From da29d88d04782ec3966607a3bcf6c328d12ea7db Mon Sep 17 00:00:00 2001
From: Victor Shnayder <victor@mitx.mit.edu>
Date: Thu, 26 Jul 2012 16:17:17 -0400
Subject: [PATCH] Add import-export round-trip check.

* add fix for the file export code--remove tag tail and text,
  not just the children.
---
 common/lib/xmodule/xmodule/xml_module.py      |  5 ++
 .../management/commands/clean_xml.py          | 82 +++++++++++++------
 2 files changed, 60 insertions(+), 27 deletions(-)

diff --git a/common/lib/xmodule/xmodule/xml_module.py b/common/lib/xmodule/xmodule/xml_module.py
index ee0b541de5..2c3fa62f6e 100644
--- a/common/lib/xmodule/xmodule/xml_module.py
+++ b/common/lib/xmodule/xmodule/xml_module.py
@@ -257,6 +257,11 @@ class XmlDescriptor(XModuleDescriptor):
             # ...and remove all of its children here
             for child in xml_object:
                 xml_object.remove(child)
+            # also need to remove the text of this object.
+            xml_object.text = ''
+            # and the tail for good measure...
+            xml_object.tail = ''
+
 
             xml_object.set('filename', self.name)
 
diff --git a/lms/djangoapps/courseware/management/commands/clean_xml.py b/lms/djangoapps/courseware/management/commands/clean_xml.py
index 1803553ba3..bd0c2b21bb 100644
--- a/lms/djangoapps/courseware/management/commands/clean_xml.py
+++ b/lms/djangoapps/courseware/management/commands/clean_xml.py
@@ -1,6 +1,8 @@
+import os
 import sys
 import traceback
 
+from filecmp import dircmp
 from fs.osfs import OSFS
 from path import path
 from lxml import etree
@@ -10,28 +12,6 @@ from django.core.management.base import BaseCommand
 from xmodule.modulestore.xml import XMLModuleStore
 
 
-def export(course, export_dir):
-    """Export the specified course to course_dir.  Creates dir if it doesn't exist.
-    Overwrites files, does not clean out dir beforehand.
-    """
-    fs = OSFS(export_dir, create=True)
-    if not fs.isdirempty('.'):
-        print ('WARNING: Directory {dir} not-empty.'
-               '  May clobber/confuse things'.format(dir=export_dir))
-
-    try:
-        xml = course.export_to_xml(fs)
-        with fs.open('course.xml', mode='w') as f:
-            f.write(xml)
-
-        return True
-    except:
-        print 'Export failed!'
-        traceback.print_exc()
-
-    return False
-
-
 def traverse_tree(course):
     '''Load every descriptor in course.  Return bool success value.'''
     queue = [course]
@@ -61,7 +41,30 @@ def make_logging_error_handler():
 
     return (error_handler, errors)
 
-def clean_xml(course_dir, export_dir, verbose=True):
+
+def export(course, export_dir):
+    """Export the specified course to course_dir.  Creates dir if it doesn't exist.
+    Overwrites files, does not clean out dir beforehand.
+    """
+    fs = OSFS(export_dir, create=True)
+    if not fs.isdirempty('.'):
+        print ('WARNING: Directory {dir} not-empty.'
+               '  May clobber/confuse things'.format(dir=export_dir))
+
+    try:
+        xml = course.export_to_xml(fs)
+        with fs.open('course.xml', mode='w') as f:
+            f.write(xml)
+
+        return True
+    except:
+        print 'Export failed!'
+        traceback.print_exc()
+
+    return False
+
+
+def import_with_checks(course_dir, verbose=True):
     all_ok = True
 
     print "Attempting to load '{0}'".format(course_dir)
@@ -105,7 +108,7 @@ def clean_xml(course_dir, export_dir, verbose=True):
 
     course = courses[0]
 
-    print course
+    #print course
     validators = (
         traverse_tree,
         )
@@ -120,10 +123,36 @@ def clean_xml(course_dir, export_dir, verbose=True):
 
     if all_ok:
         print 'Course passes all checks!'
-        export(course, export_dir)
-
     else:
         print "Course fails some checks.  See above for errors."
+    return all_ok, course
+
+
+def check_roundtrip(course_dir):
+    '''Check that import->export leaves the course the same'''
+
+    print "====== Roundtrip import ======="
+    (ok, course) = import_with_checks(course_dir)
+    if not ok:
+        raise Exception("Roundtrip import failed!")
+
+    print "====== Roundtrip export ======="
+    export_dir = course_dir + ".rt"
+    export(course, export_dir)
+
+    # dircmp doesn't do recursive diffs.
+    # diff = dircmp(course_dir, export_dir, ignore=[], hide=[])
+    print "======== Roundtrip diff: ========="
+    os.system("diff -r {0} {1}".format(course_dir, export_dir))
+    print "======== ideally there is no diff above this ======="
+
+
+def clean_xml(course_dir, export_dir):
+    (ok, course) = import_with_checks(course_dir)
+    if ok:
+        export(course, export_dir)
+        check_roundtrip(export_dir)
+    else:
         print "Did NOT export"
 
 
@@ -134,7 +163,6 @@ class Command(BaseCommand):
 
 Usage: clean_xml PATH-TO-COURSE-DIR PATH-TO-OUTPUT-DIR
 """
-
     def handle(self, *args, **options):
         if len(args) != 2:
             print Command.help