def admissionDates(fromDate, toDate):
if fromDate == '':
if toDate == '':
return
fromDate = toDate
toDate = ''
curDate = util.toTime(fromDate)
endDate = util.toTime(toDate) if toDate != '' else curDate
while curDate <= endDate:
handleStatusEvent(curDate, STATUS_IN)
curDate = util.nextDay(curDate)
def admissionDates(fromDate, toDate):
if fromDate == '':
if toDate == '':
return
fromDate = toDate
toDate = ''
curDate = util.toTime(fromDate)
endDate = util.toTime(toDate) if toDate != '' else curDate
while curDate <= endDate:
handleStatusEvent(curDate, STATUS_IN)
curDate = util.nextDay(curDate)
def emitLab(date, loinc, result, resultFlag):
# TODO add provider/physician here as well?
event = createEntry(TYPE_LABTEST, "loinc__" + loinc, claim_id,
result != '' or resultFlag != '', resultFlag,
result)
event['time'] = util.toTime(date)
obj['events'].append(event)
def dates(fromDate, toDate):
if fromDate == '':
if toDate == '':
return
fromDate = toDate
toDate = ''
curDate = util.toTime(fromDate)
endDate = util.toTime(toDate) if toDate != '' else curDate
while curDate <= endDate:
for event in handleEvent(row, claim_id):
event['time'] = curDate
handleKey(row, "claim_amount", MODE_OPTIONAL,
lambda amount: addCost(event, amount))
obj['events'].append(event)
handleStatusEvent(curDate, curStatus)
handleKey(
row, "location_flag", MODE_OPTIONAL,
lambda flag: handleStatusEvent(
curDate, STATUS_FLAG_MAP.get(flag, STATUS_UNKNOWN)))
curDate = util.nextDay(curDate)
def dates(fromDate, toDate):
if fromDate == '':
if toDate == '':
return
fromDate = toDate
toDate = ''
curDate = util.toTime(fromDate)
endDate = util.toTime(toDate) if toDate != '' else curDate
while curDate <= endDate:
for event in handleEvent(row, claim_id):
event['time'] = curDate
handleKey(row, "claim_amount", MODE_OPTIONAL, lambda amount:
addCost(event, amount)
)
obj['events'].append(event)
handleStatusEvent(curDate, curStatus)
handleKey(row, "location_flag", MODE_OPTIONAL, lambda flag:
handleStatusEvent(curDate, STATUS_FLAG_MAP.get(flag, STATUS_UNKNOWN))
)
curDate = util.nextDay(curDate)
def readShelve(pid, settings, output):
pids = [ pid ]
if pid == '--all':
pids = getAll(settings)
out = {
'delim': settings['delim'],
'quote': settings['quote'],
'out': output
}
anonymize = settings['anonymize']['do']
first = True
for patientId in pids:
if anonymize:
realId = hashlib.sha1(patientId).hexdigest()
age_shift = 0
while age_shift == 0:
age_shift = random.randint(-10, 10)
date_shift = 0
while date_shift == 0:
date_shift = random.randint(-365 * 10, 365 * 10)
else:
realId = patientId
join_id = settings['join_id']
splitter = settings['row_split']
(row_defs, length, all_hdrs) = openDB(patientId, settings, out, first)
first = False
for row_def in row_defs:
start = row_def['start']
col_num = row_def['col_num']
skip = row_def['skip']
# manipulation ixs apply before skipping
age_ixs = [ ix - start for ix in xrange(start, start + col_num) if all_hdrs[ix] in settings['anonymize']['age_columns'] ]
date_ixs = [ ix - start for ix in xrange(start, start + col_num) if all_hdrs[ix] in settings['anonymize']['date_columns'] ]
redact_ixs = [ ix - start for ix in xrange(start, start + col_num) if all_hdrs[ix] in settings['anonymize']['redact_columns'] ]
for row in row_def['data']:
if row == '':
continue
values = row.strip().split(splitter)
if anonymize:
for ix in age_ixs:
values[ix] = str(int(values[ix]) + age_shift)
for ix in date_ixs:
values[ix] = util.from_time(util.shift_days(util.toTime(values[ix]), date_shift))
for ix in redact_ixs:
values[ix] = ''
id = values.pop(skip)
if len(values) != col_num:
print("column mismatch! expected {0} got {1}: {2}".format(str(col_num), str(len(values)), row), file=sys.stderr)
continue
if id != patientId:
print("unexpected id! expected {0} got {1}: {2}".format(patientId, id, row))
continue
writeRow(values, out, start, length, realId)
def emitNDC(date, ndc):
# TODO add provider/physician here as well?
event = createEntry(TYPE_PRESCRIBED, "ndc__" + ndc, claim_id)
curDate = util.toTime(date)
event['time'] = curDate
handleKey(
row, "location_flag", MODE_OPTIONAL,
lambda flag: handleStatusEvent(
curDate, STATUS_FLAG_MAP.get(flag, STATUS_UNKNOWN)))
handleKey(row, "prescribed_amount", MODE_OPTIONAL,
lambda amount: addCost(event, amount))
obj['events'].append(event)
def emitNDC(date, ndc):
