Generate links images previews

.gitignore

@@ -12,3 +12,6 @@
 !/policies/.keep
 
 /public/dev_assets
+
+/public/media/*
+!/public/media/.keep

configuration/environment_development.php

@@ -23,6 +23,7 @@
         'brand' => $dotenv->pop('APP_BRAND', 'flusio'),
         'version' => trim(@file_get_contents($app_path . '/VERSION.txt')),
         'cache_path' => $dotenv->pop('APP_CACHE_PATH', $app_path . '/cache'),
+        'media_path' => $dotenv->pop('APP_MEDIA_PATH', $app_path . '/public/media'),
         'demo' => filter_var($dotenv->pop('APP_DEMO', false), FILTER_VALIDATE_BOOLEAN),
         'registrations_opened' => filter_var($dotenv->pop('APP_OPEN_REGISTRATIONS', true), FILTER_VALIDATE_BOOLEAN),
         'subscriptions_enabled' => $subscriptions_host !== null,

configuration/environment_production.php

@@ -24,6 +24,7 @@
         'brand' => $dotenv->pop('APP_BRAND', 'flusio'),
         'version' => trim(@file_get_contents($app_path . '/VERSION.txt')),
         'cache_path' => $dotenv->pop('APP_CACHE_PATH', $app_path . '/cache'),
+        'media_path' => $dotenv->pop('APP_MEDIA_PATH', $app_path . '/public/media'),
         'demo' => filter_var($dotenv->pop('APP_DEMO', false), FILTER_VALIDATE_BOOLEAN),
         'registrations_opened' => filter_var($dotenv->pop('APP_OPEN_REGISTRATIONS', true), FILTER_VALIDATE_BOOLEAN),
         'subscriptions_enabled' => $subscriptions_host !== null,

configuration/environment_test.php

@@ -7,9 +7,11 @@
 $temporary_directory = sys_get_temp_dir() . '/flusio/' . \flusio\utils\Random::hex(10);
 $data_directory = $temporary_directory . '/data';
 $cache_directory = $temporary_directory . '/cache';
+$media_directory = $temporary_directory . '/media';
 @mkdir($temporary_directory, 0777, true);
 @mkdir($data_directory, 0777, true);
 @mkdir($cache_directory, 0777, true);
+@mkdir($media_directory, 0777, true);
 
 $subscriptions_host = $dotenv->pop('APP_SUBSCRIPTIONS_HOST');
 
@@ -29,6 +31,7 @@
         'brand' => 'flusio',
         'version' => trim(@file_get_contents($app_path . '/VERSION.txt')),
         'cache_path' => $cache_directory,
+        'media_path' => $media_directory,
         'demo' => false,
         'registrations_opened' => true,
         'subscriptions_enabled' => false, // should be enable on a case-by-case basis

env.sample

@@ -23,6 +23,11 @@ APP_PORT=8000
 # directory presents in the current directory.
 # APP_CACHE_PATH=/tmp
 
+# You can uncomment and change the path for the media files. Default is the
+# media directory presents in the public/ directory. If you change this value,
+# make sure the /media URL path is correctly served by your Web server.
+# APP_MEDIA_PATH=/some/path/to/media
+
 ##################################
 # Database environment variables #
 ##################################

lib/SpiderBits/src/DomExtractor.php

@@ -46,6 +46,37 @@ public static function title($dom)
         return '';
     }
 
+    /**
+     * Return the illustration URL of the DOM document.
+     *
+     * @param \SpiderBits\Dom $dom
+     *
+     * @return string
+     */
+    public static function illustration($dom)
+    {
+        $xpath_queries = [
+            // Look for OpenGraph first
+            '/html/head/meta[@property = "og:image"][1]/attribute::content',
+            // Then Twitter meta tag
+            '/html/head/meta[@name = "twitter:image"][1]/attribute::content',
+            // Err, still nothing! Let's try to be more tolerant (e.g. Youtube
+            // puts the meta tags in the body :/)
+            '//meta[@property = "og:image"][1]/attribute::content',
+            '//meta[@name = "twitter:image"][1]/attribute::content',
+        ];
+
+        foreach ($xpath_queries as $query) {
+            $illustration = $dom->select($query);
+            if ($illustration) {
+                return $illustration->text();
+            }
+        }
+
+        // It's hopeless...
+        return '';
+    }
+
     /**
      * Return the main content of the DOM document.
      *

src/Links.php

@@ -472,6 +472,11 @@ public function fetch($request)
         if (isset($info['reading_time'])) {
             $link->reading_time = $info['reading_time'];
         }
+        if (isset($info['url_illustration'])) {
+            $image_service = new services\Image();
+            $image_filename = $image_service->generatePreviews($info['url_illustration']);
+            $link->image_filename = $image_filename;
+        }
 
