Import course objects first, so that later updates that try and edit the course object don't fail

common/lib/xmodule/xmodule/modulestore/xml_importer.py

@@ -89,53 +89,8 @@ def verify_content_links(module, base_dir, static_content_store, link, remap_dic
 
     return link
 
-def import_from_xml(store, data_dir, course_dirs=None, 
-                    default_class='xmodule.raw_module.RawDescriptor',
-                    load_error_modules=True, static_content_store=None, target_location_namespace=None):
-    """
-    Import the specified xml data_dir into the "store" modulestore,
-    using org and course as the location org and course.
-
-    course_dirs: If specified, the list of course_dirs to load. Otherwise, load
-    all course dirs
-
-    target_location_namespace is the namespace [passed as Location] (i.e. {tag},{org},{course}) that all modules in the should be remapped to
-    after import off disk. We do this remapping as a post-processing step because there's logic in the importing which
-    expects a 'url_name' as an identifier to where things are on disk e.g. ../policies/<url_name>/policy.json as well as metadata keys in
-    the policy.json. so we need to keep the original url_name during import
-
-    """
-    
-    module_store = XMLModuleStore(
-        data_dir,
-        default_class=default_class,
-        course_dirs=course_dirs,
-        load_error_modules=load_error_modules
-    )
-
-    # NOTE: the XmlModuleStore does not implement get_items() which would be a preferable means
-    # to enumerate the entire collection of course modules. It will be left as a TBD to implement that
-    # method on XmlModuleStore.
-    course_items = []
-    for course_id in module_store.modules.keys():
-
-        course_data_path = None
-        course_location = None
-        # Quick scan to get course Location as well as the course_data_path
-        for module in module_store.modules[course_id].itervalues():
-            if module.category == 'course':
-                course_data_path = path(data_dir) / module.data_dir
-                course_location = module.location
-
-        if static_content_store is not None:
-            _namespace_rename = target_location_namespace if target_location_namespace is not None else  module_store.modules[course_id].location
-            
-            # first pass to find everything in /static/
-            import_static_content(module_store.modules[course_id], course_location, course_data_path, static_content_store, 
-                _namespace_rename, subpath='static')
-
-        for module in module_store.modules[course_id].itervalues():
 
+def import_module_from_xml(modulestore, static_content_store, course_data_path, module, target_location_namespace=None):
     # remap module to the new namespace
     if target_location_namespace is not None:
         # This looks a bit wonky as we need to also change the 'name' of the imported course to be what
@@ -160,26 +115,6 @@ def import_from_xml(store, data_dir, course_dirs=None,
 
             module.definition['children'] = new_locs
 
-            if module.category == 'course':
-                # HACK: for now we don't support progress tabs. There's a special metadata configuration setting for this.
-                module.hide_progress_tab = True
-
-                # cdodge: more hacks (what else). Seems like we have a problem when importing a course (like 6.002) which
-                # does not have any tabs defined in the policy file. The import goes fine and then displays fine in LMS,
-                # but if someone tries to add a new tab in the CMS, then the LMS barfs because it expects that -
-                # if there is *any* tabs - then there at least needs to be some predefined ones
-                if module.tabs is None or len(module.tabs) == 0:
-                    module.tabs = [{"type": "courseware"},
-                        {"type": "course_info", "name": "Course Info"},
-                        {"type": "discussion", "name": "Discussion"},
-                        {"type": "wiki", "name": "Wiki"}]  # note, add 'progress' when we can support it on Edge
-
-                # a bit of a hack, but typically the "course image" which is shown on marketing pages is hard coded to /images/course_image.jpg
-                # so let's make sure we import in case there are no other references to it in the modules
-                verify_content_links(module, course_data_path, static_content_store, '/static/images/course_image.jpg')
-
-                course_items.append(module)
-
     if hasattr(module, 'data'):
         # cdodge: now go through any link references to '/static/' and make sure we've imported
         # it as a StaticContent asset
@@ -203,10 +138,10 @@ def import_from_xml(store, data_dir, course_dirs=None,
         except Exception:
             logging.exception("failed to rewrite links on {0}. Continuing...".format(module.location))
 
