/**
* Perform media processing for the given field if something has been uploaded
*
* @access private
* @param string $ps_field field name
* @param array options
*
* Supported options:
* delete_old_media = set to zero to prevent that old media files are deleted; defaults to 1
* these_versions_only = if set to an array of valid version names, then only the specified versions are updated with the currently updated file; ignored if no media already exists
* dont_allow_duplicate_media = if set to true, and the model as a field named "md5" then media will be rejected if a row already exists with the same MD5 signature
*/
public function _processMedia($ps_field, $pa_options = null)
{
global $AUTH_CURRENT_USER_ID;
if (!is_array($pa_options)) {
$pa_options = array();
}
if (!isset($pa_options['delete_old_media'])) {
$pa_options['delete_old_media'] = true;
}
if (!isset($pa_options['these_versions_only'])) {
$pa_options['these_versions_only'] = null;
}
$vs_sql = "";
$vn_max_execution_time = ini_get('max_execution_time');
set_time_limit(7200);
$o_tq = new TaskQueue();
$o_media_proc_settings = new MediaProcessingSettings($this, $ps_field);
# only set file if something was uploaded
# (ie. don't nuke an existing file because none
# was uploaded)
$va_field_info = $this->getFieldInfo($ps_field);
if (isset($this->_FILES_CLEAR[$ps_field]) && $this->_FILES_CLEAR[$ps_field]) {
//
// Clear files
//
$va_versions = $this->getMediaVersions($ps_field);
#--- delete files
foreach ($va_versions as $v) {
$this->_removeMedia($ps_field, $v);
}
$this->_removeMedia($ps_field, '_undo_');
$this->_FILES[$ps_field] = null;
$this->_FIELD_VALUES[$ps_field] = null;
$vs_sql = "{$ps_field} = " . $this->quote(caSerializeForDatabase($this->_FILES[$ps_field], true)) . ",";
} else {
// Don't try to process files when no file is actually set
if (isset($this->_SET_FILES[$ps_field]['tmp_name'])) {
$o_tq = new TaskQueue();
$o_media_proc_settings = new MediaProcessingSettings($this, $ps_field);
//
// Process incoming files
//
$m = new Media();
$va_media_objects = array();
// is it a URL?
$vs_url_fetched_from = null;
$vn_url_fetched_on = null;
$vb_allow_fetching_of_urls = (bool) $this->_CONFIG->get('allow_fetching_of_media_from_remote_urls');
$vb_is_fetched_file = false;
if ($vb_allow_fetching_of_urls && (bool) ini_get('allow_url_fopen') && isURL($vs_url = html_entity_decode($this->_SET_FILES[$ps_field]['tmp_name']))) {
$vs_tmp_file = tempnam(__CA_APP_DIR__ . '/tmp', 'caUrlCopy');
$r_incoming_fp = fopen($vs_url, 'r');
if (!$r_incoming_fp) {
$this->postError(1600, _t('Cannot open remote URL [%1] to fetch media', $vs_url), "BaseModel->_processMedia()", $this->tableName() . '.' . $ps_field);
set_time_limit($vn_max_execution_time);
return false;
}
$r_outgoing_fp = fopen($vs_tmp_file, 'w');
if (!$r_outgoing_fp) {
$this->postError(1600, _t('Cannot open file for media fetched from URL [%1]', $vs_url), "BaseModel->_processMedia()", $this->tableName() . '.' . $ps_field);
set_time_limit($vn_max_execution_time);
return false;
}
while (($vs_content = fgets($r_incoming_fp, 4096)) !== false) {
fwrite($r_outgoing_fp, $vs_content);
}
fclose($r_incoming_fp);
fclose($r_outgoing_fp);
$vs_url_fetched_from = $vs_url;
$vn_url_fetched_on = time();
$this->_SET_FILES[$ps_field]['tmp_name'] = $vs_tmp_file;
$vb_is_fetched_file = true;
}
// is it server-side stored user media?
if (preg_match("!^userMedia[\\d]+/!", $this->_SET_FILES[$ps_field]['tmp_name'])) {
// use configured directory to dump media with fallback to standard tmp directory
if (!is_writeable($vs_tmp_directory = $this->getAppConfig()->get('ajax_media_upload_tmp_directory'))) {
$vs_tmp_directory = caGetTempDirPath();
}
$this->_SET_FILES[$ps_field]['tmp_name'] = "{$vs_tmp_directory}/" . $this->_SET_FILES[$ps_field]['tmp_name'];
// read metadata
if (file_exists("{$vs_tmp_directory}/" . $this->_SET_FILES[$ps_field]['tmp_name'] . "_metadata")) {
if (is_array($va_tmp_metadata = json_decode(file_get_contents("{$vs_tmp_directory}/" . $this->_SET_FILES[$ps_field]['tmp_name'] . "_metadata")))) {
$this->_SET_FILES[$ps_field]['original_filename'] = $va_tmp_metadata['original_filename'];
}
}
}
if (isset($this->_SET_FILES[$ps_field]['tmp_name']) && file_exists($this->_SET_FILES[$ps_field]['tmp_name'])) {
if (!isset($pa_options['dont_allow_duplicate_media'])) {
$pa_options['dont_allow_duplicate_media'] = (bool) $this->getAppConfig()->get('dont_allow_duplicate_media');
}
if (isset($pa_options['dont_allow_duplicate_media']) && $pa_options['dont_allow_duplicate_media']) {
if ($this->hasField('md5')) {
$qr_dupe_chk = $this->getDb()->query("\n\t\t\t\t\t\t\t\tSELECT " . $this->primaryKey() . " FROM " . $this->tableName() . " WHERE md5 = ? " . ($this->hasField('deleted') ? ' AND deleted = 0' : '') . "\n\t\t\t\t\t\t\t", (string) md5_file($this->_SET_FILES[$ps_field]['tmp_name']));
if ($qr_dupe_chk->nextRow()) {
$this->postError(1600, _t("Media already exists in database"), "BaseModel->_processMedia()", $this->tableName() . '.' . $ps_field);
return false;
}
}
}
// allow adding zip and (gzipped) tape archives
$vb_is_archive = false;
$vs_original_filename = $this->_SET_FILES[$ps_field]['original_filename'];
$vs_original_tmpname = $this->_SET_FILES[$ps_field]['tmp_name'];
