def test_concat_ws_sql_col_sep_only_sep_specified():
gen = StringGen(nullable=True)
sep = StringGen('[-,*,+,!]', nullable=True)
assert_gpu_sql_fallback_collect(
lambda spark: three_col_df(spark, gen, gen, sep),
'ConcatWs',
'concat_ws_table',
'select ' +
'concat_ws(c) from concat_ws_table')
def test_window_aggs_lag_ignore_nulls_fallback(a_gen, b_gen, c_gen, d_gen):
data_gen = [('a', RepeatSeqGen(a_gen, length=20)), ('b', b_gen),
('c', c_gen), ('d', d_gen)]
assert_gpu_sql_fallback_collect(
lambda spark: gen_df(spark, data_gen), 'Lag', "window_agg_table", '''
SELECT
LAG(d, 5) IGNORE NULLS OVER (PARTITION by a ORDER BY b,c) lag_d_5
FROM window_agg_table
''')
def test_split_regexp_disabled_fallback():
conf = { 'spark.rapids.sql.regexp.enabled': 'false' }
data_gen = mk_str_gen('([bf]o{0,2}:){1,7}') \
.with_special_case('boo:and:foo')
assert_gpu_sql_fallback_collect(
lambda spark : unary_op_df(spark, data_gen),
'StringSplit',
'string_split_table',
'select ' +
'split(a, "[:]", 2), ' +
'split(a, "[o:]", 5), ' +
'split(a, "[^:]", 2), ' +
'split(a, "[^o]", 55), ' +
'split(a, "[o]{1,2}", 999), ' +
'split(a, "[bf]", 2), ' +
'split(a, "[o]", 5) from string_split_table',
conf)