standardebooks/web
1
0
Fork 0
mirror of https://github.com/standardebooks/web.git synced 2025-07-07 07:10:29 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

Add collections and authors RSS/Atom/OPDS feeds

Browse source
This commit is contained in:
Alex Cabal 2022-07-11 16:54:53 -05:00
parent e19847adac
commit 05e0f77b45
19 changed files with 280 additions and 142 deletions
Download patch file Download diff file Expand all files Collapse all files

213
scripts/generate-feeds
View file

@ -2,24 +2,58 @@
<?
require_once('/standardebooks.org/web/lib/Core.php');
use function Safe\krsort;
use Safe\DateTime;
use function Safe\getopt;
use function Safe\mkdir;
use function Safe\preg_replace;
use function Safe\sort;
function SaveFeed($feed, $force, $now = null){
function SortByUpdatedDesc($a, $b){
return $b->Updated <=> $a->Updated;
}
function SaveFeed(Feed $feed, bool $force, ?string $label = null, ?string $labelSort = null, DateTime $now = null): void{
$updateAttrs = false;
if($force){
if($now !== null){
$feed->Updated = $now;
}
$feed->Save();
$updateAttrs = true;
}
else{
$feed->SaveIfChanged();
$updateAttrs = $feed->SaveIfChanged();
}
if($updateAttrs && $label !== null && $labelSort !== null){
exec('attr -q -s se-label -V ' . escapeshellarg($label) . ' ' . escapeshellarg($feed->Path));
exec('attr -q -s se-label-sort -V ' . escapeshellarg($labelSort) . ' ' . escapeshellarg($feed->Path));
}
}
function CreateOpdsCollectionFeed(string $name, string $url, string $description, array $collections, array $ebooks, DateTime $now, string $webRoot, OpdsNavigationFeed $opdsRoot, bool $force): void{
$collator = collator_create('en_US'); // Used for sorting letters with diacritics like in author names
usort($collections, function($a, $b) use($collator){ return $collator->compare($a['sortedname'], $b['sortedname']); });
// Create the collections navigation document
$collectionNavigationEntries = [];
foreach($collections as $collection){
$collectionNavigationEntries[] = new OpdsNavigationEntry($collection['name'], str_replace('%s', $collection['name'], $description), $url . '/' . $collection['id'], $now, 'subsection', 'navigation');
}
$collectionsFeed = new OpdsNavigationFeed('Standard Ebooks by ' . ucfirst($name), 'Browse Standard Ebooks by ' . $name . '.', $url, $webRoot . $url . '/index.xml', $collectionNavigationEntries, $opdsRoot);
$collectionsFeed->Subtitle = 'Browse Standard Ebooks by collection.';
SaveFeed($collectionsFeed, $force, null, null, $now);
// Now generate each individual collection feed
foreach($collectionNavigationEntries as $collectionNavigationEntry){
$id = basename($collectionNavigationEntry->Id);
usort($ebooks[$id], 'SortByUpdatedDesc');
$collectionFeed = new OpdsAcquisitionFeed($collectionNavigationEntry->Title . ' Ebooks', $collectionNavigationEntry->Description, $url . '/' . $id, $webRoot . $url . '/' . $id . '.xml', $ebooks[$id], $collectionsFeed);
SaveFeed($collectionFeed, $force, null, null, $now);
}
}
$longopts = ['webroot:', 'force'];
$options = getopt('', $longopts);
$webRoot = $options['webroot'] ?? WEB_ROOT;
@ -29,37 +63,47 @@ $allEbooks = [];
$newestEbooks = [];
$subjects = [];
$ebooksBySubject = [];
$collections = [];
$ebooksByCollection = [];
$authors = [];
$ebooksByAuthor = [];
$ebooksPerNewestEbooksFeed = 15;
if(!is_dir($webRoot . '/feeds/opds/subjects')){
mkdir($webRoot . '/feeds/opds/subjects');
}
$dirs = [ '/feeds/opds/subjects', '/feeds/rss/subjects', '/feeds/atom/subjects',
'/feeds/opds/collections', '/feeds/rss/collections', '/feeds/atom/collections',
'/feeds/opds/authors', '/feeds/rss/authors', '/feeds/atom/authors'
];
if(!is_dir($webRoot . '/feeds/rss/subjects')){
