along with BookReader. If not, see <http://www.gnu.org/licenses/>.
*/
+require_once("BookReaderMeta.inc.php");
+
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',
+ 'x' => 'x',
+ 'y' => 'y',
+ '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
+ *
+ * Code path:
+ * - Get book metadata
+ * - Parse the requested page (e.g. cover_t.jpg, n5_r4.jpg) to determine which page type,
+ * size and format (etc) is being requested
+ * - Determine the leaf number corresponding to the page
+ * - Determine scaling values
+ * - Serve image request now that all information has been gathered
+ */
+
+ function serveLookupRequest($requestEnv) {
+ $brm = new BookReaderMeta();
+ try {
+ $metadata = $brm->buildMetadata($_REQUEST['id'], $_REQUEST['itemPath'], $_REQUEST['subPrefix'], $_REQUEST['server']);
+ } catch (Exception $e) {
+ $this->BRfatal($e->getMessage());
+ }
+
+ $page = $_REQUEST['page'];
+
+ // Index of image to return
+ $imageIndex = null;
+
+ // deal with subPrefix
+ if ($_REQUEST['subPrefix']) {
+ $parts = explode('/', $_REQUEST['subPrefix']);
+ $bookId = $parts[count($parts) - 1 ];
+ } else {
+ $bookId = $_REQUEST['id'];
+ }
+
+ $pageInfo = $this->parsePageRequest($page, $bookId);
+
+ $basePage = $pageInfo['type'];
+
+ $leaf = null;
+ $region = 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
+ // Cover page
+ // Page 0
+
+ if ( array_key_exists('date', $metadata) && array_key_exists('coverIndices', $metadata) ) {
+ if ($brm->parseYear($metadata['date']) >= 1950) {
+ $imageIndex = $metadata['coverIndices'][0];
+ break;
+ }
+ }
+ if (array_key_exists('titleIndex', $metadata)) {
+ $imageIndex = $metadata['titleIndex'];
+ break;
+ }
+ if (array_key_exists('coverIndices', $metadata)) {
+ $imageIndex = $metadata['coverIndices'][0];
+ break;
+ }
+
+ // First page
+ $imageIndex = 0;
+ break;
+
+ case 'n':
+ // Accessible index page
+ $imageIndex = intval($pageInfo['value']);
+ break;
+
+ case 'page':
+ // Named page
+ $index = array_search($pageInfo['value'], $metadata['pageNums']);
+ if ($index === FALSE) {
+ // Not found
+ $this->BRfatal("Page not found");
+ break;
+ }
+
+ $imageIndex = $index;
+ break;
+
+ case 'leaf':
+ // Leaf explicitly specified
+ $leaf = $pageInfo['value'];
+ break;
+
+ default:
+ // Shouldn't be possible
+ $this->BRfatal("Unrecognized page type requested");
+ break;
+
+ }
+
+ if (is_null($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'], $metadata['imageFormat']),
+ 'ext' => 'jpg', // XXX should pass through ext
+ );
+
+ // remove non-passthrough keys from pageInfo
+ unset($pageInfo['type']);
+ unset($pageInfo['value']);
+
+ // add pageinfo to request
+ $requestEnv = array_merge($pageInfo, $requestEnv);
+
+ // Return image data - will check privs
+ $this->serveRequest($requestEnv);
+
+ }
+
/*
+ * Returns a page image when all parameters such as the image stack location are
+ * passed in.
