From a82bb5342954655f681253ea0aacda36568d3d60 Mon Sep 17 00:00:00 2001
From: Marien Fressinaud <dev@marienfressinaud.fr>
Date: Thu, 3 Aug 2023 22:28:29 +0200
Subject: [PATCH] fix: Handle Response::utf8Data with unsupported encodings

Sometimes, a response declares an unsupported encoding.
It might be not very handy to test supported encodings as
mb_list_encodings don't return all the supported encodings.

My current solution is "try and see". If the conversion fails, force the
conversion from UTF-8 (and replace unsupported characters by "?").
---
 lib/SpiderBits/src/Response.php       |  7 +++--
 tests/lib/SpiderBits/ResponseTest.php | 38 +++++++++++++++++++++++++++
 2 files changed, 43 insertions(+), 2 deletions(-)

diff --git a/lib/SpiderBits/src/Response.php b/lib/SpiderBits/src/Response.php
index 3e737417..7e07bef2 100644
--- a/lib/SpiderBits/src/Response.php
+++ b/lib/SpiderBits/src/Response.php
@@ -138,8 +138,11 @@ public function encoding(): string
      */
     public function utf8Data(): string
     {
-        $encoding = $this->encoding();
-        return mb_convert_encoding($this->data, 'utf-8', $encoding);
+        try {
+            return mb_convert_encoding($this->data, 'utf-8', $this->encoding());
+        } catch (\ValueError $exception) {
+            return mb_convert_encoding($this->data, 'utf-8', 'utf-8');
+        }
     }
 
     /**
diff --git a/tests/lib/SpiderBits/ResponseTest.php b/tests/lib/SpiderBits/ResponseTest.php
index 3a871e92..e0a5c4dd 100644
--- a/tests/lib/SpiderBits/ResponseTest.php
+++ b/tests/lib/SpiderBits/ResponseTest.php
@@ -178,6 +178,44 @@ public function testHeaderReturnsDefaultValueIfMissing(): void
         $this->assertSame('text/html', $header);
     }
 
+    public function testUtf8Data(): void
+    {
+        $content = 'Test ëéàçï';
+        $content = mb_convert_encoding($content, 'ISO-8859-1', 'UTF-8');
+        $text = <<<TEXT
+        HTTP/2 200 OK\r
+        Content-Type: text/plain; charset="ISO-8859-1"\r
+        \r
+        {$content}
+        TEXT;
+        $response = Response::fromText($text);
+        $encoding = $response->encoding();
+        $this->assertSame('ISO-8859-1', $encoding);
+
+        $data = $response->utf8Data();
+
+        $this->assertSame('Test ëéàçï', $data);
+    }
+
+    public function testUtf8DataWithUnsupportedEncoding(): void
+    {
+        $content = 'Test ëéàçï';
+        $content = mb_convert_encoding($content, 'ISO-8859-1', 'UTF-8');
+        $text = <<<TEXT
+        HTTP/2 200 OK\r
+        Content-Type: text/plain; charset="Bad-Encoding"\r
+        \r
+        {$content}
+        TEXT;
+        $response = Response::fromText($text);
+        $encoding = $response->encoding();
+        $this->assertSame('Bad-Encoding', $encoding);
+
+        $data = $response->utf8Data();
+
+        $this->assertSame('Test ?????', $data);
+    }
+
     public function testEncodingWithNoSpecifiedEncoding(): void
     {
         $text = <<<TEXT