edx_olxformat_merge_courses.py

'''
Merge/combine courses in the OpenedX OLX format.
'''

import sys
import os
from distutils.dir_util import copy_tree
import json

# Example:
# Place un-tared course exports in folder and create output folder
# $python edx_olxformat_merge_courses.py /Users/username/projects/edx_olx_merge/to_process/MOOCS /Users/username/projects/edx_olx_merge/processed/

def getCourses(folder_with_merged_courses):
    course_folders = next(os.walk(folder_with_merged_courses))[1]
    course_folders = course_folders[::-1]
    return course_folders

print('Number of arguments:', len(sys.argv), 'arguments.')
print('Argument List:', str(sys.argv))

folder_with_merged_courses = sys.argv[1]

output_path = sys.argv[2]

print("Folder with course exports to merge", folder_with_merged_courses)

course_folders = getCourses(folder_with_merged_courses)

print('Number of courses:', len(course_folders))

print(course_folders)

# Make first course parent and copy into output folder

parent_course_path = folder_with_merged_courses + '/' + course_folders[0] + '/course/' 
output_course_path = output_path + 'course/'
copy_tree(parent_course_path, output_course_path)

# Copy Problems, HTML, Verticals, Sequentials and Chapters from Child to Parent

for crs in course_folders[1:]:
    folders_to_copy = ['chapter', 'html', 'problem', 'sequential', 'vertical', 'video', 'static']

    for fld in folders_to_copy:
        # copy child files to parent 
        #print(folder_with_merged_courses + '/' + crs + '/course/' + fld + '/')
        #print(output_course_path + fld + '/')
        copy_tree(folder_with_merged_courses + '/' + crs + '/course/' + fld + '/' , parent_course_path + fld + '/')

# Merge the static files json between the child and parent courses

# load the parent asset file as a dict

parent_assets = parent_course_path + 'policies/assets.json'
#print(parent_assets)
parent_policy_dict = {} #json.loads(parent_assets)

with open(parent_assets, encoding='utf-8', errors='ignore') as json_data:
     parent_policy_dict = json.load(json_data, strict=False)

for crs in course_folders[1:]:
    child_policy = folder_with_merged_courses + '/' + crs + '/course/' + 'policies/assets.json'
    child_policy_dict = {}
    with open(child_policy, encoding='utf-8', errors='ignore') as json_data:
        child_policy_dict = json.load(json_data, strict=False)
    # pure magic - merges assets dictionary
    parent_policy_dict = { k : child_policy_dict[k] for k in set(child_policy_dict) - set(parent_policy_dict) }

with open(output_course_path +'assets.json', 'w') as fp:
    json.dump(parent_policy_dict, fp)

# Finally
# 1 - Copy assets.json to policies folder
# 2 - Manually copy chapter.xml contents from child courses and append to parents chapter.xml
# 3 - Compress Parent course
# $tar -cvzf mergedcourse.tar.gz /path/to/exportedmergedcourse
# 4 - Upload to edx