' => array(
+ 'file_filehash_idx' => array('filehash'),
+ ),
);
}
- function isProtected($url) {
- return 'http://www.facebook.com/login.php' === $url;
+ public static function isProtected($url) {
+
+ $protected_urls_exps = array(
+ 'https://www.facebook.com/login.php',
+ common_path('main/login')
+ );
+
+ foreach ($protected_urls_exps as $protected_url_exp) {
+ if (preg_match('!^'.preg_quote($protected_url_exp).'(.*)$!i', $url) === 1) {
+ return true;
+ }
+ }
+
+ return false;
}
/**
* @param string $given_url
* @return File
*/
- public static function saveNew(array $redir_data, $given_url) {
-
- // I don't know why we have to keep doing this but I'm adding this last check to avoid
- // uniqueness bugs.
+ public static function saveNew(array $redir_data, $given_url)
+ {
+ $file = null;
+ try {
+ // I don't know why we have to keep doing this but we run a last check to avoid
+ // uniqueness bugs.
+ $file = File::getByUrl($given_url);
+ return $file;
+ } catch (NoResultException $e) {
+ // We don't have the file's URL since before, so let's continue.
+ }
+
+ // if the given url is an local attachment url and the id already exists, don't
+ // save a new file record. This should never happen, but let's make it foolproof
+ // FIXME: how about attachments servers?
+ $u = parse_url($given_url);
+ if (isset($u['host']) && $u['host'] === common_config('site', 'server')) {
+ $r = Router::get();
+ // Skip the / in the beginning or $r->map won't match
+ try {
+ $args = $r->map(mb_substr($u['path'], 1));
+ if ($args['action'] === 'attachment') {
+ try {
+ // $args['attachment'] should always be set if action===attachment, given our routing rules
+ $file = File::getByID($args['attachment']);
+ return $file;
+ } catch (EmptyPkeyValueException $e) {
+ // ...but $args['attachment'] can also be 0...
+ } catch (NoResultException $e) {
+ // apparently this link goes to us, but is _not_ an existing attachment (File) ID?
+ }
+ }
+ } catch (Exception $e) {
+ // Some other exception was thrown from $r->map, likely a
+ // ClientException (404) because of some malformed link to
+ // our own instance. It's still a valid URL however, so we
+ // won't abort anything... I noticed this when linking:
+ // https://social.umeahackerspace.se/mmn/foaf' (notice the
+ // apostrophe in the end, making it unrecognizable for our
+ // URL routing.
+ // That specific issue (the apostrophe being part of a link
+ // is something that may or may not have been fixed since,
+ // in lib/util.php in common_replace_urls_callback().
+ }
+ }
- $file = File::getKV('urlhash', self::hashurl($given_url));
-
- if (!$file instanceof File) {
- $file = new File;
- $file->urlhash = self::hashurl($given_url);
- $file->url = $given_url;
- if (!empty($redir_data['protected'])) $file->protected = $redir_data['protected'];
- if (!empty($redir_data['title'])) $file->title = $redir_data['title'];
- if (!empty($redir_data['type'])) $file->mimetype = $redir_data['type'];
- if (!empty($redir_data['size'])) $file->size = intval($redir_data['size']);
- if (isset($redir_data['time']) && $redir_data['time'] > 0) $file->date = intval($redir_data['time']);
- $file_id = $file->insert();
- }
-
- Event::handle('EndFileSaveNew', array($file, $redir_data, $given_url));
- assert ($file instanceof File);
+ $file = new File;
+ $file->url = $given_url;
+ if (!empty($redir_data['protected'])) $file->protected = $redir_data['protected'];
+ if (!empty($redir_data['title'])) $file->title = $redir_data['title'];
+ if (!empty($redir_data['type'])) $file->mimetype = $redir_data['type'];
+ if (!empty($redir_data['size'])) $file->size = intval($redir_data['size']);
+ if (isset($redir_data['time']) && $redir_data['time'] > 0) $file->date = intval($redir_data['time']);
+ $file->saveFile();
return $file;
}
+ public function saveFile() {
+ $this->urlhash = self::hashurl($this->url);
+
+ if (!Event::handle('StartFileSaveNew', array(&$this))) {
+ throw new ServerException('File not saved due to an aborted StartFileSaveNew event.');
+ }
+
+ $this->id = $this->insert();
+
+ if ($this->id === false) {
+ throw new ServerException('File/URL metadata could not be saved to the database.');
+ }
+
+ Event::handle('EndFileSaveNew', array($this));
+ }
+
/**
* Go look at a URL and possibly save data about it if it's new:
* - follow redirect chains and store them in file_redirection
* - optionally save a file_to_post record
* - return the File object with the full reference
*
- * @fixme refactor this mess, it's gotten pretty scary.
