From 981f5cee45507bff72f85215696de74db5937359 Mon Sep 17 00:00:00 2001
From: Chris Dodge <cdodge@edx.org>
Date: Thu, 8 Nov 2012 16:07:17 -0500
Subject: [PATCH] initial buildout of a 'xlint' test to verify legacy coursewar

---
 .../contentstore/management/commands/xlint.py | 26 +++++++++++
 .../xmodule/modulestore/xml_importer.py       | 43 +++++++++++++++++--
 rakefile                                      | 12 ++++++
 3 files changed, 77 insertions(+), 4 deletions(-)
 create mode 100644 cms/djangoapps/contentstore/management/commands/xlint.py

diff --git a/cms/djangoapps/contentstore/management/commands/xlint.py b/cms/djangoapps/contentstore/management/commands/xlint.py
new file mode 100644
index 0000000000..355b639f2d
--- /dev/null
+++ b/cms/djangoapps/contentstore/management/commands/xlint.py
@@ -0,0 +1,26 @@
+from django.core.management.base import BaseCommand, CommandError
+from xmodule.modulestore.xml_importer import import_from_xml
+from xmodule.modulestore.django import modulestore
+from xmodule.contentstore.django import contentstore
+
+
+unnamed_modules = 0
+
+
+class Command(BaseCommand):
+    help = \
+'''Verify the structure of courseware as to it's suitability for import'''
+
+    def handle(self, *args, **options):
+        if len(args) == 0:
+            raise CommandError("import requires at least one argument: <data directory> [<course dir>...]")
+
+        data_dir = args[0]
+        if len(args) > 1:
+            course_dirs = args[1:]
+        else:
+            course_dirs = None
+        print "Importing.  Data_dir={data}, course_dirs={courses}".format(
+            data=data_dir,
+            courses=course_dirs)
+        import_from_xml(None, data_dir, course_dirs, load_error_modules=False, validate_only=True)
diff --git a/common/lib/xmodule/xmodule/modulestore/xml_importer.py b/common/lib/xmodule/xmodule/modulestore/xml_importer.py
index 00ddb6a948..4a3526b53e 100644
--- a/common/lib/xmodule/xmodule/modulestore/xml_importer.py
+++ b/common/lib/xmodule/xmodule/modulestore/xml_importer.py
@@ -88,7 +88,7 @@ def verify_content_links(module, base_dir, static_content_store, link, remap_dic
 
 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):
+                    load_error_modules=True, static_content_store=None, target_location_namespace=None, validate_only=False):
     """
     Import the specified xml data_dir into the "store" modulestore,
     using org and course as the location org and course.
@@ -110,6 +110,10 @@ def import_from_xml(store, data_dir, course_dirs=None,
         load_error_modules=load_error_modules
     )
 
+    if validate_only:
+        validate_module_structure(module_store)
+        return module_store, []
+
     # 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.
@@ -192,7 +196,6 @@ def import_from_xml(store, data_dir, course_dirs=None,
 
                 store.update_item(module.location, module_data)
 
-
             if 'children' in module.definition:
                 store.update_children(module.location, module.definition['children'])
 
@@ -200,6 +203,38 @@ def import_from_xml(store, data_dir, course_dirs=None,
             # inherited metadata everywhere.
             store.update_metadata(module.location, dict(module.own_metadata))
 
-
-
     return module_store, course_items
+
+
+def validate_category_hierarcy(module_store, course_id, parent_category, expected_child_category):
+    err_cnt = 0
+
+    parents = []
+    # get all modules of parent_category
+    for module in module_store.modules[course_id].itervalues():
+        if module.location.category == parent_category:
+            parents.append(module)
+
+    for parent in parents:
+        for child_loc in [Location(child) for child in parent.definition.get('children', [])]:
+            if child_loc.category != expected_child_category:
+                err_cnt += 1
+                print 'ERROR: child {0} of parent {1} was expected to be category of {2} but was {3}'.format(
+                    child_loc, parent.location, expected_child_category, child_loc.category)
+
+    return err_cnt
+
+def validate_module_structure(module_store):
+    err_cnt = 0
+    warn_cnt = 0
+    for course_id in module_store.modules.keys():
+        # constrain that courses only have 'chapter' children
+        err_cnt += validate_category_hierarcy(module_store, course_id, "course", "chapter")
+        # constrain that chapters only have 'sequentials'
+        err_cnt += validate_category_hierarcy(module_store, course_id, "chapter", "sequential")
+        # constrain that sequentials only have 'verticals'
+        err_cnt += validate_category_hierarcy(module_store, course_id, "sequential", "vertical")
+
+    print "SUMMARY: {0} Errors   {1} Warnings".format(err_cnt, warn_cnt)
+       
+
diff --git a/rakefile b/rakefile
index 4f1c15321f..beb787c8c3 100644
--- a/rakefile
+++ b/rakefile
@@ -364,6 +364,18 @@ namespace :cms do
   end
 end
 
+namespace :cms do
+  desc "Import course data within the given DATA_DIR variable"
+  task :xlint do
+    if ENV['DATA_DIR']
+      sh(django_admin(:cms, :dev, :xlint, ENV['DATA_DIR']))
+    else
+      raise "Please specify a DATA_DIR variable that point to your data directory.\n" +
+        "Example: \`rake cms:import DATA_DIR=../data\`"
+    end
+  end
+end
+
 desc "Build a properties file used to trigger autodeploy builds"
 task :autodeploy_properties do
     File.open("autodeploy.properties", "w") do |file|