def fix_collections(options):
"""Make HTTP requests to fix the order of forms in the OLD's collections.
"""
# Get raw LingSync JSON data.
ls_json_file = getattr(options, 'ls_json_file')
try:
ls_data = json.load(open(ls_json_file))
except:
sys.exit(u'%sUnable to locate file %s. Aborting.%s' % (ANSI_FAIL,
ls_json_file, ANSI_ENDC))
# Get converted OLD data.
old_json_file = getattr(options, 'old_json_file')
try:
old_data = json.load(open(old_json_file))
except:
sys.exit(u'%sUnable to locate file %s. Aborting.%s' % (ANSI_FAIL,
old_json_file, ANSI_ENDC))
forms = old_data['forms']
# `datums` holds the raw LingSync dicts representing all of the datums.
datumid2dateentered = {}
for datum in (r['doc'] for r in ls_data['rows']
if get_collection_for_lingsync_doc(r['doc']) == 'datums'):
datumid2dateentered[datum['_id']] = datum['dateEntered']
# Get an OLD client.
old_url = getattr(options, 'old_url', None)
old_username = getattr(options, 'old_username', None)
old_password = getattr(options, 'old_password', None)
c = OLDClient(old_url)
# Log in to the OLD.
logged_in = c.login(old_username, old_password)
if not logged_in:
sys.exit(u'%sUnable to log in to %s with username %s and password %s.'
u' Aborting.%s' % (ANSI_FAIL, old_url, old_username, old_password,
ANSI_ENDC))
# Populate the `formid2dateentered` dict, so that it maps OLD form ids to
# date entered values taken from the raw LingSync data.
formid2dateentered = {}
patt3 = re.compile('This form was created from LingSync datum (\w+)')
for form in c.get('forms'):
form_id = form['id']
datum_id = patt3.findall(form['comments'])
if len(datum_id) == 0:
print '%sUnable to find LingSync datum id for OLD form %d: %s.%s' % (
ANSI_WARNING, form_id, form['transcription'], ANSI_ENDC)
datum_id = None
else:
if len(datum_id) > 1:
print ('%sWarning: found multiple LingSync datum ids for OLD'
' form %d.%s' % (ANSI_WARNING, form_id, ANSI_ENDC))
datum_id = datum_id[0]
if datum_id:
date_entered = datumid2dateentered[datum_id]
else:
date_entered = '0'
formid2dateentered[form_id] = date_entered
# Issue the requests to fix each of the OLD collections, in turn.
collections = c.get('collections')
# print len(collections)
patt1 = re.compile('^(form\[\d+\])*$')
patt2 = re.compile('form\[(\d+)\]')
manualfix = {}
for collection in collections:
# print collection['contents']
# If there's anything besides form references in the collection, then
# we know the user has manually updated it and we can't fix it
# automatedly; best we can do is tell the user the order of form
# references that matches the LingSync version.
tmp = collection['contents'].replace(' ', '').replace('\n', '')
if patt1.search(tmp) or collection['contents'].strip() == '':
contents_modified = False
else:
contents_modified = True
# print '\n%d' % collection['id']
current_form_ids = map(int, patt2.findall(tmp))
sorted_form_ids = [x[1] for x in sorted(
[(formid2dateentered[id_], id_) for id_ in current_form_ids])]
new_contents = '\n'.join(['form[%d]' % x for x in sorted_form_ids])
if contents_modified:
manualfix[collection['id']] = new_contents
else:
if current_form_ids == sorted_form_ids:
print ('Collection %d already has its forms in the correct'
' order.' % collection['id'])
else:
print 'Fixing collection %d.' % collection['id']
collection['contents'] = new_contents
# We must fix any relational data and or date elicited values
# for the update request.
if collection['elicitor']:
collection['elicitor'] = collection['elicitor']['id']
if collection['speaker']:
collection['speaker'] = collection['speaker']['id']
if collection['source']:
collection['source'] = collection['source']['id']
if collection['tags']:
collection['tags'] = [t['id'] for t in collection['tags']]
if collection['files']:
collection['files'] = [t['id'] for t in collection['files']]
if collection['date_elicited']:
# Convert yyyy-mm-dd to mm/dd/yyyy format
parts = collection['date_elicited'].split('-')
collection['date_elicited'] = '%s/%s/%s' % (parts[1],
parts[2], parts[0])
resp = c.put('collections/%d' % collection['id'], collection)
if resp.get('contents') != new_contents:
print ('Something went wrong when attempting to update the'
' contents of collection %d. It should have the following'
' contents value\n%s' % (collection['id'], new_contents))
p(resp)
for id in manualfix:
new_contents = manualfix[id]
print ('Collection %d has been altered by a user on the OLD so we'
' can\'t fix its form order here. You will have to do it. Please make'
' sure that the order of form references matches the following:\n%s.' % (
id, new_contents))
print 'Done.'