* @param string $given_url the URL we're looking at
- * @param int $notice_id (optional)
+ * @param Notice $notice (optional)
* @param bool $followRedirects defaults to true
*
* @return mixed File on success, -1 on some errors
*
* @throws ServerException on failure
*/
- public static function processNew($given_url, $notice_id=null, $followRedirects=true) {
+ public static function processNew($given_url, Notice $notice=null, $followRedirects=true) {
if (empty($given_url)) {
throw new ServerException('No given URL to process');
}
throw new ServerException('No canonical URL from given URL to process');
}
- $file = File::getKV('url', $given_url);
- if (!$file instanceof File) {
- // First check if we have a lookup trace for this URL already
- $file_redir = File_redirection::getKV('url', $given_url);
- if ($file_redir instanceof File_redirection) {
- $file = File::getKV('id', $file_redir->file_id);
- if (!$file instanceof File) {
- // File did not exist, let's clean up the File_redirection entry
- $file_redir->delete();
- }
- }
-
- // If we still don't have a File object, let's create one now!
- if (!$file instanceof File) {
- // @fixme for new URLs this also looks up non-redirect data
- // such as target content type, size, etc, which we need
- // for File::saveNew(); so we call it even if not following
- // new redirects.
- $redir_data = File_redirection::where($given_url);
- if (is_array($redir_data)) {
- $redir_url = $redir_data['url'];
- } elseif (is_string($redir_data)) {
- $redir_url = $redir_data;
- $redir_data = array();
- } else {
- // TRANS: Server exception thrown when a URL cannot be processed.
- throw new ServerException(sprintf(_("Cannot process URL '%s'"), $given_url));
- }
+ $redir = File_redirection::where($given_url);
+ $file = $redir->getFile();
- // TODO: max field length
- if ($redir_url === $given_url || strlen($redir_url) > 255 || !$followRedirects) {
- // Save the File object based on our lookup trace
- $file = File::saveNew($redir_data, $given_url);
- } else {
- // This seems kind of messed up... for now skipping this part
- // if we're already under a redirect, so we don't go into
- // horrible infinite loops if we've been given an unstable
- // redirect (where the final destination of the first request
- // doesn't match what we get when we ask for it again).
- //
- // Seen in the wild with clojure.org, which redirects through
- // wikispaces for auth and appends session data in the URL params.
- $file = self::processNew($redir_url, $notice_id, /*followRedirects*/false);
- File_redirection::saveNew($redir_data, $file->id, $given_url);
- }
- }
-
- if (!$file instanceof File) {
- // This should only happen if File::saveNew somehow did not return a File object,
- // though we have an assert for that in case the event there might've gone wrong.
- // If anything else goes wrong, there should've been an exception thrown.
- throw new ServerException('URL processing failed without new File object');
- }
+ if (!$file instanceof File || empty($file->id)) {
+ // This should not happen
+ throw new ServerException('URL processing failed without new File object');
}
- if (!empty($notice_id)) {
- File_to_post::processNew($file->id, $notice_id);
+ if ($notice instanceof Notice) {
+ File_to_post::processNew($file, $notice);
}
+
return $file;
}
return true;
}
+ public function getFilename()
+ {
+ return self::tryFilename($this->filename);
+ }
+
+ public function getSize()
+ {
+ return intval($this->size);
+ }
+
// where should the file go?
static function filename(Profile $profile, $origname, $mimetype)
{
- try {
- $ext = common_supported_mime_to_ext($mimetype);
- } catch (Exception $e) {
- // We don't support this mimetype, but let's guess the extension
- $ext = substr(strrchr($mimetype, '/'), 1);
- }
+ $ext = self::guessMimeExtension($mimetype, $origname);
// Normalize and make the original filename more URL friendly.