+ *
* Approach:
*
* Get info about requested image (input)
* Return image data
* Clean up temporary files
*/
-
function serveRequest($requestEnv) {
+
// Process some of the request parameters
$zipPath = $requestEnv['zip'];
$file = $requestEnv['file'];
}
if ( !file_exists($zipPath) ) {
- $this->BRfatal('Image stack does not exist');
+ $this->BRfatal('Image stack does not exist at ' . $zipPath);
}
// Make sure the image stack is readable - return 403 if not
$this->checkPrivs($zipPath);
// Get the image size and depth
$imageInfo = $this->getImageInfo($zipPath, $file);
+ $region = array();
+ foreach (array('x', 'y', 'width', 'height') as $key) {
+ if (array_key_exists($key, $requestEnv)) {
+ $region[$key] = $requestEnv[$key];
+ }
+ }
+ $regionDimensions = $this->getRegionDimensions($imageInfo, $region);
+
+ /*
+ print('imageInfo');
+ print_r($imageInfo);
+ print('region');
+ print_r($region);
+ print('regionDimensions');
+ print_r($regionDimensions);
+ print('asFloat');
+ print_r($this->getRegionDimensionsAsFloat($imageInfo, $region));
+ die(-1);
+ */
+
// Output json if requested
if ('json' == $ext) {
// $$$ we should determine the output size first based on requested scale
$jpegOptions = '-quality 75';
// The pbmreduce reduction factor produces an image with dimension 1/n
- // The kakadu reduction factor produceds an image with dimension 1/(2^n)
- if (isset($requestEnv['height'])) {
+ // The kakadu reduction factor produces an image with dimension 1/(2^n)
+
+ // We interpret the requested size and scale, look at image format, and determine the
+ // actual scaling to be returned to the client. We generally return the largest
+ // power of 2 reduction that is larger than the requested size in order to reduce
+ // image processing load on our cluster. The client should then scale to their final
+ // needed size.
+
+ // Set scale from height or width if set and no x or y specified
+ if ( isset($requestEnv['height']) && !isset($requestEnv['x']) && !isset($requestEnv['y']) ) {
$powReduce = $this->nearestPow2Reduce($requestEnv['height'], $imageInfo['height']);
$scale = pow(2, $powReduce);
- } else if (isset($requestEnv['width'])) {
+ } else if ( isset($requestEnv['width']) && !isset($requestEnv['x']) && !isset($requestEnv['y']) ) {
$powReduce = $this->nearestPow2Reduce($requestEnv['width'], $imageInfo['width']);
$scale = pow(2, $powReduce);
} else {
- // $$$ 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
// $$$ consider doing a 302 here instead, to make better use of the browser cache
// Limit scaling for 1-bit images. See https://bugs.edge.launchpad.net/bookreader/+bug/486011
if (1 == $imageInfo['bits']) {
+
if ($scale > 1) {
$scale /= 2;
$powReduce -= 1;
$unzipCmd = $this->getUnarchiveCommand($zipPath, $file);
- $decompressCmd = $this->getDecompressCmd($imageInfo['type'], $powReduce, $rotate, $scale, $stdoutLink);
+ $decompressCmd = $this->getDecompressCmd($imageInfo, $powReduce, $rotate, $scale, $region, $stdoutLink);
// Non-integer scaling is currently disabled on the cluster
// if (isset($_REQUEST['height'])) {
$filenameForClient = $this->filenameForClient($file, $ext);
- $headers = array('Content-type: '. $MIMES[$ext], // XXX is nginx swallowing this?
+ $headers = array('Content-type: '. self::$MIMES[$ext],
'Cache-Control: max-age=15552000',
'Content-disposition: inline; filename=' . $filenameForClient);
$errorMessage = '';
+
if (! $this->passthruIfSuccessful($headers, $cmd, $errorMessage)) { // $$$ move to BookReaderRequest
// $$$ automated reporting
trigger_error('BookReader Processing Error: ' . $cmd . ' -- ' . $errorMessage, E_USER_WARNING);
// Try some content-specific recovery
- $recovered = false;
+ $recovered = false;
if ($imageInfo['type'] == 'jp2') {
$records = $this->getJp2Records($zipPath, $file);
- if ($powReduce > intval($records['Clevels'])) {
- $powReduce = $records['Clevels'];
- $reduce = pow(2, $powReduce);
+ if (array_key_exists('Clevels', $records)) {
+ $maxReduce = intval($records['Clevels']);
+ trigger_error("BookReader using max reduce $maxReduce from jp2 records");