$va_matches = array();
if (preg_match("/(\\.zip|\\.tar\\.gz|\\.tgz)\$/", $vs_original_filename, $va_matches)) {
$vs_archive_extension = $va_matches[1];
// add file extension to temporary file if necessary; otherwise phar barfs when handling the archive
$va_tmp = array();
preg_match("/[.]*\\.([a-zA-Z0-9]+)\$/", $vs_original_tmpname, $va_tmp);
if (!isset($va_tmp[1])) {
@rename($this->_SET_FILES[$ps_field]['tmp_name'], $this->_SET_FILES[$ps_field]['tmp_name'] . $vs_archive_extension);
$this->_SET_FILES[$ps_field]['tmp_name'] = $this->_SET_FILES[$ps_field]['tmp_name'] . $vs_archive_extension;
}
if (caIsArchive($this->_SET_FILES[$ps_field]['tmp_name'])) {
$va_archive_files = caGetDirectoryContentsAsList('phar://' . $this->_SET_FILES[$ps_field]['tmp_name']);
if (sizeof($va_archive_files) > 0) {
// get first file we encounter in the archive and use it to generate them previews
$vb_is_archive = true;
$vs_archive = $this->_SET_FILES[$ps_field]['tmp_name'];
$va_tmp = array();
// copy primary file from the archive to temporary file (with extension so that *Magick can identify properly)
// this is basically a fallback. if the code below fails, we still have a 'fake' original
preg_match("/[.]*\\.([a-zA-Z0-9]+)\$/", $va_archive_files[0], $va_tmp);
$vs_primary_file_tmp = tempnam(caGetTempDirPath(), "caArchivePrimary");
@unlink($vs_primary_file_tmp);
$vs_primary_file_tmp = $vs_primary_file_tmp . "." . $va_tmp[1];
if (!@copy($va_archive_files[0], $vs_primary_file_tmp)) {
$this->postError(1600, _t("Couldn't extract first file from archive. There is probably a invalid character in a directory or file name inside the archive"), "BaseModel->_processMedia()", $this->tableName() . '.' . $ps_field);
set_time_limit($vn_max_execution_time);
if ($vb_is_fetched_file) {
@unlink($vs_tmp_file);
}
if ($vb_is_archive) {
@unlink($vs_archive);
@unlink($vs_primary_file_tmp);
}
return false;
}
$this->_SET_FILES[$ps_field]['tmp_name'] = $vs_primary_file_tmp;
// prepare to join archive contents to form a better downloadable "original" than the zip/tgz archive (e.g. a multi-page tiff)
// to do that, we have to 'extract' the archive so that command-line utilities like *Magick can operate
@caRemoveDirectory(caGetTempDirPath() . '/caArchiveExtract');
// remove left-overs, just to be sure we don't include other files in the tiff
$va_archive_tmp_files = array();
@mkdir(caGetTempDirPath() . '/caArchiveExtract');
foreach ($va_archive_files as $vs_archive_file) {
$vs_basename = basename($vs_archive_file);
$vs_tmp_file_name = caGetTempDirPath() . '/caArchiveExtract/' . str_replace(" ", "_", $vs_basename);
$va_archive_tmp_files[] = $vs_tmp_file_name;
@copy($vs_archive_file, $vs_tmp_file_name);
}
}
}
if (!$vb_is_archive) {
// something went wrong (i.e. no valid or empty archive) -> restore previous state
@rename($this->_SET_FILES[$ps_field]['tmp_name'] . $vs_archive_extension, $vs_original_tmpname);
$this->_SET_FILES[$ps_field]['tmp_name'] = $vs_original_tmpname;
}
}
// ImageMagick partly relies on file extensions to properly identify images (RAW images in particular)
// therefore we rename the temporary file here (using the extension of the original filename, if any)
$va_matches = array();
$vb_renamed_tmpfile = false;
preg_match("/[.]*\\.([a-zA-Z0-9]+)\$/", $this->_SET_FILES[$ps_field]['tmp_name'], $va_matches);
if (!isset($va_matches[1])) {
// file has no extension, i.e. is probably PHP upload tmp file
$va_matches = array();
preg_match("/[.]*\\.([a-zA-Z0-9]+)\$/", $this->_SET_FILES[$ps_field]['original_filename'], $va_matches);
if (strlen($va_matches[1]) > 0) {
$va_parts = explode("/", $this->_SET_FILES[$ps_field]['tmp_name']);
$vs_new_filename = sys_get_temp_dir() . "/" . $va_parts[sizeof($va_parts) - 1] . "." . $va_matches[1];
if (!move_uploaded_file($this->_SET_FILES[$ps_field]['tmp_name'], $vs_new_filename)) {
rename($this->_SET_FILES[$ps_field]['tmp_name'], $vs_new_filename);
}
$this->_SET_FILES[$ps_field]['tmp_name'] = $vs_new_filename;
$vb_renamed_tmpfile = true;
}
}
$input_mimetype = $m->divineFileFormat($this->_SET_FILES[$ps_field]['tmp_name']);
if (!($input_type = $o_media_proc_settings->canAccept($input_mimetype))) {
# error - filetype not accepted by this field
$this->postError(1600, $input_mimetype ? _t("File type %1 not accepted by %2", $input_mimetype, $ps_field) : _t("Unknown file type not accepted by %1", $ps_field), "BaseModel->_processMedia()", $this->tableName() . '.' . $ps_field);
set_time_limit($vn_max_execution_time);
if ($vb_is_fetched_file) {
@unlink($vs_tmp_file);
}
if ($vb_is_archive) {
@unlink($vs_archive);
@unlink($vs_primary_file_tmp);
}
return false;
}
# ok process file...
if (!$m->read($this->_SET_FILES[$ps_field]['tmp_name'])) {
$this->errors = array_merge($this->errors, $m->errors());
// copy into model plugin errors
set_time_limit($vn_max_execution_time);
if ($vb_is_fetched_file) {
@unlink($vs_tmp_file);
}
if ($vb_is_archive) {
@unlink($vs_archive);
@unlink($vs_primary_file_tmp);
}
return false;
}
// if necessary, join archive contents to form a better downloadable "original" than the zip/tgz archive (e.g. a multi-page tiff)
// by this point, a backend plugin was picked using the first file of the archive. this backend is also used to prepare the new original.
