Method to get all html parts and complete html body (#557)

Co-authored-by: Arwin <[email protected]>
ddeboer · Feb 17, 2023 · 392fa57 · 392fa57
1 parent 92647fd
commit 392fa57
Show file tree

Hide file tree

Showing 6 changed files with 336 additions and 12 deletions.
diff --git a/.gitignore b/.gitignore
@@ -7,3 +7,5 @@ composer.lock
 phpunit.xml
 wait-for-it
 coverage/
+
+.DS_Store
diff --git a/composer.json b/composer.json
@@ -23,17 +23,19 @@
     ],
     "require": {
         "php": "~8.1.0 || ~8.2.0",
+        "ext-dom": "*",
         "ext-iconv": "*",
         "ext-imap": "*",
+        "ext-libxml": "*",
         "ext-mbstring": "*"
     },
     "require-dev": {
-        "friendsofphp/php-cs-fixer": "^3.13.2",
-        "laminas/laminas-mail": "^2.21.1",
-        "phpstan/phpstan": "^1.9.8",
-        "phpstan/phpstan-phpunit": "^1.3.3",
-        "phpstan/phpstan-strict-rules": "^1.4.4",
-        "phpunit/phpunit": "^9.5.27"
+        "friendsofphp/php-cs-fixer": "^3.14.4",
+        "laminas/laminas-mail": "^2.22.0",
+        "phpstan/phpstan": "^1.9.17",
+        "phpstan/phpstan-phpunit": "^1.3.4",
+        "phpstan/phpstan-strict-rules": "^1.4.5",
+        "phpunit/phpunit": "^9.6.3"
     },
     "autoload": {
         "psr-4": {

diff --git a/src/Message/AbstractMessage.php b/src/Message/AbstractMessage.php
@@ -203,23 +203,75 @@ final public function getReferences(): array
     }
 
     /**
-     * Get body HTML.
+     * Get first body HTML part.
      */
     final public function getBodyHtml(): ?string
     {
-        $iterator = new \RecursiveIteratorIterator($this, \RecursiveIteratorIterator::SELF_FIRST);
+        $htmlParts = $this->getBodyHtmlParts();
+
+        return $htmlParts[0] ?? null;
+    }
+
+    /**
+     * Get body HTML parts.
+     *
+     * @return string[]
+     */
+    final public function getBodyHtmlParts(): array
+    {
+        $iterator  = new \RecursiveIteratorIterator($this, \RecursiveIteratorIterator::SELF_FIRST);
+        $htmlParts = [];
         foreach ($iterator as $part) {
             if (self::SUBTYPE_HTML === $part->getSubtype()) {
-                return $part->getDecodedContent();
+                $htmlParts[] = $part->getDecodedContent();
             }
         }
+        if (\count($htmlParts) > 0) {
+            return $htmlParts;
+        }
 
         // If message has no parts and is HTML, return content of message itself.
         if (self::SUBTYPE_HTML === $this->getSubtype()) {
-            return $this->getDecodedContent();
+            return [$this->getDecodedContent()];
         }
 
-        return null;
+        return [];
+    }
+
+    /**
+     * Get all body HTML parts merged into 1 html.
+     */
+    final public function getCompleteBodyHtml(): ?string
+    {
+        $htmlParts = $this->getBodyHtmlParts();
+
+        if (1 === \count($htmlParts)) {
+            return $htmlParts[0];
+        }
+        if (0 === \count($htmlParts)) {
+            return null;
+        }
+        \libxml_use_internal_errors(true); // Suppress parse errors, get errors with libxml_get_errors();
+
+        $newDom = new \DOMDocument();
+
+        $newBody = '';
+        $newDom->loadHTML(\implode('', $htmlParts));
+
+        $bodyTags = $newDom->getElementsByTagName('body');
+
+        foreach ($bodyTags as $body) {
+            foreach ($body->childNodes as $node) {
+                $newBody .= $newDom->saveHTML($node);
+            }
+        }
+
+        $newDom = new \DOMDocument();
+        $newDom->loadHTML($newBody);
+
+        $completeHtml = $newDom->saveHTML();
+
+        return false === $completeHtml ? null : $completeHtml;
     }
 
     /**

diff --git a/src/Message/BasicMessageInterface.php b/src/Message/BasicMessageInterface.php
@@ -116,12 +116,26 @@ public function getInReplyTo(): array;
     public function getReferences(): array;
 
     /**
-     * Get body HTML.
+     * Get first body HTML part.
      *
      * @return null|string Null if message has no HTML message part
      */
     public function getBodyHtml(): ?string;
 
+    /**
+     * Get all body HTML parts as array.
+     *
+     * @return string[]
+     */
+    public function getBodyHtmlParts(): array;
+
+    /**
+     * Get all body HTML parts merged into 1 html.
+     *
+     * @return null|string Null if message has no HTML message part
+     */
+    public function getCompleteBodyHtml(): ?string;
+
     /**
      * Get body text.
      */

diff --git a/tests/MessageTest.php b/tests/MessageTest.php
@@ -1029,6 +1029,57 @@ public function testNestesEmbeddedWithAttachment(): void
         }
     }
 
+    public function testMultipleHtmlParts(): void
+    {
+        $this->mailbox->addMessage($this->getFixture('multiple_html_parts_and_attachments'));
+
+        $message = $this->mailbox->getMessage(1);
+
+        // Test attachments
+        $expectedFileNames = [
+            'attachment1.pdf',
+            'attachment2.pdf',
+        ];
+        $attachments = $message->getAttachments();
+        static::assertCount(2, $attachments);
+        foreach ($attachments as $attachment) {
+            static::assertContains($attachment->getFilename(), $expectedFileNames);
+        }
+
+        // Test html parts
+        static::assertCount(3, $message->getBodyHtmlParts());
+
+        // Test html parts
+        $completeBody = $message->getCompleteBodyHtml();
+        $completeBody = null === $completeBody ? '' : $completeBody;
+
+        static::assertStringContainsString('first', $completeBody);
+        static::assertStringContainsString('second', $completeBody);
+        static::assertStringContainsString('last', $completeBody);
+    }
+
+    public function testBodyHtmlEmpty(): void
+    {
+        $this->mailbox->addMessage($this->getFixture('plain_only'));
+
+        $message = $this->mailbox->getMessage(1);
+
+        static::assertCount(0, $message->getBodyHtmlParts());
+
+        static::assertNull($message->getCompleteBodyHtml());
+    }
+
+    public function testBodyHtmlOnePart(): void
+    {
+        $this->mailbox->addMessage($this->getFixture('html_only'));
+
+        $message = $this->mailbox->getMessage(1);
+
+        static::assertCount(1, $message->getBodyHtmlParts());
+
+        static::assertNotNull($message->getCompleteBodyHtml());
+    }
+
     public function testImapMimeHeaderDecodeReturnsFalse(): void
     {
         $this->mailbox->addMessage($this->getFixture('imap_mime_header_decode_returns_false'));