standardebooks/web
1
0
Fork 0
mirror of https://github.com/standardebooks/web.git synced 2025-07-09 16:20:27 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

Flesh out OPDS generation backend to be more robust and to support generic Atom feeds

Browse source
This commit is contained in:
Alex Cabal 2022-06-20 21:55:34 -05:00
parent 35188195f1
commit f6df03cfca
23 changed files with 1549 additions and 267 deletions
Download patch file Download diff file Expand all files Collapse all files

122
scripts/generate-feeds Executable file
View file

@ -0,0 +1,122 @@
#!/usr/bin/php
<?
require_once('/standardebooks.org/web/lib/Core.php');
use function Safe\krsort;
use function Safe\getopt;
use function Safe\preg_replace;
use function Safe\sort;
$longopts = ["webroot:", "weburl:"];
$options = getopt("", $longopts);
$webRoot = $options["webroot"] ?? "/standardebooks.org/web";
$webUrl = $options["weburl"] ?? "https://standardebooks.org";
$contentFiles = explode("\n", trim(shell_exec('find ' . escapeshellarg($webRoot . '/www/ebooks/') . ' -name "content.opf" | sort') ?? ''));
$allEbooks = [];
$newestEbooks = [];
$subjects = [];
$ebooksBySubject = [];
$ebooksPerNewestEbooksFeed = 30;
// Iterate over all ebooks to build the various feeds
foreach($contentFiles as $path){
if($path == '')
continue;
$ebookWwwFilesystemPath = '';
try{
$ebookWwwFilesystemPath = preg_replace('|/content\.opf|ius', '', $path);
$ebook = new Ebook($ebookWwwFilesystemPath);
$allEbooks[$ebook->ModifiedTimestamp->format('Y-m-d\TH:i:s\Z') . ' ' . $ebook->Identifier] = $ebook;
$newestEbooks[$ebook->Timestamp->format('Y-m-d\TH:i:s\Z') . ' ' . $ebook->Identifier] = $ebook;
foreach($ebook->Tags as $tag){
// Add the book's subjects to the main subjects list
if(!in_array($tag->Name, $subjects)){
$subjects[] = $tag->Name;
}
// Sort this ebook by subject
$ebooksBySubject[$tag->Name][$ebook->Timestamp->format('Y-m-d\TH:i:s\Z') . ' ' . $ebook->Identifier] = $ebook;
}
}
catch(\Exception $ex){
print('Failed to generate OPDS entry for `' . $ebookWwwFilesystemPath . '`. Exception: ' . $ex->getMessage());
continue;
}
}
$now = new DateTime();
// Create OPDS feeds
$opdsRootEntries = [
new OpdsNavigationEntry(
'/opds/new-releases',
'http://opds-spec.org/sort/new',
'acquisition',
$now,
'Newest ' . number_format($ebooksPerNewestEbooksFeed) . ' Standard Ebooks',
'A list of the ' . number_format($ebooksPerNewestEbooksFeed) . ' newest Standard Ebooks, most-recently-released first.'),
new OpdsNavigationEntry(
'/opds/subjects',
'subsection',
'navigation',
$now,
'Standard Ebooks by Subject',
'Browse Standard Ebooks by subject.'),
new OpdsNavigationEntry(
'/opds/all',
'http://opds-spec.org/crawlable',
'acquisition',
$now,
'All Standard Ebooks',
'A list of all Standard Ebooks, most-recently-updated first. This is a Complete Acquisition Feed as defined in OPDS 1.2 §2.5.')
];
$opdsRoot = new OpdsNavigationFeed('/opds', 'Standard Ebooks', WEB_ROOT . '/opds/index.xml', $opdsRootEntries, null);
$opdsRoot->Save();
// Create the subjects navigation document
sort($subjects);
$subjectNavigationEntries = [];
foreach($subjects as $subject){
$summary = number_format(sizeof($ebooksBySubject[$subject])) . ' Standard Ebook';
if(sizeof($ebooksBySubject[$subject]) != 1){
$summary .= 's';
}
$summary .= ' tagged with “' . strtolower($subject) . ',” most-recently-released first.';
// We leave the updated timestamp blank, as it will be filled in when we generate the individual feeds
$subjectNavigationEntries[] = new OpdsNavigationEntry('/opds/subjects/' . Formatter::MakeUrlSafe($subject), 'subsection', 'navigation', $now, $subject, $summary);
}
$subjectsFeed = new OpdsNavigationFeed('/opds/subjects', 'Standard Ebooks by Subject', WEB_ROOT . '/opds/subjects/index.xml', $subjectNavigationEntries, $opdsRoot);
$subjectsFeed->Save();
// Now generate each individual subject feed
foreach($ebooksBySubject as $subject => $ebooks){
krsort($ebooks);
$subjectFeed = new OpdsAcquisitionFeed('/opds/subjects/' . Formatter::MakeUrlSafe((string)$subject), (string)$subject, WEB_ROOT . '/opds/subjects/' . Formatter::MakeUrlSafe((string)$subject) . '.xml', $ebooks, $subjectsFeed);
$subjectFeed->Save();
}
// Create the 'all' feed
krsort($allEbooks);
$allFeed = new OpdsAcquisitionFeed('/opds/all', 'All Standard Ebooks', WEB_ROOT . '/opds/all.xml', $allEbooks, $opdsRoot, true);
$allFeed->Save();
// Create the 'newest' feed
krsort($newestEbooks);
$newestEbooks = array_slice($newestEbooks, 0, $ebooksPerNewestEbooksFeed);
$newestFeed = new OpdsAcquisitionFeed('/opds/new-releases', 'Newest ' . number_format($ebooksPerNewestEbooksFeed) . ' Standard Ebooks', WEB_ROOT . '/opds/new-releases.xml', $newestEbooks, $opdsRoot);
$newestFeed->Save();
// Now create RSS feeds
// Create the 'newest' feed
$newestFeed = new RssFeed('/rss/new-releases', 'Newest ' . number_format($ebooksPerNewestEbooksFeed) . ' Standard Ebooks', WEB_ROOT . '/rss/new-releases.xml', 'A list of the ' . number_format($ebooksPerNewestEbooksFeed) . ' latest Standard Ebooks ebook releases, most-recently-released first.', $newestEbooks);
$newestFeed->Save();
?>