standardebooks/web
1
0
Fork 0
mirror of https://github.com/standardebooks/web.git synced 2025-07-06 06:40:33 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

Make MARC roles an enum and restructure how contributors are fetched from the DB to reduce queries

Browse source
This commit is contained in:
Alex Cabal 2024-11-11 12:01:53 -06:00
parent daf8e16ef4
commit d6a2bdcbc8
8 changed files with 91 additions and 75 deletions
Download patch file Download diff file Expand all files Collapse all files

9
config/apache/rewrites/ebooks.conf
View file

@ -1,3 +1,12 @@
# Work around some naughty hotlinkers
RewriteCond %{HTTP_REFERER} !^$
RewriteCond %{HTTP_REFERER} ^https?://.*medialibrary.it [NC]
RewriteRule ^(/ebooks/.+/downloads/|/images/covers/) /images/do-not-hotlink.jpg [R=301,L,NC]
# Redirect cover images with caching sha's to the root image
# We do this because some sites like Google cache the cover image path, so changing it results in lots of 404s
RewriteRule ^/images/covers/(.+?)\-[a-z0-9]{8}\-(cover|hero)(@2x)?\.(jpg|avif)$ /images/covers/$1-$2$3.$4
# Rewrite ebook downloads
RewriteRule ^/ebooks/(.+?)/download$ /ebooks/download.php?url-path=$1 [QSA]

9
config/apache/rewrites/misc.conf
View file

@ -7,15 +7,6 @@ RewriteRule ^/tools$ https://github.com/standardebooks/tools [R=302,L]
# Redirect latest version of the manual
RewriteRule ^/manual/latest(.*) /manual/index.php?url=$1 [L]
# Work around some naughty hotlinkers
RewriteCond %{HTTP_REFERER} !^$
RewriteCond %{HTTP_REFERER} ^https?://.*medialibrary.it [NC]
RewriteRule ^(/ebooks/.+/downloads/|/images/covers/) /images/do-not-hotlink.jpg [R=301,L,NC]
# Redirect cover images with caching sha's to the root image
# We do this because some sites like Google cache the cover image path, so changing it results in lots of 404s
RewriteRule ^/images/covers/(.+?)\-[a-z0-9]{8}\-(cover|hero)(@2x)?\.(jpg|avif)$ /images/covers/$1-$2$3.$4
# Rewrite rules for bulk downloads
RewriteRule ^/bulk-downloads/(.+\.zip)$ /bulk-downloads/download.php?path=$1
RewriteRule ^/bulk-downloads/([^/\.]+)$ /bulk-downloads/collection.php?class=$1

2
config/sql/se/Contributors.sql
View file

@ -4,7 +4,7 @@ CREATE TABLE IF NOT EXISTS `Contributors` (
`UrlName` varchar(255) NOT NULL,
`SortName` varchar(255) NULL,
`WikipediaUrl` varchar(255) NULL,
`MarcRole` varchar(10) NULL,
`MarcRole` ENUM('aut', 'ctb', 'ill', 'trl') NOT NULL,
`FullName` varchar(255) NULL,
`NacoafUrl` varchar(255) NULL,
`SortOrder` tinyint(3) unsigned NOT NULL,

18
lib/Contributor.php
View file

@ -7,7 +7,7 @@ class Contributor{
public string $UrlName;
public ?string $SortName = null;
public ?string $WikipediaUrl = null;
public ?string $MarcRole = null;
public Enums\MarcRole $MarcRole;
public ?string $FullName = null;
public ?string $NacoafUrl = null;
public int $SortOrder;
@ -32,19 +32,14 @@ class Contributor{
if($this->Name == ''){
$error->Add(new Exceptions\ContributorNameRequiredException());
}
$this->UrlName = Formatter::MakeUrlSafe($this->Name);
}
else{
$error->Add(new Exceptions\ContributorNameRequiredException());
}
if(isset($this->UrlName)){
$this->UrlName = trim($this->UrlName);
if($this->UrlName == ''){
$error->Add(new Exceptions\ContributorUrlNameRequiredException());
}
}
else{
if(!isset($this->UrlName)){
$error->Add(new Exceptions\ContributorUrlNameRequiredException());
}
@ -73,11 +68,6 @@ class Contributor{
}
}
$this->MarcRole = trim($this->MarcRole ?? '');
if($this->MarcRole == ''){
$this->MarcRole = null;
}
$this->NacoafUrl = trim($this->NacoafUrl ?? '');
if($this->NacoafUrl == ''){
$this->NacoafUrl = null;

