From 37b078013e1be34e86d99f122175b430da87415e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20Bar=C3=A1=C5=A1ek?= <janbarasek@gmail.com>
Date: Thu, 14 Oct 2021 09:41:52 +0200
Subject: [PATCH] ImageStorage: Use strict policy.

---
 src/ImageStorage.php | 41 ++++++++++++++++++++++++++++++++++++++---
 1 file changed, 38 insertions(+), 3 deletions(-)

diff --git a/src/ImageStorage.php b/src/ImageStorage.php
index aec33e3..32e47f5 100644
--- a/src/ImageStorage.php
+++ b/src/ImageStorage.php
@@ -12,6 +12,8 @@
 
 final class ImageStorage
 {
+	public const IMAGE_MIME_TYPES = ['image/gif', 'image/png', 'image/jpeg', 'image/webp'];
+
 	private string $storagePath;
 
 	private string $relativeStoragePath;
@@ -33,29 +35,55 @@ public function __construct(?string $storagePath = null, string $relativeStorage
 	}
 
 
+	/**
+	 * The method downloads the image from the physical URL and saves it to the internal storage.
+	 * The download is checked to ensure that it is a valid data type for the image.
+	 */
 	public function save(string $url): void
 	{
 		if (Validators::isUrl($url) === false) {
 			throw new \InvalidArgumentException('Given input is not valid absolute URL, because "' . $url . '" given.');
 		}
 		$storagePath = $this->getInternalPath($url);
-		if (\is_file($storagePath) === false) {
-			FileSystem::copy($url, $storagePath);
+		if (is_file($storagePath) === false) { // image does not exist in local storage
+			$content = FileSystem::read($url); // download image
+			$contentType = strtolower(finfo_file(finfo_open(FILEINFO_MIME_TYPE), $content));
+			if (in_array($contentType, self::IMAGE_MIME_TYPES, true) === false) {
+				throw new \RuntimeException(
+					'Security issue: Downloaded file "' . $url . '" is not valid image, '
+					. 'because content type "' . $contentType . '" has been detected.',
+				);
+			}
+			FileSystem::write($storagePath, $content);
 		}
 	}
 
 
+	/**
+	 * Returns the absolute path for the internal data store.
+	 * Specifies the absolute physical disk path where the image will be written to (or read from) based on the URL.
+	 * The disk path is calculated by a deterministic algorithm for future content readability.
+	 */
 	public function getInternalPath(string $url): string
 	{
 		return $this->storagePath . '/' . $this->getRelativeInternalUrl($url);
 	}
 
 
+	/**
+	 * Returns the physical URL to download the image directly to local disk storage.
+	 */
 	public function getAbsoluteInternalUrl(string $url): string
 	{
 		try {
 			$baseUrl = Url::get()->getBaseUrl();
 		} catch (\Throwable) {
+			if (PHP_SAPI === 'cli') {
+				throw new \LogicException(
+					__METHOD__ . ': Absolute URL is not available in CLI. '
+					. 'Did you set context URL to "' . Url::class . '" service?',
+				);
+			}
 			$baseUrl = '';
 		}
 
@@ -63,6 +91,9 @@ public function getAbsoluteInternalUrl(string $url): string
 	}
 
 
+	/**
+	 * Returns the relative URL to retrieve the image.
+	 */
 	public function getRelativeInternalUrl(string $url): string
 	{
 		$originalFileName = (string) preg_replace_callback(
@@ -76,10 +107,14 @@ public function getRelativeInternalUrl(string $url): string
 	}
 
 
+	/**
+	 * Generate an automatic unique hash based on the image URL
+	 * so that file names from many different sources cannot collide.
+	 */
 	private function resolvePrefixDir(string $url): string
 	{
 		if ($url === '') {
-			throw new \InvalidArgumentException('URL can not be empty string.');
+			throw new \LogicException('URL can not be empty string.');
 		}
 		if (preg_match('/wp-content.+(\d{4})\/(\d{2})/', $url, $urlParser)) {
 			return $urlParser[1] . '-' . $urlParser[2];