form: HTML sanitization and remove Source from CKEditor

Author: zubeydecivelek

cds/modules/deposit/static/json/cds_deposit/forms/project.json

@@ -45,9 +45,6 @@
                   "Replace",
                   "-",
                   "RemoveFormat"
-                ],
-                [
-                  "Source"
                 ]
               ],
               "disableNativeSpellChecker": false,

cds/modules/deposit/static/json/cds_deposit/forms/video.json

@@ -44,9 +44,6 @@
                 "Replace",
                 "-",
                 "RemoveFormat"
-              ],
-              [
-                "Source"
               ]
             ],
             "disableNativeSpellChecker": false,

cds/modules/records/serializers/json.py

@@ -32,6 +32,7 @@
     has_read_record_permission,
 )
 from ..utils import HTMLTagRemover, remove_html_tags
+from marshmallow_utils.html import sanitize_html
 
 
 class CDSJSONSerializer(JSONSerializer):
@@ -46,6 +47,44 @@ def dump(self, obj, context=None):
         """Serialize object with schema."""
         return self.schema_class(context=context).dump(obj)
 
+    def _sanitize_metadata(self, metadata):
+        """Sanitize title, description and translations in metadata."""
+        try:
+            # top-level title
+            if "title" in metadata and "title" in metadata["title"]:
+                title = metadata["title"]["title"]
+                title = self.html_tag_remover.unescape(title)
+                metadata["title"]["title"] = remove_html_tags(
+                    self.html_tag_remover, title
+                )
+
+            # top-level description
+            if "description" in metadata:
+                description = metadata["description"]
+                description = self.html_tag_remover.unescape(description)
+                metadata["description"] = sanitize_html(description)
+
+            # translations
+            if "translations" in metadata:
+                for t in metadata["translations"]:
+                    if "title" in t and "title" in t["title"]:
+                        t_title = t["title"]["title"]
+                        t_title = self.html_tag_remover.unescape(t_title)
+                        t["title"]["title"] = remove_html_tags(
+                            self.html_tag_remover, t_title
+                        )
+
+                    if "description" in t:
+                        t_desc = t["description"]
+                        t_desc = self.html_tag_remover.unescape(t_desc)
+                        t["description"] = sanitize_html(t_desc)
+
+        except KeyError:
+            # ignore error if keys are missing
+            pass
+
+        return metadata
+
     def preprocess_record(self, pid, record, links_factory=None):
         """Include ``_eos_library_path`` for single record retrievals."""
         result = super(CDSJSONSerializer, self).preprocess_record(
@@ -62,16 +101,7 @@ def preprocess_record(self, pid, record, links_factory=None):
 
             # sanitize title by unescaping and stripping html tags
             try:
-                title = metadata["title"]["title"]
-                title = self.html_tag_remover.unescape(title)
-                metadata["title"]["title"] = remove_html_tags(
-                    self.html_tag_remover, title
-                )
-
-                # decode html entities
-                metadata["description"] = self.html_tag_remover.unescape(
-                    metadata["description"]
-                )
+                metadata = self._sanitize_metadata(metadata)
                 if has_request_context():
                     metadata["videos"] = [
                         video
@@ -93,19 +123,6 @@ def preprocess_search_hit(self, pid, record_hit, links_factory=None):
 
         if "metadata" in result:
             metadata = result["metadata"]
-
-            try:
-                title = metadata["title"]["title"]
-                title = self.html_tag_remover.unescape(title)
-                metadata["title"]["title"] = remove_html_tags(
-                    self.html_tag_remover, title
-                )
-
-                metadata["description"] = self.html_tag_remover.unescape(
-                    metadata["description"]
-                )
-            except KeyError:
-                # ignore error if keys are missing in the metadata
-                pass
+            result["metadata"] = self._sanitize_metadata(result["metadata"])
 
         return result

cds/modules/records/serializers/schemas/common.py

@@ -21,6 +21,7 @@
 
 from marshmallow import RAISE, Schema, ValidationError, fields, validates_schema
 from marshmallow.validate import Length
+from marshmallow_utils.fields import SanitizedHTML
 
 from ...api import Keyword
 from ...resolver import keyword_resolver
@@ -140,7 +141,7 @@ class TranslationsSchema(StrictKeysSchema):
     """Translations schema."""
 
     title = fields.Nested(TitleSchema)
-    description = fields.Str()
+    description = SanitizedHTML()
     language = fields.Str()
 