107
lib/Ebook.php
View file

@ -199,18 +199,55 @@ class Ebook{
return $this->_Sources;
}
/**
* Fill all contributor properties for this ebook, e.g. authors, translators, etc.
*
* We do this in a single database query to prevent 4+ queries for each ebook.
*/
protected function GetAllContributors(): void{
$contributors = Db::Query('
SELECT *
from Contributors
where EbookId = ?
order by MarcRole asc, SortOrder asc
', [$this->EbookId], Contributor::class);
$this->_Authors = [];
$this->_Translators = [];
$this->_Illustrators = [];
$this->_Contributors = [];
foreach($contributors as $contributor){
switch($contributor->MarcRole){
case Enums\MarcRole::Author:
$this->_Authors[] = $contributor;
break;
case Enums\MarcRole::Translator:
$this->_Translators[] = $contributor;
break;
case Enums\MarcRole::Illustrator:
$this->_Illustrators[] = $contributor;
break;
case Enums\MarcRole::Contributor:
$this->_Contributors[] = $contributor;
break;
}
}
}
/**
* @return array<Contributor>
*/
protected function GetAuthors(): array{
if(!isset($this->_Authors)){
$this->_Authors = Db::Query('
SELECT *
from Contributors
where EbookId = ?
and MarcRole = ?
order by SortOrder asc
', [$this->EbookId, 'aut'], Contributor::class);
$this->GetAllContributors();
}
return $this->_Authors;
@ -221,13 +258,7 @@ class Ebook{
*/
protected function GetIllustrators(): array{
if(!isset($this->_Illustrators)){
$this->_Illustrators = Db::Query('
SELECT *
from Contributors
where EbookId = ?
and MarcRole = ?
order by SortOrder asc
', [$this->EbookId, 'ill'], Contributor::class);
$this->GetAllContributors();
}
return $this->_Illustrators;
@ -238,13 +269,7 @@ class Ebook{
*/
protected function GetTranslators(): array{
if(!isset($this->_Translators)){
$this->_Translators = Db::Query('
SELECT *
from Contributors
where EbookId = ?
and MarcRole = ?
order by SortOrder asc
', [$this->EbookId, 'trl'], Contributor::class);
$this->GetAllContributors();
}
return $this->_Translators;
@ -255,13 +280,7 @@ class Ebook{
*/
protected function GetContributors(): array{
if(!isset($this->_Contributors)){
$this->_Contributors = Db::Query('
SELECT *
from Contributors
where EbookId = ?
and MarcRole = ?
order by SortOrder asc
', [$this->EbookId, 'ctb'], Contributor::class);
$this->GetAllContributors();
}
return $this->_Contributors;
@ -317,13 +336,9 @@ class Ebook{
return $this->_UrlSafeIdentifier;
}
private function GetLatestCommitHash(): string{
return substr(sha1($this->GitCommits[0]->Hash), 0, 8);
}
protected function GetHeroImageUrl(): string{
if(!isset($this->_HeroImageUrl)){
$this->_HeroImageUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . $this->GetLatestCommitHash() . '-hero.jpg';
$this->_HeroImageUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . substr(sha1($this->Updated->format(Enums\DateTimeFormat::UnixTimestamp->value)), 0, 8) . '-hero.jpg';
}
return $this->_HeroImageUrl;
@ -332,7 +347,7 @@ class Ebook{
protected function GetHeroImageAvifUrl(): ?string{
if(!isset($this->_HeroImageAvifUrl)){
if(file_exists(WEB_ROOT . '/images/covers/' . $this->UrlSafeIdentifier . '-hero.avif')){
$this->_HeroImageAvifUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . $this->GetLatestCommitHash() . '-hero.avif';
$this->_HeroImageAvifUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . substr(sha1($this->Updated->format(Enums\DateTimeFormat::UnixTimestamp->value)), 0, 8) . '-hero.avif';
}
}
@ -341,7 +356,7 @@ class Ebook{
protected function GetHeroImage2xUrl(): string{
if(!isset($this->_HeroImage2xUrl)){
$this->_HeroImage2xUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . $this->GetLatestCommitHash() . '-hero@2x.jpg';
$this->_HeroImage2xUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . substr(sha1($this->Updated->format(Enums\DateTimeFormat::UnixTimestamp->value)), 0, 8) . '-hero@2x.jpg';
}
return $this->_HeroImage2xUrl;
@ -350,7 +365,7 @@ class Ebook{
protected function GetHeroImage2xAvifUrl(): ?string{
if(!isset($this->_HeroImage2xAvifUrl)){
if(file_exists(WEB_ROOT . '/images/covers/' . $this->UrlSafeIdentifier . '-hero@2x.avif')){
$this->_HeroImage2xAvifUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . $this->GetLatestCommitHash() . '-hero@2x.avif';
