ClickHouse/tests/queries/0_stateless/00921_datetime64_compatibility_long.python

#!/usr/bin/env python3
# encoding: utf-8

import re
import itertools
import sys
import argparse

# Create SQL statement to verify dateTime64 is accepted as argument to functions taking DateTime.
FUNCTIONS="""
toTimeZone(N, 'UTC')
toYear(N, 'Europe/Moscow')
toQuarter(N, 'Europe/Moscow')
toMonth(N, 'Europe/Moscow')
toDayOfYear(N, 'Europe/Moscow')
toDayOfMonth(N, 'Europe/Moscow')
toDayOfWeek(N, 'Europe/Moscow')
toHour(N, 'Europe/Moscow')
toMinute(N, 'Europe/Moscow')
toSecond(N, 'Europe/Moscow')
toUnixTimestamp(N)
toStartOfYear(N, 'Europe/Moscow')
toStartOfISOYear(N, 'Europe/Moscow')
toStartOfQuarter(N, 'Europe/Moscow')
toStartOfMonth(N, 'Europe/Moscow')
toMonday(N, 'Europe/Moscow')
toStartOfWeek(N, 'Europe/Moscow')
toStartOfDay(N, 'Europe/Moscow')
toStartOfHour(N, 'Europe/Moscow')
toStartOfMinute(N, 'Europe/Moscow')
toStartOfFiveMinute(N, 'Europe/Moscow')
toStartOfTenMinutes(N, 'Europe/Moscow')
toStartOfFifteenMinutes(N, 'Europe/Moscow')
toStartOfInterval(N, INTERVAL 1 year, 'Europe/Moscow')
toStartOfInterval(N, INTERVAL 1 month, 'Europe/Moscow')
toStartOfInterval(N, INTERVAL 1 day, 'Europe/Moscow')
toStartOfInterval(N, INTERVAL 15 minute, 'Europe/Moscow')
date_trunc('year', N, 'Europe/Moscow')
date_trunc('month', N, 'Europe/Moscow')
date_trunc('day', N, 'Europe/Moscow')
date_trunc('minute', N, 'Europe/Moscow')
toTime(N, 'Europe/Moscow')
toRelativeYearNum(N, 'Europe/Moscow')
toRelativeQuarterNum(N, 'Europe/Moscow')
toRelativeMonthNum(N, 'Europe/Moscow')
toRelativeWeekNum(N, 'Europe/Moscow')
toRelativeDayNum(N, 'Europe/Moscow')
toRelativeHourNum(N, 'Europe/Moscow')
toRelativeMinuteNum(N, 'Europe/Moscow')
toRelativeSecondNum(N, 'Europe/Moscow')
toISOYear(N, 'Europe/Moscow')
toISOWeek(N, 'Europe/Moscow')
toWeek(N, 'Europe/Moscow')
toYearWeek(N, 'Europe/Moscow')
timeSlot(N, 'Europe/Moscow')
toYYYYMM(N, 'Europe/Moscow')
toYYYYMMDD(N, 'Europe/Moscow')
toYYYYMMDDhhmmss(N, 'Europe/Moscow')
addYears(N, 1, 'Europe/Moscow')
addMonths(N, 1, 'Europe/Moscow')
addWeeks(N, 1, 'Europe/Moscow')
addDays(N, 1, 'Europe/Moscow')
addHours(N, 1, 'Europe/Moscow')
addMinutes(N, 1, 'Europe/Moscow')
addSeconds(N, 1, 'Europe/Moscow')
addQuarters(N, 1, 'Europe/Moscow')
subtractYears(N, 1, 'Europe/Moscow')
subtractMonths(N, 1, 'Europe/Moscow')
subtractWeeks(N, 1, 'Europe/Moscow')
subtractDays(N, 1, 'Europe/Moscow')
subtractHours(N, 1, 'Europe/Moscow')
subtractMinutes(N, 1, 'Europe/Moscow')
subtractSeconds(N, 1, 'Europe/Moscow')
subtractQuarters(N, 1, 'Europe/Moscow')
CAST(N as DateTime('Europe/Minsk'))
CAST(N as Date)
CAST(N as UInt64)
CAST(N as DateTime64(0, 'Europe/Minsk'))
CAST(N as DateTime64(3, 'Europe/Minsk'))
CAST(N as DateTime64(6, 'Europe/Minsk'))
CAST(N as DateTime64(9, 'Europe/Minsk'))
# Casting our test values to DateTime(12) will cause an overflow and hence will fail the test under UB sanitizer.
# CAST(N as DateTime64(12, 'Europe/Moscow'))
# DateTime64(18) will always fail due to zero precision, but it is Ok to test here:
# CAST(N as DateTime64(18, 'Europe/Moscow'))
formatDateTime(N, '%C %d %D %e %F %H %I %j %m %M %p %R %S %T %u %V %w %y %Y %%', 'Europe/Moscow')
""".splitlines()