 

cds/modules/records/serializers/schemas/project.py

@@ -20,6 +20,7 @@
 
 from invenio_jsonschemas import current_jsonschemas
 from marshmallow import Schema, fields, pre_load, post_load
+from marshmallow_utils.fields import SanitizedHTML
 
 from ....deposit.api import Project, deposit_video_resolver
 from .common import (
@@ -76,7 +77,7 @@ class ProjectSchema(StrictKeysSchema):
     _deposit = fields.Nested(ProjectDepositSchema, required=True)
     _cds = fields.Nested(_CDSSSchema, required=True)
     title = fields.Nested(TitleSchema, required=True)
-    description = fields.Str()
+    description = SanitizedHTML()
     category = fields.Str(required=True)
     type = fields.Str(required=True)
     note = fields.Str()

cds/modules/records/serializers/schemas/video.py

@@ -20,7 +20,7 @@
 
 from invenio_jsonschemas import current_jsonschemas
 from marshmallow import Schema, fields, pre_load, post_load
-
+from marshmallow_utils.fields import SanitizedHTML
 from ....deposit.api import Video
 from ..fields.datetime import DateString
 from .common import (
@@ -126,7 +126,7 @@ class VideoSchema(StrictKeysSchema):
     contributors = fields.Nested(ContributorSchema, many=True, required=True)
     copyright = fields.Nested(CopyrightSchema)
     date = DateString(required=True)
-    description = fields.Str(required=True)
+    description = SanitizedHTML(required=True)
     doi = DOI()
     duration = fields.Str()
     external_system_identifiers = fields.Nested(

requirements.txt

@@ -134,6 +134,7 @@ lxml_html_clean==0.4.1
 Mako==1.3.8
 MarkupSafe==3.0.2
 marshmallow==3.23.1
+marshmallow-utils==0.13.0
 matplotlib-inline==0.1.7
 maxminddb==2.6.2
 maxminddb-geolite2==2018.703

setup.cfg

@@ -116,6 +116,7 @@ install_requires =
     invenio-sequencegenerator==1.0.0a3
     requests-toolbelt>=1.0.0,<2.0.0
     python-ldap>=3.4.0,<3.5.0
+    marshmallow-utils>=0.13.0,<1.0.0
 
 [options.extras_require]
 tests =

tests/unit/test_serializer.py

@@ -29,6 +29,9 @@
 
 from cds.modules.deposit.api import Video
 from cds.modules.records.serializers.drupal import VideoDrupal
+from cds.modules.records.serializers.json import CDSJSONSerializer
+from cds.modules.records.api import CDSRecord
+from unittest.mock import Mock
 from cds.modules.records.serializers.smil import Smil
 from cds.modules.records.serializers.vtt import VTT
 
@@ -149,3 +152,59 @@ def test_drupal_serializer(video_record_metadata, deposit_metadata):
     data = serializer.format()["entries"][0]["entry"]
     data = {k: data[k] for k in data if k in expected}
     assert data == expected
+
+
+def test_cds_json_serializer_sanitization(video_record_metadata):
+    """Test HTML sanitization in CDSJSONSerializer."""
+    record = CDSRecord.create(video_record_metadata)
+    
+    # Add malicious HTML
+    record['description'] = '<script>alert("xss")</script>Safe content <b>bold</b>'
+    record['title']['title'] = 'Test <script>alert("title")</script> Title  <b>bold</b>'
+    record['translations'] = [
+        {
+            'language': 'en',
+            'description': '<script>alert("desc")</script>Translated <i>italic</i>',
+            'title': {'title': '<b>Translated</b> <script>alert("title")</script> Title'}
+        },
+        {
+            'language': 'fr',
+            'description': 'Bonjour <script>alert("desc")</script> <u>underline</u>',
+            'title': {'title': '<script>alert("bad")</script> Titre'}
+        }
+    ]
+    
+    # Test the serializer
+    serializer = CDSJSONSerializer()
+    
+    # Create a mock PID (required by the serializer)
+    mock_pid = Mock()
+    mock_pid.pid_value = '1'
+    
+    # Test preprocess_record method
+    result = serializer.preprocess_record(mock_pid, record)
+    
+    # Check sanitization 
+    description = result['metadata']['description']
+    assert '<script>' not in description
+    assert '</script>' not in description
+    assert 'Safe content' in description
+    # Keep safe HTML tags like <b>
+    assert '<b>bold</b>' in description
+    
+    # Remove everything in title
+    title = result['metadata']['title']['title']
+    assert '<script>' not in title
+    assert '</script>' not in title
+    assert 'Test' in title and 'Title' in title
+    assert '<b>' not in title
+    
+    # --- Translations checks ---
+    translations = result['metadata']['translations']
+    for tr in translations:
+        # description
+        assert '<script>' not in tr['description']
+        # title
+        assert '<script>' not in tr['title']['title']
+        assert '<b>' not in tr['title']['title']
+