# Extract data fields
output = pd.DataFrame()
for fn_idx, fn in enumerate(filenames):
  cfs = re.compile(r"Call For Service Detail Report - CFS ([0-9]+)\n")
  cfs_matches = np.array(cfs.findall(data[fn_idx]))
  this_output = pd.DataFrame(index=np.arange(len(cfs_matches)),data={'CFS':cfs_matches})

  this_output['Filename'] = fn

  address = re.compile(r"Address\n\n(.+)\n\n")
  address_matches = address.findall(data[fn_idx])
  this_output['Original address'] = [capwords(s) for s in address_matches]
  this_output[['Original address','Town/city']] = \
    this_output['Original address'].str.split(', ',expand=True)
  this_output[['Primary street','Cross street']] = \
    this_output['Original address'].str.split('@ ',expand=True)
  this_output.loc[~this_output['Original address'].str.contains('@'),'Address'] = \
    this_output.loc[~this_output['Original address'].str.contains('@'),'Primary street']
  this_output.loc[~this_output['Original address'].str.contains('@'),'Primary street'] = ''
  this_output.loc[this_output['Address'].isnull(),'Address'] = ''

  date = re.compile(r"Create Date\n\n(.+) (.+ .+)\n\n")
  date_matches = date.findall(data[fn_idx])
  date_matches_dist = []
  date_match_idx = 0
  for sub_cad_log in data[fn_idx].split('Call For Service Detail Report - CFS')[1:]:
    if date_match_idx >= len(date_matches):
      date_matches_dist.append(('',''))
    elif (date_matches[date_match_idx][0] in sub_cad_log) and \
        (date_matches[date_match_idx][1] in sub_cad_log):
      date_matches_dist.append(date_matches[date_match_idx])
      date_match_idx += 1
    else:
      date_matches_dist.append(('',''))
  this_output['Date'] = pd.to_datetime(np.array(date_matches_dist)[:,0])
  this_output['Time'] = pd.to_datetime(np.array(date_matches_dist)[:,1])
  this_output['Time'] = this_output['Time'].dt.floor('Min').dt.time

  incnum = re.compile(r"Incident Number\n(.+(?:\n.+)*)\n\nJurisdiction")
  incnum_matches = [s.rstrip("\nUnits") for s in incnum.findall(data[fn_idx])]
  incnum_matches_dist = []
  incnum_match_idx = 0
  for sub_cad_log in data[fn_idx].split('Call For Service Detail Report - CFS')[1:]:
    if incnum_match_idx >= len(incnum_matches):
      incnum_matches_dist.append('')
    elif incnum_matches[incnum_match_idx] in sub_cad_log:
      incnum_matches_dist.append(incnum_matches[incnum_match_idx])
      incnum_match_idx += 1
    else:
      incnum_matches_dist.append('')
  incnum_matches_dist = ['CHECK_INC_NUM' if s == '' else s.replace('\n',', ') \
                         for s in incnum_matches_dist]
  this_output['Incident'] = incnum_matches_dist

  this_output['IDs'] = this_output['CFS'].str.cat(this_output['Incident'],sep=', ')

  this_output = this_output[['Filename','IDs','Town/city','Address',
                             'Primary street','Cross street','Date','Time']]
  output = output.append(this_output)

output = output.reset_index().drop(columns=['index'])

output.to_excel(cad_dir + 'SNO911_parsed.xlsx')
display(output)