web/scripts/upsert-to-cover-art-database

#!/usr/bin/php
<?
require_once('/standardebooks.org/web/lib/Core.php');

$longopts = ['repoDir:', 'workDir:', 'ebookWwwFilesystemPath:', 'verbose'];
$options = getopt('v', $longopts);

$repoDir = $options['repoDir'] ?? false;
$workDir = $options['workDir'] ?? false;
$ebookWwwFilesystemPath = $options['ebookWwwFilesystemPath'] ?? false;

$verbose = false;
if(isset($options['v']) || isset($options['verbose'])){
	$verbose = true;
}

/**
 * Coverts SimpleXMLElement objects with inner tags like this:
 *   '<abbr>Mr.</abbr> Smith'
 * to:
 *   'Mr. Smith'
 */
function StripInnerTags($elements): ?string{
	if($elements === false){
		return null;
	}

	if(isset($elements[0])){
		return strip_tags($elements[0]->asXML());
	}

	return null;
}

if(!$repoDir || !$workDir || !$ebookWwwFilesystemPath){
	print("Expected usage: upsert-to-cover-art-database [-v] --repoDir <dir> --workDir <dir> --ebookWwwFilesystemPath <path>\n");
	exit(1);
}

if($verbose){
	print("\nrepoDir: $repoDir\n");
	print("workDir: $workDir\n");
	print("ebookWwwFilesystemPath: $ebookWwwFilesystemPath\n");
}

chdir($repoDir);
$contentOpf = shell_exec("git show HEAD:src/epub/content.opf");
$xml = new SimpleXMLElement(str_replace('xmlns=', 'ns=', $contentOpf));
$xml->registerXPathNamespace('dc', 'http://purl.org/dc/elements/1.1/');
$artistName = StripInnerTags($xml->xpath('/package/metadata/dc:contributor[@id="artist"]'));
if($artistName === null){
	// Some ebooks have an artist-1 and artist-2. Take artist-1, which isn't ideal, but is usually correct.
	$artistName = StripInnerTags($xml->xpath('/package/metadata/dc:contributor[@id="artist-1"]'));
	if($artistName === null){
		print($repoDir . " Error: Could not find artist name in content.opf\n");
		exit($repoDir . " Error: missing artistName\n");
	}
}

if(!file_exists($ebookWwwFilesystemPath . '/text/colophon.xhtml')){
	exit($repoDir . ' Error: no text/colophon.xhtml at ' . $ebookWwwFilesystemPath . "\n");
}

$rawColophon = file_get_contents($ebookWwwFilesystemPath . '/text/colophon.xhtml');
if(empty($rawColophon)){
	exit($repoDir . ' Error: empty colophon at ' . $ebookWwwFilesystemPath . "\n");
}

preg_match('|a painting completed \w+ (\d+)|ius', $rawColophon, $matches);
$completedYear = null;
if(sizeof($matches) == 2){
	$completedYear = (int)$matches[1];
}

$colophonXml = new SimpleXMLElement(str_replace('xmlns=', 'ns=', $rawColophon));
$artworkName = StripInnerTags($colophonXml->xpath('/html/body/main/section/p/i[@epub:type="se:name.visual-art.painting"]'));
if($artworkName === null){
	print($repoDir . " Error: Could not find artwork name in colophon.xhtml\n");
	exit($repoDir . " Error: missing artworkName");
}

$artistUrlName = Formatter::MakeUrlSafe($artistName);
$artworkUrlName = Formatter::MakeUrlSafe($artworkName);
$artwork = null;

if($verbose){
	print("artistName: $artistName\n");
	print("artistUrlName: $artistUrlName\n");
	print("completedYear: $completedYear\n");
	print("artworkName: $artworkName\n");
	print("artworkUrlName: $artworkUrlName\n");
}

try{
	$artwork = Artwork::GetByUrlAndIsApproved($artistUrlName, $artworkUrlName);
}
catch(Exceptions\ArtworkNotFoundException){
	// $artwork is null by default, just continue
}

if($artwork === null){
	if($verbose){
		printf($repoDir . " No existing artwork found at %s/%s, inserting new artwork.\n", $artistUrlName, $artworkUrlName);
	}

	// The ebook colophon provides the artist's name, but not their death year.
	// Prefer matching an existing artist to creating a new record with a null death year if possible.
	$artist = Artist::FindMatch($artistName);
	if($artist === null){
		$artist = new Artist();
		$artist->Name = $artistName;
	}

	$artwork = new Artwork();
	$artwork->Artist = new Artist();
	$artwork->Artist = $artist;
	$artwork->Name = $artworkName;
	$artwork->CompletedYear = $completedYear;
	$artwork->CompletedYearIsCirca = false;
	$artwork->Created = new DateTime();
	$artwork->Status = COVER_ARTWORK_STATUS_IN_USE;
	$artwork->EbookWwwFilesystemPath = $ebookWwwFilesystemPath;
	$artwork->MimeType = ImageMimeType::JPG;

	$coverSourceFile = tempnam($workDir, 'cover.source.');
	// Search for JPEG, PNG, and TIFF source files, in that order.
	exec("git show HEAD:images/cover.source.jpg > $coverSourceFile.jpg", $shellOutput, $resultCode);
	if($resultCode !== 0){
		// No JPEG, try PNG.
		exec("git show HEAD:images/cover.source.png > $coverSourceFile.png", $shellOutput, $resultCode);
		if($resultCode == 0){
			// Found PNG, convert it to JPEG.
			exec("convert $coverSourceFile.png -resize '3750x>' -sampling-factor 4:2:0 -strip -quality 80 -colorspace RGB -interlace JPEG $coverSourceFile.jpg", $shellOutput, $resultCode);
			if($resultCode !== 0){
				exit($repoDir . " Error: Failed to convert images/cover.source.png to JPEG\n");
			}
		}else{
			// No JPEG or PNG, try TIFF.
			exec("git show HEAD:images/cover.source.tif > $coverSourceFile.tif", $shellOutput, $resultCode);
			if($resultCode == 0){
				// Found TIFF, convert it to JPEG.
				exec("convert $coverSourceFile.tif -resize '3750x>' -sampling-factor 4:2:0 -strip -quality 80 -colorspace RGB -interlace JPEG $coverSourceFile.jpg", $shellOutput, $resultCode);
				if($resultCode !== 0){
					exit($repoDir . " Error: Failed to convert images/cover.source.tif to JPEG\n");
				}
			}else{
				exit($repoDir . " Error: no images/cover.source.jpg or images/cover.source.png or images/cover.source.tif\n");

			}
		}
	}

	$uploadedFile = ['tmp_name' => $coverSourceFile . '.jpg', 'error' => UPLOAD_ERR_OK];
	$artwork->Create($uploadedFile);
}
else{
	if($verbose){
		printf($repoDir . " Existing artwork found at %s/%s, updating its status.\n", $artistUrlName, $artworkUrlName);
	}

	if($artwork->CompletedYear != $completedYear){
		printf($repoDir . " Error: Existing database artwork completed year, %d, does not match ebook colophon completed year, %d. Not updating database.\n", $artwork->CompletedYear, $completedYear);
		exit($repoDir . " Error: completed year\n");
	}

	if($artwork->Status === COVER_ARTWORK_STATUS_IN_USE){
		printf($repoDir . " Error: Existing database artwork already marked as 'in_use' by ebook '%s'. Not updating database.\n", $artwork->EbookWwwFilesystemPath);
		exit($repoDir . " Error: in_use\n");
	}

	$artwork->MarkInUse($ebookWwwFilesystemPath);
}