if ($vb_is_archive && sizeof($va_archive_tmp_files) > 0) {
if ($vs_archive_original = $m->joinArchiveContents($va_archive_tmp_files)) {
// mangle filename, so that the uploaded archive.zip becomes archive.tif or archive.gif or whatever extension the Media plugin prefers
$va_new_original_pathinfo = pathinfo($vs_archive_original);
$va_archive_pathinfo = pathinfo($this->_SET_FILES[$ps_field]['original_filename']);
$this->_SET_FILES[$ps_field]['original_filename'] = $va_archive_pathinfo['filename'] . "." . $va_new_original_pathinfo['extension'];
// this is now our original, disregard the uploaded archive
$this->_SET_FILES[$ps_field]['tmp_name'] = $vs_archive_original;
// re-run mimetype detection and read on the new original
$m->reset();
$input_mimetype = $m->divineFileFormat($vs_archive_original);
$input_type = $o_media_proc_settings->canAccept($input_mimetype);
$m->read($vs_archive_original);
}
@caRemoveDirectory(caGetTempDirPath() . '/caArchiveExtract');
}
$va_media_objects['_original'] = $m;
// preserve center setting from any existing media
$va_center = null;
if (is_array($va_tmp = $this->getMediaInfo($ps_field))) {
$va_center = caGetOption('_CENTER', $va_tmp, array());
}
$media_desc = array("ORIGINAL_FILENAME" => $this->_SET_FILES[$ps_field]['original_filename'], "_CENTER" => $va_center, "_SCALE" => caGetOption('_SCALE', $va_tmp, array()), "_SCALE_UNITS" => caGetOption('_SCALE_UNITS', $va_tmp, array()), "INPUT" => array("MIMETYPE" => $m->get("mimetype"), "WIDTH" => $m->get("width"), "HEIGHT" => $m->get("height"), "MD5" => md5_file($this->_SET_FILES[$ps_field]['tmp_name']), "FILESIZE" => filesize($this->_SET_FILES[$ps_field]['tmp_name']), "FETCHED_FROM" => $vs_url_fetched_from, "FETCHED_ON" => $vn_url_fetched_on));
if (isset($this->_SET_FILES[$ps_field]['options']['TRANSFORMATION_HISTORY']) && is_array($this->_SET_FILES[$ps_field]['options']['TRANSFORMATION_HISTORY'])) {
$media_desc['TRANSFORMATION_HISTORY'] = $this->_SET_FILES[$ps_field]['options']['TRANSFORMATION_HISTORY'];
}
#
# Extract metadata from file
#
$media_metadata = $m->getExtractedMetadata();
# get versions to create
$va_versions = $this->getMediaVersions($ps_field, $input_mimetype);
$error = 0;
# don't process files that are not going to be processed or converted
# we don't want to waste time opening file we're not going to do anything with
# also, we don't want to recompress JPEGs...
$media_type = $o_media_proc_settings->canAccept($input_mimetype);
$version_info = $o_media_proc_settings->getMediaTypeVersions($media_type);
$va_default_queue_settings = $o_media_proc_settings->getMediaTypeQueueSettings($media_type);
if (!($va_media_write_options = $this->_FILES[$ps_field]['options'])) {
$va_media_write_options = $this->_SET_FILES[$ps_field]['options'];
}
# Is an "undo" version set in options?
if (isset($this->_SET_FILES[$ps_field]['options']['undo']) && file_exists($this->_SET_FILES[$ps_field]['options']['undo'])) {
if ($volume = $version_info['original']['VOLUME']) {
$vi = $this->_MEDIA_VOLUMES->getVolumeInformation($volume);
if ($vi["absolutePath"] && strlen($dirhash = $this->_getDirectoryHash($vi["absolutePath"], $this->getPrimaryKey()))) {
$magic = rand(0, 99999);
$vs_filename = $this->_genMediaName($ps_field) . "_undo_";
$filepath = $vi["absolutePath"] . "/" . $dirhash . "/" . $magic . "_" . $vs_filename;
if (copy($this->_SET_FILES[$ps_field]['options']['undo'], $filepath)) {
$media_desc['_undo_'] = array("VOLUME" => $volume, "FILENAME" => $vs_filename, "HASH" => $dirhash, "MAGIC" => $magic, "MD5" => md5_file($filepath));
}
}
}
}
$va_process_these_versions_only = array();
if (isset($pa_options['these_versions_only']) && is_array($pa_options['these_versions_only']) && sizeof($pa_options['these_versions_only'])) {
$va_tmp = $this->_FIELD_VALUES[$ps_field];
foreach ($pa_options['these_versions_only'] as $vs_this_version_only) {
if (in_array($vs_this_version_only, $va_versions)) {
if (is_array($this->_FIELD_VALUES[$ps_field])) {
$va_process_these_versions_only[] = $vs_this_version_only;
}
}
}
// Copy metadata for version we're not processing
if (sizeof($va_process_these_versions_only)) {
foreach (array_keys($va_tmp) as $v) {
if (!in_array($v, $va_process_these_versions_only)) {
$media_desc[$v] = $va_tmp[$v];
}
}
}
}
$va_files_to_delete = array();
$va_queued_versions = array();
$queue_enabled = !sizeof($va_process_these_versions_only) && $this->getAppConfig()->get('queue_enabled') ? true : false;
$vs_path_to_queue_media = null;
foreach ($va_versions as $v) {
$vs_use_icon = null;
if (sizeof($va_process_these_versions_only) && !in_array($v, $va_process_these_versions_only)) {
// only processing certain versions... and this one isn't it so skip
continue;
}
$queue = $va_default_queue_settings['QUEUE'];
$queue_threshold = isset($version_info[$v]['QUEUE_WHEN_FILE_LARGER_THAN']) ? intval($version_info[$v]['QUEUE_WHEN_FILE_LARGER_THAN']) : (int) $va_default_queue_settings['QUEUE_WHEN_FILE_LARGER_THAN'];
$rule = isset($version_info[$v]['RULE']) ? $version_info[$v]['RULE'] : '';
$volume = isset($version_info[$v]['VOLUME']) ? $version_info[$v]['VOLUME'] : '';
$basis = isset($version_info[$v]['BASIS']) ? $version_info[$v]['BASIS'] : '';
if (isset($media_desc[$basis]) && isset($media_desc[$basis]['FILENAME'])) {
if (!isset($va_media_objects[$basis])) {
$o_media = new Media();
$basis_vi = $this->_MEDIA_VOLUMES->getVolumeInformation($media_desc[$basis]['VOLUME']);
if ($o_media->read($p = $basis_vi['absolutePath'] . "/" . $media_desc[$basis]['HASH'] . "/" . $media_desc[$basis]['MAGIC'] . "_" . $media_desc[$basis]['FILENAME'])) {