$origname = basename($origname, ".$ext");
return $filename;
}
+ /**
+ * @param $mimetype The mimetype we've discovered for this file.
+ * @param $filename An optional filename which we can use on failure.
+ */
+ static function guessMimeExtension($mimetype, $filename=null)
+ {
+ try {
+ // first see if we know the extension for our mimetype
+ $ext = common_supported_mime_to_ext($mimetype);
+ // we do, so use it!
+ return $ext;
+ } catch (UnknownMimeExtensionException $e) {
+ // We don't know the extension for this mimetype, but let's guess.
+
+ // If we can't recognize the extension from the MIME, we try
+ // to guess based on filename, if one was supplied.
+ if (!is_null($filename) && preg_match('/^.+\.([A-Za-z0-9]+)$/', $filename, $matches)) {
+ // we matched on a file extension, so let's see if it means something.
+ $ext = mb_strtolower($matches[1]);
+
+ $blacklist = common_config('attachments', 'extblacklist');
+ // If we got an extension from $filename we want to check if it's in a blacklist
+ // so we avoid people uploading .php files etc.
+ if (array_key_exists($ext, $blacklist)) {
+ if (!is_string($blacklist[$ext])) {
+ // we don't have a safe replacement extension
+ throw new ClientException(_('Blacklisted file extension.'));
+ }
+ common_debug('Found replaced extension for filename '._ve($filename).': '._ve($ext));
+
+ // return a safe replacement extension ('php' => 'phps' for example)
+ return $blacklist[$ext];
+ }
+ // the attachment extension based on its filename was not blacklisted so it's ok to use it
+ return $ext;
+ }
+ } catch (Exception $e) {
+ common_log(LOG_INFO, 'Problem when figuring out extension for mimetype: '._ve($e));
+ }
+
+ // If nothing else has given us a result, try to extract it from
+ // the mimetype value (this turns .jpg to .jpeg for example...)
+ $matches = array();
+ // FIXME: try to build a regexp that will get jpeg from image/jpeg as well as json from application/jrd+json
+ if (!preg_match('/\/([a-z0-9]+)/', mb_strtolower($mimetype), $matches)) {
+ throw new Exception('Malformed mimetype: '.$mimetype);
+ }
+ return mb_strtolower($matches[1]);
+ }
+
/**
* Validation for as-saved base filenames
*/
return preg_match('/^[A-Za-z0-9._-]+$/', $filename);
}
+ static function tryFilename($filename)
+ {
+ if (!self::validFilename($filename))
+ {
+ throw new InvalidFilenameException($filename);
+ }
+ // if successful, return the filename for easy if-statementing
+ return $filename;
+ }
+
/**
* @throws ClientException on invalid filename
*/
static function path($filename)
{
- if (!self::validFilename($filename)) {
- // TRANS: Client exception thrown if a file upload does not have a valid name.
- throw new ClientException(_("Invalid filename."));
- }
+ self::tryFilename($filename);
+
$dir = common_config('attachments', 'dir');
- if ($dir[strlen($dir)-1] != '/') {
- $dir .= '/';
+ if (!in_array($dir[mb_strlen($dir)-1], ['/', '\\'])) {
+ $dir .= DIRECTORY_SEPARATOR;
}
return $dir . $filename;
static function url($filename)
{
- if (!self::validFilename($filename)) {
- // TRANS: Client exception thrown if a file upload does not have a valid name.
- throw new ClientException(_("Invalid filename."));
- }
+ self::tryFilename($filename);
if (common_config('site','private')) {
}
- if (StatusNet::useHTTPS()) {
+ if (GNUsocial::useHTTPS()) {
$sslserver = common_config('attachments', 'sslserver');
return $protocol.'://'.$server.$path.$filename;
}
+ static $_enclosures = array();
+
function getEnclosure(){
+ if (isset(self::$_enclosures[$this->getID()])) {
+ return self::$_enclosures[$this->getID()];
+ }
+
$enclosure = (object) array();
- foreach (array('title', 'url', 'date', 'modified', 'size', 'mimetype') as $key) {
- $enclosure->$key = $this->$key;
+ foreach (array('title', 'url', 'date', 'modified', 'size', 'mimetype', 'width', 'height') as $key) {
+ if ($this->$key !== '') {
+ $enclosure->$key = $this->$key;
+ }
}
- $needMoreMetadataMimetypes = array(null, 'application/xhtml+xml');
+ $needMoreMetadataMimetypes = array(null, 'application/xhtml+xml', 'text/html');
if (!isset($this->filename) && in_array(common_bare_mime($enclosure->mimetype), $needMoreMetadataMimetypes)) {
// This fetches enclosure metadata for non-local links with unset/HTML mimetypes,
// which may be enriched through oEmbed or similar (implemented as plugins)
Event::handle('FileEnclosureMetadata', array($this, &$enclosure));
}
- if (empty($enclosure->mimetype) || in_array(common_bare_mime($enclosure->mimetype), $needMoreMetadataMimetypes)) {
+ if (empty($enclosure->mimetype)) {
// This means we either don't know what it is, so it can't
// be shown as an enclosure, or it is an HTML link which
// does not link to a resource with further metadata.
