Пример #1
0
def test_valid_data_parsing_type_preserved(input_data, input_format):
    if input_format:
        func = DatetimeParser(formats=[input_format],
                              enforce_type=False).build()
    else:
        func = DatetimeParser(enforce_type=False).build()
    assert func(input_data) == input_data
Пример #2
0
def test_enforce_type():
    func = DatetimeParser(enforce_type=False).build()
    data = "20200101"
    assert func(data) == "20200101"
    func = DatetimeParser(enforce_type=True).build()
    data = "20200101"
    assert func(data) == datetime.datetime.strptime('20200101', '%Y%m%d')
Пример #3
0
def test_invalid_data_parsing(input_data, input_format):
    if input_format:
        func = DatetimeParser(formats=[input_format],
                              enforce_type=False).build()
    else:
        func = DatetimeParser(enforce_type=False).build()
    with pytest.raises(DateTimeParsingException):
        assert func(input_data)
Пример #4
0
def test_min_value_validator():
    func = DatetimeParser(formats=['%Y-%m-%d']).min_value(
        value='01/01/2020', format='%d/%m/%Y').build()
    assert func('2020-12-31')
    assert func('2020-01-01')
    with pytest.raises(MinimumValueConstraintException):
        assert func('2019-12-31')

    with pytest.raises(UnexpectedParsingException):
        DatetimeParser().max_value("300").build()
Пример #5
0
def test_not_null_validator():
    func = DatetimeParser(quoted=0, enforce_type=False).not_null().build()
    assert func("20200101") == "20200101"
    with pytest.raises(NullValueInNotNullFieldException):
        assert func(None)
        assert func("") == ""

    # Default value assignment check
    func = DatetimeParser().not_null(default_value="20200101",
                                     format='%Y%m%d').build()
    assert func(None) == datetime.datetime.strptime('20200101', '%Y%m%d')
    assert func("") == datetime.datetime.strptime('20200101', '%Y%m%d')
Пример #6
0
def test_valid_data_parsing_type_converted(input_data, input_format):
    if input_format:
        func = DatetimeParser(formats=[input_format]).build()
    else:
        func = DatetimeParser().build()
    if not input_format:
        if len(input_data) == 8:
            input_format = '%Y%m%d'
        else:
            input_format = '%Y%m%d%H%M%S'
    assert func(input_data) == datetime.datetime.strptime(
        input_data, input_format)
Пример #7
0
def test_value_set_validator():
    allowed_values = [
        datetime.datetime.strptime('20200101', '%Y%m%d'),
        datetime.datetime.strptime('20200102', '%Y%m%d')
    ]
    func = DatetimeParser().value_set(allowed_values).build()
    assert func('20200101') == datetime.datetime.strptime('20200101', '%Y%m%d')
    with pytest.raises(ValidValueCheckException):
        assert func('20200103')
Пример #8
0
def test_range_validator():
    func = DatetimeParser(formats=['%Y-%m-%d']).range(
        lower_bound='01/01/2020',
        upper_bound=datetime.datetime.strptime('20201231', '%Y%m%d'),
        format='%d/%m/%Y').build()
    assert func('2020-01-01')
    assert func('2020-12-31')
    assert func('2020-06-30')
    with pytest.raises(MinimumValueConstraintException):
        assert func('2019-12-31')
    with pytest.raises(MaximumValueConstraintException):
        assert func('2021-01-01')
def _schema(custom_function):
    return [
        ('ID', StringParser(quoted=1)),
        ('RUN_ID', StringParser().regex_match(r'\w+_\d{4}-\d{2}-\d{2}').change_case('u')),
        ('CLASS', StringParser(start=1, end=1).value_set(['a', 'b', 'A'])),
        ('INITIATED_ON', DatetimeParser(formats=['%Y%m%d', '%Y-%m-%d %H:%M:%S'])
         .convert('%Y/%m/%d').max_value(datetime.datetime.now())
         .min_value(value='20000101', format='%Y%m%d')
         .not_null(datetime.datetime.strptime('19001231', '%Y%m%d'))
         ),
        ('ASKED_AMOUNT', IntegerParser().max_value(2000).not_null(default_value=0)),
        ('ADJUSTED_AMOUNT', FloatParser().min_value(10.0).not_null(0.0)),
        ('ROLE_MODEL', ConstantParser('Leo Messi')),
        ('BLOCK_NUMBER', IntegerParser().add_func(custom_function).range(0, 40))
    ]
Пример #10
0
def test_single_quoted_data():
    func = DatetimeParser(quoted=2).build()
    data = "'20200101'"
    assert func(data) == datetime.datetime.strptime('20200101', '%Y%m%d')
Пример #11
0
def test_double_quoted_data():
    func = DatetimeParser(quoted=1).build()
    data = '"20200101"'
    assert func(data) == datetime.datetime.strptime('20200101', '%Y%m%d')
Пример #12
0
def test_non_quoted_data():
    func = DatetimeParser(enforce_type=False).build()
    data = "20200101"
    assert func(data) == "20200101"
    int_data = 20200101
    assert func(int_data) == 20200101
Пример #13
0
def test_convert():
    func = DatetimeParser().convert('%Y|%m|%d').build()
    assert func("20200501") == "2020|05|01"
Пример #14
0
def test_add_func_validator():
    func = DatetimeParser().add_func(_get_month_value).build()
    assert func("20200501") == 5
Пример #15
0
# Import required parser class
from parseval.parser import DatetimeParser
import datetime
import logging

