Merge pull request #42 from kiwilan/develop

2.0.20
kiwilan · Aug 28, 2023 · cef5002 · cef5002
2 parents 875c43e + 384cf23
commit cef5002
Show file tree

Hide file tree

Showing 15 changed files with 277 additions and 46 deletions.
diff --git a/.github/workflows/codecov.yml b/.github/workflows/codecov.yml
@@ -15,9 +15,10 @@ jobs:
     steps:
       - name: Install for Linux
         run: |
+          sudo apt update
           sudo apt -y install p7zip-full ghostscript imagemagick
-          sudo apt-get install -y unrar
-          sudo apt-get install -y libunrar-dev
+          sudo apt install -y unrar
+          sudo apt install -y libunrar-dev
           sudo sed -i '/disable ghostscript format types/,+6d' /etc/ImageMagick-6/policy.xml
         shell: bash
 

diff --git a/.github/workflows/run-tests.yml b/.github/workflows/run-tests.yml
@@ -17,9 +17,10 @@ jobs:
     steps:
       - name: Install
         run: |
+          sudo apt update
           sudo apt -y install p7zip-full ghostscript imagemagick
-          sudo apt-get install -y unrar
-          sudo apt-get install -y libunrar-dev
+          sudo apt install -y unrar
+          sudo apt install -y libunrar-dev
           sudo sed -i '/disable ghostscript format types/,+6d' /etc/ImageMagick-6/policy.xml
         shell: bash
 

diff --git a/README.md b/README.md
@@ -98,8 +98,6 @@ composer require kiwilan/php-ebook
 
 ## Usage
 
-### Main
-
 With eBook files (`.epub`, `.cbz`, `.cba`, `.cbr`, `.cb7`, `.cbt`, `.pdf`) or audiobook files (`mp3`, `m4a`, `m4b`, `flac`, `ogg`).
 
 ```php
@@ -114,6 +112,7 @@ $ebook->getTitle(); // string
 $ebook->getAuthors(); // BookAuthor[] (`name`: string, `role`: string)
 $ebook->getAuthorMain(); // ?BookAuthor => First BookAuthor (`name`: string, `role`: string)
 $ebook->getDescription(); // ?string
+$ebook->getDescriptionHtml(); // ?string
 $ebook->getCopyright(); // ?string
 $ebook->getPublisher(); // ?string
 $ebook->getIdentifiers(); // BookIdentifier[] (`value`: string, `scheme`: string)
@@ -142,13 +141,22 @@ $ebook->getExtras(); // array<string, mixed> => additional data for book
 $ebook->getExtra(string $key); // mixed => safely extract data from `extras` array
 ```
 
+To know if eBook is valid, you can use `isValid()` static method, before `read()`.
+
+```php
+use Kiwilan\Ebook\Ebook;
+
+$isValid = Ebook::isValid('path/to/ebook.epub');
+```
+
 To get additional data, you can use these methods:
 
 ```php
 $ebook->getMetadata(); // ?EbookMetadata => metadata with parsers
 $ebook->getMetaTitle(); // ?MetaTitle, with slug and sort properties for `title` and `series`
 $ebook->getFormat(); // ?EbookFormatEnum => `epub`, `pdf`, `cba`
 $ebook->getCover(); // ?EbookCover => cover of book
+$ebook->getArchive(); // ?BaseArchive => archive of book from `kiwilan/php-archive`
 ```
 
 And to test if some data exists:
@@ -263,6 +271,7 @@ Please see [CHANGELOG](CHANGELOG.md) for more information on what has changed re
 
 -   [`spatie`](https://github.com/spatie) for `spatie/package-skeleton-php`
 -   [`kiwilan`](https://github.com/kiwilan) for `kiwilan/php-archive`, `kiwilan/php-audio`, `kiwilan/php-xml-reader`
+-   [All Contributors](../../contributors)
 
 ## License
 

diff --git a/composer.json b/composer.json
@@ -1,7 +1,7 @@
 {
     "name": "kiwilan/php-ebook",
     "description": "PHP package to read metadata and extract covers from eBooks (.epub, .cbz, .cbr, .cb7, .cbt, .pdf) and audiobooks (.mp3, .m4a, .m4b, .flac, .ogg).",
-    "version": "2.0.12",
+    "version": "2.0.20",
     "keywords": [
         "php",
         "ebook",

diff --git a/src/Ebook.php b/src/Ebook.php
@@ -30,6 +30,8 @@ class Ebook
 
     protected ?string $description = null;
 
+    protected ?string $descriptionHtml = null;
+
     protected ?string $publisher = null;
 