mkdir($webRoot . '/feeds/rss/subjects');
}
if(!is_dir($webRoot . '/feeds/atom/subjects')){
mkdir($webRoot . '/feeds/atom/subjects');
foreach($dirs as $dir){
if(!is_dir($webRoot . $dir)){
mkdir($webRoot . $dir);
}
}
// Iterate over all ebooks to build the various feeds
foreach(Library::GetEbooksFromFilesystem($webRoot) as $ebook){
$allEbooks[$ebook->Updated->format('Y-m-d\TH:i:s\Z') . ' ' . $ebook->Identifier] = $ebook;
$newestEbooks[$ebook->Created->format('Y-m-d\TH:i:s\Z') . ' ' . $ebook->Identifier] = $ebook;
$allEbooks[] = $ebook;
$newestEbooks[] = $ebook;
foreach($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->Created->format('Y-m-d\TH:i:s\Z') . ' ' . $ebook->Identifier] = $ebook;
$urlName = Formatter::MakeUrlSafe($tag->Name);
$ebooksBySubject[$urlName][] = $ebook;
$subjects[$urlName] = ['id' => $urlName, 'name' => $tag->Name, 'sortedname' => $tag->Name];
}
foreach($ebook->Collections as $collection){
$urlName = Formatter::MakeUrlSafe($collection->Name);
$ebooksByCollection[$urlName][] = $ebook;
$collections[$urlName] = ['id' => $urlName, 'name' => $collection->Name, 'sortedname' => $collection->GetSortedName()];
}
$authorsUrl = preg_replace('|^/ebooks/|', '', $ebook->AuthorsUrl);
$ebooksByAuthor[$authorsUrl][] = $ebook;
$authors[$authorsUrl] = ['id' => $authorsUrl, 'name' => strip_tags($ebook->AuthorsHtml), 'sortedname' => $ebook->Authors[0]->SortName];
}
krsort($newestEbooks);
usort($allEbooks, 'SortByUpdatedDesc');
usort($newestEbooks, function($a, $b){ return $b->Created <=> $a->Created; });
$newestEbooks = array_slice($newestEbooks, 0, $ebooksPerNewestEbooksFeed);
$now = new DateTime();
@ -81,6 +125,20 @@ $opdsRootEntries = [
$now,
'subsection',
'navigation'),
new OpdsNavigationEntry(
'Standard Ebooks by Collection',
'Browse Standard Ebooks by collection.',
'/feeds/opds/collections',
$now,
'subsection',
'navigation'),
new OpdsNavigationEntry(
'Standard Ebooks by Author',
'Browse Standard Ebooks by author.',
'/feeds/opds/authors',
$now,
'subsection',
'navigation'),
new OpdsNavigationEntry(
'All Standard Ebooks',
'All Standard Ebooks, most-recently-updated first. This is a Complete Acquisition Feed as defined in OPDS 1.2 §2.5.',
@ -91,64 +149,85 @@ $opdsRootEntries = [
];
$opdsRoot = new OpdsNavigationFeed('Standard Ebooks', 'The Standard Ebooks catalog.', '/feeds/opds', $webRoot . '/feeds/opds/index.xml', $opdsRootEntries, null);
SaveFeed($opdsRoot, $force, $now);
SaveFeed($opdsRoot, $force, null, null, $now);
// Create the subjects navigation document
sort($subjects);
$subjectNavigationEntries = [];
foreach($subjects as $subject){
$subjectNavigationEntries[] = new OpdsNavigationEntry($subject, 'Standard Ebooks tagged with “' . strtolower($subject) . ',” most-recently-released first.', '/feeds/opds/subjects/' . Formatter::MakeUrlSafe($subject), $now, 'subsection', 'navigation');
}
$subjectsFeed = new OpdsNavigationFeed('Standard Ebooks by Subject', 'Browse Standard Ebooks by subject.', '/feeds/opds/subjects', $webRoot . '/feeds/opds/subjects/index.xml', $subjectNavigationEntries, $opdsRoot);
$subjectsFeed->Subtitle = 'Browse Standard Ebooks by subject.';
SaveFeed($subjectsFeed, $force, $now);
// Create the Subjects feeds
CreateOpdsCollectionFeed('subject', '/feeds/opds/subjects', 'Standard Ebooks in the “%s” subject, most-recently-released first.', $subjects, $ebooksBySubject, $now, $webRoot, $opdsRoot, $force);
// Now generate each individual subject feed
foreach($subjectNavigationEntries as $subjectNavigationEntry){
krsort($ebooksBySubject[$subjectNavigationEntry->Title]);
