def get_file_format(the_line): email_field = -1 delim = None sep = None fields = parse(the_line) debug(fields) i = 0 for f in fields: if validate_email(f): email_field = i break i += 1 ff = file_format() ff.field_count = len(fields) ff.email_field = email_field if ff.email_field == -1 and i == 1: #only one field, assume email ff.email_field = 0 return ff
def process_file(infile):
global config
good = 0
bad = 0
cleaned = 0
total = 0
pf = ProcessedFile()
outfile = outfile_name(infile)
#open file streams
instream = open(infile, 'r')
outstream = open(outfile, 'aw')
ff = None #file_format obj
for line in instream:
total += 1
if (total == 1 and config['has_header']):
ff = get_file_format(line)
if ff.email_field == -1:
#get it from user
fields = parse(line)
email_field = ask_user_which_field_is_email(fields)
email_field -= 1 #comp for 0 index
ff.email_field = email_field
outstream.write(line) #write header
continue
fields = parse(line, ff)
email = fields[ff.email_field]
email = clean_email(email)
print total, " -- ", len(fields)
if validate_email(email):
good += 1
if email != fields[ff.email_field]:
cleaned += 1
line = line.replace(fields[ff.email_field], email)
#TODO: need to get the line back from feilds, but for now just write the line
outstream.write(line)
else:
bad += 1
pf.good_emails = good
pf.bad_emails = bad
if config['has_header']:
total -= 1
pf.total_emails = total
#close streams
instream.close()
outstream.close()
ow = get_opt('overwrite_files')
if ow:
os.rename(outfile, infile)
return pf