     /** @var BookIdentifier[] */
@@ -66,6 +68,7 @@ class Ebook
     protected function __construct(
         protected string $path,
         protected string $filename,
+        protected string $basename,
         protected string $extension,
         protected ?BaseArchive $archive = null,
         protected ?Audio $audio = null,
@@ -83,7 +86,46 @@ protected function __construct(
     public static function read(string $path): ?self
     {
         $start = microtime(true);
-        $filename = pathinfo($path, PATHINFO_BASENAME);
+        $self = self::parseFile($path);
+
+        $format = match ($self->format) {
+            EbookFormatEnum::EPUB => $self->epub(),
+            EbookFormatEnum::MOBI => $self->mobi(),
+            EbookFormatEnum::CBA => $self->cba(),
+            EbookFormatEnum::PDF => $self->pdf(),
+            EbookFormatEnum::AUDIOBOOK => $self->audiobook(),
+            default => null,
+        };
+
+        if ($format === null) {
+            return null;
+        }
+
+        $self->metaTitle = MetaTitle::make($self);
+
+        $time = microtime(true) - $start;
+        $self->execTime = (float) number_format((float) $time, 5, '.', '');
+
+        return $self;
+    }
+
+    /**
+     * Check if an ebook file is valid.
+     */
+    public static function isValid(string $path): bool
+    {
+        $self = self::parseFile($path);
+
+        return ! $self->isBadFile;
+    }
+
+    /**
+     * Parse an ebook file.
+     */
+    private static function parseFile(string $path): Ebook
+    {
+        $basename = pathinfo($path, PATHINFO_BASENAME);
+        $filename = pathinfo($path, PATHINFO_FILENAME);
         $extension = pathinfo($path, PATHINFO_EXTENSION);
 
         $cbaExtensions = ['cbz', 'cbr', 'cb7', 'cbt'];
@@ -100,7 +142,7 @@ public static function read(string $path): ?self
             throw new \Exception("Unknown archive type: {$extension}");
         }
 
-        $self = new self($path, $filename, $extension);
+        $self = new self($path, $filename, $basename, $extension);
 
         $self->format = match ($extension) {
             'epub' => $self->format = EbookFormatEnum::EPUB,
@@ -141,24 +183,6 @@ public static function read(string $path): ?self
             $self->audio = Audio::get($path);
         }
 
-        $format = match ($self->format) {
-            EbookFormatEnum::EPUB => $self->epub(),
-            EbookFormatEnum::MOBI => $self->mobi(),
-            EbookFormatEnum::CBA => $self->cba(),
-            EbookFormatEnum::PDF => $self->pdf(),
-            EbookFormatEnum::AUDIOBOOK => $self->audiobook(),
-            default => null,
-        };
-
-        if ($format === null) {
-            return null;
-        }
-
-        $self->metaTitle = MetaTitle::make($self);
-
-        $time = microtime(true) - $start;
-        $self->execTime = (float) number_format((float) $time, 5, '.', '');
-
         return $self;
     }
 
@@ -216,6 +240,7 @@ private function convertEbook(): self
         $this->authorMain = $ebook->getAuthorMain();
         $this->authors = $ebook->getAuthors();
         $this->description = $ebook->getDescription();
+        $this->descriptionHtml = $ebook->getDescriptionHtml();
         $this->publisher = $ebook->getPublisher();
         $this->identifiers = $ebook->getIdentifiers();
         $this->publishDate = $ebook->getPublishDate();
@@ -294,7 +319,12 @@ public function getAuthors(): array
     }
 
     /**
-     * Description of the book.
+     * Description of the book, without HTML.
+     *
+     * If original description has HTML, all HTML will be removed and text will be trimmed.
+     * You can use `getDescriptionHtml()` to get the original description sanitized.
+     *
+     * @param  int|null  $limit  Limit the length of the description.
      */
     public function getDescription(int $limit = null): ?string
     {
@@ -305,6 +335,16 @@ public function getDescription(int $limit = null): ?string
         return $this->description;
     }
 
+    /**
+     * Description of the book with HTML sanitized.
+     *
+     * If original description doesn't have HTML, it will be the same as `getDescription()`.
+     */
+    public function getDescriptionHtml(): ?string
+    {
+        return $this->descriptionHtml;
+    }
+
     /**
      * Publisher of the book.
      */
@@ -388,31 +428,48 @@ public function getPath(): string
     }
 
     /**
-     * Filename of the ebook.
+     * Filename of the ebook, e.g. `The Clan of the Cave Bear`.
      */
     public function getFilename(): string
     {
         return $this->filename;
     }
 
     /**
-     * Extension of the ebook.
+     * Basename of the ebook, e.g. `The Clan of the Cave Bear.epub`.
+     */
+    public function getBasename(): string
+    {
+        return $this->basename;
+    }
+
+    /**
+     * Extension of the ebook, e.g. `epub`.
      */
     public function getExtension(): string
     {
         return $this->extension;
     }
 
     /**
-     * Archive reader.
+     * Archive reader, from `kiwilan/php-archive`.
+     *
+     * @docs https://github.com/kiwilan/php-archive
      */
     public function getArchive(): ?BaseArchive
     {
+        // if (! $this->archive) {
+        //     error_log("{$this->path} can't be read as archive.");
+        //     throw new \Exception("{$this->path} can't be read as archive.");
+        // }
+
         return $this->archive;
     }
 