$va_media_objects[$basis] = $o_media;
} else {
$m = $va_media_objects['_original'];
}
} else {
$m = $va_media_objects[$basis];
}
} else {
$m = $va_media_objects['_original'];
}
$m->reset();
# get volume
$vi = $this->_MEDIA_VOLUMES->getVolumeInformation($volume);
if (!is_array($vi)) {
print "Invalid volume '{$volume}'<br>";
exit;
}
// Send to queue if it's too big to process here
if ($queue_enabled && $queue && $queue_threshold > 0 && $queue_threshold < (int) $media_desc["INPUT"]["FILESIZE"] && $va_default_queue_settings['QUEUE_USING_VERSION'] != $v) {
$va_queued_versions[$v] = array('VOLUME' => $volume);
$media_desc[$v]["QUEUED"] = $queue;
if ($version_info[$v]["QUEUED_MESSAGE"]) {
$media_desc[$v]["QUEUED_MESSAGE"] = $version_info[$v]["QUEUED_MESSAGE"];
} else {
$media_desc[$v]["QUEUED_MESSAGE"] = $va_default_queue_settings['QUEUED_MESSAGE'] ? $va_default_queue_settings['QUEUED_MESSAGE'] : _t("Media is being processed and will be available shortly.");
}
if ($pa_options['delete_old_media']) {
$va_files_to_delete[] = array('field' => $ps_field, 'version' => $v);
}
continue;
}
# get transformation rules
$rules = $o_media_proc_settings->getMediaTransformationRule($rule);
if (sizeof($rules) == 0) {
$output_mimetype = $input_mimetype;
$m->set("version", $v);
#
# don't process this media, just copy the file
#
$ext = $m->mimetype2extension($output_mimetype);
if (!$ext) {
$this->postError(1600, _t("File could not be copied for %1; can't convert mimetype '%2' to extension", $ps_field, $output_mimetype), "BaseModel->_processMedia()", $this->tableName() . '.' . $ps_field);
$m->cleanup();
set_time_limit($vn_max_execution_time);
if ($vb_is_fetched_file) {
@unlink($vs_tmp_file);
}
if ($vb_is_archive) {
@unlink($vs_archive);
@unlink($vs_primary_file_tmp);
@unlink($vs_archive_original);
}
return false;
}
if (($dirhash = $this->_getDirectoryHash($vi["absolutePath"], $this->getPrimaryKey())) === false) {
$this->postError(1600, _t("Could not create subdirectory for uploaded file in %1. Please ask your administrator to check the permissions of your media directory.", $vi["absolutePath"]), "BaseModel->_processMedia()", $this->tableName() . '.' . $ps_field);
set_time_limit($vn_max_execution_time);
if ($vb_is_fetched_file) {
@unlink($vs_tmp_file);
}
if ($vb_is_archive) {
@unlink($vs_archive);
@unlink($vs_primary_file_tmp);
@unlink($vs_archive_original);
}
return false;
}
if ((bool) $version_info[$v]["USE_EXTERNAL_URL_WHEN_AVAILABLE"]) {
$filepath = $this->_SET_FILES[$ps_field]['tmp_name'];
if ($pa_options['delete_old_media']) {
$va_files_to_delete[] = array('field' => $ps_field, 'version' => $v);
}
$media_desc[$v] = array("VOLUME" => $volume, "MIMETYPE" => $output_mimetype, "WIDTH" => $m->get("width"), "HEIGHT" => $m->get("height"), "PROPERTIES" => $m->getProperties(), "EXTERNAL_URL" => $media_desc['INPUT']['FETCHED_FROM'], "FILENAME" => null, "HASH" => null, "MAGIC" => null, "EXTENSION" => $ext, "MD5" => md5_file($filepath));
} else {
$magic = rand(0, 99999);
$filepath = $vi["absolutePath"] . "/" . $dirhash . "/" . $magic . "_" . $this->_genMediaName($ps_field) . "_" . $v . "." . $ext;
if (!copy($this->_SET_FILES[$ps_field]['tmp_name'], $filepath)) {
$this->postError(1600, _t("File could not be copied. Ask your administrator to check permissions and file space for %1", $vi["absolutePath"]), "BaseModel->_processMedia()", $this->tableName() . '.' . $ps_field);
$m->cleanup();
set_time_limit($vn_max_execution_time);
if ($vb_is_fetched_file) {
@unlink($vs_tmp_file);
}
if ($vb_is_archive) {
@unlink($vs_archive);
@unlink($vs_primary_file_tmp);
@unlink($vs_archive_original);
}
return false;
}
if ($v === $va_default_queue_settings['QUEUE_USING_VERSION']) {
$vs_path_to_queue_media = $filepath;
}
if ($pa_options['delete_old_media']) {
$va_files_to_delete[] = array('field' => $ps_field, 'version' => $v, 'dont_delete_path' => $vi["absolutePath"] . "/" . $dirhash . "/" . $magic . "_" . $this->_genMediaName($ps_field) . "_" . $v, 'dont_delete_extension' => $ext);
}
if (is_array($vi["mirrors"]) && sizeof($vi["mirrors"]) > 0) {
$vs_entity_key = join("/", array($this->tableName(), $ps_field, $this->getPrimaryKey(), $v));
$vs_row_key = join("/", array($this->tableName(), $this->getPrimaryKey()));
foreach ($vi["mirrors"] as $vs_mirror_code => $va_mirror_info) {
$vs_mirror_method = $va_mirror_info["method"];
$vs_queue = $vs_mirror_method . "mirror";
if (!$o_tq->cancelPendingTasksForEntity($vs_entity_key)) {
//$this->postError(560, _t("Could not cancel pending tasks: %1", $this->error),"BaseModel->_processMedia()");
//$m->cleanup();
//return false;
}
if ($o_tq->addTask($vs_queue, array("MIRROR" => $vs_mirror_code, "VOLUME" => $volume, "FIELD" => $ps_field, "TABLE" => $this->tableName(), "VERSION" => $v, "FILES" => array(array("FILE_PATH" => $filepath, "ABS_PATH" => $vi["absolutePath"], "HASH" => $dirhash, "FILENAME" => $magic . "_" . $this->_genMediaName($ps_field) . "_" . $v . "." . $ext)), "MIRROR_INFO" => $va_mirror_info, "PK" => $this->primaryKey(), "PK_VAL" => $this->getPrimaryKey()), array("priority" => 100, "entity_key" => $vs_entity_key, "row_key" => $vs_row_key, 'user_id' => $AUTH_CURRENT_USER_ID))) {
continue;
} else {
$this->postError(100, _t("Couldn't queue mirror using '%1' for version '%2' (handler '%3')", $vs_mirror_method, $v, $queue), "BaseModel->_processMedia()");
}
}
}
$media_desc[$v] = array("VOLUME" => $volume, "MIMETYPE" => $output_mimetype, "WIDTH" => $m->get("width"), "HEIGHT" => $m->get("height"), "PROPERTIES" => $m->getProperties(), "FILENAME" => $this->_genMediaName($ps_field) . "_" . $v . "." . $ext, "HASH" => $dirhash, "MAGIC" => $magic, "EXTENSION" => $ext, "MD5" => md5_file($filepath));
