/**
* Indexing a single file.
* Use indexUsingCurrentSetup() or indexFiles() instead.
*
* @param string $filepath: file path or array: array('processFile' => 'path to file that should be indexed', 'metaFile' => 'additional file that holds meta data for the file to be indexed')
* @param integer $crdate: timestamp of the index run
* @param integer $pid: The sysfolder to store the meta data record
* @param mixed $metaCallbackFunc Will be called to process the meta data
* @param mixed $filePreprocessingCallbackFunc Will be called to allow preprocessing of the file before indexing
* @param array $metaPreset: Meta data preset. $meta['fields'] has the record data.
* @return array Meta data array. $meta['fields'] has the record data.
*/
function indexFile($filepath, $crdate = 0, $pid = NULL, $metaCallbackFunc = NULL, $filePreprocessingCallbackFunc = NULL, $metaPreset = array())
{
global $TYPO3_CONF_VARS;
$pathname = $this->getFilePath($filepath);
$pathname = tx_dam::file_absolutePath($pathname);
// locks the indexing for the current file
// If the file is currently indexed this will return false
if (!$this->lock($pathname)) {
return FALSE;
}
$pid = is_null($pid) ? $this->pid : $pid;
if ($filePreprocessingCallbackFunc) {
call_user_func($filePreprocessingCallbackFunc, 'filePreprocessing', $pathname, $this);
if ($this->writeDevLog) {
t3lib_div::devLog('indexFile(): call filePreprocessingCallbackFunc', 'tx_dam_indexing', 0, $filePreprocessingCallbackFunc);
}
}
// might be possible to have $pathname call by reference and change the filename - usable for copying files before indexing??? Needs to be tested.
// Answer: Note that the parameters for call_user_func() are not passed by reference.
$meta = $this->getFileNodeInfo($pathname, true);
if ($metaFile = $this->getMetaFilePath($filepath)) {
$meta['metaFile'] = $metaFile;
}
if ($this->skipThisFile($meta['file'])) {
unset($meta);
$this->log('Skipped file: ' . $pathname, 1, 0);
if ($this->writeDevLog) {
t3lib_div::devLog('indexFile(): file skipped: ' . $pathname, 'tx_dam_indexing');
}
} elseif (is_array($meta)) {
if ($this->writeDevLog) {
t3lib_div::devLog('indexFile() - got file node info: ' . $pathname, 'tx_dam_indexing', 0, $meta);
}
if ($this->writeDevLog and $metaPreset) {
t3lib_div::devLog('indexFile: use meta preset', 'tx_dam_indexing', 0, $metaPreset);
}
$meta = t3lib_div::array_merge_recursive_overrule($metaPreset, $meta);
$status = tx_dam::index_check($meta['fields'], $meta['fields']['file_hash']);
$uid = intval($status['meta']['uid']);
if ($uid) {
if ($this->writeDevLog) {
t3lib_div::devLog('indexFile(): file already indexed (uid:' . $uid . ')', 'tx_dam_indexing', 0, $status);
}
$res = $GLOBALS['TYPO3_DB']->exec_SELECTquery('*', 'tx_dam', 'uid=' . intval($uid));
$row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($res);
if ($this->writeDevLog) {
t3lib_div::devLog('indexFile(): fetch index data for reindexing', 'tx_dam_indexing', 0, $row);
}
// this is needed for fields like group/MM
require_once PATH_t3lib . 'class.t3lib_transferdata.php';
$processData = t3lib_div::makeInstance('t3lib_transferData');
$row = $processData->renderRecordRaw('tx_dam', $row['uid'], $row['pid'], $row);
if ($this->writeDevLog) {
t3lib_div::devLog('indexFile(): call t3lib_transferdata->renderRecordRaw() for data from index', 'tx_dam_indexing', 0, $row);
}
// index rule use 'row' for merging
$meta['row'] = $row;
$meta['indexExist'] = true;
$meta['reindexed'] = $this->doReindexing;
} else {
$uid = 'NEW';
$meta['indexExist'] = false;
$meta['reindexed'] = false;
}
// TODO handle TXDAM_file_missing and reconnect file to index
if ($status['__status'] == TXDAM_file_unknown or $status['__status'] > TXDAM_file_unknown and $this->doReindexing) {
$mimeType = array();
$mimeType['fields'] = $this->getFileMimeType($pathname);
$meta = t3lib_div::array_merge_recursive_overrule(array('fields' => $this->getDefaultRecord()), $meta);
$meta = t3lib_div::array_merge_recursive_overrule($meta, $mimeType);
$meta['fields']['uid'] = $uid;
$meta['fields']['pid'] = $pid;
$meta['fields']['index_type'] = $this->indexRunType;
$meta = $this->getFileMetaInfo($pathname, $meta);
if ($meta['textExtract']) {
$meta['textExtract'] = $this->processTextExcerpt($meta['textExtract']);
} else {
$meta['textExtract'] = $this->getFileTextExcerpt($pathname, $meta['fields']['file_type']);
}
$meta['fields']['search_content'] = $meta['textExtract'];
$meta['fields']['abstract'] = $meta['fields']['abstract'] ? $meta['fields']['abstract'] : trim($meta['fields']['search_content']);
$meta['fields']['language'] = $this->getMetaLanguage($meta);
$meta['fields']['file_dl_name'] = $meta['fields']['file_dl_name'] ? $meta['fields']['file_dl_name'] : $meta['fields']['file_name'];
$meta['fields']['crdate'] = $crdate ? $crdate : time();
$meta['fields']['tstamp'] = time();
$meta['fields']['cruser_id'] = intval($GLOBALS['BE_USER']->user['uid']);
$meta['fields']['date_cr'] = $meta['fields']['date_cr'] ? $meta['fields']['date_cr'] : time();
$meta['fields']['date_mod'] = $meta['fields']['date_mod'] ? $meta['fields']['date_mod'] : $meta['fields']['date_cr'];
// TODO category handling - merging?