         $link_dao = new models\dao\Link();
         $link_dao->save($link);

src/migrations/Migration202010220001AddImageFilenameToLinks.php

@@ -0,0 +1,30 @@
+<?php
+
+namespace flusio\migrations;
+
+class Migration202010220001AddImageFilenameToLinks
+{
+    public function migrate()
+    {
+        $database = \Minz\Database::get();
+
+        $database->exec(<<<'SQL'
+            ALTER TABLE links
+            ADD COLUMN image_filename TEXT;
+        SQL);
+
+        return true;
+    }
+
+    public function rollback()
+    {
+        $database = \Minz\Database::get();
+
+        $database->exec(<<<'SQL'
+            ALTER TABLE links
+            DROP COLUMN image_filename;
+        SQL);
+
+        return true;
+    }
+}

src/models/Image.php

@@ -0,0 +1,203 @@
+<?php
+
+namespace flusio\models;
+
+/**
+ * @author  Marien Fressinaud <[email protected]>
+ * @license http://www.gnu.org/licenses/agpl-3.0.en.html AGPL
+ */
+class Image
+{
+    /** @var resource */
+    private $resource;
+
+    /** @var integer */
+    private $width;
+
+    /** @var integer */
+    private $height;
+
+    /** @var string */
+    private $type;
+
+    /**
+     * @param resource $resource The image as GD resource
+     * @param string $type jpeg, png or webp
+     */
+    public function __construct($resource, $type)
+    {
+        $this->resource = $resource;
+        $this->type = $type;
+        $this->width = imagesx($resource);
+        $this->height = imagesy($resource);
+    }
+
+    /**
+     * Initialize an image from a string
+     *
+     * @param string $string_image
+     *
+     * @return \flusio\models\Image
+     */
+    public static function fromString($string_image)
+    {
+        $mime = finfo_buffer(finfo_open(FILEINFO_MIME_TYPE), $string_image);
+        switch (strtolower($mime)) {
+            case 'image/jpg':
+            case 'image/jpeg':
+            case 'image/pjpeg':
+                $type = 'jpeg';
+                break;
+            case 'image/png':
+            case 'image/x-png':
+                $type = 'png';
+                break;
+            case 'image/webp':
+            case 'image/x-webp':
+                $type = 'webp';
+                break;
+            default:
+                throw new \DomainException("The given string doesn’t look like a supported image ({$mime})");
+        }
+
+        $resource = @imagecreatefromstring($string_image);
+        if (!$resource) {
+            throw new \DomainException('Cannot create an image from the given string');
+        }
+
+        return new self($resource, $type);
+    }
+
+    /**
+     * @return string
+     */
+    public function type()
+    {
+        return $this->type;
+    }
+
+    /**
+     * Resize the current image to the given size.
+     *
+     * The image is cropped in the middle to keep the proportion of the image.
+     *
+     * @param integer $width
+     * @param integer $height
+     */
+    public function resize($width, $height)
+    {
+        $new_resource = imagecreatetruecolor($width, $height);
+
+        // Preserve transparency
+        // Code from the Intervention Image library
+        // @see https://github.com/Intervention/image/blob/8ee5f346ce8c6dcbdc7dec443486bd5f6ad924ff/src/Intervention/Image/Gd/Commands/ResizeCommand.php
+        $transparent_index = imagecolortransparent($this->resource);
+        if ($transparent_index !== -1) {
+            $rgba = imagecolorsforindex($new_resource, $transparent_index);
+            $transparent_color = imagecolorallocatealpha(
+                $new_resource,
+                $rgba['red'],
+                $rgba['green'],
+                $rgba['blue'],
+                127
+            );
+            imagefill($new_resource, 0, 0, $transparent_color);
+            imagecolortransparent($new_resource, $transparent_color);
+        } else {
+            imagealphablending($new_resource, false);
+            imagesavealpha($new_resource, true);
+        }
+
+        $src_rect = self::resizeRectangle(
+            [$this->width, $this->height],
+            [$width, $height]
+        );
+
+        imagecopyresampled(
+            $new_resource,
+            $this->resource,
+            0,
+            0,
+            $src_rect['x'],
+            $src_rect['y'],
+            $width,
+            $height,
+            $src_rect['width'],
+            $src_rect['height']
+        );
+
+        imagedestroy($this->resource);
+        $this->resource = $new_resource;
+        $this->width = $width;
+        $this->height = $height;
+    }
+
+    /**
+     * Save the image on disk.
+     *
+     * @param string $filepath
+     */
+    public function save($filepath)
+    {
+        switch ($this->type) {
+            case 'jpeg':
+                imagejpeg($this->resource, $filepath);
+                break;
+            case 'png':
+                imagepng($this->resource, $filepath);
+                break;
+            case 'webp':
+                imagewebp($this->resource, $filepath);
+                break;
+        }
+    }
+
+    /**
+     * Get the src rectangle to be used to resize the initial image
+     *
+     * This function return the bigger rectangle with destination ratio that
+     * fits in the initial rectangle.
+     *
+     * @see https://www.php.net/manual/function.imagecopyresampled
+     *
+     * @param array $initial_size
+     *     The size of the initial image (first value is width, second is height)
+     * @param array $destination_size
+     *     The size of the desired image (first value is width, second is height)
+     *
+     * @return array
+     *     The src rectange to use in imagecopyresampled, array indexes are:
+     *     x, y, width and height.
+     */
+    public static function resizeRectangle($initial_size, $destination_size)
+    {
+        list($initial_width, $initial_height) = $initial_size;
+        list($destination_width, $destination_height) = $destination_size;
+
+        $initial_ratio = $initial_width / $initial_height;
+        $destination_ratio = $destination_width / $destination_height;
+        if ($initial_ratio <= $destination_ratio) {
+            // The current width will entirely fit in the future image, so we
+            // can take the full width. We have to crop the height though
+            // (taking the middle of the image).
+            $src_width = $initial_width;
+            $src_height = $initial_width / $destination_ratio;
+            $src_x = 0;
+            $src_y = ($initial_height - $src_height) / 2;
+        } else {
+            // Same thing, but with height which entirely fit in the future
+            // image.
+            $src_width = $initial_height * $destination_ratio;
+            $src_height = $initial_height;
+            $src_x = ($initial_width - $src_width) / 2;
+            $src_y = 0;
+        }
+
+        return [
+            'x' => $src_x,
+            'y' => $src_y,
+            'width' => $src_width,
+            'height' => $src_height,
+        ];
+    }
+}

