[Backport to 3.1.x #7105] Remove HTML tags from metadata detail

geonode/base/models.py

@@ -22,6 +22,7 @@
 import re
 import math
 import uuid
+import html
 import logging
 import traceback
 
@@ -40,7 +41,7 @@
 from django.contrib.contenttypes.models import ContentType
 from django.contrib.staticfiles.templatetags import staticfiles
 from django.core.files.storage import default_storage as storage
-
+from django.utils.html import strip_tags
 from mptt.models import MPTTModel, TreeForeignKey
 
 from PIL import Image, ImageOps
@@ -896,11 +897,16 @@ def __str__(self):
         return "{0}".format(self.title)
 
     def _remove_html_tags(self, attribute_str):
+        _attribute_str = attribute_str
         try:
             pattern = re.compile('<.*?>')
-            return re.sub(pattern, '', attribute_str)
+            _attribute_str = html.unescape(
+                re.sub(pattern, '', attribute_str).replace('\n', ' ').replace('\r', '').strip())
         except Exception:
-            return attribute_str
+            if attribute_str:
+                _attribute_str = html.unescape(
+                    attribute_str.replace('\n', ' ').replace('\r', '').strip())
+        return strip_tags(_attribute_str)
 
     @property
     def raw_abstract(self):

geonode/base/tests.py

@@ -905,6 +905,19 @@ def test_complex_tags_in_attribute(self):
         filtered_value = r._remove_html_tags(tagged_value)
         self.assertEqual(filtered_value, attribute_target_value)
 
+    def test_converted_html_in_tags_with_char_references(self):
+        tagged_value = """<p>&lt;p&gt;Abstract value &amp; some text&lt;/p&gt;</p>"""
+        attribute_target_value = """Abstract value & some text"""
+        r = ResourceBase()
+        filtered_value = r._remove_html_tags(tagged_value)
+        self.assertEqual(filtered_value, attribute_target_value)
+
+    def test_converted_html_in_tags_with_with_multiple_tags(self):
+        tagged_value = """<p><p><p><p>Abstract value &amp; some text</p></p></p></p>"""
+        attribute_target_value = """Abstract value & some text"""
+        r = ResourceBase()
+        filtered_value = r._remove_html_tags(tagged_value)
+        self.assertEqual(filtered_value, attribute_target_value)
 
 class TestTagThesaurus(TestCase):
     #  loading test thesausurs

geonode/catalogue/templates/geonode_metadata_full.html

@@ -59,7 +59,7 @@ <h2 class="page-title">{{ resource.title }}</h2>
 							<dd>{{resource.date}}, {% trans resource.date_type|title %} </dd>
 
 						<dt>{% trans "Abstract" %}</dt>
-							<dd>{{resource.abstract}}</dd>
+							<dd>{{resource.raw_abstract}}</dd>
 
 						<dt>{% trans "Edition" %}</dt>
 							{% if resource.edition %}
@@ -79,7 +79,7 @@ <h2 class="page-title">{{ resource.title }}</h2>
 
 						<dt>{% trans "Purpose" %}</dt>
 								{% if resource.purpose %}
-									<dd>{{resource.purpose}}</dd>
+									<dd>{{resource.raw_purpose}}</dd>
 								{% else %}
 									<dd> -- </dd>
 								{% endif %}
@@ -93,7 +93,7 @@ <h2 class="page-title">{{ resource.title }}</h2>
 
 						<dt>{% trans "Restrictions" %}</dt>
 							<dd>{{resource.restriction_code_type}}</dd>
-							<dd>{{resource.constraints_other}}</dd>
+							<dd>{{resource.raw_constraints_other}}</dd>
 
 						<dt>{% trans "License" %}</dt>
 							<dd>{{resource.license}}</dd>
@@ -117,11 +117,11 @@ <h2 class="page-title">{{ resource.title }}</h2>
 
 
 						<dt>{% trans "Supplemental Information" %}</dt>
-							<dd>{{resource.supplemental_information}}</dd>
+							<dd>{{resource.raw_supplemental_information}}</dd>
 
 						<dt>{% trans "Data Quality" %} </dt>
 							{% if resource.data_quality_statement %}
-							<dd>{{resource.data_quality_statement}}</dd>
+							<dd>{{resource.raw_data_quality_statement}}</dd>
 							{% else %}
 								<dd> -- </dd>
 							{% endif %}