From ecbf148688fc3d54c9ccf4ea268bbf4216f1fc29 Mon Sep 17 00:00:00 2001
From: Chris Dodge <cdodge@edx.org>
Date: Thu, 20 Jun 2013 11:46:16 -0400
Subject: [PATCH] add a new django-admin command to dump out a course structure
 document. This is response to an emergency request from Harvard researchers.

---
 .../commands/dump_course_structure.py         | 55 +++++++++++++++++++
 1 file changed, 55 insertions(+)
 create mode 100644 cms/djangoapps/contentstore/management/commands/dump_course_structure.py

diff --git a/cms/djangoapps/contentstore/management/commands/dump_course_structure.py b/cms/djangoapps/contentstore/management/commands/dump_course_structure.py
new file mode 100644
index 0000000000..d9b7c55cbd
--- /dev/null
+++ b/cms/djangoapps/contentstore/management/commands/dump_course_structure.py
@@ -0,0 +1,55 @@
+from django.core.management.base import BaseCommand, CommandError
+from xmodule.course_module import CourseDescriptor
+from xmodule.modulestore.django import modulestore
+from json import dumps
+from xmodule.modulestore.inheritance import own_metadata
+from django.conf import settings
+
+filter_list = ['xml_attributes', 'checklists']
+
+
+class Command(BaseCommand):
+    help = '''Write out to stdout a structural and metadata information about a course in a flat dictionary serialized
+              in a JSON format. This can be used for analytics.'''
+
+    def handle(self, *args, **options):
+        if len(args) < 2 or len(args) > 3:
+            raise CommandError("dump_course_structure requires two or more arguments: <location> <outfile> |<db>|")
+
+        course_id = args[0]
+        outfile = args[1]
+
+        # use a user-specified database name, if present
+        # this is useful for doing dumps from databases restored from prod backups
+        if len(args) == 3:
+            settings.MODULESTORE['direct']['OPTIONS']['db'] = args[2]
+
+        loc = CourseDescriptor.id_to_location(course_id)
+
+        store = modulestore()
+
+        course = None
+        try:
+            course = store.get_item(loc, depth=4)
+        except:
+            print 'Could not find course at {0}'.format(course_id)
+            return
+
+        info = {}
+
+        def dump_into_dict(module, info):
+            filtered_metadata = dict((key, value) for key, value in own_metadata(module).iteritems()
+                                     if key not in filter_list)
+            info[module.location.url()] = {
+                'category': module.location.category,
+                'children': module.children if hasattr(module, 'children') else [],
+                'metadata': filtered_metadata
+            }
+
+            for child in module.get_children():
+                dump_into_dict(child, info)
+
+        dump_into_dict(course, info)
+
+        with open(outfile, 'w') as f:
+            f.write(dumps(info))