From beda4f95b26694cf17bcf8cf22a9e98db4392970 Mon Sep 17 00:00:00 2001
From: Victor Shnayder <victor@mitx.mit.edu>
Date: Wed, 25 Jul 2012 14:12:16 -0400
Subject: [PATCH] Initial version of import/check/export script.

* uses xml modulestore, and new error_handler hook
---
 .../management/commands/clean_xml.py          | 143 ++++++++++++++++++
 1 file changed, 143 insertions(+)
 create mode 100644 lms/djangoapps/courseware/management/commands/clean_xml.py

diff --git a/lms/djangoapps/courseware/management/commands/clean_xml.py b/lms/djangoapps/courseware/management/commands/clean_xml.py
new file mode 100644
index 0000000000..1803553ba3
--- /dev/null
+++ b/lms/djangoapps/courseware/management/commands/clean_xml.py
@@ -0,0 +1,143 @@
+import sys
+import traceback
+
+from fs.osfs import OSFS
+from path import path
+from lxml import etree
+
+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]
+    while len(queue) > 0:
+        node = queue.pop()
+#        print '{0}:'.format(node.location)
+#        if 'data' in node.definition:
+#            print '{0}'.format(node.definition['data'])
+        queue.extend(node.get_children())
+
+    return True
+
+def make_logging_error_handler():
+    '''Return a tuple (handler, error_list), where
+    the handler appends the message and any exc_info
+    to the error_list on every call.
+    '''
+    errors = []
+
+    def error_handler(msg, exc_info=None):
+        '''Log errors'''
+        if exc_info is None:
+            if sys.exc_info() != (None, None, None):
+                exc_info = sys.exc_info()
+
+        errors.append((msg, exc_info))
+
+    return (error_handler, errors)
+
+def clean_xml(course_dir, export_dir, verbose=True):
+    all_ok = True
+
+    print "Attempting to load '{0}'".format(course_dir)
+
+    course_dir = path(course_dir)
+    data_dir = course_dir.dirname()
+    course_dirs = [course_dir.basename()]
+
+    (error_handler, errors) = make_logging_error_handler()
+    # No default class--want to complain if it doesn't find plugins for any
+    # module.
+    modulestore = XMLModuleStore(data_dir,
+                   default_class=None,
+                   eager=True,
+                   course_dirs=course_dirs,
+                   error_handler=error_handler)
+
+    def str_of_err(tpl):
+        (msg, exc_info) = tpl
+        if exc_info is None:
+            return msg
+
+        exc_str = '\n'.join(traceback.format_exception(*exc_info))
+        return '{msg}\n{exc}'.format(msg=msg, exc=exc_str)
+
+    courses = modulestore.get_courses()
+    if len(errors) != 0:
+        all_ok = False
+        print '\n'
+        print "=" * 40
+        print 'ERRORs during import:'
+        print '\n'.join(map(str_of_err,errors))
+        print "=" * 40
+        print '\n'
+
+    n = len(courses)
+    if n != 1:
+        print 'ERROR: Expect exactly 1 course.  Loaded {n}: {lst}'.format(
+            n=n, lst=courses)
+        return
+
+    course = courses[0]
+
+    print course
+    validators = (
+        traverse_tree,
+        )
+
+    print "=" * 40
+    print "Running validators..."
+
+    for validate in validators:
+        print 'Running {0}'.format(validate.__name__)
+        all_ok = validate(course) and all_ok
+
+
+    if all_ok:
+        print 'Course passes all checks!'
+        export(course, export_dir)
+
+    else:
+        print "Course fails some checks.  See above for errors."
+        print "Did NOT export"
+
+
+
+class Command(BaseCommand):
+    help = """Imports specified course.xml, validate it, then exports in
+    a canonical format.
+
+Usage: clean_xml PATH-TO-COURSE-DIR PATH-TO-OUTPUT-DIR
+"""
+
+    def handle(self, *args, **options):
+        if len(args) != 2:
+            print Command.help
+            return
+
+        clean_xml(args[0], args[1])