HTML-to-Text conversion removed linebreaks and links (#35)

Author: GitRon

CHANGES.md

@@ -1,5 +1,8 @@
 # Changelog
 
+**2.7.1** (2025-05-29)
+  * Fixed a bug where html-converted plain text would miss links
+
 **2.7.0** (2025-05-21)
   * Replaced `html2text` with `beautifulsoup4` to make package available under MIT license
   * Switched back to MIT license
@@ -13,7 +16,7 @@
   * Maintenance updates via ambient-package-update
 
 * *2.6.1* (2025-03-18)
-  * Fixed a bug, where translations were not deactivated after sending an email
+  * Fixed a bug where translations were not deactivated after sending an email
 
 * *2.6.0* (2025-03-17)
   * Added check for email structure validity

django_pony_express/__init__.py

@@ -1,3 +1,3 @@
 """Class-based emails including a test suite for Django"""
 
-__version__ = "2.7.0"
+__version__ = "2.7.1"

django_pony_express/services/base.py

@@ -242,8 +242,15 @@ def _generate_html_content(self, mail_attributes: dict) -> str:
     def _generate_text_content(self, mail_attributes: dict, html_content: str) -> str:
         # Render TXT body part if a template is explicitly set, otherwise convert HTML template to plain text
         if not self.template_txt_name:
-            soup = BeautifulSoup(html_content)
-            return soup.get_text(strip=True)
+            soup = BeautifulSoup(html_content, "html.parser")
+
+            # Convert links to this pattern: "[LINK NAME] ([LINK URL])"
+            for a in soup.find_all("a"):
+                text = a.get_text()
+                href = a.get("href", "")
+                a.replace_with(f"{text} ({href})")
+
+            return soup.get_text(separator="\n", strip=True)
         else:
             return render_to_string(self.template_txt_name, mail_attributes)
 

testapp/templates/testapp/test_email.html

@@ -3,6 +3,9 @@
 <div>
     Current date test: {% now "l, d. F Y" %}
 </div>
+<div>
+    <a href="{{ link_url }}">I am a link</a>
+</div>
 <div>
     Variable test: {{ my_var }}
 </div>

tests/services/base/test_base_mail_service.py

@@ -475,23 +475,29 @@ def test_text_templates_rendering(self):
         my_var = "Lorem ipsum dolor!"
         service = BaseEmailService()
         service.template_txt_name = "testapp/test_email.txt"
-        msg_html = service._generate_text_content({"my_var": my_var}, "")
+        msg_text = service._generate_text_content({"my_var": my_var}, "")
 
         # Assertions
-        self.assertIsInstance(msg_html, str)
+        self.assertIsInstance(msg_text, str)
 
-        self.assertIn("Lorem ipsum dolor", msg_html)
-        self.assertIn("I am a different content", msg_html)
-        self.assertNotIn("Current date test", msg_html)
+        self.assertIn("Lorem ipsum dolor", msg_text)
+        self.assertIn("I am a different content", msg_text)
+        self.assertNotIn("Current date test", msg_text)
 
-    def test_text_templates_rendering_fallback(self):
+    def test_generate_text_content_html_conversion(self):
         my_var = "Lorem ipsum dolor!"
         service = BaseEmailService()
-        msg_html = service._generate_text_content({"my_var": my_var}, "Lorem ipsum dolor")
+        service.template_name = "testapp/test_email.html"
+
+        msg_html = service._generate_html_content({"my_var": my_var, "link_url": "https//example.com?pony=horse"})
+        msg_txt = service._generate_text_content(mail_attributes={}, html_content=msg_html)
 
         # Assertions
         self.assertIsInstance(msg_html, str)
 
-        self.assertIn("Lorem ipsum dolor", msg_html)
-        self.assertNotIn("I am a different content", msg_html)
-        self.assertNotIn("Current date test", msg_html)
+        self.assertIn("Lorem ipsum dolor", msg_txt)
+        # Check linebreaks are not removed
+        self.assertIn("29. May 2025\n", msg_txt)
+
+        # Check links are converted correctly
+        self.assertIn("I am a link (https//example.com?pony=horse)", msg_txt)