class BookReaderImages
{
- public $MIMES = array('gif' => 'image/gif',
+ public static $MIMES = array('gif' => 'image/gif',
'jp2' => 'image/jp2',
'jpg' => 'image/jpeg',
'jpeg' => 'image/jpeg',
'tif' => 'image/tiff',
'tiff' => 'image/tiff');
- public $EXTENSIONS = array('gif' => 'gif',
+ public static $EXTENSIONS = array('gif' => 'gif',
'jp2' => 'jp2',
'jpeg' => 'jpeg',
'jpg' => 'jpeg',
'tiff' => 'tiff');
// Width when generating thumbnails
- public $imageSizes = array(
+ public static $imageSizes = array(
'thumb' => 100,
- 'small' => 240,
- 'medium' => 500,
- 'large' => 1024,
+ 'small' => 256,
+ 'medium' => 512,
+ 'large' => 2048,
+ );
+
+ // Keys in the image permalink urls, e.g. http://www.archive.org/download/itemid/page/cover_{keyval}_{keyval}.jpg
+ public static $imageUrlKeys = array(
+ //'r' => 'reduce', // pow of 2 reduction
+ 's' => 'scale', // $$$ scale is downscaling factor in BookReaderImages but most people call this "reduce"
+ 'region' => 'region',
+ 'tile' => 'tile',
+ 'w' => 'width',
+ 'h' => 'height',
+ 'rotate' => 'rotate'
);
// Paths to command-line tools
var $exiftool = '/petabox/sw/books/exiftool/exiftool';
var $kduExpand = '/petabox/sw/bin/kdu_expand';
+ // Name of temporary files, to be cleaned at exit
+ var $tempFiles = array();
+
/*
* Serve an image request that requires looking up the book metadata
*
try {
$metadata = $brm->buildMetadata($_REQUEST['id'], $_REQUEST['itemPath'], $_REQUEST['subPrefix'], $_REQUEST['server']);
} catch (Exception $e) {
- $this->BRfatal($e->getMessage);
+ $this->BRfatal($e->getMessage());
}
$page = $_REQUEST['page'];
$basePage = $pageInfo['type'];
+ $leaf = null;
switch ($basePage) {
+
case 'title':
if (! array_key_exists('titleIndex', $metadata)) {
$this->BRfatal("No title page asserted in book");
}
$imageIndex = $metadata['titleIndex'];
break;
-
+
+ /* Old 'cover' behaviour where it would show cover 0 if it exists or return 404.
+ Could be re-added as cover0, cover1, etc
case 'cover':
if (! array_key_exists('coverIndices', $metadata)) {
$this->BRfatal("No cover asserted in book");
}
$imageIndex = $metadata['coverIndices'][0]; // $$$ TODO add support for other covers
break;
-
+ */
+
case 'preview':
+ case 'cover': // Show our best guess if cover is requested
// Preference is:
// Cover page if book was published >= 1950
// Title page
$imageIndex = $index;
break;
+ case 'leaf':
+ // Leaf explicitly specified
+ $leaf = $pageInfo['value'];
+ break;
+
default:
// Shouldn't be possible
$this->BRfatal("Unrecognized page type requested");
}
- $leaf = $brm->leafForIndex($imageIndex, $metadata['leafNums']);
+ if (is_null($leaf)) {
+ // Leaf was not explicitly set -- look it up
+ $leaf = $brm->leafForIndex($imageIndex, $metadata['leafNums']);
+ }
$requestEnv = array(
'zip' => $metadata['zip'],
'file' => $brm->imageFilePath($leaf, $metadata['subPrefix'], $metadata['imageFormat']),
- 'ext' => 'jpg',
+ 'ext' => 'jpg', // XXX should pass through ext
);
+
+ // remove non-passthrough keys from pageInfo
+ unset($pageInfo['type']);
+ unset($pageInfo['value']);
+
+ // add pageinfo to request
+ $requestEnv = array_merge($pageInfo, $requestEnv);
// Return image data - will check privs
$this->serveRequest($requestEnv);
* Clean up temporary files
*/
function serveRequest($requestEnv) {
+
// Process some of the request parameters
$zipPath = $requestEnv['zip'];
$file = $requestEnv['file'];
$jpegOptions = '-quality 75';
// The pbmreduce reduction factor produces an image with dimension 1/n
- // The kakadu reduction factor produceds an image with dimension 1/(2^n)
+ // The kakadu reduction factor produces an image with dimension 1/(2^n)
+
+ // We interpret the requested size and scale, look at image format, and determine the
+ // actual scaling to be returned to the client. We generally return the largest
+ // power of 2 reduction that is larger than the requested size in order to reduce
+ // image processing load on our cluster. The client should then scale to their final
+ // needed size.