} else {
- $reduce = 1;
- $powReduce = 0;
+ $maxReduce = 0;
}
-
- $cmd = $unzipCmd . $this->getDecompressCmd($imageInfo['type'], $powReduce, $rotate, $scale, $stdoutLink) . $compressCmd;
+
+ $powReduce = min($powReduce, $maxReduce);
+ $reduce = pow(2, $powReduce);
+
+ $cmd = $unzipCmd . $this->getDecompressCmd($imageInfo, $powReduce, $rotate, $scale, $region, $stdoutLink) . $compressCmd;
+ trigger_error('BookReader rerunning with new cmd: ' . $cmd, E_USER_WARNING);
if ($this->passthruIfSuccessful($headers, $cmd, $errorMessage)) { // $$$ 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');
}
$bits = intval($tags["BitDepth"]);
break;
default:
- $this->BRfatal("Unsupported image type");
+ $this->BRfatal("Unsupported image type $type for file $file in $zipPath");
break;
}
echo $jsonOutput;
}
- function getDecompressCmd($imageType, $powReduce, $rotate, $scale, $stdoutLink) {
+ function getDecompressCmd($srcInfo, $powReduce, $rotate, $scale, $region, $stdoutLink) {
- switch ($imageType) {
+ switch ($srcInfo['type']) {
case 'jp2':
+ $regionAsFloat = $this->getRegionDimensionsAsFloat($srcInfo, $region);
+ $regionString = sprintf("{%f,%f},{%f,%f}", $regionAsFloat['y'], $regionAsFloat['x'], $regionAsFloat['h'], $regionAsFloat['w']);
$decompressCmd =
- " | " . $this->kduExpand . " -no_seek -quiet -reduce $powReduce -rotate $rotate -i /dev/stdin -o " . $stdoutLink;
+ " | " . $this->kduExpand . " -no_seek -quiet -reduce $powReduce -rotate $rotate -region $regionString -i /dev/stdin -o " . $stdoutLink;
if ($this->decompressToBmp) {
// We suppress output since bmptopnm always outputs on stderr
$decompressCmd .= ' | (bmptopnm 2>/dev/null)';
// 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);
// $$$ look at bit depth when reducing
$decompressCmd =
return $decompressCmd;
}
+
// If the command has its initial output on stdout the headers will be emitted followed
// by the stdout output. If initial output is on stderr an error message will be
// returned.
$read = array($stdout, $stderr);
$write = NULL;
$except = NULL;
+
$numChanged = stream_select($read, $write, $except, NULL); // $$$ 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: $string");
- //echo "alert('$string');\n";
- //die(-1);
}
// 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($filename)) {
header('HTTP/1.1 403 Forbidden');
exit(0);
return strlen($binStr) - 1;
}
+ /*
+ * Parses a page request like "page5_r2.jpg" or "cover_t.jpg" to corresponding
+ * page type, size, reduce, and format
+ */
+ 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];
+ $extension = $matches[2];
+ if ($extension == 'jpeg') {
+ $extension = 'jpg';
+ }
+ } else {
+ $extension = 'jpg';
+ }
+ $pageInfo['extension'] = $extension;
+
+ // Split parts out
+ $parts = explode('_', $pageRequest);
+
+ // Remove book prefix if it was included (historical)
+ if ($parts[0] == $bookPrefix) {
+ array_shift($parts);
+ }
+
+ if (count($parts) === 0) {
+ $this->BRfatal('No page type specified');
+ }
+ $page = array_shift($parts);
+
+ $pageTypes = array(
+ 'page' => 'str',
+ 'n' => 'num',
+ 'cover' => 'single',
+ 'preview' => 'single',
+ 'title' => 'single',
+ 'leaf' => 'num'
+ );
+
+ // Look for known page types
+ foreach ( $pageTypes as $pageName => $kind ) {
+ if ( preg_match('#^(' . $pageName . ')(.*)#', $page, $matches) === 1 ) {
+ $pageInfo['type'] = $matches[1];
+ switch ($kind) {
+ case 'str':
+ $pageInfo['value'] = $matches[2];
+ break;
+ case 'num':
+ $pageInfo['value'] = intval($matches[2]);
+ break;
+ case 'single':
+ break;
+ }
+ }
+ }
+
+ if ( !array_key_exists('type', $pageInfo) ) {
+ $this->BRfatal('Unrecognized page type');
+ }
+
+ // Look for other known parts
+ foreach ($parts as $part) {
+ if ( array_key_exists($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;
+ }
+
+ $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 $pageInfo;
+ }
+
+ function getRegionDimensions($sourceDimensions, $regionDimensions) {
+ // Return region dimensions as { 'x' => xOffset, 'y' => yOffset, 'w' => width, 'h' => height }
+ // in terms of full resolution image.