logging.basicConfig(format='%(levelname)s:%(asctime)s:: %(message)s',
                    level=logging.DEBUG)
basic_parser = DatetimeParser(formats=['%Y%m%d'])  # Create basic parser object
basic_parse_func = basic_parser.build()  # Build the parser function
input_data = "20200101"  # Input Data
basic_parsed_output = basic_parse_func(input_data)  # Parse data

logging.info(('#' * 50) + " DATETIME PARSING " + ('#' * 50))
logging.info("====> Simple Data Parsing example:")
logging.info("Input 1: {}".format(input_data))
logging.info("Output 1: {}".format(basic_parsed_output))
logging.info('\n')

# Now let's see some available validators, to get the idea of how to use those
# Note, we will not go through all available validators, because all validators work in same fashion
# Syntax and description of all validators are available in documentation
default_date = datetime.datetime.strptime("20200101", "%Y%m%d")
min_date = datetime.datetime.strptime("20200101", "%Y%m%d")
max_date = datetime.datetime.strptime("20200831", "%Y%m%d")
validation_parser = DatetimeParser()\
    .range(min_date, max_date)\
    .not_null(default_value=default_date)  # null check validation and allowed values validation is added
validation_parse_func = validation_parser(
)  # Yes, you can directly call the object to build the parser

more_generic_parser = DatetimeParser(formats=['%Y%m%d', '%Y%m%d %H%M%S'
Пример #16
0
        if i_data % 2 != 0:
            raise Exception("The data has to be even!")
    return data


# The cursors has to provided in a list, where each element of the list is a tuple.
# First element of each element is the column name, this is just for reference, no internal usage
# Second element of each tuple is the actual parser (parser objects, not built parser function)
schema = [
    ('ID', StringParser(quoted=1)),
    ('RUN_ID',
     StringParser().regex_match(r'\w+_\d{4}-\d{2}-\d{2}').change_case('u')),
    ('CLASS', StringParser(start=1, end=1).value_set(['a', 'b', 'A'])),
    ('INITIATED_ON', DatetimeParser(
        formats=['%Y%m%d', '%Y-%m-%d %H:%M:%S']).convert('%Y/%m/%d').max_value(
            datetime.datetime.now()).min_value(value='20000101',
                                               format='%Y%m%d').not_null(
                                                   datetime.datetime.strptime(
                                                       '19001231', '%Y%m%d'))),
    ('ASKED_AMOUNT',
     IntegerParser().max_value(2000).not_null(default_value=0)),
    ('ADJUSTED_AMOUNT', FloatParser().min_value(10.0).not_null(0.0)),
    ('ROLE_MODEL', ConstantParser('Iron-Man')),
    ('BLOCK_NUMBER', IntegerParser().add_func(_parity_check).range(0, 40))
]

p = Parser(schema=schema, stop_on_error=1, parsed_row_format='dict')
# Creating temporary file for the example
with tempfile.NamedTemporaryFile() as tf:
    with open(tf.name, 'w') as sf:
        sf.writelines('""|Trig2020-23-12|A|20200123|2000|21.0934||10\n')
        sf.writelines('"DEF"||abc|||||34\n')