$line_of_text = fgets($file_handle);
$parts = explode('/n', $line_of_text);
foreach ($parts as $img_capts) {
list($img_filename, $img_caption) = explode('|', $img_capts);
$img_captions[$img_filename] = $img_caption;
}
}
fclose($file_handle);
}
// 3. LOAD FILES
if ($file != "." && $file != ".." && $file != "folder.jpg") {
// JPG, GIF and PNG
if (preg_match("/.jpg\$|.gif\$|.png\$/i", $file)) {
//Read EXIF
if ($display_exif == 1) {
$img_captions[$file] .= readEXIF($currentdir . "/" . $file);
}
checkpermissions($currentdir . "/" . $file);
list($img_width, $img_height, $img_type, $img_attr) = getimagesize($currentdir . "/" . $file);
$files[] = array("name" => $file, "date" => filemtime($currentdir . "/" . $file), "size" => filesize($currentdir . "/" . $file), "html" => "<li class='thumbnail'>" . "<a href='" . $currentdir . "/" . $file . "' rel='gallery' title='{$img_captions[$file]}'>" . "<img class='detalle' data-original-title='" . $file . "' " . "data-content='Dimensiones: {$img_width} por {$img_height} pixels' src='" . GALLERY_ROOT . "phpThumb.php?src=" . $thumbdir . "/" . $file . "&w={$thumb_size}&h={$thumb_size}&zc=1' alt='{$label_loading}' />" . "</a>" . "</li>");
}
// Other filetypes
$extension = "";
if (preg_match("/.pdf\$/i", $file)) {
$extension = "PDF";
}
// PDF
if (preg_match("/.zip\$/i", $file)) {
$extension = "ZIP";
}
// ZIP archive
}
}
}
// 3. LOAD FILES
if ($file != "." && $file != ".." && $file != "folder.jpg") {
if ($display_filename) {
$filename_caption = "<em>" . padstring($file, $label_max_length) . "</em>";
} else {
$filename_caption = "";
}
// JPG, GIF and PNG
if (preg_match("/.jpg\$|.gif\$|.png\$/i", $file)) {
//Read EXIF
if (!array_key_exists($file, $img_captions)) {
if ($display_exif == 1) {
$exifReaden = readEXIF($currentdir . "/" . $file);
//Add to the caption all the EXIF information
$img_captions[$file] = $file . $exifReaden;
} else {
//If no EXIF, just use the filename as caption
$img_captions[$file] = $file;
}
}
// Read the optionnal image title and caption in html file (image.jpg --> image.jpg.html)
// Format: title::caption
// Example: My cat::My cat like to <i>roll</i> on the floor.
// If file is not provided, image filename will be used instead.
checkpermissions($currentdir . "/" . $file);
if (is_file($currentdir . '/' . $file . '.html')) {
$img_captions[$file] = $file . '::' . htmlspecialchars(file_get_contents($currentdir . '/' . $file . '.html'), ENT_QUOTES);
}
/**
*
* @global type $rep_counter
* @global string $rep_issues
* @global array $fieldhelper_to_heurist_map
* @param type $dir
*/
function doHarvestInDir($dir)
{
global $rep_issues, $fieldhelper_to_heurist_map, $mediaExts, $progress_divid, $geoDT, $fileDT, $titleDT, $startdateDT, $enddateDT, $descriptionDT;
$rep_processed = 0;
$rep_added = 0;
$rep_updated = 0;
$rep_processed_dir = 0;
$rep_ignored = 0;
$f_items = null;
//reference to items element
$progress_divid++;
ob_start();
print "<div><b>{$dir}</b><span id='progress{$progress_divid}'></span></div>";
ob_flush();
flush();
if (!is_writable($dir)) {
print "<div style=\"color:red\">Folder is not writeable. Check permissions</div>";
return 0;
}
$manifest_file = $dir . "fieldhelper.xml";
//list of all files in given folder - need to treat new files that are not mentioned in manifest file
$all_files = scandir($dir);
if (file_exists($manifest_file)) {
//read fieldhelper.xml
if (is_readable($manifest_file)) {
//check write permission
if (!is_writable($manifest_file)) {