throw new ServerException('Unknown enclosure mimetype, not enough metadata');
}
+
+ self::$_enclosures[$this->getID()] = $enclosure;
return $enclosure;
}
+ public function hasThumbnail()
+ {
+ try {
+ $this->getThumbnail();
+ } catch (Exception $e) {
+ return false;
+ }
+ return true;
+ }
+
/**
* Get the attachment's thumbnail record, if any.
* Make sure you supply proper 'int' typed variables (or null).
* @param $width int Max width of thumbnail in pixels. (if null, use common_config values)
* @param $height int Max height of thumbnail in pixels. (if null, square-crop to $width)
* @param $crop bool Crop to the max-values' aspect ratio
+ * @param $force_still bool Don't allow fallback to showing original (such as animated GIF)
+ * @param $upscale mixed Whether or not to scale smaller images up to larger thumbnail sizes. (null = site default)
*
* @return File_thumbnail
+ *
+ * @throws UseFileAsThumbnailException if the file is considered an image itself and should be itself as thumbnail
+ * @throws UnsupportedMediaException if, despite trying, we can't understand how to make a thumbnail for this format
+ * @throws ServerException on various other errors
*/
- public function getThumbnail($width=null, $height=null, $crop=false, $force_still=true)
+ public function getThumbnail($width=null, $height=null, $crop=false, $force_still=true, $upscale=null)
{
// Get some more information about this file through our ImageFile class
$image = ImageFile::fromFileObject($this);
// null means "always use file as thumbnail"
// false means you get choice between frozen frame or original when calling getThumbnail
if (is_null(common_config('thumbnail', 'animated')) || !$force_still) {
- throw new UseFileAsThumbnailException($this->id);
+ try {
+ // remote files with animated GIFs as thumbnails will match this
+ return File_thumbnail::byFile($this);
+ } catch (NoResultException $e) {
+ // and if it's not a remote file, it'll be safe to use the locally stored File
+ throw new UseFileAsThumbnailException($this);
+ }
}
}
- if ($width === null) {
- $width = common_config('thumbnail', 'width');
- $height = common_config('thumbnail', 'height');
- $crop = common_config('thumbnail', 'crop');
- }
-
- if ($height === null) {
- $height = $width;
- $crop = true;
- }
-
- // Get proper aspect ratio width and height before lookup
- // We have to do it through an ImageFile object because of orientation etc.
- // Only other solution would've been to rotate + rewrite uploaded files.
- list($width, $height, $x, $y, $w, $h) =
- $image->scaleToFit($width, $height, $crop);
+ return $image->getFileThumbnail($width, $height, $crop,
+ !is_null($upscale) ? $upscale : common_config('thumbnail', 'upscale'));
+ }
- $params = array('file_id'=> $this->id,
- 'width' => $width,
- 'height' => $height);
- $thumb = File_thumbnail::pkeyGet($params);
- if ($thumb instanceof File_thumbnail) {
- return $thumb;
+ public function getPath()
+ {
+ $filepath = self::path($this->filename);
+ if (!file_exists($filepath)) {
+ throw new FileNotFoundException($filepath);
}
+ return $filepath;
+ }
- // throws exception on failure to generate thumbnail
- $outname = "thumb-{$width}x{$height}-" . $image->filename;
- $outpath = self::path($outname);
-
- // The boundary box for our resizing
- $box = array('width'=>$width, 'height'=>$height,
- 'x'=>$x, 'y'=>$y,
- 'w'=>$w, 'h'=>$h);
+ public function getAttachmentUrl()
+ {
+ return common_local_url('attachment', array('attachment'=>$this->getID()));
+ }
- // Doublecheck that parameters are sane and integers.