}
} else {
$m->set("version", $v);
while (list($operation, $parameters) = each($rules)) {
if ($operation === 'SET') {
foreach ($parameters as $pp => $pv) {
if ($pp == 'format') {
$output_mimetype = $pv;
} else {
$m->set($pp, $pv);
}
}
} else {
if (is_array($this->_FIELD_VALUES[$ps_field]) && is_array($va_media_center = $this->getMediaCenter($ps_field))) {
$parameters['_centerX'] = caGetOption('x', $va_media_center, 0.5);
$parameters['_centerY'] = caGetOption('y', $va_media_center, 0.5);
if ($parameters['_centerX'] < 0 || $parameters['_centerX'] > 1) {
$parameters['_centerX'] = 0.5;
}
if ($parameters['_centerY'] < 0 || $parameters['_centerY'] > 1) {
$parameters['_centerY'] = 0.5;
}
}
if (!$m->transform($operation, $parameters)) {
$error = 1;
$error_msg = "Couldn't do transformation '{$operation}'";
break 2;
}
}
}
if (!$output_mimetype) {
$output_mimetype = $input_mimetype;
}
if (!($ext = $m->mimetype2extension($output_mimetype))) {
$this->postError(1600, _t("File could not be processed for %1; can't convert mimetype '%2' to extension", $ps_field, $output_mimetype), "BaseModel->_processMedia()", $this->tableName() . '.' . $ps_field);
$m->cleanup();
set_time_limit($vn_max_execution_time);
if ($vb_is_fetched_file) {
@unlink($vs_tmp_file);
}
if ($vb_is_archive) {
@unlink($vs_archive);
@unlink($vs_primary_file_tmp);
@unlink($vs_archive_original);
}
return false;
}
if (($dirhash = $this->_getDirectoryHash($vi["absolutePath"], $this->getPrimaryKey())) === false) {
$this->postError(1600, _t("Could not create subdirectory for uploaded file in %1. Please ask your administrator to check the permissions of your media directory.", $vi["absolutePath"]), "BaseModel->_processMedia()", $this->tableName() . '.' . $ps_field);
set_time_limit($vn_max_execution_time);
if ($vb_is_fetched_file) {
@unlink($vs_tmp_file);
}
if ($vb_is_archive) {
@unlink($vs_archive);
@unlink($vs_primary_file_tmp);
@unlink($vs_archive_original);
}
return false;
}
$magic = rand(0, 99999);
$filepath = $vi["absolutePath"] . "/" . $dirhash . "/" . $magic . "_" . $this->_genMediaName($ps_field) . "_" . $v;
if (!($vs_output_file = $m->write($filepath, $output_mimetype, $va_media_write_options))) {
$this->postError(1600, _t("Couldn't write file: %1", join("; ", $m->getErrors())), "BaseModel->_processMedia()", $this->tableName() . '.' . $ps_field);
$m->cleanup();
set_time_limit($vn_max_execution_time);
if ($vb_is_fetched_file) {
@unlink($vs_tmp_file);
}
if ($vb_is_archive) {
@unlink($vs_archive);
@unlink($vs_primary_file_tmp);
@unlink($vs_archive_original);
}
return false;
break;
} else {
if ($vs_output_file === __CA_MEDIA_VIDEO_DEFAULT_ICON__ || $vs_output_file === __CA_MEDIA_AUDIO_DEFAULT_ICON__ || $vs_output_file === __CA_MEDIA_DOCUMENT_DEFAULT_ICON__ || $vs_output_file === __CA_MEDIA_3D_DEFAULT_ICON__) {
$vs_use_icon = $vs_output_file;
}
}
if ($v === $va_default_queue_settings['QUEUE_USING_VERSION']) {
$vs_path_to_queue_media = $vs_output_file;
$vs_use_icon = __CA_MEDIA_QUEUED_ICON__;
}
if ($pa_options['delete_old_media'] && !$error) {
if ($vs_old_media_path = $this->getMediaPath($ps_field, $v)) {
$va_files_to_delete[] = array('field' => $ps_field, 'version' => $v, 'dont_delete_path' => $filepath, 'dont_delete_extension' => $ext);
}
}
if (is_array($vi["mirrors"]) && sizeof($vi["mirrors"]) > 0) {
$vs_entity_key = join("/", array($this->tableName(), $ps_field, $this->getPrimaryKey(), $v));
$vs_row_key = join("/", array($this->tableName(), $this->getPrimaryKey()));
foreach ($vi["mirrors"] as $vs_mirror_code => $va_mirror_info) {
$vs_mirror_method = $va_mirror_info["method"];
$vs_queue = $vs_mirror_method . "mirror";
if (!$o_tq->cancelPendingTasksForEntity($vs_entity_key)) {
//$this->postError(560, _t("Could not cancel pending tasks: %1", $this->error),"BaseModel->_processMedia()");
//$m->cleanup();
//return false;
}
if ($o_tq->addTask($vs_queue, array("MIRROR" => $vs_mirror_code, "VOLUME" => $volume, "FIELD" => $ps_field, "TABLE" => $this->tableName(), "VERSION" => $v, "FILES" => array(array("FILE_PATH" => $filepath . "." . $ext, "ABS_PATH" => $vi["absolutePath"], "HASH" => $dirhash, "FILENAME" => $magic . "_" . $this->_genMediaName($ps_field) . "_" . $v . "." . $ext)), "MIRROR_INFO" => $va_mirror_info, "PK" => $this->primaryKey(), "PK_VAL" => $this->getPrimaryKey()), array("priority" => 100, "entity_key" => $vs_entity_key, "row_key" => $vs_row_key, 'user_id' => $AUTH_CURRENT_USER_ID))) {
continue;
} else {
$this->postError(100, _t("Couldn't queue mirror using '%1' for version '%2' (handler '%3')", $vs_mirror_method, $v, $queue), "BaseModel->_processMedia()");
}
}
}
if ($vs_use_icon) {
$media_desc[$v] = array("MIMETYPE" => $output_mimetype, "USE_ICON" => $vs_use_icon, "WIDTH" => $m->get("width"), "HEIGHT" => $m->get("height"));
} else {
$media_desc[$v] = array("VOLUME" => $volume, "MIMETYPE" => $output_mimetype, "WIDTH" => $m->get("width"), "HEIGHT" => $m->get("height"), "PROPERTIES" => $m->getProperties(), "FILENAME" => $this->_genMediaName($ps_field) . "_" . $v . "." . $ext, "HASH" => $dirhash, "MAGIC" => $magic, "EXTENSION" => $ext, "MD5" => md5_file($vi["absolutePath"] . "/" . $dirhash . "/" . $magic . "_" . $this->_genMediaName($ps_field) . "_" . $v . "." . $ext));
}
$m->reset();
}
}
if (sizeof($va_queued_versions)) {
$vs_entity_key = md5(join("/", array_merge(array($this->tableName(), $ps_field, $this->getPrimaryKey()), array_keys($va_queued_versions))));
