sanitizer: allow style for description

Author: zubeydecivelek

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

@@ -19,6 +19,11 @@
               "enterMode": 2,
               "entities": false,
               "height": 200,
+              "allowedContent": true,
+              "forcePasteAsPlainText": false,
+              "extraAllowedContent": "*{color}; *[style]",
+              "disableAutoInline": true,
+              "extraPlugins": "colorbutton",
               "toolbar": [
                 [
                   "PasteText",
@@ -32,6 +37,9 @@
                   "Subscript",
                   "Superscript"
                 ],
+                [
+                  "TextColor"
+                ],
                 [
                   "NumberedList",
                   "BulletedList",

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

@@ -18,6 +18,11 @@
             "enterMode": 2,
             "entities": false,
             "height": 200,
+            "allowedContent": true,
+            "forcePasteAsPlainText": false,
+            "extraAllowedContent": "*{color}; *[style]",
+            "disableAutoInline": true,
+            "extraPlugins": "colorbutton",
             "toolbar": [
               [
                 "PasteText",
@@ -31,6 +36,9 @@
                 "Subscript",
                 "Superscript"
               ],
+              [
+                "TextColor"
+              ],
               [
                 "NumberedList",
                 "BulletedList",

cds/modules/records/serializers/json.py

@@ -32,7 +32,15 @@
     has_read_record_permission,
 )
 from ..utils import HTMLTagRemover, parse_video_chapters, remove_html_tags
-from marshmallow_utils.html import sanitize_html
+from marshmallow_utils.html import sanitize_html, ALLOWED_HTML_ATTRS, ALLOWED_CSS_STYLES
+
+CUSTOM_ALLOWED_ATTRS = {
+    **ALLOWED_HTML_ATTRS,
+    "span": ALLOWED_HTML_ATTRS.get("span", []) + ["style"],
+    "p": ALLOWED_HTML_ATTRS.get("p", []) + ["style"],
+}
+
+CUSTOM_ALLOWED_CSS = ALLOWED_CSS_STYLES + ["color"]
 
 
 class CDSJSONSerializer(JSONSerializer):
@@ -60,7 +68,11 @@ def _sanitize_metadata(self, metadata):
             if "description" in metadata:
                 description = metadata["description"]
                 description = self.html_tag_remover.unescape(description)
-                metadata["description"] = sanitize_html(description)
+                metadata["description"] = sanitize_html(
+                    description,
+                    attrs=CUSTOM_ALLOWED_ATTRS,
+                    css_styles=CUSTOM_ALLOWED_CSS,
+                )
 
             if "translations" in metadata:
                 for t in metadata["translations"]:
@@ -108,12 +120,12 @@ def preprocess_record(self, pid, record, links_factory=None):
             except KeyError:
                 # ignore error if keys are missing in the metadata
                 pass
-            
-            description = metadata.get('description', '')
+
+            description = metadata.get("description", "")
             if description:
-                metadata['chapters'] = parse_video_chapters(description)
+                metadata["chapters"] = parse_video_chapters(description)
             else:
-                metadata['chapters'] = []
+                metadata["chapters"] = []
 
         return result
 

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

@@ -22,6 +22,7 @@
 from marshmallow import RAISE, Schema, ValidationError, fields, validates_schema
 from marshmallow.validate import Length
 from marshmallow_utils.fields import SanitizedHTML
+from marshmallow_utils.html import sanitize_html
 
 from ...api import Keyword
 from ...resolver import keyword_resolver
@@ -219,4 +220,42 @@ class RelatedIdentifiersSchema(Schema):
     identifier = fields.Str(required=True)
     scheme = fields.Str(required=True)
     relation_type = fields.Str(required=True)
-    resource_type = fields.Str()
+    resource_type = fields.Str()
+
+
+class SanitizedHTMLWithCSS(fields.String):
+    """Enhanced SanitizedHTML supporting inline CSS sanitization.
+
+    Fully compatible with marshmallow_utils.fields.SanitizedHTML,
+    but adds CSS.
+    """
+
+    def __init__(
+        self,
+        tags=None,
+        attrs=None,
+        css_styles=None,
+        *args,
+        **kwargs,
+    ):
+        """
+        :param tags: Allowed HTML tags.
+        :param attrs: Allowed HTML attributes per tag.
+        :param css_styles: List of allowed CSS properties (e.g., ["color"]).
+        """
+        super().__init__(*args, **kwargs)
+
+        self.tags = tags
+        self.attrs = attrs
+        self.css_styles = css_styles
+
+    def _deserialize(self, value, attr, data, **kwargs):
+        """Run bleach sanitize with CSS support."""
+        value = super()._deserialize(value, attr, data, **kwargs)
+
+        return sanitize_html(
+            value,
+            tags=self.tags,
+            attrs=self.attrs,
+            css_styles=self.css_styles,
+        )