# $fieldsUpdated = tx_dam_db::getUpdateData($meta['fields'], $this->replaceData, $this->appendData);
foreach ($this->dataPreset as $field => $value) {
if ($value and !$meta['fields'][$field]) {
$meta['fields'][$field] = $value;
}
}
$fieldsUpdated = tx_dam_db::getUpdateData($meta['fields'], $this->dataPostset, $this->dataAppend);
$meta['fields'] = array_merge($meta['fields'], $fieldsUpdated);
$meta = $this->rulesCallback('process', $meta, $pathname);
if ($metaCallbackFunc) {
$meta = call_user_func($metaCallbackFunc, 'process', $meta, $pathname, $this);
if ($this->writeDevLog) {
t3lib_div::devLog('indexFile(): call_user_func: ' . @get_class($metaCallbackFunc[0]) . '->' . $metaCallbackFunc[1] . ' (process)', 'tx_dam_indexing', 0, $meta);
}
}
$meta['failure'] = false;
if (!$this->dryRun) {
if ($this->writeDevLog) {
t3lib_div::devLog('indexFile(): call tx_dam_db::insertUpdateData()', 'tx_dam_indexing', 0, $meta['fields']);
}
$meta['fields']['uid'] = tx_dam_db::insertUpdateData($meta['fields']);
if (!intval($meta['fields']['uid'])) {
list($error, $errorMsg) = tx_dam_db::getLastError();
$meta['failure'] = 'Meta record could not be inserted: ' . $errorMsg;
$this->log('Meta record could not be inserted: ' . $pathname . '. ' . $errorMsg, 1, 1);
} else {
if ($this->writeDevLog) {
$res = $GLOBALS['TYPO3_DB']->exec_SELECTquery('*', 'tx_dam', 'uid=' . intval($meta['fields']['uid']));
}
if ($this->writeDevLog) {
$row = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($res);
}
if ($this->writeDevLog) {
t3lib_div::devLog('indexFile(): written to index', 'tx_dam_indexing', 0, $row);
}
}
}
if (intval($meta['fields']['uid'])) {
$meta['isIndexed'] = true;
$this->infoList[] = array('uid' => $meta['fields']['uid'], 'title' => $meta['fields']['title'], 'file_name' => $meta['fields']['file_name'], 'file_path' => $meta['fields']['file_path'], 'reindexed' => $meta['reindexed']);
$this->stat['newIndexed'] += $meta['reindexed'] ? 0 : 1;
$this->stat['reIndexed'] += $meta['reindexed'] ? 1 : 0;
}
$meta = $this->rulesCallback('post', $meta, $pathname);
if ($metaCallbackFunc) {
$meta = call_user_func($metaCallbackFunc, 'post', $meta, $pathname, $this);
if ($this->writeDevLog) {
t3lib_div::devLog('indexFile(): call_user_func: ' . @get_class($metaCallbackFunc[0]) . '->' . $metaCallbackFunc[1] . ' (post)', 'tx_dam_indexing', 0, $meta);
}
}
$currentUid = intval($meta['fields']['uid']) ? $meta['fields']['uid'] : '_NO_UID_' . (string) intval($this->noIdCounter++);
if ($this->collectMeta) {
$this->meta[$currentUid] = $meta;
}
// todo: indexing of childs to this file - eg. images from a OpenOffice file
if (is_array($meta['childs'])) {
foreach ($meta['childs'] as $fileDef) {
$pathname = $fileDef['pathname'];
if (file_exists($pathname)) {
if ($meta['fields']['file_hash'] and $fileDef['fileStorageType'] === 'moveToInternal') {
$storageFolder = PATH_site . 'uploads/tx_dam/storage/' . $meta['fields']['file_hash'] . '/';
$targetFile = $storageFolder . tx_dam::file_basename($pathname);
if (!is_dir($storageFolder)) {
t3lib_div::mkdir($storageFolder);
}
@unlink($targetFile);
rename($pathname, $targetFile);
$pathname = $targetFile;
}
$metaPreset = is_array($fileDef['metaPreset']) ? $fileDef['metaPreset'] : array();
$metaPreset['fields']['parent_id'] = $currentUid;
$this->indexFile($pathname, $crdate, $pid, $metaCallbackFunc, $filePreprocessingCallbackFunc, $metaPreset);
}
}
}
$this->statMeta($meta);
$this->unlock();
return $meta;
} elseif (is_array($meta['row'])) {
$meta['fields'] = $meta['row'];
$this->statMeta($meta);
$this->unlock();
return $meta;
}
} else {
if (!@is_file($pathname)) {
$this->log('Is not a file: ' . $pathname, 1, 1);
} elseif (!@is_readable($pathname)) {
$this->log('Is not readable: ' . $pathname, 1, 1);
}
}
$this->unlock();
return FALSE;
}
