fix: Update url capture regex for course id

openedx/core/constants.py

@@ -6,9 +6,14 @@
 # Note: these intentionally greedily grab all chars up to the next slash includingny pluses
 # DHM: I really wanted to ensure the separators were the same (+ or /) but all patts tried had
 # too many inadvertent side effects :-(
-
 COURSE_KEY_PATTERN = r'(?P<course_key_string>[^/+]+(/|\+)[^/+]+(/|\+)[^/?]+)'
 COURSE_ID_PATTERN = COURSE_KEY_PATTERN.replace('course_key_string', 'course_id')
 COURSE_KEY_REGEX = COURSE_KEY_PATTERN.replace('P<course_key_string>', ':')
+
+# This constant can be extended to take into account future urls to negate with the following pattern
+# (?=\/(enroll|unenroll|other_patterns)$)
+POSTFIXED_PATTERNS_TO_NEGATE = r'(?=\/enroll)$'
+CAPTURED_CLEAN_COURSE_ID_PATTERN = r'^(?P<course_id>.*?)'
+
 COURSE_PUBLISHED = 'published'
 COURSE_UNPUBLISHED = 'unpublished'

openedx/core/lib/request_utils.py

@@ -13,10 +13,12 @@
 from opaque_keys.edx.keys import CourseKey
 from rest_framework.views import exception_handler
 
+from openedx.core.constants import CAPTURED_CLEAN_COURSE_ID_PATTERN, POSTFIXED_PATTERNS_TO_NEGATE
 from openedx.core.djangoapps.site_configuration import helpers as configuration_helpers
 
 # accommodates course api urls, excluding any course api routes that do not fall under v*/courses, such as v1/blocks.
 COURSE_REGEX = re.compile(fr'^(.*?/course(s)?/)(?!v[0-9]+/[^/]+){settings.COURSE_ID_PATTERN}')
+CLEANED_COURSE_ID_REGEX = re.compile(f'{CAPTURED_CLEAN_COURSE_ID_PATTERN}{POSTFIXED_PATTERNS_TO_NEGATE}')
 
 log = logging.getLogger(__name__)
 
@@ -77,15 +79,22 @@ def course_id_from_url(url):
 
     if match is None:
         return None
-
     course_id = match.group('course_id')
 
     if course_id is None:
         return None
 
+    cleaned_course_id = CLEANED_COURSE_ID_REGEX.match(course_id)
+    if cleaned_course_id:
+        cleaned_course_id = cleaned_course_id.group('course_id')
+        if cleaned_course_id:
+            course_id = cleaned_course_id
+    print(course_id, 'course_id_from_url 1')
     try:
+        print(course_id, 'course_id_from_url 2')
         return CourseKey.from_string(course_id)
     except InvalidKeyError:
+        print(course_id, 'course_id_from_url 3')
         return None
 
 

openedx/core/lib/tests/test_request_utils.py

@@ -95,8 +95,16 @@ def test_course_id_from_url(self):
         course_id = course_id_from_url('/api/courses/v1/courses/edX/maths/2020')
         self.assertCourseIdFieldsMatch(course_id=course_id, org='edX', course='maths', run='2020')
 
+        course_id = course_id_from_url('/enterprise/5d566680-12a8-4b85-89d8-d9eacbf0f9eb/course/edX+math/enroll/')
+        print(course_id)
+        self.assertCourseIdFieldsMatch(course_id=course_id, org='edX', course='maths', run=None)
+
+        course_id = course_id_from_url('/enterprise/5d566680-12a8-4b85-89d8-d9eacbf0f9eb/course/edX+math+2020/enroll/')
+        self.assertCourseIdFieldsMatch(course_id=course_id, org='edX', course='maths', run='2020')
+
     def assertCourseIdFieldsMatch(self, course_id, org, course, run):
         """ Asserts that the passed-in course id matches the specified fields"""
+        print(course_id, org, course, run)
         assert course_id.org == org
         assert course_id.course == course
         assert course_id.run == run