- if ($box['width'] < 1 || $box['width'] > common_config('thumbnail', 'maxsize')
- || $box['height'] < 1 || $box['height'] > common_config('thumbnail', 'maxsize')
- || $box['w'] < 1 || $box['x'] >= $image->width
- || $box['h'] < 1 || $box['y'] >= $image->height) {
- // Fail on bad width parameter. If this occurs, it's due to algorithm in ImageFile->scaleToFit
- common_debug("Boundary box parameters for resize of {$image->filepath} : ".var_export($box,true));
- throw new ServerException('Bad thumbnail size parameters.');
+ /**
+ * @param mixed $use_local true means require local, null means prefer local, false means use whatever is stored
+ */
+ public function getUrl($use_local=null)
+ {
+ if ($use_local !== false) {
+ if (is_string($this->filename) || !empty($this->filename)) {
+ // A locally stored file, so let's generate a URL for our instance.
+ return self::url($this->getFilename());
+ }
+ if ($use_local) {
+ // if the file wasn't stored locally (has filename) and we require a local URL
+ throw new FileNotStoredLocallyException($this);
+ }
}
- common_debug(sprintf('Generating a thumbnail of File id==%u of size %ux%u', $this->id, $width, $height));
- // Perform resize and store into file
- $image->resizeTo($outpath, $box);
- // Avoid deleting the original
- if ($image->getPath() != self::path($image->filename)) {
- $image->unlink();
- }
- return File_thumbnail::saveThumbnail($this->id,
- self::url($outname),
- $width, $height,
- $outname);
+ // No local filename available, return the URL we have stored
+ return $this->url;
}
- public function getPath()
+ static public function getByUrl($url)
{
- return self::path($this->filename);
+ $file = new File();
+ $file->urlhash = self::hashurl($url);
+ if (!$file->find(true)) {
+ throw new NoResultException($file);
+ }
+ return $file;
}
- public function getUrl()
+ /**
+ * @param string $hashstr String of (preferrably lower case) hexadecimal characters, same as result of 'hash_file(...)'
+ */
+ static public function getByHash($hashstr)
{
- if (!empty($this->filename)) {
- // A locally stored file, so let's generate a URL for our instance.
- $url = self::url($this->filename);
- if (self::urlhash($url) !== $this->urlhash) {
- // For indexing purposes, in case we do a lookup on the 'url' field.
- // also we're fixing possible changes from http to https, or paths
- $this->updateUrl($url);
- }
- return $url;
+ $file = new File();
+ $file->filehash = strtolower($hashstr);
+ if (!$file->find(true)) {
+ throw new NoResultException($file);
}
-
- // No local filename available, return the URL we have stored
- return $this->url;
+ return $file;
}
public function updateUrl($url)
throw new ServerException('URL already exists in DB');
}
$sql = 'UPDATE %1$s SET urlhash=%2$s, url=%3$s WHERE urlhash=%4$s;';
- $result = $this->query(sprintf($sql, $this->__table,
+ $result = $this->query(sprintf($sql, $this->tableName(),
$this->_quote((string)self::hashurl($url)),
$this->_quote((string)$url),
$this->_quote((string)$this->urlhash)));
if ($result === false) {
common_log_db_error($this, 'UPDATE', __FILE__);
- throw new ServerException("Could not UPDATE {$this->__table}.url");
+ throw new ServerException("Could not UPDATE {$this->tableName()}.url");
}
return $result;
function blowCache($last=false)
{
- self::blow('file:notice-ids:%s', $this->urlhash);
+ self::blow('file:notice-ids:%s', $this->id);
if ($last) {
- self::blow('file:notice-ids:%s;last', $this->urlhash);
+ self::blow('file:notice-ids:%s;last', $this->id);
}
self::blow('file:notice-count:%d', $this->id);
}
function stream($offset=0, $limit=NOTICES_PER_PAGE, $since_id=0, $max_id=0)
{
- $stream = new FileNoticeStream($this);
+ // FIXME: Try to get the Profile::current() here in some other way to avoid mixing
+ // the current session user with possibly background/queue processing.