     /**
-     * Audio reader.
+     * Audio reader, from `kiwilan/php-audio`.
+     *
+     * @docs https://github.com/kiwilan/php-audio
      */
     public function getAudio(): ?Audio
     {
@@ -601,6 +658,13 @@ public function setDescription(?string $description): self
         return $this;
     }
 
+    public function setDescriptionHtml(?string $descriptionHtml): self
+    {
+        $this->descriptionHtml = $descriptionHtml;
+
+        return $this;
+    }
+
     public function setPublisher(?string $publisher): self
     {
         $this->publisher = $publisher;
@@ -705,6 +769,7 @@ public function toArray(): array
             'authorMain' => $this->authorMain?->getName(),
             'authors' => array_map(fn (BookAuthor $author) => $author->getName(), $this->authors),
             'description' => $this->description,
+            'descriptionHtml' => $this->descriptionHtml,
             'publisher' => $this->publisher,
             'identifiers' => array_map(fn (BookIdentifier $identifier) => $identifier->toArray(), $this->identifiers),
             'date' => $this->publishDate?->format('Y-m-d H:i:s'),
@@ -716,6 +781,7 @@ public function toArray(): array
             'pagesCount' => $this->pagesCount,
             'path' => $this->path,
             'filename' => $this->filename,
+            'basename' => $this->basename,
             'extension' => $this->extension,
             'format' => $this->format,
             'metadata' => $this->metadata?->toArray(),

diff --git a/src/Formats/EbookModule.php b/src/Formats/EbookModule.php
@@ -22,6 +22,59 @@ abstract public function toCounts(): Ebook;
 
     abstract public function toArray(): array;
 
+    /**
+     * Convert HTML to string, remove all tags.
+     */
+    protected function htmlToString(?string $html): ?string
+    {
+        if (! $html) {
+            return null;
+        }
+
+        $html = strip_tags($html);
+        $html = $this->formatText($html);
+
+        return $html;
+    }
+
+    /**
+     * Sanitize HTML, remove all tags except div, p, br, b, i, u, strong, em.
+     */
+    protected function sanitizeHtml(?string $html): ?string
+    {
+        if (! $html) {
+            return null;
+        }
+
+        $html = strip_tags($html, [
+            'div',
+            'p',
+            'br',
+            'b',
+            'i',
+            'u',
+            'strong',
+            'em',
+        ]);
+        $html = $this->formatText($html);
+
+        return $html;
+    }
+
+    /**
+     * Clean string, remove tabs, new lines, carriage returns, and multiple spaces.
+     */
+    private function formatText(string $text): string
+    {
+        $text = str_replace("\n", '', $text);
+        $text = str_replace("\r", '', $text);
+        $text = str_replace("\t", '', $text);
+        $text = trim($text);
+        $text = preg_replace('/\s+/', ' ', $text);
+
+        return $text;
+    }
+
     public function toJson(): string
     {
         return json_encode($this->toArray(), JSON_PRETTY_PRINT);

diff --git a/src/Formats/Epub/EpubMetadata.php b/src/Formats/Epub/EpubMetadata.php
@@ -75,9 +75,8 @@ public function toEbook(): Ebook
 
         $authors = array_values($this->opf->getDcCreators());
         $this->ebook->setAuthors($authors);
-        if ($this->opf->getDcDescription()) {
-            $this->ebook->setDescription(strip_tags($this->opf->getDcDescription()));
-        }
+        $this->ebook->setDescription($this->htmlToString($this->opf->getDcDescription()));
+        $this->ebook->setDescriptionHtml($this->sanitizeHtml($this->opf->getDcDescription()));
         $this->ebook->setCopyright(! empty($this->opf->getDcRights()) ? implode(', ', $this->opf->getDcRights()) : null);
         $this->ebook->setPublisher($this->opf->getDcPublisher());
         $this->ebook->setIdentifiers($this->opf->getDcIdentifiers());
@@ -97,14 +96,14 @@ public function toEbook(): Ebook
         $rating = null;
         if (! empty($this->opf->getMeta())) {
             foreach ($this->opf->getMeta() as $meta) {
-                if ($meta->name() === 'calibre:series') {
-                    $this->ebook->setSeries($meta->content());
+                if ($meta->getName() === 'calibre:series') {
+                    $this->ebook->setSeries($meta->getContent());
                 }
-                if ($meta->name() === 'calibre:series_index') {
-                    $this->ebook->setVolume((int) $meta->content());
+                if ($meta->getName() === 'calibre:series_index') {
+                    $this->ebook->setVolume((int) $meta->getContent());
                 }
-                if ($meta->name() === 'calibre:rating') {
-                    $rating = (float) $meta->content();
+                if ($meta->getName() === 'calibre:rating') {
+                    $rating = (float) $meta->getContent();
                 }
             }
         }