$vs_row_key = join("/", array($this->tableName(), $this->getPrimaryKey()));
if (!$o_tq->cancelPendingTasksForEntity($vs_entity_key, $queue)) {
// TODO: log this
}
if (!($filename = $vs_path_to_queue_media)) {
// if we're not using a designated not-queued representation to generate the queued ones
// then copy the uploaded file to the tmp dir and use that
$filename = $o_tq->copyFileToQueueTmp($va_default_queue_settings['QUEUE'], $this->_SET_FILES[$ps_field]['tmp_name']);
}
if ($filename) {
if ($o_tq->addTask($va_default_queue_settings['QUEUE'], array("TABLE" => $this->tableName(), "FIELD" => $ps_field, "PK" => $this->primaryKey(), "PK_VAL" => $this->getPrimaryKey(), "INPUT_MIMETYPE" => $input_mimetype, "FILENAME" => $filename, "VERSIONS" => $va_queued_versions, "OPTIONS" => $va_media_write_options, "DONT_DELETE_OLD_MEDIA" => $filename == $vs_path_to_queue_media ? true : false), array("priority" => 100, "entity_key" => $vs_entity_key, "row_key" => $vs_row_key, 'user_id' => $AUTH_CURRENT_USER_ID))) {
if ($pa_options['delete_old_media']) {
foreach ($va_queued_versions as $vs_version => $va_version_info) {
$va_files_to_delete[] = array('field' => $ps_field, 'version' => $vs_version);
}
}
} else {
$this->postError(100, _t("Couldn't queue processing for version '%1' using handler '%2'", !$v, $queue), "BaseModel->_processMedia()");
}
} else {
$this->errors = $o_tq->errors;
}
} else {
// Generate preview frames for media that support that (Eg. video)
// and add them as "multifiles" assuming the current model supports that (ca_object_representations does)
if (!sizeof($va_process_these_versions_only) && ((bool) $this->_CONFIG->get('video_preview_generate_frames') || (bool) $this->_CONFIG->get('document_preview_generate_pages')) && method_exists($this, 'addFile')) {
if (method_exists($this, 'removeAllFiles')) {
$this->removeAllFiles();
// get rid of any previously existing frames (they might be hanging ar
}
$va_preview_frame_list = $m->writePreviews(array('width' => $m->get("width"), 'height' => $m->get("height"), 'minNumberOfFrames' => $this->_CONFIG->get('video_preview_min_number_of_frames'), 'maxNumberOfFrames' => $this->_CONFIG->get('video_preview_max_number_of_frames'), 'numberOfPages' => $this->_CONFIG->get('document_preview_max_number_of_pages'), 'frameInterval' => $this->_CONFIG->get('video_preview_interval_between_frames'), 'pageInterval' => $this->_CONFIG->get('document_preview_interval_between_pages'), 'startAtTime' => $this->_CONFIG->get('video_preview_start_at'), 'endAtTime' => $this->_CONFIG->get('video_preview_end_at'), 'startAtPage' => $this->_CONFIG->get('document_preview_start_page'), 'outputDirectory' => __CA_APP_DIR__ . '/tmp'));
if (is_array($va_preview_frame_list)) {
foreach ($va_preview_frame_list as $vn_time => $vs_frame) {
$this->addFile($vs_frame, $vn_time, true);
// the resource path for each frame is it's time, in seconds (may be fractional) for video, or page number for documents
@unlink($vs_frame);
// clean up tmp preview frame file
}
}
}
}
if (!$error) {
#
# --- Clean up old media from versions that are not supported in the new media
#
if ($pa_options['delete_old_media']) {
foreach ($this->getMediaVersions($ps_field) as $old_version) {
if (!is_array($media_desc[$old_version])) {
$this->_removeMedia($ps_field, $old_version);
}
}
}
foreach ($va_files_to_delete as $va_file_to_delete) {
$this->_removeMedia($va_file_to_delete['field'], $va_file_to_delete['version'], $va_file_to_delete['dont_delete_path'], $va_file_to_delete['dont_delete_extension']);
}
# Remove old _undo_ file if defined
if ($vs_undo_path = $this->getMediaPath($ps_field, '_undo_')) {
@unlink($vs_undo_path);
}
$this->_FILES[$ps_field] = $media_desc;
$this->_FIELD_VALUES[$ps_field] = $media_desc;
$vs_serialized_data = caSerializeForDatabase($this->_FILES[$ps_field], true);
$vs_sql = "{$ps_field} = " . $this->quote($vs_serialized_data) . ",";
if (($vs_metadata_field_name = $o_media_proc_settings->getMetadataFieldName()) && $this->hasField($vs_metadata_field_name)) {
$this->set($vs_metadata_field_name, $media_metadata);
$vs_sql .= " " . $vs_metadata_field_name . " = " . $this->quote(caSerializeForDatabase($media_metadata, true)) . ",";
}
if (($vs_content_field_name = $o_media_proc_settings->getMetadataContentName()) && $this->hasField($vs_content_field_name)) {
$this->_FIELD_VALUES[$vs_content_field_name] = $this->quote($m->getExtractedText());
$vs_sql .= " " . $vs_content_field_name . " = " . $this->_FIELD_VALUES[$vs_content_field_name] . ",";
}
if (is_array($va_locs = $m->getExtractedTextLocations())) {
MediaContentLocationIndexer::clear($this->tableNum(), $this->getPrimaryKey());
foreach ($va_locs as $vs_content => $va_loc_list) {
foreach ($va_loc_list as $va_loc) {
MediaContentLocationIndexer::index($this->tableNum(), $this->getPrimaryKey(), $vs_content, $va_loc['p'], $va_loc['x1'], $va_loc['y1'], $va_loc['x2'], $va_loc['y2']);
}
}
MediaContentLocationIndexer::write();
}
} else {
# error - invalid media
$this->postError(1600, _t("File could not be processed for %1: %2", $ps_field, $error_msg), "BaseModel->_processMedia()");
# return false;
}
$m->cleanup();
if ($vb_renamed_tmpfile) {
@unlink($this->_SET_FILES[$ps_field]['tmp_name']);
}
} elseif (is_array($this->_FIELD_VALUES[$ps_field])) {
// Just set field values in SQL (assume in-place update of media metadata) because no tmp_name is set
// [This generally should not happen]
$this->_FILES[$ps_field] = $this->_FIELD_VALUES[$ps_field];
$vs_sql = "{$ps_field} = " . $this->quote(caSerializeForDatabase($this->_FILES[$ps_field], true)) . ",";