print "<div style=\"color:red\">Manifest is not writeable. Check permissions.</div>";
return 0;
}
} else {
print "<div style=\"color:red\">Manifest is not readable. Check permissions.</div>";
return 0;
}
$fh_data = simplexml_load_file($manifest_file);
if ($fh_data == null || is_string($fh_data)) {
print "<div style=\"color:red\">Manifest is corrupted</div>";
return 0;
}
//MAIN LOOP in manifest
$not_found = true;
//true if manifest is empty
foreach ($fh_data->children() as $f_gen) {
if ($f_gen->getName() == "items") {
$f_items = $f_gen;
$not_found = false;
$tot_files = count($f_gen->children());
$cnt_files = 0;
foreach ($f_gen->children() as $f_item) {
$recordId = null;
$recordType = RT_MEDIA_RECORD;
//media by default
$recordURL = null;
$recordNotes = null;
$el_heuristid = null;
$lat = null;
$lon = null;
$filename = null;
$filename_base = null;
//filename only
$details = array();
$file_id = null;
$old_md5 = null;
foreach ($f_item->children() as $el) {
//$key=>$value
$content = strval($el);
// (string)$el;
$key = $el->getName();
$value = $content;
if ($key == "md5") {
$old_md5 = $value;
} else {
if (@$fieldhelper_to_heurist_map[$key]) {
$key2 = $fieldhelper_to_heurist_map[$key];
if ($key2 == "file") {
$filename = $dir . $value;
$filename_base = $value;
$key3 = $fieldhelper_to_heurist_map['file_name'];
if ($key3 > 0) {
$details["t:" . $key3] = array("1" => $value);
}
$key3 = $fieldhelper_to_heurist_map['file_path'];
if ($key3 > 0) {
$relative_path = getRelativePath(HEURIST_FILESTORE_DIR, $dir);
$details["t:" . $key3] = array("1" => $relative_path);
//change to relative path
}
} else {
if ($key2 == "lat") {
$lat = floatval($value);
} else {
if ($key2 == "lon") {
$lon = floatval($value);
} else {
if ($key2 == "recordId") {
$recordId = $value;
$el_heuristid = $el;
} else {
if (intval($key2) > 0) {
//add to details
$details["t:" . $key2] = array("1" => $value);
}
}
}
}
}
// else field type not defined in this instance
}
}
}
//for item keys
if ($filename) {
//exclude from the list of all files in this folder
if (in_array($filename_base, $all_files)) {
$ind = array_search($filename_base, $all_files, true);
unset($all_files[$ind]);
}
}
if ($recordId != null) {
//veify that this record exists
$res = mysql__select_array("Records", "rec_ID", "rec_ID=" . $recordId);
if (!(is_array($res) && count($res) > 0)) {
print "<div>File: <i>{$filename_base}</i> was indexed as rec# {$recordId}. " . "This record was not found. File will be reindexed</div>";
$recordId = null;
}
}
if ($recordId == null) {
//import only new
if ($filename) {
if (file_exists($filename)) {
$currfile = $filename;
//assign to global
//add-update the uploaded file
$file_id = register_file($filename, null, false);
if (is_numeric($file_id)) {
$details["t:" . $fileDT] = array("1" => $file_id);
//read EXIF data for JPEG images
$recordNotes = readEXIF($filename);
} else {
print "<div>File: <i>{$filename_base}</i> <span style=\"color:red\">" . "Error: Failed to register. No record created</span></div>";
$file_id = null;
}
} else {
print "<div>File: <i>{$filename_base}</i> <span style=\"color:red\">" . "File is referenced in fieldhelper.xml but was not found on disk." . "No record was created.</span></div>";
}
}
if (!$file_id) {
continue;
//add with valid file only
}
if (is_numeric($lat) && is_numeric($lon) && ($lat != 0 || $lon != 0)) {
$details["t:" . $geoDT] = array("1" => "p POINT ({$lon} {$lat})");
}
//set title by default
if (!array_key_exists("t:" . $titleDT, $details)) {
$details["t:" . $titleDT] = array("1" => $filename);
