feat: Serialize tag data in OLX for blocks (#34145)

cms/envs/common.py

@@ -129,6 +129,7 @@
 
 from lms.djangoapps.lms_xblock.mixin import LmsBlockMixin
 from cms.lib.xblock.authoring_mixin import AuthoringMixin
+from cms.lib.xblock.tagging.tagged_block_mixin import TaggedBlockMixin
 from xmodule.modulestore.edit_info import EditInfoMixin
 from openedx.core.djangoapps.theming.helpers_dirs import (
     get_themes_unchecked,
@@ -975,6 +976,7 @@
     XModuleMixin,
     EditInfoMixin,
     AuthoringMixin,
+    TaggedBlockMixin,
 )
 XBLOCK_EXTRA_MIXINS = ()
 

cms/lib/xblock/tagging/tagged_block_mixin.py

@@ -0,0 +1,57 @@
+# lint-amnesty, pylint: disable=missing-module-docstring
+from urllib.parse import quote
+
+
+class TaggedBlockMixin:
+    """
+    Mixin containing XML serializing and parsing functionality for tagged blocks
+    """
+
+    def serialize_tag_data(self):
+        """
+        Serialize block's tag data to include in the xml, escaping special characters
+
+        Example tags:
+            LightCast Skills Taxonomy: ["Typing", "Microsoft Office"]
+            Open Canada Skills Taxonomy: ["MS Office", "<some:;,skill/|=>"]
+
+        Example serialized tags:
+            lightcast-skills:Typing,Microsoft Office;open-canada-skills:MS Office,%3Csome%3A%3B%2Cskill%2F%7C%3D%3E
+        """
+        # This import is done here since we import and use TaggedBlockMixin in the cms settings, but the
+        # content_tagging app wouldn't have loaded yet, so importing it outside causes an error
+        from openedx.core.djangoapps.content_tagging.api import get_object_tags
+        content_tags = get_object_tags(self.scope_ids.usage_id)
+
+        serialized_tags = []
+        taxonomies_and_tags = {}
+        for tag in content_tags:
+            taxonomy_export_id = tag.taxonomy.export_id
+
+            if not taxonomies_and_tags.get(taxonomy_export_id):
+                taxonomies_and_tags[taxonomy_export_id] = []
+
+            # Escape special characters in tag values, except spaces (%20) for better readability
+            escaped_tag = quote(tag.value).replace("%20", " ")
+            taxonomies_and_tags[taxonomy_export_id].append(escaped_tag)
+
+        for taxonomy in taxonomies_and_tags:
+            merged_tags = ','.join(taxonomies_and_tags.get(taxonomy))
+            serialized_tags.append(f"{taxonomy}:{merged_tags}")
+
+        return ";".join(serialized_tags)
+
+    def add_tags_to_node(self, node):
+        """
+        Serialize and add tag data (if any) to node
+        """
+        tag_data = self.serialize_tag_data()
+        if tag_data:
+            node.set('tags-v1', tag_data)
+
+    def add_xml_to_node(self, node):
+        """
+        Include the serialized tag data in XML when adding to node
+        """
+        super().add_xml_to_node(node)
+        self.add_tags_to_node(node)

openedx/core/lib/xblock_serializer/block_serializer.py

@@ -7,6 +7,8 @@
 
 from lxml import etree
 
+from cms.lib.xblock.tagging.tagged_block_mixin import TaggedBlockMixin
+
 from .data import StaticFile
 from . import utils
 
@@ -113,6 +115,10 @@ def _serialize_html_block(self, block) -> etree.Element:
         if block.use_latex_compiler:
             olx_node.attrib["use_latex_compiler"] = "true"
 
+        # Serialize and add tag data if any
+        if isinstance(block, TaggedBlockMixin):
+            block.add_tags_to_node(olx_node)
+
         # Escape any CDATA special chars
         escaped_block_data = block.data.replace("]]>", "]]>")
         olx_node.text = etree.CDATA("\n" + escaped_block_data + "\n")