+
+ // Set scale from height or width if set
if (isset($requestEnv['height'])) {
$powReduce = $this->nearestPow2Reduce($requestEnv['height'], $imageInfo['height']);
$scale = pow(2, $powReduce);
$scale = pow(2, $powReduce);
} else {
- // $$$ could be cleaner
- // Provide next smaller power of two reduction
- $scale = $requestEnv['scale'];
- if (!$scale) {
- $scale = 1;
- }
- if (array_key_exists($scale, $this->imageSizes)) {
+ // Set scale from named size (e.g. 'large') if set
+ $size = $requestEnv['size'];
+ if ( $size && array_key_exists($size, self::$imageSizes)) {
$srcRatio = floatval($imageInfo['width']) / floatval($imageInfo['height']);
if ($srcRatio > 1) {
// wide
} else {
$dimension = 'height';
}
- $powReduce = $this->nearestPow2Reduce($this->imageSizes[$scale], $imageInfo[$dimension]);
+ $powReduce = $this->nearestPow2Reduce(self::$imageSizes[$size], $imageInfo[$dimension]);
+ $scale = pow(2, $powReduce);
+
} else {
+ // No named size - use explicit scale, if given
+ $scale = $requestEnv['scale'];
+ if (!$scale) {
+ $scale = 1;
+ }
$powReduce = $this->nearestPow2ForScale($scale);
- }
- $scale = pow(2, $powReduce);
+ // ensure integer scale
+ $scale = pow(2, $powReduce);
+ }
}
// Override depending on source image format
// $$$ consider doing a 302 here instead, to make better use of the browser cache
// Limit scaling for 1-bit images. See https://bugs.edge.launchpad.net/bookreader/+bug/486011
if (1 == $imageInfo['bits']) {
+
if ($scale > 1) {
$scale /= 2;
$powReduce -= 1;
$filenameForClient = $this->filenameForClient($file, $ext);
- $headers = array('Content-type: '. $MIMES[$ext], // XXX is nginx swallowing this?
+ $headers = array('Content-type: '. self::$MIMES[$ext],
'Cache-Control: max-age=15552000',
'Content-disposition: inline; filename=' . $filenameForClient);
$errorMessage = '';
+
if (! $this->passthruIfSuccessful($headers, $cmd, $errorMessage)) { // $$$ move to BookReaderRequest
// $$$ automated reporting
trigger_error('BookReader Processing Error: ' . $cmd . ' -- ' . $errorMessage, E_USER_WARNING);
// Try some content-specific recovery
- $recovered = false;
+ $recovered = false;
if ($imageInfo['type'] == 'jp2') {
$records = $this->getJp2Records($zipPath, $file);
- if ($powReduce > intval($records['Clevels'])) {
- $powReduce = $records['Clevels'];
- $reduce = pow(2, $powReduce);
+ if (array_key_exists('Clevels', $records)) {
+ $maxReduce = intval($records['Clevels']);
+ trigger_error("BookReader using max reduce $maxReduce from jp2 records");
} else {
- $reduce = 1;
- $powReduce = 0;
+ $maxReduce = 0;
}
-
+
+ $powReduce = min($powReduce, $maxReduce);
+ $reduce = pow(2, $powReduce);
+
$cmd = $unzipCmd . $this->getDecompressCmd($imageInfo['type'], $powReduce, $rotate, $scale, $stdoutLink) . $compressCmd;
+ trigger_error('BookReader rerunning with new cmd: ' . $cmd, E_USER_WARNING);
if ($this->passthruIfSuccessful($headers, $cmd, $errorMessage)) { // $$$ move to BookReaderRequest
$recovered = true;
} else {
+ $this->cleanup();
trigger_error('BookReader fallback image processing also failed: ' . $errorMessage, E_USER_WARNING);
}
}
}
}
- if (isset($tempFile)) {
- unlink($tempFile);
- }
+ $this->cleanup();
}
function getUnarchiveCommand($archivePath, $file)
function imageExtensionToType($extension)
{
- if (array_key_exists($extension, $this->EXTENSIONS)) {
- return $this->EXTENSIONS[$extension];
+ if (array_key_exists($extension, self::$EXTENSIONS)) {
+ return self::$EXTENSIONS[$extension];
} else {
$this->BRfatal('Unknown image extension');
}
// We use the BookReaderTiff prefix to give a hint in case things don't
// get cleaned up.