print "<div>File: <i>{$filename_base}</i> <span style=\"color:#ff8844\">" . "Warning: there was no title recorded in the XML manifest for this file." . "Using file path + file name as title.</span></div>";
}
$new_md5 = null;
$key = $fieldhelper_to_heurist_map['md5'];
if ($key > 0) {
$new_md5 = md5_file($filename);
$details["t:" . $key] = array("1" => $new_md5);
}
$out['error'] = 'test';
//add-update Heurist record
$out = saveRecord($recordId, $recordType, $recordURL, $recordNotes, null, null, null, null, null, null, null, $details, null, null, null, null, null);
if (@$out['error']) {
print "<div>File: <i>{$filename_base}</i> <span style='color:red'>Error: " . implode("; ", $out["error"]) . "</span></div>";
} else {
if ($new_md5 == null) {
$new_md5 = md5_file($filename);
}
//update xml
if ($recordId == null) {
if ($old_md5 != $new_md5) {
print "<div>File: <i>{$filename_base}</i> <span style=\"color:#ff8844\">" . "Warning: Checksum differs from value in manifest; " . "data file may have been changed</span></div>";
}
$f_item->addChild("heurist_id", $out["bibID"]);
$f_item->addChild("md5", $new_md5);
$f_item->addChild("filesize", filesize($filename));
$rep_added++;
} else {
$el_heuristid["heurist_id"] = $out["bibID"];
$rep_updated++;
}
if (@$out['warning']) {
print "<div>File: <i>{$filename_base}</i> <span style=\"color:#ff8844\">Warning: " . implode("; ", $out["warning"]) . "</span></div>";
}
}
$rep_processed++;
} else {
$rep_ignored++;
}
$cnt_files++;
if ($cnt_files % 5 == 0) {
ob_start();
print '<script type="text/javascript">update_counts(' . $progress_divid . ',' . $cnt_files . ',' . $rep_processed . ',' . $tot_files . ')</script>' . "\n";
ob_flush();
flush();
}
}
//for items
}
//if has items
}
//for all children in manifest
if ($not_found) {
print "<div style=\"color:red\">Manifest is either corrupted or empty</div>";
} else {
if ($rep_processed > 0) {
print "<div>{$rep_processed} files processed. {$rep_added} added. {$rep_updated} updated.</div>";
}
if ($rep_ignored > 0) {
print "<div>{$rep_ignored} files already indexed.</div>";
}
}
} else {
//create empty manifest XML - TODO!!!!
$s_manifest = <<<XML
<?xml version="1.0" encoding="UTF-8"?>
<fieldhelper version="1">
<info>
<AppName>Heurist</AppName>
<AppVersion>v 3.0.0 2012-01-01</AppVersion>
<AppCopyright>© ArtEresearch, University of Sydney</AppCopyright>
<date></date>
</info>
<formatOutput>1</formatOutput>
</fieldhelper>
XML;
$fh_data = simplexml_load_string($s_manifest);
}
// add new empty items element
if ($f_items == null) {
$f_items = $fh_data->addChild("items");
}
$tot_files = count($all_files);
$cnt_files = 0;
$cnt_added = 0;
//for files in folder that are not specified in the manifest file
foreach ($all_files as $filename) {
if (!($filename == "." || $filename == ".." || is_dir($dir . $filename) || $filename == "fieldhelper.xml")) {
$filename_base = $filename;
$filename = $dir . $filename;
$currfile = $filename;
$flleinfo = pathinfo($filename);
$recordNotes = null;
//checks for allowed extensions
if (in_array(strtolower(@$flleinfo['extension']), $mediaExts)) {
$details = array();
$file_id = register_file($filename, null, false);
if (is_numeric($file_id)) {
$details["t:" . $fileDT] = array("1" => $file_id);
//read EXIF data for JPEG images
$recordNotes = readEXIF($filename);
} else {
print "<div>File: <i>{$filename_base}</i> <span style=\"color:#ff8844\">" . "Warning: failed to register. No record created for: .{$file_id}</span></div>";
//$rep_issues = $rep_issues."<br/>Can't register file:".$filename.". ".$file_id;
