From e295044255624245c1b6a0bdf46e16920ed318e6 Mon Sep 17 00:00:00 2001
From: Maik Schneider <mas@xima.de>
Date: Fri, 6 Sep 2024 12:39:31 +0200
Subject: [PATCH 1/3] feat: adjust test fixtures for line breaks

Co-authored-by: Markus Hackel <mha@xima.de>
---
 Tests/Fixtures/TestMailTemplate.html | 4 +++-
 Tests/Fixtures/TestMailTemplate.txt  | 2 ++
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/Tests/Fixtures/TestMailTemplate.html b/Tests/Fixtures/TestMailTemplate.html
index bc9c364..5d1c750 100644
--- a/Tests/Fixtures/TestMailTemplate.html
+++ b/Tests/Fixtures/TestMailTemplate.html
@@ -1 +1,3 @@
-https://domain/sign-in/registration/confirmation?tx_getaccess_confirmation%5Baction%5D=user&tx_getaccess_confirmation%5Bcontroller%5D=Confirmation&tx_getaccess_confirmation%5Bhash%5D=2720ba93a3007066&cHash=3179e61306999063853b2247a1bd4d
+<h1>Hallo Welt!</h1><br />
+<br />
+<a href="#">https://domain/sign-in/registration/confirmation?tx_getaccess_confirmation%5Baction%5D=user&tx_getaccess_confirmation%5Bcontroller%5D=Confirmation&tx_getaccess_confirmation%5Bhash%5D=2720ba93a3007066&cHash=3179e61306999063853b2247a1bd4d</a>
diff --git a/Tests/Fixtures/TestMailTemplate.txt b/Tests/Fixtures/TestMailTemplate.txt
index bc9c364..898b333 100644
--- a/Tests/Fixtures/TestMailTemplate.txt
+++ b/Tests/Fixtures/TestMailTemplate.txt
@@ -1 +1,3 @@
+Hallo Welt!
+
 https://domain/sign-in/registration/confirmation?tx_getaccess_confirmation%5Baction%5D=user&tx_getaccess_confirmation%5Bcontroller%5D=Confirmation&tx_getaccess_confirmation%5Bhash%5D=2720ba93a3007066&cHash=3179e61306999063853b2247a1bd4d

From d9499ff641c27719ecaae46041ba4a1d1159c9be Mon Sep 17 00:00:00 2001
From: Maik Schneider <mas@xima.de>
Date: Fri, 6 Sep 2024 12:39:57 +0200
Subject: [PATCH 2/3] feat: add plain text-only and html-only mails to test

Co-authored-by: Markus Hackel <mha@xima.de>
---
 .../Utility/LogParserUtilityTest.php          | 20 +++++++++++++++++--
 1 file changed, 18 insertions(+), 2 deletions(-)

diff --git a/Tests/Functional/Utility/LogParserUtilityTest.php b/Tests/Functional/Utility/LogParserUtilityTest.php
index 841734e..37fbd9c 100644
--- a/Tests/Functional/Utility/LogParserUtilityTest.php
+++ b/Tests/Functional/Utility/LogParserUtilityTest.php
@@ -64,21 +64,37 @@ public static function mailDataProvider(): array
             'format' => FluidEmail::FORMAT_BOTH,
         ];
 
+        $htmlOnlyMail = $defaultMail;
+        $htmlOnlyMail['format'] = FluidEmail::FORMAT_HTML;
+
+        $plainOnlyMail = $defaultMail;
+        $plainOnlyMail['format'] = FluidEmail::FORMAT_PLAIN;
+
         return [
             [
                 [$defaultMail],
             ],
             [
                 [
+                    $plainOnlyMail,
+                    $plainOnlyMail,
                     $defaultMail,
-                    $defaultMail,
+                    $plainOnlyMail,
                 ],
             ],
             [
                 [
+                    $htmlOnlyMail,
+                    $htmlOnlyMail,
                     $defaultMail,
+                    $htmlOnlyMail,
                 ],
             ],
+            [
+                [$defaultMail],
+                [$htmlOnlyMail],
+                [$plainOnlyMail],
+            ],
         ];
     }
 
@@ -89,7 +105,7 @@ public static function mailDataProvider(): array
      */
     public function testEmailEncoding(array $exampleMails): void
     {
-        foreach ($exampleMails as $key => $exampleMail) {
+        foreach ($exampleMails as $exampleMail) {
             $this->createTestMail($exampleMail);
         }
 

From eaea0d6a67b91c87e14052c0a50e9a922d308532 Mon Sep 17 00:00:00 2001
From: Maik Schneider <mas@xima.de>
Date: Fri, 6 Sep 2024 12:40:10 +0200
Subject: [PATCH 3/3] feat: replace email splitting by boundary with preg_split
 by header key words

Co-authored-by: Markus Hackel <mha@xima.de>
---
 Classes/Utility/LogParserUtility.php | 27 ++++++++-------------------
 1 file changed, 8 insertions(+), 19 deletions(-)

diff --git a/Classes/Utility/LogParserUtility.php b/Classes/Utility/LogParserUtility.php
index 97ab946..40e2db9 100644
--- a/Classes/Utility/LogParserUtility.php
+++ b/Classes/Utility/LogParserUtility.php
@@ -51,26 +51,11 @@ public function extractMessages(): void
             return;
         }
 
-        preg_match_all(
-            '/(?:; boundary=)(.+)(?:\r\n)/Ums',
-            $this->fileContent,
-            $boundaries
-        );
+        $mails = preg_split('/(?=(To:\s(.+)\nFrom:\s(.+)\nSubject:\s))/Ums', $this->fileContent);
+        $mails = array_filter($mails);
 
-        if (!isset($boundaries[1])) {
-            return;
-        }
-
-        foreach ($boundaries[1] as $boundary) {
-            $separator = '--' . $boundary . '--';
-            $messageParts = explode($separator, $this->fileContent);
-
-            if (!str_contains($messageParts[0], 'boundary=')) {
-                continue;
-            }
-
-            $messageString = trim($messageParts[0]);
-            $this->fileContent = $messageParts[1] ?: '';
+        foreach ($mails as $mailContent) {
+            $messageString = trim($mailContent);
             $this->messages[] = self::convertToDto((string)$messageString);
         }
     }
@@ -131,6 +116,10 @@ protected static function convertToDto(string $msg): MailMessage
         $dto->bodyPlain = @mb_convert_encoding($message->getTextContent() ?? '', 'UTF-8', 'auto');
         $dto->bodyHtml = @mb_convert_encoding($message->getHtmlContent() ?? '', 'UTF-8', 'auto');
 
+        // remove "\r\n" from content, replace double line breaks with single line break and remove trailing whitespaces
+        $dto->bodyPlain = preg_replace(['/\r\n\r\n/', '/\s$/'], [' ', ''], $dto->bodyPlain);
+        $dto->bodyHtml = preg_replace(['/\r\n/', '/\s$/'], [' ', ''], $dto->bodyHtml);
+
         $folder = self::getTempPath() . $dto->messageId;
         if (!file_exists($folder)) {
             GeneralUtility::mkdir($folder);