Merge pull request #24680 from open-craft/ahmed/fix-capa-dropdown-indexing

[TNL-6993] [BD-29]: Handle edge case where dropdown values are not being indexed

common/lib/xmodule/xmodule/capa_module.py

@@ -7,6 +7,7 @@ import re
 import sys
 
 import six
+from bleach.sanitizer import Cleaner
 from lxml import etree
 from pkg_resources import resource_string
 from web_fragments.fragment import Fragment
@@ -17,7 +18,6 @@ from xmodule.contentstore.django import contentstore
 from xmodule.editing_module import EditingMixin
 from xmodule.exceptions import NotFoundError, ProcessingError
 from xmodule.raw_module import RawMixin
-from xmodule.util.misc import escape_html_characters
 from xmodule.util.sandboxing import get_python_lib_zip
 from xmodule.util.xmodule_django import add_webpack_to_fragment
 from xmodule.x_module import (
@@ -302,6 +302,10 @@ class ProblemBlock(
         Return dictionary prepared with module content and type for indexing.
         """
         xblock_body = super(ProblemBlock, self).index_dictionary()
+
+        # Make optioninput's options index friendly by replacing the actual tag with the values
+        capa_content = re.sub(r'<optioninput options="\(([^"]+)\)".*?>\s*|\S*<\/optioninput>', r'\1', self.data)
+
         # Removing solutions and hints, as well as script and style
         capa_content = re.sub(
             re.compile(
@@ -314,9 +318,14 @@ class ProblemBlock(
                 re.DOTALL |
                 re.VERBOSE),
             "",
-            self.data
+            capa_content
         )
-        capa_content = escape_html_characters(capa_content)
+        capa_content = re.sub(
+            r"(\s|&nbsp;|//)+",
+            " ",
+            Cleaner(tags=[], strip=True).clean(capa_content)
+        )
+
         capa_body = {
             "capa_content": capa_content,
             "display_name": self.display_name,

common/lib/xmodule/xmodule/tests/test_capa_module.py

@@ -2536,14 +2536,21 @@ class ProblemBlockXMLTest(unittest.TestCase):
         name = "Other Test Capa Problem"
         descriptor = self._create_descriptor(xml, name=name)
         self.assertEqual(descriptor.problem_types, {"multiplechoiceresponse", "optionresponse"})
-        six.assertCountEqual(
-            self, descriptor.index_dictionary(), {
+
+        # We are converting problem_types to a set to compare it later without taking into account the order
+        # the reasoning behind is that the problem_types (property) is represented by dict and when it is converted
+        # to list its ordering is different everytime.
+
+        indexing_result = descriptor.index_dictionary()
+        indexing_result['problem_types'] = set(indexing_result['problem_types'])
+        self.assertDictEqual(
+            indexing_result, {
                 'content_type': ProblemBlock.INDEX_CONTENT_TYPE,
-                'problem_types': ["optionresponse", "multiplechoiceresponse"],
+                'problem_types': set(["optionresponse", "multiplechoiceresponse"]),
                 'content': {
                     'display_name': name,
-                    'capa_content': ' Label Some comment Donut Buggy '
-                }
+                    'capa_content': " Label Some comment Donut Buggy '1','2' "
+                },
             }
         )
 
@@ -2619,7 +2626,7 @@ class ProblemBlockXMLTest(unittest.TestCase):
             Dropdown problems allow learners to select only one option from a list of options.
             Description
             You can use the following example problem as a model.
-            Which of the following countries celebrates its independence on August 15?
+            Which of the following countries celebrates its independence on August 15? 'India','Spain','China','Bermuda'
         """)
         self.assertEqual(descriptor.problem_types, {"optionresponse"})
         self.assertEqual(