-                store.update_item(module.location, module.data)
+        modulestore.update_item(module.location, module.data)
 
     if module.has_children:
-                store.update_children(module.location, module.children)
+        modulestore.update_children(module.location, module.children)
 
     metadata = {}
     for field in module.fields + module.lms.fields:
@@ -216,7 +151,83 @@ def import_from_xml(store, data_dir, course_dirs=None,
             field.name in module._model_data):
 
             metadata[field.name] = module._model_data[field.name]
-            store.update_metadata(module.location, metadata)
+    modulestore.update_metadata(module.location, metadata)
+
+
+def import_course_from_xml(modulestore, static_content_store, course_data_path, module, target_location_namespace=None):
+    # HACK: for now we don't support progress tabs. There's a special metadata configuration setting for this.
+    module.hide_progress_tab = True
+
+    # cdodge: more hacks (what else). Seems like we have a problem when importing a course (like 6.002) which
+    # does not have any tabs defined in the policy file. The import goes fine and then displays fine in LMS,
+    # but if someone tries to add a new tab in the CMS, then the LMS barfs because it expects that -
+    # if there is *any* tabs - then there at least needs to be some predefined ones
+    if module.tabs is None or len(module.tabs) == 0:
+        module.tabs = [{"type": "courseware"},
+            {"type": "course_info", "name": "Course Info"},
+            {"type": "discussion", "name": "Discussion"},
+            {"type": "wiki", "name": "Wiki"}]  # note, add 'progress' when we can support it on Edge
+
+    # a bit of a hack, but typically the "course image" which is shown on marketing pages is hard coded to /images/course_image.jpg
+    # so let's make sure we import in case there are no other references to it in the modules
+    verify_content_links(module, course_data_path, static_content_store, '/static/images/course_image.jpg')
+    import_module_from_xml(modulestore, static_content_store, course_data_path, module, target_location_namespace)
+
+
+def import_from_xml(store, data_dir, course_dirs=None, 
+                    default_class='xmodule.raw_module.RawDescriptor',
+                    load_error_modules=True, static_content_store=None, target_location_namespace=None):
+    """
+    Import the specified xml data_dir into the "store" modulestore,
+    using org and course as the location org and course.
+
+    course_dirs: If specified, the list of course_dirs to load. Otherwise, load
+    all course dirs
+
+    target_location_namespace is the namespace [passed as Location] (i.e. {tag},{org},{course}) that all modules in the should be remapped to
+    after import off disk. We do this remapping as a post-processing step because there's logic in the importing which
+    expects a 'url_name' as an identifier to where things are on disk e.g. ../policies/<url_name>/policy.json as well as metadata keys in
+    the policy.json. so we need to keep the original url_name during import
+
+    """
+    
+    module_store = XMLModuleStore(
+        data_dir,
+        default_class=default_class,
+        course_dirs=course_dirs,
+        load_error_modules=load_error_modules
+    )
+
+    # NOTE: the XmlModuleStore does not implement get_items() which would be a preferable means
+    # to enumerate the entire collection of course modules. It will be left as a TBD to implement that
+    # method on XmlModuleStore.
+    course_items = []
+    for course_id in module_store.modules.keys():
+
+        course_data_path = None
+        course_location = None
+        # Quick scan to get course Location as well as the course_data_path
+        for module in module_store.modules[course_id].itervalues():
+            if module.category == 'course':
+                course_data_path = path(data_dir) / module.data_dir
+                course_location = module.location
+
+        if static_content_store is not None:
+            _namespace_rename = target_location_namespace if target_location_namespace is not None else  module_store.modules[course_id].location
+            
+            # first pass to find everything in /static/
+            import_static_content(module_store.modules[course_id], course_location, course_data_path, static_content_store, 
+                _namespace_rename, subpath='static')
+
+        # Import course modules first, because importing some of the children requires the course to exist
+        for module in module_store.modules[course_id].itervalues():
+            if module.category == 'course':
+                import_course_from_xml(store, static_content_store, course_data_path, module, target_location_namespace)
+
+        # Import the rest of the modules
+        for module in module_store.modules[course_id].itervalues():
+            if module.category != 'course':
+                import_module_from_xml(store, static_content_store, course_data_path, module, target_location_namespace)
 
     return module_store, course_items