+ $stream = new FileNoticeStream($this, Profile::current());
return $stream->getNotices($offset, $limit, $since_id, $max_id);
}
$thumbs->delete();
}
}
+
+ $f2p = new File_to_post();
+ $f2p->file_id = $this->id;
+ if ($f2p->find()) {
+ while ($f2p->fetch()) {
+ $f2p->delete();
+ }
+ }
}
// And finally remove the entry from the database
}
return hash(self::URLHASH_ALG, $url);
}
+
+ static public function beforeSchemaUpdate()
+ {
+ $table = strtolower(get_called_class());
+ $schema = Schema::get();
+ $schemadef = $schema->getTableDef($table);
+
+ // 2015-02-19 We have to upgrade our table definitions to have the urlhash field populated
+ if (isset($schemadef['fields']['urlhash']) && isset($schemadef['unique keys']['file_urlhash_key'])) {
+ // We already have the urlhash field, so no need to migrate it.
+ return;
+ }
+ echo "\nFound old $table table, upgrading it to contain 'urlhash' field...";
+
+ $file = new File();
+ $file->query(sprintf('SELECT id, LEFT(url, 191) AS shortenedurl, COUNT(*) AS c FROM %1$s WHERE LENGTH(url)>191 GROUP BY shortenedurl HAVING c > 1', $schema->quoteIdentifier($table)));
+ print "\nFound {$file->N} URLs with too long entries in file table\n";
+ while ($file->fetch()) {
+ // We've got a URL that is too long for our future file table
+ // so we'll cut it. We could save the original URL, but there is
+ // no guarantee it is complete anyway since the previous max was 255 chars.
+ $dupfile = new File();
+ // First we find file entries that would be duplicates of this when shortened
+ // ... and we'll just throw the dupes out the window for now! It's already so borken.
+ $dupfile->query(sprintf('SELECT * FROM file WHERE LEFT(url, 191) = "%1$s"', $file->shortenedurl));
+ // Leave one of the URLs in the database by using ->find(true) (fetches first entry)
+ if ($dupfile->find(true)) {
+ print "\nShortening url entry for $table id: {$file->id} [";
+ $orig = clone($dupfile);
+ $dupfile->url = $file->shortenedurl; // make sure it's only 191 chars from now on
+ $dupfile->update($orig);
+ print "\nDeleting duplicate entries of too long URL on $table id: {$file->id} [";
+ // only start deleting with this fetch.
+ while($dupfile->fetch()) {
+ print ".";
+ $dupfile->delete();
+ }
+ print "]\n";
+ } else {
+ print "\nWarning! URL suddenly disappeared from database: {$file->url}\n";
+ }
+ }
+ echo "...and now all the non-duplicates which are longer than 191 characters...\n";
+ $file->query('UPDATE file SET url=LEFT(url, 191) WHERE LENGTH(url)>191');
+
+ echo "\n...now running hacky pre-schemaupdate change for $table:";
+ // We have to create a urlhash that is _not_ the primary key,
+ // transfer data and THEN run checkSchema
+ $schemadef['fields']['urlhash'] = array (
+ 'type' => 'varchar',
+ 'length' => 64,
+ 'not null' => false, // this is because when adding column, all entries will _be_ NULL!
+ 'description' => 'sha256 of destination URL (url field)',
+ );
+ $schemadef['fields']['url'] = array (
+ 'type' => 'text',
+ 'description' => 'destination URL after following possible redirections',
+ );
+ unset($schemadef['unique keys']);
+ $schema->ensureTable($table, $schemadef);
+ echo "DONE.\n";
+
+ $classname = ucfirst($table);
+ $tablefix = new $classname;
+ // urlhash is hash('sha256', $url) in the File table
+ echo "Updating urlhash fields in $table table...";
+ // Maybe very MySQL specific :(
+ $tablefix->query(sprintf('UPDATE %1$s SET %2$s=%3$s;',
+ $schema->quoteIdentifier($table),
+ 'urlhash',
+ // The line below is "result of sha256 on column `url`"
+ 'SHA2(url, 256)'));
+ echo "DONE.\n";
+ echo "Resuming core schema upgrade...";
+ }
}