standardebooks/web
1
0
Fork 0
mirror of https://github.com/standardebooks/web.git synced 2025-07-07 23:30:35 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

Split bulk downloads into file type and cache output

Browse source
This commit is contained in:
Alex Cabal 2022-07-09 17:22:13 -05:00
parent 55985b0c2f
commit 12b79b5dcd
11 changed files with 395 additions and 178 deletions
Download patch file Download diff file Expand all files Collapse all files

146
scripts/generate-bulk-downloads Executable file
View file

@ -0,0 +1,146 @@
#!/usr/bin/php
<?
require_once('/standardebooks.org/web/lib/Core.php');
use function Safe\mkdir;
$longopts = ['webroot:'];
$options = getopt('', $longopts);
$webRoot = $options['webroot'] ?? WEB_ROOT;
$ebooksByMonth = [];
$lastUpdatedTimestampsByMonth = [];
$subjects = [];
$ebooksBySubject = [];
$lastUpdatedTimestampsBySubject = [];
function CreateZip(string $filePath, array $ebooks, string $type, string $webRoot, ?string $subject = null, ?string $month = null): void{
$tempFilename = tempnam(sys_get_temp_dir(), "se-ebooks");
$zip = new ZipArchive();
if($zip->open($tempFilename, ZipArchive::CREATE) !== true){
print('Can\'t open file: ' . $tempFilename . "\n");
}
foreach($ebooks as $ebook){
if($type == 'epub' && $ebook->EpubUrl !== null){
$ebookFilePath = $webRoot . '/' . $ebook->EpubUrl;
$zip->addFile($ebookFilePath, basename($ebookFilePath));
}
if($type == 'azw3' && $ebook->Azw3Url !== null){
$ebookFilePath = $webRoot . '/' . $ebook->Azw3Url;
$zip->addFile($ebookFilePath, basename($ebookFilePath));
}
if($type == 'kepub' && $ebook->KepubUrl !== null){
$ebookFilePath = $webRoot . '/' . $ebook->KepubUrl;
$zip->addFile($ebookFilePath, basename($ebookFilePath));
}
if($type == 'epub-advanced' && $ebook->AdvancedEpubUrl !== null){
$ebookFilePath = $webRoot . '/' . $ebook->AdvancedEpubUrl;
$zip->addFile($ebookFilePath, basename($ebookFilePath));
}
if($type == 'xhtml' && $ebook->TextSinglePageUrl !== null){
$ebookFilePath = $webRoot . '/' . $ebook->TextSinglePageUrl . '.xhtml';
// Strip the navigation header that was added as part of the deploy process
$xhtml = file_get_contents($ebookFilePath);
$xhtml = preg_replace('|<body><header><nav>.+?</nav></header>|ius', '<body>', $xhtml);
$zip->addFromString(str_replace('single-page', $ebook->UrlSafeIdentifier, basename($ebookFilePath)), $xhtml);
}
}
$zip->close();
$dir = dirname($filePath);
if(!is_dir($dir)){
mkdir($dir, 0775, true);
}
rename($tempFilename, $filePath);
// Set a filesystem attribute for the number of ebooks in the file. This will be used
// to display that number on the downloads page.
exec('attr -q -s se-ebook-count -V ' . escapeshellarg(sizeof($ebooks)) . ' ' . escapeshellarg($filePath));
exec('attr -q -s se-ebook-type -V ' . escapeshellarg($type) . ' ' . escapeshellarg($filePath));
// If we're passed a subject, add it as a file attribute too
if($subject !== null){
exec('attr -q -s se-subject -V ' . escapeshellarg($subject) . ' ' . escapeshellarg($filePath));
}
if($month !== null){
exec('attr -q -s se-month -V ' . escapeshellarg($month) . ' ' . escapeshellarg($filePath));
}
}
// Iterate over all ebooks and arrange them by publication month
foreach(Library::GetEbooksFromFilesystem($webRoot) as $ebook){
$timestamp = $ebook->Created->format('Y-m');
$updatedTimestamp = $ebook->Updated->getTimestamp();
if(!isset($ebooksByMonth[$timestamp])){
$ebooksByMonth[$timestamp] = [];
$lastUpdatedTimestampsByMonth[$timestamp] = $updatedTimestamp;
}
// Add to the 'ebooks by month' list
$ebooksByMonth[$timestamp][] = $ebook;
if($updatedTimestamp > $lastUpdatedTimestampsByMonth[$timestamp]){
$lastUpdatedTimestampsByMonth[$timestamp] = $updatedTimestamp;
}
// Add to the 'books by subject' list
foreach($ebook->Tags as $tag){
// Add the book's subjects to the main subjects list
if(!in_array($tag->Name, $subjects)){
$subjects[] = $tag->Name;
$lastUpdatedTimestampsBySubject[$tag->Name] = $updatedTimestamp;
}
// Sort this ebook by subject
$ebooksBySubject[$tag->Name][] = $ebook;
if($updatedTimestamp > $lastUpdatedTimestampsBySubject[$tag->Name]){
$lastUpdatedTimestampsBySubject[$tag->Name] = $updatedTimestamp;
}
}
}
$types = ['epub', 'epub-advanced', 'azw3', 'kepub', 'xhtml'];
foreach($ebooksByMonth as $month => $ebooks){
foreach($types as $type){
$filename = 'se-ebooks-' . $month . '-' . $type . '.zip';
$filePath = $webRoot . '/patrons-circle/downloads/months/' . $month . '/' . $filename;
// If the file doesn't exist, or if the content.opf last updated time is newer than the file modification time
if(!file_exists($filePath) || filemtime($filePath) < $lastUpdatedTimestampsByMonth[$month]){
print('Creating ' . $filePath . "\n");
CreateZip($filePath, $ebooks, $type, $webRoot, null, $month);
}
}
}
foreach($ebooksBySubject as $subject => $ebooks){
foreach($types as $type){
$urlSafeSubject = Formatter::MakeUrlSafe($subject);
$filename = 'se-ebooks-' . $urlSafeSubject . '-' . $type . '.zip';
$filePath = $webRoot . '/patrons-circle/downloads/subjects/' . $urlSafeSubject . '/'. $filename;
// If the file doesn't exist, or if the content.opf last updated time is newer than the file modification time
if(!file_exists($filePath) || filemtime($filePath) < $lastUpdatedTimestampsBySubject[$subject]){
print('Creating ' . $filePath . "\n");
CreateZip($filePath, $ebooks, $type, $webRoot, $subject, null);
}
}
}