}
$this->_SET_FILES[$ps_field] = null;
} elseif (is_array($this->_FIELD_VALUES[$ps_field])) {
// Just set field values in SQL (usually in-place update of media metadata)
$this->_FILES[$ps_field] = $this->_FIELD_VALUES[$ps_field];
$vs_sql = "{$ps_field} = " . $this->quote(caSerializeForDatabase($this->_FILES[$ps_field], true)) . ",";
}
}
set_time_limit($vn_max_execution_time);
if ($vb_is_fetched_file) {
@unlink($vs_tmp_file);
}
if ($vb_is_archive) {
@unlink($vs_archive);
@unlink($vs_primary_file_tmp);
@unlink($vs_archive_original);
}
return $vs_sql;
}
/**
*
*/
public function SearchWithinMedia()
{
$pn_representation_id = $this->request->getParameter('representation_id', pInteger);
$ps_q = $this->request->getParameter('q', pString);
$va_results = MediaContentLocationIndexer::SearchWithinMedia($ps_q, 'ca_object_representations', $pn_representation_id, 'media');
$this->view->setVar('results', $va_results);
$this->render('Details/object_representation_within_media_search_results_json.php');
}
/**
* Reindex PDF media by content for in-PDF search
*/
public static function reindex_pdfs($po_opts = null)
{
require_once __CA_LIB_DIR__ . "/core/Db.php";
require_once __CA_MODELS_DIR__ . "/ca_object_representations.php";
if (!caPDFMinerInstalled()) {
CLIUtils::addError(_t("Can't reindex PDFs: PDFMiner is not installed."));
return false;
}
$o_db = new Db();
$t_rep = new ca_object_representations();
$t_rep->setMode(ACCESS_WRITE);
$va_versions = array("original");
$va_kinds = ($vs_kinds = $po_opts->getOption("kinds")) ? explode(",", $vs_kinds) : array();
if (!is_array($va_kinds) || !sizeof($va_kinds)) {
$va_kinds = array('all');
}
$va_kinds = array_map('strtolower', $va_kinds);
if (in_array('all', $va_kinds) || in_array('ca_object_representations', $va_kinds)) {
if (!($vn_start = (int) $po_opts->getOption('start_id'))) {
$vn_start = null;
}
if (!($vn_end = (int) $po_opts->getOption('end_id'))) {
$vn_end = null;
}
if ($vn_id = (int) $po_opts->getOption('id')) {
$vn_start = $vn_id;
$vn_end = $vn_id;
}
$va_ids = array();
if ($vs_ids = (string) $po_opts->getOption('ids')) {
if (sizeof($va_tmp = explode(",", $vs_ids))) {
foreach ($va_tmp as $vn_id) {
if ((int) $vn_id > 0) {
$va_ids[] = (int) $vn_id;
}
}
}
}
$vs_sql_where = null;
$va_params = array();
if (sizeof($va_ids)) {
$vs_sql_where = "WHERE representation_id IN (?)";
$va_params[] = $va_ids;
} else {
if ($vn_start > 0 && $vn_end > 0 && $vn_start <= $vn_end || $vn_start > 0 && $vn_end == null) {
$vs_sql_where = "WHERE representation_id >= ?";
$va_params[] = $vn_start;
if ($vn_end) {
$vs_sql_where .= " AND representation_id <= ?";
$va_params[] = $vn_end;
}
}
}
if ($vs_sql_where) {
$vs_sql_where .= " AND mimetype = 'application/pdf'";
} else {
$vs_sql_where = " WHERE mimetype = 'application/pdf'";
}
$qr_reps = $o_db->query("\n\t\t\t\t\tSELECT * \n\t\t\t\t\tFROM ca_object_representations \n\t\t\t\t\t{$vs_sql_where}\n\t\t\t\t\tORDER BY representation_id\n\t\t\t\t", $va_params);
print CLIProgressBar::start($qr_reps->numRows(), _t('Reindexing PDF representations'));
$vn_rep_table_num = $t_rep->tableNum();
while ($qr_reps->nextRow()) {
$va_media_info = $qr_reps->getMediaInfo('media');
$vs_original_filename = $va_media_info['ORIGINAL_FILENAME'];
print CLIProgressBar::next(1, _t("Reindexing PDF %1", $vs_original_filename ? $vs_original_filename . " (" . $qr_reps->get('representation_id') . ")" : $qr_reps->get('representation_id')));
$t_rep->load($qr_reps->get('representation_id'));
$vn_rep_id = $t_rep->getPrimaryKey();
$m = new Media();
if ($m->read($vs_path = $t_rep->getMediaPath('media', 'original')) && is_array($va_locs = $m->getExtractedTextLocations())) {
MediaContentLocationIndexer::clear($vn_rep_table_num, $vn_rep_id);
foreach ($va_locs as $vs_content => $va_loc_list) {
foreach ($va_loc_list as $va_loc) {
MediaContentLocationIndexer::index($vn_rep_table_num, $vn_rep_id, $vs_content, $va_loc['p'], $va_loc['x1'], $va_loc['y1'], $va_loc['x2'], $va_loc['y2']);
}
}
MediaContentLocationIndexer::write();
} else {
//CLIUtils::addError(_t("[Warning] No content to reindex for PDF representation: %1", $vs_path));
}
}
print CLIProgressBar::finish();
}
if (in_array('all', $va_kinds) || in_array('ca_attributes', $va_kinds)) {
// get all Media elements
$va_elements = ca_metadata_elements::getElementsAsList(false, null, null, true, false, true, array(16));
// 16=media
$qr_c = $o_db->query("\n\t\t\t\t\tSELECT count(*) c \n\t\t\t\t\tFROM ca_attribute_values\n\t\t\t\t\tWHERE\n\t\t\t\t\t\telement_id in (?)\n\t\t\t\t", caExtractValuesFromArrayList($va_elements, 'element_id', array('preserveKeys' => false)));
if ($qr_c->nextRow()) {
$vn_count = $qr_c->get('c');
} else {
$vn_count = 0;
}
$t_attr_val = new ca_attribute_values();
$vn_attr_table_num = $t_attr_val->tableNum();
print CLIProgressBar::start($vn_count, _t('Reindexing metadata attribute media'));
foreach ($va_elements as $vs_element_code => $va_element_info) {
$qr_vals = $o_db->query("SELECT value_id FROM ca_attribute_values WHERE element_id = ?", (int) $va_element_info['element_id']);
$va_vals = $qr_vals->getAllFieldValues('value_id');
foreach ($va_vals as $vn_value_id) {
$t_attr_val = new ca_attribute_values($vn_value_id);
if ($t_attr_val->getPrimaryKey()) {
$t_attr_val->setMode(ACCESS_WRITE);
$t_attr_val->useBlobAsMediaField(true);
$va_media_info = $t_attr_val->getMediaInfo('value_blob');
$vs_original_filename = $va_media_info['ORIGINAL_FILENAME'];
if (!is_array($va_media_info) || $va_media_info['MIMETYPE'] !== 'application/pdf') {
continue;