# TODO add provider/physician here as well?
event = createEntry(TYPE_PRESCRIBED, "ndc__" + ndc, claim_id)
curDate = util.toTime(date)
event['time'] = curDate
handleKey(row, "location_flag", MODE_OPTIONAL, lambda flag:
handleStatusEvent(curDate, STATUS_FLAG_MAP.get(flag, STATUS_UNKNOWN))
)
handleKey(row, "prescribed_amount", MODE_OPTIONAL, lambda amount:
addCost(event, amount)
)
obj['events'].append(event)
def to_time(self, value):
return util.toTime(value.strftime("%Y%m%d"))
def date():
lit = literal()
try:
return util.toTime(lit)
except Exception, e:
err(State.pos - len(lit), State.pos, "cannot convert to date: " + str(e))
def to_time(self, value):
return util.toTime(value.strftime("%Y%m%d"))
while args:
arg = args.pop(0)
if arg == '--':
break
if arg == '-h':
usage()
if arg == '--num-cutoff':
if not args or args[0] == '--':
print('--num-cutoff requires number', file=sys.stderr)
usage()
num_cutoff = int(args.pop(0))
elif arg == '--age-time':
if not args or args[0] == '--':
print('--age-time requires a date', file=sys.stderr)
usage()
age_time = util.toTime(args.pop(0))
elif arg == '--from':
if not args or args[0] == '--':
print('--from requires a date', file=sys.stderr)
usage()
from_time = util.toTime(args.pop(0))
elif arg == '--to':
if not args or args[0] == '--':
print('--to requires a date', file=sys.stderr)
usage()
to_time = util.toTime(args.pop(0))
elif arg == '-w':
if not args or args[0] == '--':
print('-w requires whitelist file', file=sys.stderr)
usage()
if whitelist is None:
def emitLab(date, loinc, result, resultFlag):
# TODO add provider/physician here as well?
event = createEntry(TYPE_LABTEST, "loinc__" + loinc, claim_id, result != '' or resultFlag != '', resultFlag, result)
event['time'] = util.toTime(date)
obj['events'].append(event)
while args:
arg = args.pop(0)
if arg == '--':
break
if arg == '-h':
usage()
if arg == '--num-cutoff':
if not args or args[0] == '--':
print('--num-cutoff requires number', file=sys.stderr)
usage()
num_cutoff = int(args.pop(0))
elif arg == '--age-time':
if not args or args[0] == '--':
print('--age-time requires a date', file=sys.stderr)
usage()
age_time = util.toTime(args.pop(0))
elif arg == '--from':
if not args or args[0] == '--':
print('--from requires a date', file=sys.stderr)
usage()
from_time = util.toTime(args.pop(0))
elif arg == '--to':
if not args or args[0] == '--':
print('--to requires a date', file=sys.stderr)
usage()
to_time = util.toTime(args.pop(0))
elif arg == '-w':
if not args or args[0] == '--':
print('-w requires whitelist file', file=sys.stderr)
usage()
if whitelist is None:
def readShelve(pid, settings, output):
pids = [pid]
if pid == '--all':
pids = getAll(settings)
out = {
'delim': settings['delim'],
'quote': settings['quote'],
'out': output
}
anonymize = settings['anonymize']['do']
first = True
for patientId in pids:
if anonymize:
realId = hashlib.sha1(patientId).hexdigest()
age_shift = 0
while age_shift == 0:
age_shift = random.randint(-10, 10)
date_shift = 0
while date_shift == 0:
date_shift = random.randint(-365 * 10, 365 * 10)
else:
realId = patientId
join_id = settings['join_id']
splitter = settings['row_split']
(row_defs, length, all_hdrs) = openDB(patientId, settings, out, first)
first = False
for row_def in row_defs:
start = row_def['start']
col_num = row_def['col_num']
skip = row_def['skip']
# manipulation ixs apply before skipping
age_ixs = [
ix - start for ix in xrange(start, start + col_num)
if all_hdrs[ix] in settings['anonymize']['age_columns']
]
date_ixs = [
ix - start for ix in xrange(start, start + col_num)
if all_hdrs[ix] in settings['anonymize']['date_columns']
]
redact_ixs = [
ix - start for ix in xrange(start, start + col_num)
if all_hdrs[ix] in settings['anonymize']['redact_columns']
]
for row in row_def['data']:
if row == '':
continue
values = row.strip().split(splitter)
if anonymize:
for ix in age_ixs:
values[ix] = str(int(values[ix]) + age_shift)
for ix in date_ixs:
values[ix] = util.from_time(
util.shift_days(util.toTime(values[ix]),
date_shift))
for ix in redact_ixs:
values[ix] = ''
id = values.pop(skip)
if len(values) != col_num:
print("column mismatch! expected {0} got {1}: {2}".format(
str(col_num), str(len(values)), row),
file=sys.stderr)
continue
if id != patientId:
print("unexpected id! expected {0} got {1}: {2}".format(
patientId, id, row))
continue
writeRow(values, out, start, length, realId)