src/models/Link.php

@@ -39,6 +39,10 @@ class Link extends \Minz\Model
             'required' => true,
         ],
 
+        'image_filename' => [
+            'type' => 'string',
+        ],
+
         'fetched_at' => [
             'type' => 'datetime',
         ],

src/schema.sql

@@ -55,6 +55,7 @@ CREATE TABLE links (
     url TEXT NOT NULL,
     is_public BOOLEAN NOT NULL DEFAULT false,
     reading_time INTEGER NOT NULL DEFAULT 0,
+    image_filename TEXT,
     fetched_at TIMESTAMPTZ,
     fetched_code INTEGER NOT NULL DEFAULT 0,
     fetched_error TEXT,

src/services/Fetch.php

@@ -41,6 +41,7 @@ public function __construct()
      *     - error
      *     - title
      *     - reading_time
+     *     - url_illustration
      */
     public function fetch($url)
     {
@@ -87,6 +88,13 @@ public function fetch($url)
         $words = array_filter(explode(' ', $content));
         $info['reading_time'] = intval(count($words) / 200);
 
+        // Get the illustration URL if any
+        $url_illustration = \SpiderBits\DomExtractor::illustration($dom);
+        $url_illustration = \SpiderBits\Url::sanitize($url_illustration);
+        if ($url_illustration) {
+            $info['url_illustration'] = $url_illustration;
+        }
+
         return $info;
     }
 }