Add custom ExcerptCreator and use it within the API reponses

integreat_cms/api/v3/events.py

@@ -6,6 +6,7 @@
 from django.http import JsonResponse
 from django.utils import timezone
 
+from .utils import ExcerptGenerator
 from ...cms.models.events.event_translation import EventTranslation
 from ...cms.utils.slug_utils import generate_unique_slug
 from ..decorators import json_response
@@ -46,6 +47,8 @@ def transform_event_translation(event_translation):
     """
 
     event = event_translation.event
+    excerpt_generator = ExcerptGenerator()
+    excerpt_generator.feed(event_translation.content)
     if event.location:
         location_translation = (
             event.location.get_public_translation(event_translation.language.slug)
@@ -60,7 +63,7 @@ def transform_event_translation(event_translation):
         "path": event_translation.get_absolute_url(),
         "title": event_translation.title,
         "modified_gmt": event_translation.last_updated.strftime("%Y-%m-%d %H:%M:%S"),
-        "excerpt": event_translation.content,
+        "excerpt": excerpt_generator.text,
         "content": event_translation.content,
         "available_languages": event_translation.available_languages,
         "thumbnail": event.icon.url if event.icon else None,

integreat_cms/api/v3/imprint.py

@@ -6,6 +6,7 @@
 from django.conf import settings
 from django.http import JsonResponse
 
+from .utils import ExcerptGenerator
 from ..decorators import json_response
 
 logger = logging.getLogger(__name__)
@@ -21,14 +22,16 @@ def transform_imprint(imprint_translation):
     :return: data necessary for API
     :rtype: dict
     """
+    excerpt_generator = ExcerptGenerator()
+    excerpt_generator.feed(imprint_translation.content)
     absolute_url = imprint_translation.get_absolute_url()
     return {
         "id": imprint_translation.id,
         "url": settings.BASE_URL + absolute_url,
         "path": absolute_url,
         "title": imprint_translation.title,
         "modified_gmt": imprint_translation.last_updated,
-        "excerpt": imprint_translation.content,
+        "excerpt": excerpt_generator.text,
         "content": imprint_translation.content,
         "parent": None,
         "available_languages": imprint_translation.available_languages,

integreat_cms/api/v3/pages.py

@@ -9,6 +9,7 @@
 from django.shortcuts import get_object_or_404
 from django.views.decorators.csrf import csrf_exempt
 
+from .utils import ExcerptGenerator
 from ...cms.models import Page
 from ...cms.forms import PageTranslationForm
 from ..decorators import json_response, matomo_tracking
@@ -61,14 +62,16 @@ def transform_page(page_translation):
         # use tree id of mptt model for ordering of root pages
         order = page_translation.page.tree_id
 
+    excerpt_generator = ExcerptGenerator()
+    excerpt_generator.feed(page_translation.combined_text)
     absolute_url = page_translation.get_absolute_url()
     return {
         "id": page_translation.id,
         "url": settings.BASE_URL + absolute_url,
         "path": absolute_url,
         "title": page_translation.title,
         "modified_gmt": page_translation.combined_last_updated,
-        "excerpt": page_translation.content,
+        "excerpt": excerpt_generator.text,
         "content": page_translation.combined_text,
         "parent": parent,
         "order": order,

integreat_cms/api/v3/utils.py

@@ -0,0 +1,32 @@
+"""
+This module includes some utility classes and functions used within the API module.
+"""
+import logging
+
+from html.parser import HTMLParser
+
+logger = logging.getLogger(__name__)
+
+
+class ExcerptGenerator(HTMLParser):
+    """
+    Class that can be used to convert a string containing HTML into a pure text string.
+    """
+
+    text = ""
+
+    def handle_data(self, data):
+        """
+        Function that overwrites the HTMLParser function and returns a a pure text representations.
+
+        :param data: The original HTML content.
+        :type data: str
+        """
+        self.text += data
+
+    def error(self, message):
+        logger.warning(
+            "The ExcerptGenerator couldn't create a pure text for %s and returns the following error message: %s",
+            self.text,
+            message,
+        )