$tempFile = tempnam("/tmp", "BookReaderTiff");
+ array_push($this->tempFiles, $tempFile);
// $$$ look at bit depth when reducing
$decompressCmd =
$read = array($stdout, $stderr);
$write = NULL;
$except = NULL;
+
$numChanged = stream_select($read, $write, $except, NULL); // $$$ no timeout
if (false === $numChanged) {
// select failed
$errorMessage = 'Select failed';
$retVal = false;
- }
- if ($read[0] == $stdout && (1 == $numChanged)) {
- // Got output first on stdout (only)
- // $$$ make sure we get all stdout
+ error_log('BookReader select failed!');
+ } else {
+ if (in_array($stderr, $read)) {
+ // Either content in stderr, or stderr is closed (could read 0 bytes)
+ $error = stream_get_contents($stderr);
+ if ($error) {
+
+ $errorMessage = $error;
+ $retVal = false;
+
+ fclose($stderr);
+ fclose($stdout);
+ fclose($stdin);
+
+ // It is important that you close any pipes before calling
+ // proc_close in order to avoid a deadlock
+ proc_close($process);
+ return $retVal;
+
+ }
+ }
+
$output = fopen('php://output', 'w');
foreach($headers as $header) {
header($header);
stream_copy_to_stream($pipes[1], $output);
fclose($output); // okay since tied to special php://output
$retVal = true;
- } else {
- // Got output on stderr
- // $$$ make sure we get all stderr
- $errorMessage = stream_get_contents($stderr);
- $retVal = false;
}
fclose($stderr);
}
function BRfatal($string) {
+ $this->cleanup();
throw new Exception("Image error: $string");
}
// Returns true if using a power node
+ // XXX change to "on red box" - not working for new Xeon
function onPowerNode() {
exec("lspci | fgrep -c Realtek", $output, $return);
if ("0" != $output[0]) {
}
function checkPrivs($filename) {
+ // $$$ we assume here that requests for the title, cover or preview
+ // come in via BookReaderPreview.php which will be re-run with
+ // privileges after we return the 403
if (!is_readable($filename)) {
header('HTTP/1.1 403 Forbidden');
exit(0);
*/
function parsePageRequest($pageRequest, $bookPrefix) {
+ // Will hold parsed results
$pageInfo = array();
+ // Normalize
+ $pageRequest = strtolower($pageRequest);
+
// Pull off extension
if (preg_match('#(.*)\.([^.]+)$#', $pageRequest, $matches) === 1) {
$pageRequest = $matches[1];
'n' => 'num',
'cover' => 'single',
'preview' => 'single',
- 'title' => 'single'
+ 'title' => 'single',
+ 'leaf' => 'num'
);
// Look for known page types
// Look for other known parts
foreach ($parts as $part) {
- $start = substr($part, 0, 1);
+ if ( array_key_exists($part, self::$imageSizes) ) {
+ $pageInfo['size'] = $part;
+ continue;
+ }
+
+ // Key must be alpha, value must start with digit and contain digits, alpha, ',' or '.'
+ // Should prevent injection of strange values into the redirect to datanode
+ if ( preg_match('#^([a-z]+)(\d[a-z0-9,.]*)#', $part, $matches) === 0) {
+ // Not recognized
+ continue;
+ }
- switch ($start) {
- case 't':
- $pageInfo['size'] = $start;
- break;
- case 'r':
- $pageInfo['reduce'] = substr($part, 0);
- break;
- default:
- // Unrecognized... just let it pass
- break;
+ $key = $matches[1];
+ $value = $matches[2];
+
+ if ( array_key_exists($key, self::$imageUrlKeys) ) {
+ $pageInfo[self::$imageUrlKeys[$key]] = $value;
+ continue;
}
+
+ // If we hit here, was unrecognized (no action)
}
return $pageInfo;
}
+ // Clean up temporary files and resources
+ function cleanup() {
+ foreach($this->tempFiles as $tempFile) {
+ unlink($tempFile);
+ }
+ $this->tempFiles = array();
+ }
+
}
?>
\ No newline at end of file
projects / bookreader.git / blobdiff