+ // Note: this will clip the returned dimensions to fit within the source image
+
+ $sourceX = 0;
+ if (array_key_exists('x', $regionDimensions)) {
+ $sourceX = $this->intAmount($regionDimensions['x'], $sourceDimensions['width']);
+ }
+ $sourceX = $this->clamp(0, $sourceDimensions['width'] - 2, $sourceX); // Allow at least one pixel
+
+ $sourceY = 0;
+ if (array_key_exists('y', $regionDimensions)) {
+ $sourceY = $this->intAmount($regionDimensions['y'], $sourceDimensions['height']);
+ }
+ $sourceY = $this->clamp(0, $sourceDimensions['height'] - 2, $sourceY); // Allow at least one pixel
+
+ $sourceWidth = $sourceDimensions['width'] - $sourceX;
+ if (array_key_exists('width', $regionDimensions)) {
+ $sourceWidth = $this->intAmount($regionDimensions['width'], $sourceDimensions['width']);
+ }
+ $sourceWidth = $this->clamp(1, max(1, $sourceDimensions['width'] - $sourceX), $sourceWidth);
+
+ $sourceHeight = $sourceDimensions['height'] - $sourceY;
+ if (array_key_exists('height', $regionDimensions)) {
+ $sourceHeight = $this->intAmount($regionDimensions['height'], $sourceDimensions['height']);
+ }
+ $sourceHeight = $this->clamp(1, max(1, $sourceDimensions['height'] - $sourceY), $sourceHeight);
+
+ return array('x' => $sourceX, 'y' => $sourceY, 'w' => $sourceWidth, 'h' => $sourceHeight);
+ }
+
+ function getRegionDimensionsAsFloat($sourceDimensions, $regionDimensions) {
+ // Return region dimensions as { 'x' => xOffset, 'y' => yOffset, 'w' => width, 'h' => height }
+ // in terms of full resolution image.
+ // Note: this will clip the returned dimensions to fit within the source image
+
+ $sourceX = 0;
+ if (array_key_exists('x', $regionDimensions)) {
+ $sourceX = $this->floatAmount($regionDimensions['x'], $sourceDimensions['width']);
+ }
+ $sourceX = $this->clamp(0.0, 1.0, $sourceX);
+
+ $sourceY = 0;
+ if (array_key_exists('y', $regionDimensions)) {
+ $sourceY = $this->floatAmount($regionDimensions['y'], $sourceDimensions['height']);
+ }
+ $sourceY = $this->clamp(0.0, 1.0, $sourceY);
+
+ $sourceWidth = 1 - $sourceX;
+ if (array_key_exists('width', $regionDimensions)) {
+ $sourceWidth = $this->floatAmount($regionDimensions['width'], $sourceDimensions['width']);
+ }
+ $sourceWidth = $this->clamp(0.0, 1.0, $sourceWidth);
+
+ $sourceHeight = 1 - $sourceY;
+ if (array_key_exists('height', $regionDimensions)) {
+ $sourceHeight = $this->floatAmount($regionDimensions['height'], $sourceDimensions['height']);
+ }
+ $sourceHeight = $this->clamp(0.0, 1.0, $sourceHeight);
+
+ return array('x' => $sourceX, 'y' => $sourceY, 'w' => $sourceWidth, 'h' => $sourceHeight);
+ }
+
+ function intAmount($stringValue, $maximum) {
+ // Returns integer amount for string like "5" (5 units) or "0.5" (50%)
+ if (strpos($stringValue, '.') === false) {
+ // No decimal, assume int
+ return intval($stringValue);
+ }
+
+ return floatval($stringValue) * $maximum + 0.5;
+ }
+
+ function floatAmount($stringValue, $maximum) {
+ // Returns float amount (0.0 to 1.0) for string like "0.4" (40%) or "4" (40% if max is 10)
+ if (strpos($stringValue, ".") === false) {
+ // No decimal, assume int value out of maximum
+ return floatval($stringValue) / $maximum;
+ }
+
+ // Given float - just pass through
+ return floatval($stringValue);
+ }
+
+ function clamp($minValue, $maxValue, $observedValue) {
+ if ($observedValue < $minValue) {
+ return $minValue;
+ }
+
+ if ($observedValue > $maxValue) {
+ return $maxValue;
+ }
+
+ return $observedValue;
+ }
+
+ // Clean up temporary files and resources
+ function cleanup() {
+ foreach($this->tempFiles as $tempFile) {
+ unlink($tempFile);
+ }
+ $this->tempFiles = array();
+ }
+
}
?>
\ No newline at end of file
projects / bookreader.git / blobdiff