$this->_HeroImage2xAvifUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . substr(sha1($this->Updated->format(Enums\DateTimeFormat::UnixTimestamp->value)), 0, 8) . '-hero@2x.avif';
}
}
@ -359,7 +374,7 @@ class Ebook{
protected function GetCoverImageUrl(): string{
if(!isset($this->_CoverImageUrl)){
$this->_CoverImageUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . $this->GetLatestCommitHash() . '-cover.jpg';
$this->_CoverImageUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . substr(sha1($this->Updated->format(Enums\DateTimeFormat::UnixTimestamp->value)), 0, 8) . '-cover.jpg';
}
return $this->_CoverImageUrl;
@ -368,7 +383,7 @@ class Ebook{
protected function GetCoverImageAvifUrl(): ?string{
if(!isset($this->_CoverImageAvifUrl)){
if(file_exists(WEB_ROOT . '/images/covers/' . $this->UrlSafeIdentifier . '-cover.avif')){
$this->_CoverImageAvifUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . $this->GetLatestCommitHash() . '-cover.avif';
$this->_CoverImageAvifUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . substr(sha1($this->Updated->format(Enums\DateTimeFormat::UnixTimestamp->value)), 0, 8) . '-cover.avif';
}
}
@ -377,7 +392,7 @@ class Ebook{
protected function GetCoverImage2xUrl(): string{
if(!isset($this->_CoverImage2xUrl)){
$this->_CoverImage2xUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . $this->GetLatestCommitHash() . '-cover@2x.jpg';
$this->_CoverImage2xUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . substr(sha1($this->Updated->format(Enums\DateTimeFormat::UnixTimestamp->value)), 0, 8) . '-cover@2x.jpg';
}
return $this->_CoverImage2xUrl;
@ -386,7 +401,7 @@ class Ebook{
protected function GetCoverImage2xAvifUrl(): ?string{
if(!isset($this->_CoverImage2xAvifUrl)){
if(file_exists(WEB_ROOT . '/images/covers/' . $this->UrlSafeIdentifier . '-cover@2x.avif')){
$this->_CoverImage2xAvifUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . $this->GetLatestCommitHash() . '-cover@2x.avif';
$this->_CoverImage2xAvifUrl = '/images/covers/' . $this->UrlSafeIdentifier . '-' . substr(sha1($this->Updated->format(Enums\DateTimeFormat::UnixTimestamp->value)), 0, 8) . '-cover@2x.avif';
}
}
@ -823,7 +838,7 @@ class Ebook{
$contributor->SortName = $fileAs;
$contributor->FullName = Ebook::NullIfEmpty($xml->xpath('/package/metadata/meta[@property="se:name.person.full-name"][@refines="#' . $id . '"]'));
$contributor->WikipediaUrl = Ebook::NullIfEmpty($xml->xpath('/package/metadata/meta[@property="se:url.encyclopedia.wikipedia"][@refines="#' . $id . '"]'));
$contributor->MarcRole = 'aut';
$contributor->MarcRole = Enums\MarcRole::Author;
$contributor->NacoafUrl = Ebook::NullIfEmpty($xml->xpath('/package/metadata/meta[@property="se:url.authority.nacoaf"][@refines="#' . $id . '"]'));
$authors[] = $contributor;
@ -850,7 +865,7 @@ class Ebook{
$c->SortName = Ebook::NullIfEmpty($xml->xpath('/package/metadata/meta[@property="file-as"][@refines="#' . $id . '"]'));
$c->FullName = Ebook::NullIfEmpty($xml->xpath('/package/metadata/meta[@property="se:name.person.full-name"][@refines="#' . $id . '"]'));
$c->WikipediaUrl = Ebook::NullIfEmpty($xml->xpath('/package/metadata/meta[@property="se:url.encyclopedia.wikipedia"][@refines="#' . $id . '"]'));
$c->MarcRole = $role;
$c->MarcRole = Enums\MarcRole::tryFrom((string)$role) ?? Enums\MarcRole::Contributor;
$c->NacoafUrl = Ebook::NullIfEmpty($xml->xpath('/package/metadata/meta[@property="se:url.authority.nacoaf"][@refines="#' . $id . '"]'));
// A display-sequence of 0 indicates that we don't want to process this contributor.
@ -1437,10 +1452,10 @@ class Ebook{
foreach($contributors as $contributor){
$role = 'schema:contributor';
switch($contributor->MarcRole){
case 'trl':
case Enums\MarcRole::Translator:
$role = 'schema:translator';
break;
case 'ill':
case Enums\MarcRole::Illustrator:
$role = 'schema:illustrator';
break;
}
@ -1497,10 +1512,10 @@ class Ebook{
foreach($this->Translators as $contributor){
$role = 'schema:contributor';
switch($contributor->MarcRole){
case 'trl':
case Enums\MarcRole::Translator:
$role = 'schema:translator';
break;
case 'ill':
case Enums\MarcRole::Illustrator:
$role = 'schema:illustrator';
break;
}