$subjectFeed = new OpdsAcquisitionFeed($subjectNavigationEntry->Title . ' Ebooks', $subjectNavigationEntry->Description, '/feeds/opds/subjects/' . Formatter::MakeUrlSafe($subjectNavigationEntry->Title), $webRoot . '/feeds/opds/subjects/' . Formatter::MakeUrlSafe($subjectNavigationEntry->Title) . '.xml', $ebooksBySubject[$subjectNavigationEntry->Title], $subjectsFeed);
SaveFeed($subjectFeed, $force, $now);
}
// Create the Collections feeds
CreateOpdsCollectionFeed('collection', '/feeds/opds/collections', 'Standard Ebooks in the “%s” collection, most-recently-released first.', $collections, $ebooksByCollection, $now, $webRoot, $opdsRoot, $force);
// Create the 'all' feed
krsort($allEbooks);
// Create the Author feeds
CreateOpdsCollectionFeed('author', '/feeds/opds/authors', 'Standard Ebooks by %s, most-recently-released first.', $authors, $ebooksByAuthor, $now, $webRoot, $opdsRoot, $force);
// Create the All feed
$allFeed = new OpdsAcquisitionFeed('All Standard Ebooks', 'All Standard Ebooks, most-recently-updated first. This is a Complete Acquisition Feed as defined in OPDS 1.2 §2.5.', '/feeds/opds/all', $webRoot . '/feeds/opds/all.xml', $allEbooks, $opdsRoot, true);
SaveFeed($allFeed, $force, $now);
SaveFeed($allFeed, $force, null, null, $now);
// Create the 'newest' feed
// Create the Newest feed
$newestFeed = new OpdsAcquisitionFeed('Newest Standard Ebooks', 'The ' . number_format($ebooksPerNewestEbooksFeed) . ' latest Standard Ebooks, most-recently-released first.', '/feeds/opds/new-releases', $webRoot . '/feeds/opds/new-releases.xml', $newestEbooks, $opdsRoot);
SaveFeed($newestFeed, $force, $now);
SaveFeed($newestFeed, $force, null, null, $now);
// Now create RSS feeds
// Create the 'newest' feed
$newestRssFeed = new RssFeed('Standard Ebooks - Newest Ebooks', 'The ' . number_format($ebooksPerNewestEbooksFeed) . ' latest Standard Ebooks, most-recently-released first.', '/feeds/rss/new-releases', $webRoot . '/feeds/rss/new-releases.xml', $newestEbooks);
SaveFeed($newestRssFeed, $force);
// Create the 'all' feed
// Create RSS/Atom feeds
// Create the RSS All feed
$allRssFeed = new RssFeed('Standard Ebooks - All Ebooks', 'All Standard Ebooks, most-recently-released first.', '/feeds/rss/all', $webRoot . '/feeds/rss/all.xml', $allEbooks);
SaveFeed($allRssFeed, $force);
SaveFeed($allRssFeed, $force, null, null);
// Generate each individual subject feed
foreach($ebooksBySubject as $subject => $ebooks){
krsort($ebooks);
$subjectRssFeed = new RssFeed('Standard Ebooks - ' . (string)$subject . ' Ebooks', 'Standard Ebooks tagged with “' . strtolower($subject) . ',” most-recently-released first.', '/feeds/rss/subjects/' . Formatter::MakeUrlSafe((string)$subject), $webRoot . '/feeds/rss/subjects/' . Formatter::MakeUrlSafe((string)$subject) . '.xml', $ebooks);
SaveFeed($subjectRssFeed, $force);
}
// Create the RSS Newest feed
$newestRssFeed = new RssFeed('Standard Ebooks - Newest Ebooks', 'The ' . number_format($ebooksPerNewestEbooksFeed) . ' latest Standard Ebooks, most-recently-released first.', '/feeds/rss/new-releases', $webRoot . '/feeds/rss/new-releases.xml', $newestEbooks);
SaveFeed($newestRssFeed, $force, null, null);
// Now create the Atom feeds
// Create the 'newest' feed
$newestAtomFeed = new AtomFeed('Standard Ebooks - Newest Ebooks', 'The ' . number_format($ebooksPerNewestEbooksFeed) . ' latest Standard Ebooks, most-recently-released first.', '/feeds/atom/new-releases', $webRoot . '/feeds/atom/new-releases.xml', $newestEbooks);
SaveFeed($newestAtomFeed, $force, $now);
// Create the 'all' feed
// Create the Atom All feed
$allAtomFeed = new AtomFeed('Standard Ebooks - All Ebooks', 'All Standard Ebooks, most-recently-released first.', '/feeds/atom/all', $webRoot . '/feeds/atom/all.xml', $allEbooks);
SaveFeed($allAtomFeed, $force, $now);
