def parse(self, file_contents):
assert file_contents is not None
xlsx_file = StringIO.StringIO(file_contents)
workbook = load_workbook(xlsx_file, use_iterators=True)
all_sheets = workbook.worksheets
if len(all_sheets) == 1:
raise CodeSheetMissingException()
elif len(all_sheets) > 1:
codes_sheet = self._get_code_sheet(all_sheets)
worksheet = self._get_worksheet(workbook.worksheets)
rows = []
for cs in codes_sheet.iter_rows():
rows = [self._get_value(x.value) for x in cs]
header, header_found = self._is_header_row(rows)
parsed_data = []
form_code = REGISTRATION_FORM_CODE
header = header[1:]
for row in worksheet.iter_rows(row_offset=1):
row_values = [self._get_value(x.value) for x in row]
values = dict(zip(header, row_values))
values.update({"t": "reporter"})
parsed_data.append((form_code, values))
if not header_found:
raise XlsParserInvalidHeaderFormatException()
return parsed_data
def parse(self, xls_contents):
assert xls_contents is not None
workbook = xlrd.open_workbook(file_contents=xls_contents)
all_sheets = workbook.sheets()
if len(all_sheets) == 1:
raise CodeSheetMissingException()
elif len(all_sheets)>1:
codes_sheet = self._get_code_sheet(all_sheets)
worksheet = self._get_worksheet(all_sheets)
row = codes_sheet.row_values(0)
header, header_found = self._is_header_row(row)
form_code = header[0]
header = header[1:]
empty_indexes = self.get_empty_indexes(header)
header = self.remove_empty_indexes(header, empty_indexes)
parsed_data = []
for row_num in range(1, worksheet.nrows):
row = worksheet.row_values(row_num)
row = self._clean(row)
row = self.remove_empty_indexes(row, empty_indexes)
values = OrderedDict(zip(header, row))
parsed_data.append((form_code, values))
if not header_found:
raise XlsParserInvalidHeaderFormatException()
return parsed_data
def _get_code_sheet(self, all_sheets):
code_sheets = [
work_sheet for work_sheet in all_sheets
if work_sheet.title == 'codes' and work_sheet.max_column > 1
]
if len(code_sheets):
return code_sheets[0]
raise CodeSheetMissingException()
def _get_code_sheet(self, all_sheets):
code_sheets = [
work_sheet for work_sheet in all_sheets
if work_sheet.name == 'codes' and len(work_sheet._cell_values)
]
if len(code_sheets):
return code_sheets[0]
raise CodeSheetMissingException()