# Expanded later to cartesian product of all arguments, using format string.
extra_ops = [
    # With same type:
    (
        ['N {op} N'],
        {
            'op':
            [
                '- ', # does not work, but should it?
                '+ ', # does not work, but should it?
                '!=', '==', # equality and inequality supposed to take sub-second part in account
                '< ',
                '<=',
                '> ',
                '>='
            ]
        }
    ),
    # With other DateTime types:
    (
        [
            'N {op} {arg}',
            '{arg} {op} N'
        ],
        {
            'op':
            [
                '-', # does not work, but should it?
                '!=', '==',
                # these are naturally expected to work, but they don't:
                '< ',
                '<=',
                '> ',
                '>='
            ],
            'arg': ['DT', 'D', 'DT64'],
        }
    ),
    # With arithmetic types
    (
        [
            'N {op} {arg}',
            '{arg} {op} N'
        ],
        {
            'op':
            [
                '+ ',
                '- ',
                '==',
                '!=',
                '< ',
                '<=',
                '> ',
                '>='
            ],
            'arg':
            [
                'toUInt8(1)',
                'toInt8(-1)',
                'toUInt16(1)',
                'toInt16(-1)',
                'toUInt32(1)',
                'toInt32(-1)',
                'toUInt64(1)',
                'toInt64(-1)'
            ],
        },
    ),
]

# Expand extra_ops here
for funcs, args in extra_ops:
    args_keys = list(args.keys())
    for args_vals in itertools.product(*list(args.values())):
        for func in funcs:
            result_func = func.format(**dict(list(zip(args_keys, args_vals))))
            FUNCTIONS.append(result_func)

# filter out empty lines and commented out lines
COMMENTED_OUT_LINE_RE = re.compile(r"^\s*#")
FUNCTIONS = list([f for f in FUNCTIONS if len(f) != 0 and COMMENTED_OUT_LINE_RE.match(f) == None])
TYPES = ['D', 'DT', 'DT64']

def escape_string(s):
    if sys.version_info[0] > 2:
        return s.encode('unicode_escape').decode('utf-8').replace("'", "\\'")
    else:
        return s.encode('string-escape').decode('utf-8')


def execute_functions_for_types(functions, types):
    # NOTE: use string.Template here to allow lines with missing keys, like type, e.g. SELECT CAST(toDateTime64(1234567890), 'DateTime64')
    for func in functions:
        print(("""SELECT 'SELECT {func}';""".format(func=escape_string(func))))
        for dt in types:
            prologue = "\
WITH \
toDateTime64('2019-09-16 19:20:11.234', 3, 'Europe/Minsk') as DT64, \
toDateTime('2019-09-16 19:20:11', 'Europe/Minsk') as DT, \
toDate('2019-09-16') as D, {X} as N".format(X=dt)
            print(("""{prologue} SELECT toTypeName(r), {func} as r FORMAT CSV;""".format(prologue=prologue, func=func)))
        print("""SELECT '------------------------------------------';""")

def main():
    def parse_args():
        parser = argparse.ArgumentParser()
        parser.add_argument('--functions_re', type=re.compile, help="RE to enable functions", default=None)
        parser.add_argument('--types_re',
                type=lambda s: re.compile('^(' + s + ')$'),
                help="RE to enable types, supported types: " + ",".join(TYPES), default=None)
        parser.add_argument('--list_functions', action='store_true', help="List all functions to be tested and exit")
        return parser.parse_args()

    args = parse_args()

    functions = FUNCTIONS
    types = TYPES

    if args.functions_re:
        functions = list([f for f in functions if args.functions_re.search(f)])
        if len(functions) == 0:
            print("functions list is empty")
            return -1

    if args.types_re:
        types = list([t for t in types if args.types_re.match(t)])
        if len(types) == 0:
            print("types list is empty")
            return -1

    if args.list_functions:
        print(("\n".join(functions)))
        return 0

    execute_functions_for_types(functions, types)

if __name__ == '__main__':
    exit(main())