$file_id = null;
continue;
}
$details["t:" . $titleDT] = array("1" => $flleinfo['basename']);
/* TODO - extract these data from exif
$details["t:".$descriptionDT] = array("1"=>$file_id);
$details["t:".$startdateDT] = array("1"=>$file_id);
$details["t:".$enddateDT] = array("1"=>$file_id);
$details["t:".$geoDT] = array("1"=>$file_id);
*/
$new_md5 = md5_file($filename);
$key = $fieldhelper_to_heurist_map['md5'];
if ($key > 0) {
$details["t:" . $key] = array("1" => $new_md5);
}
$key = $fieldhelper_to_heurist_map['file_name'];
if ($key > 0) {
$details["t:" . $key] = array("1" => $flleinfo['basename']);
}
$key = $fieldhelper_to_heurist_map['file_path'];
if ($key > 0) {
$targetPath = $flleinfo['dirname'];
$rel_path = getRelativePath(HEURIST_FILESTORE_DIR, $targetPath);
//getRelativePath2($targetPath);
$details["t:" . $key] = array("1" => $rel_path);
/*print "<div>".HEURIST_FILESTORE_DIR."</div>";
print "<div>file path :".$targetPath."</div>";
print "<div>relative path :".strpos($targetPath, HEURIST_FILESTORE_DIR)."--".$rel_path."</div>";
print "<div>relative path old :".getRelativePath(HEURIST_FILESTORE_DIR, $targetPath)."<br><br></div>";*/
}
$key = $fieldhelper_to_heurist_map['extension'];
if ($key > 0) {
$details["t:" . $key] = array("1" => $flleinfo['extension']);
}
$key = $fieldhelper_to_heurist_map['filesize'];
if ($key > 0) {
$details["t:" . $key] = array("1" => filesize($filename));
}
//add-update Heurist record
$out['error'] = 'test2';
$out = saveRecord(null, RT_MEDIA_RECORD, null, $recordNotes, null, null, null, null, null, null, null, $details, null, null, null, null, null);
$f_item = $f_items->addChild("item");
$f_item->addChild("filename", htmlspecialchars($flleinfo['basename']));
$f_item->addChild("nativePath", htmlspecialchars($filename));
$f_item->addChild("folder", htmlspecialchars($flleinfo['dirname']));
$f_item->addChild("extension", $flleinfo['extension']);
//$f_item->addChild("DateTime", );
//$f_item->addChild("DateTimeOriginal", );
$f_item->addChild("filedate", date("Y/m/d H:i:s.", filemtime($filename)));
$f_item->addChild("typeContent", "image");
$f_item->addChild("device", "image");
$f_item->addChild("duration", "2000");
$f_item->addChild("original_metadata", "chk");
$f_item->addChild("Name0", htmlspecialchars($flleinfo['basename']));
$f_item->addChild("md5", $new_md5);
$f_item->addChild("filesize", filesize($filename));
if (@$out['error']) {
print "<div>Fle: <i>{$filename_base}</i> <span style='color:red'>Error: " . implode("; ", $out["error"]) . "</span></div>";
} else {
$f_item->addChild("heurist_id", $out["bibID"]);
$cnt_added++;
}
$rep_processed_dir++;
}
//check ext
$cnt_files++;
if ($cnt_files % 5 == 0) {
ob_start();
print '<script type="text/javascript">update_counts(' . $progress_divid . ',' . $cnt_files . ',' . $cnt_added . ',' . $tot_files . ')</script>' . "\n";
ob_flush();
flush();
}
}
}
//for files in folder that are not specified in the directory
ob_start();
if ($rep_processed_dir > 0) {
print "<div style=\"color:green\">{$rep_processed_dir} processed. {$cnt_added} records created (new entries added to manifests)</div>";
}
print '<script type="text/javascript">update_counts(' . $progress_divid . ',' . $cnt_files . ',' . $cnt_added . ',0)</script>' . "\n";
ob_flush();
flush();
if ($rep_processed + $rep_processed_dir > 0) {
//save modified xml (with updated heurist_id tags)
$fh_data->formatOutput = true;
$fh_data->saveXML($manifest_file);
}
return $rep_processed + $rep_processed_dir;
}
