Script I wrote to fix a problem that we found in VCDB where a few of the incidents did not have a valid confidentiality attribute. We created new json files, but we couldn't overwrite the old ones because other changes would have been lost. This script maps the reference field to the filename so we can replace just the confidentiality section of…

prc_fix.py

import os
import json

vcdb_path = 'Documents/development/python/VCDB/incidents'
prc_fix_path = 'Downloads/prc-fix'
mapping = {}

for filename in os.listdir(vcdb_path):
  if filename.endswith('.json'):
    i = json.loads(open(os.path.join(vcdb_path,filename)).read())
    if 'reference' in i.keys():
      mapping[i['reference']] = i['incident_id']

for filename in os.listdir(prc_fix_path):
  if filename.endswith('.json'):
    print "reading file",filename
    i = json.loads(open(os.path.join(prc_fix_path,filename)).read())
    fix_incident = json.loads(open(os.path.join(vcdb_path,mapping[i['reference']] + '.json')).read())
    print "\tFixing file",mapping[i['reference']]
    fix_incident['attribute']['confidentiality'] = i['attribute']['confidentiality']
    try:
        fix_incident['attribute'].pop(' confidentiality')
    except:
        next
    outfile = open(os.path.join(vcdb_path,mapping[i['reference']] + '.json'),'w')
    outfile.write(json.dumps(fix_incident,indent=2, sort_keys=True, separators=(',', ': ')))