9
lib/Enums/MarcRole.php Normal file
View file

@ -0,0 +1,9 @@
<?
namespace Enums;
enum MarcRole: string{
case Author = 'aut';
case Contributor = 'ctb';
case Illustrator = 'ill';
case Translator = 'trl';
}

2
scripts/deploy-ebook-to-www
View file

@ -354,7 +354,7 @@ do
find "${workDir}"/src/epub \( -type d -name .git -prune \) -o -type f -name "*.xhtml" -print0 | xargs -0 sed --in-place --regexp-extended "s|<body([^>]*)>|<body\1><header><nav><ul><li><a href=\"/\">Standard Ebooks</a></li><li><a href=\"${bookUrl}\">Back to ebook</a></li><li><a href=\"${bookUrl}/text\">Table of contents</a></li></ul></nav></header>|"
# Add a chapter navigation footer to each page.
"${scriptsDir}"/inject-chapter-navigation-footer "${workDir}" "${bookUrl}"
#"${scriptsDir}"/inject-chapter-navigation-footer "${workDir}" "${bookUrl}"
# Adjust sponsored links in the colophon.
sed --in-place 's|<p><a href="http|<p><a rel="nofollow" href="http|g' "${workDir}"/src/epub/text/colophon.xhtml

10
scripts/generate-bulk-downloads
View file

@ -23,8 +23,10 @@ $types = ['epub', 'epub-advanced', 'azw3', 'kepub', 'xhtml'];
$groups = ['collections', 'subjects', 'authors', 'months'];
$ebooksByGroup = ['collections' => [], 'subjects' => [], 'authors' => [], 'months' => []];
/**
* @see https://www.php.net/manual/en/function.rmdir.php#117354
*/
function rrmdir(string $src): void{
// See <https://www.php.net/manual/en/function.rmdir.php#117354>.
$dir = opendir($src);
while(false !== ($file = readdir($dir))){
if (($file != '.') && ($file != '..')){
@ -94,8 +96,8 @@ function CreateZip(string $filePath, array $ebooks, string $type, string $webRoo
$zip->close();
// We have to do a copy, then unlink because rename() can't rename across file systems.
// If the bulk downloads are symlinked to a storage volume, the rename() won't work.
// We have to do a copy, then unlink because `rename()` can't rename across file systems.
// If the bulk downloads are symlinked to a storage volume, then `rename()` won't work.
copy($tempFilename, $filePath);
unlink($tempFilename);
@ -167,7 +169,7 @@ foreach(Ebook::GetAll() as $ebook){
}
// Add to the 'books by author' list.
// We have to index by UrlName for cases like `Samuel Butler` whose UrlName is `samuel-butler-1612-1680`.
// We have to index by `UrlName` for cases like `Samuel Butler` whose `UrlName` is `samuel-butler-1612-1680`.
$authorsUrl = preg_replace('|^/ebooks/|', '', $ebook->AuthorsUrl);
if(!isset($ebooksByGroup['authors'][$authorsUrl])){
$obj = new stdClass();