SaveFeed($allAtomFeed, $force, null, null, $now);
// Create the Atom Newest feed
$newestAtomFeed = new AtomFeed('Standard Ebooks - Newest Ebooks', 'The ' . number_format($ebooksPerNewestEbooksFeed) . ' latest Standard Ebooks, most-recently-released first.', '/feeds/atom/new-releases', $webRoot . '/feeds/atom/new-releases.xml', $newestEbooks);
SaveFeed($newestAtomFeed, $force, null, null, $now);
// Generate each individual subject feed
foreach($ebooksBySubject as $subject => $ebooks){
krsort($ebooks);
$subjectAtomFeed = new AtomFeed('Standard Ebooks - ' . (string)$subject . ' Ebooks', 'Standard Ebooks tagged with “' . strtolower($subject) . ',” most-recently-released first.', '/feeds/atom/subjects/' . Formatter::MakeUrlSafe((string)$subject), $webRoot . '/feeds/atom/subjects/' . Formatter::MakeUrlSafe((string)$subject) . '.xml', $ebooks);
SaveFeed($subjectAtomFeed, $force, $now);
usort($ebooks, 'SortByUpdatedDesc');
$title = 'Standard Ebooks - ' . $subjects[$subject]['name'] . ' Ebooks';
$subtitle = 'Standard Ebooks in the “' . strtolower($subjects[$subject]['name']) . '” subject, most-recently-released first.';
$subjectRssFeed = new RssFeed($title, $subtitle, '/feeds/rss/subjects/' . Formatter::MakeUrlSafe($subject), $webRoot . '/feeds/rss/subjects/' . Formatter::MakeUrlSafe($subject) . '.xml', $ebooks);
SaveFeed($subjectRssFeed, $force, $subjects[$subject]['name'], $subjects[$subject]['sortedname']);
$subjectAtomFeed = new AtomFeed($title, $subtitle, '/feeds/atom/subjects/' . Formatter::MakeUrlSafe($subject), $webRoot . '/feeds/atom/subjects/' . Formatter::MakeUrlSafe($subject) . '.xml', $ebooks);
SaveFeed($subjectAtomFeed, $force, $subjects[$subject]['name'], $subjects[$subject]['sortedname'], $now);
}
// Generate each individual collection feed
foreach($ebooksByCollection as $collection => $ebooks){
usort($ebooks, 'SortByUpdatedDesc');
$titleName = preg_replace('/^The /ius', '', $collections[$collection]['name']);
$title ='Standard Ebooks - Ebooks in the ' . $titleName . ' collection';
$subtitle = 'Standard Ebooks in the ' . $titleName . ' collection, most-recently-released first.';
$collectionRssFeed = new RssFeed($title, $subtitle, '/feeds/rss/collections/' . Formatter::MakeUrlSafe($collection), $webRoot . '/feeds/rss/collections/' . Formatter::MakeUrlSafe($collection) . '.xml', $ebooks);
SaveFeed($collectionRssFeed, $force, $collections[$collection]['name'], $collections[$collection]['sortedname']);
$collectionAtomFeed = new AtomFeed($title, $subtitle, '/feeds/atom/collections/' . Formatter::MakeUrlSafe($collection), $webRoot . '/feeds/atom/collections/' . Formatter::MakeUrlSafe($collection) . '.xml', $ebooks);
SaveFeed($collectionAtomFeed, $force, $collections[$collection]['name'], $collections[$collection]['sortedname'], $now);
}
// Generate each individual author feed
foreach($ebooksByAuthor as $collection => $ebooks){
usort($ebooks, 'SortByUpdatedDesc');
$title = 'Standard Ebooks - Ebooks by ' . $authors[$collection]['name'];
$subtitle = 'Standard Ebooks by ' . $authors[$collection]['name'] . ', most-recently-released first.';
$collectionRssFeed = new RssFeed($title, $subtitle, '/feeds/rss/authors/' . $authors[$collection]['id'], $webRoot . '/feeds/rss/authors/' . $authors[$collection]['id'] . '.xml', $ebooks);
SaveFeed($collectionRssFeed, $force, $authors[$collection]['name'], $authors[$collection]['sortedname']);
$collectionAtomFeed = new AtomFeed($title, $subtitle, '/feeds/atom/authors/' . $authors[$collection]['id'], $webRoot . '/feeds/atom/authors/' . $authors[$collection]['id'] . '.xml', $ebooks);
SaveFeed($collectionAtomFeed, $force, $authors[$collection]['name'], $authors[$collection]['sortedname'], $now);
}
?>