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

@@ -18,9 +18,12 @@
 # 59 Temple Place, Suite 330, Boston, MA 02D111-1307, USA.
 """Project JSON schema."""
 
+from cds.modules.records.serializers.json import (
+    CUSTOM_ALLOWED_ATTRS,
+    CUSTOM_ALLOWED_CSS,
+)
 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 (
@@ -32,6 +35,7 @@
     KeywordsSchema,
     LicenseSchema,
     OaiSchema,
+    SanitizedHTMLWithCSS,
     StrictKeysSchema,
     TitleSchema,
     TranslationsSchema,
@@ -77,7 +81,10 @@ class ProjectSchema(StrictKeysSchema):
     _deposit = fields.Nested(ProjectDepositSchema, required=True)
     _cds = fields.Nested(_CDSSSchema, required=True)
     title = fields.Nested(TitleSchema, required=True)
-    description = SanitizedHTML()
+    description = SanitizedHTMLWithCSS(
+        attrs=CUSTOM_ALLOWED_ATTRS,
+        css_styles=CUSTOM_ALLOWED_CSS,
+    )
     category = fields.Str(required=True)
     type = fields.Str(required=True)
     note = fields.Str()

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

@@ -18,9 +18,12 @@
 # 59 Temple Place, Suite 330, Boston, MA 02D111-1307, USA.
 """Video JSON schema."""
 
+from cds.modules.records.serializers.json import (
+    CUSTOM_ALLOWED_ATTRS,
+    CUSTOM_ALLOWED_CSS,
+)
 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 (
@@ -38,6 +41,7 @@
     OaiSchema,
     RelatedIdentifiersSchema,
     RelatedLinksSchema,
+    SanitizedHTMLWithCSS,
     StrictKeysSchema,
     TitleSchema,
     TranslationsSchema,
@@ -131,7 +135,9 @@ class VideoSchema(StrictKeysSchema):
     contributors = fields.Nested(ContributorSchema, many=True, required=True)
     copyright = fields.Nested(CopyrightSchema)
     date = DateString(required=True)
-    description = SanitizedHTML(required=True)
+    description = SanitizedHTMLWithCSS(
+        attrs=CUSTOM_ALLOWED_ATTRS, css_styles=CUSTOM_ALLOWED_CSS, required=True
+    )
     doi = DOI()
     duration = fields.Str()
     external_system_identifiers = fields.Nested(
@@ -147,7 +153,7 @@ class VideoSchema(StrictKeysSchema):
     note = fields.Str()
     publication_date = fields.Str()
     recid = fields.Number()
-    legacy_recid =fields.Number()
+    legacy_recid = fields.Number()
     related_links = fields.Nested(RelatedLinksSchema, many=True)
     report_number = fields.List(fields.Str, many=True)
     schema = fields.Str(attribute="$schema", data_key="$schema")
@@ -158,15 +164,11 @@ class VideoSchema(StrictKeysSchema):
     _curation = fields.Nested(CurationSchema)
     additional_titles = fields.List(fields.Nested(AdditionalTitlesSchema))
     additional_descriptions = fields.List(fields.Nested(AdditionalDescriptionsSchema))
-    alternate_identifiers = fields.Nested(
-        AlternateIdentifiersSchema, many=True
-    )
-    related_identifiers = fields.Nested(
-        RelatedIdentifiersSchema, many=True
-    )
+    alternate_identifiers = fields.Nested(AlternateIdentifiersSchema, many=True)
+    related_identifiers = fields.Nested(RelatedIdentifiersSchema, many=True)
     collections = fields.List(fields.Str, many=True)
     additional_languages = fields.List(fields.Str, many=True)
-    
+
     # Preservation fields
     location = fields.Str()
     original_source = fields.Str()