}
print CLIProgressBar::next(1, _t("Reindexing %1", $vs_original_filename ? $vs_original_filename . " ({$vn_value_id})" : $vn_value_id));
$m = new Media();
if ($m->read($vs_path = $t_attr_val->getMediaPath('value_blob', 'original')) && is_array($va_locs = $m->getExtractedTextLocations())) {
MediaContentLocationIndexer::clear($vn_attr_table_num, $vn_attr_table_num);
foreach ($va_locs as $vs_content => $va_loc_list) {
foreach ($va_loc_list as $va_loc) {
MediaContentLocationIndexer::index($vn_attr_table_num, $vn_value_id, $vs_content, $va_loc['p'], $va_loc['x1'], $va_loc['y1'], $va_loc['x2'], $va_loc['y2']);
}
}
MediaContentLocationIndexer::write();
} else {
//CLIUtils::addError(_t("[Warning] No content to reindex for PDF in metadata attribute: %1", $vs_path));
}
}
}
}
print CLIProgressBar::finish();
}
return true;
}
/**
*
*/
static function SearchWithinMedia($ps_query, $pm_table, $pn_row_id, $ps_field)
{
$o_dm = Datamodel::load();
$o_config = Configuration::load();
$o_search_config = Configuration::load($o_config->get('search_config'));
$vs_indexing_regex = $o_search_config->get('indexing_tokenizer_regex');
$va_words = preg_split("![{$vs_indexing_regex}]!", $ps_query);
$va_results = array('matches' => 0, 'results' => array(), 'locations' => array(), 'query' => $ps_query);
if (!($t_instance = is_numeric($pm_table) ? $o_dm->getInstanceByTableNum($pm_table) : $o_dm->getInstanceByTableName($pm_table))) {
throw new Exception(_t("Invalid table %1", $pm_table));
}
if (!$t_instance->load($pn_row_id)) {
throw new Exception(_t("Invalid row %2 for table %1", $pm_table, $pn_row_id));
}
if (!$t_instance->hasField($ps_field)) {
throw new Exception(_t("Invalid field %2 for table %1", $pm_table, $ps_field));
}
$va_media_info = $t_instance->getMediaInfo($ps_field);
$vn_page_width = $va_media_info['INPUT']['WIDTH'];
$vn_page_height = $va_media_info['INPUT']['HEIGHT'];
$vn_page_image_width = $va_media_info['large']['WIDTH'];
$vn_page_image_height = $va_media_info['large']['HEIGHT'];
$va_hit_acc = array();
foreach ($va_words as $vn_word_num => $vs_word) {
$va_hits = MediaContentLocationIndexer::find($t_instance->tableName(), $pn_row_id, $vs_word);
$va_hit_acc_matched = array();
if (is_array($va_hits) && sizeof($va_hits)) {
$va_pages = array();
foreach ($va_hits as $va_hit) {
$x1_percent = $va_hit['x1'] / $vn_page_width;
$x2_percent = $va_hit['x2'] / $vn_page_width;
$y1_percent = ($vn_page_height - $va_hit['y2']) / $vn_page_height;
$y2_percent = ($vn_page_height - $va_hit['y1']) / $vn_page_height;
// Is this hit part of a phrase match?
if ($vn_word_num > 0) {
if (!is_array($va_hit_acc[$va_hit['p']]) || !sizeof($va_hit_acc[$va_hit['p']])) {
continue;
}
// if page is empty after the first word is processed then we can skip checking the rest of the words
$vn_i = -1;
foreach ($va_hit_acc[$va_hit['p']] as $vn_i => $va_existing_loc) {
if ($va_existing_loc['c'] >= $vn_word_num + 1) {
continue;
}
// Check if word is to the right of and on the same line as the previous one
if (abs($x1_percent - $va_existing_loc['x2']) > 0.08 || abs($y1_percent - $va_existing_loc['y1']) > 0.005) {
// Check if word to to the bottom and right of the previous one
if (abs($y2_percent - ($va_existing_loc['y2'] + ($va_existing_loc['y2'] - $va_existing_loc['y1']))) < 0.003 && $va_existing_loc['x1'] > $x1_percent) {
if (isset($va_hit_acc[$va_hit['p']][$vn_i]['isNewlined']) && $va_hit_acc[$va_hit['p']][$vn_i]['isNewlined']) {
continue;
}
$va_hit_acc[$va_hit['p']][$vn_i]['c'] = $vn_word_num + 1;
$va_hit_acc[$va_hit['p']][$vn_i]['isNewlined'] = true;
// word is part of phrase but on new line so create new hit for it
array_splice($va_hit_acc[$va_hit['p']], $vn_i, 0, array(array('word' => "newline {$vs_word}", 'x1' => $x1_percent, 'y1' => $y1_percent, 'x2' => $x2_percent, 'y2' => $y2_percent, 'c' => $vn_word_num + 1, 'previousLine' => &$va_hit_acc[$va_hit['p']][$vn_i])));
continue;
} else {
// Word isn't near the previous one so skip the hit
continue;
}
}
// extend selection to encompass this word in the same line as previous
$va_hit_acc[$va_hit['p']][$vn_i]['x2'] = $x2_percent;
$va_hit_acc[$va_hit['p']][$vn_i]['y2'] = $y2_percent;
$va_hit_acc[$va_hit['p']][$vn_i]['c'] = $vn_word_num + 1;
$va_hit_acc[$va_hit['p']][$vn_i]['word'] .= "extend {$vs_word}";
if ($va_hit_acc[$va_hit['p']][$vn_i]['previousLine']) {
$va_hit_acc[$va_hit['p']][$vn_i]['previousLine']['c'] = $vn_word_num + 1;
}
break;
}
} else {
$va_hit_acc[$va_hit['p']][] = array('word' => "add {$vs_word}", 'x1' => $x1_percent, 'y1' => $y1_percent, 'x2' => $x2_percent, 'y2' => $y2_percent, 'c' => $vn_word_num + 1);
}
}
}
// Remove all previous hits that didn't line up with the current word
foreach ($va_hit_acc as $vn_p => $va_hits_per_page) {
foreach ($va_hits_per_page as $vn_i => $va_hit) {
if ($va_hit['c'] < $vn_word_num + 1) {
unset($va_hit_acc[$vn_p][$vn_i]);
}
}
}
}
// Copy hits into final locations list with coordinates translated into pixel values
foreach ($va_hit_acc as $vn_p => $va_hits_per_page) {
foreach ($va_hits_per_page as $vn_i => $va_hit) {
if ($va_hit['c'] < sizeof($va_words)) {
continue;
}
$va_results['results'][] = $vn_p;
$x1r = $va_hit['x1'] * $vn_page_image_width + 2;
$x2r = $va_hit['x2'] * $vn_page_image_width + 12;
$y1r = $va_hit['y1'] * $vn_page_image_height;
$y2r = $va_hit['y2'] * $vn_page_image_height;
$va_results['locations'][$vn_p][] = array('c' => $va_hit['c'], 'word' => $va_hit['word'], 'x1' => $x1r, 'y1' => $y1r, 'x2' => $x2r, 'y2' => $y2r);
}
}
$va_results['matches'] = sizeof($va_results['results']);
return $va_results;
}
