ops.RegexSearch:
_regex_search,
ops.RegexExtract:
_regex_extract,
ops.RegexReplace:
_regex_replace,
ops.GroupConcat:
_reduction('STRING_AGG'),
ops.IfNull:
fixed_arity('IFNULL', 2),
ops.Cast:
_cast,
ops.StructField:
_struct_field,
ops.ArrayCollect:
unary('ARRAY_AGG'),
ops.ArrayConcat:
_array_concat,
ops.ArrayIndex:
_array_index,
ops.ArrayLength:
unary('ARRAY_LENGTH'),
ops.HLLCardinality:
_reduction('APPROX_COUNT_DISTINCT'),
ops.Log:
_log,
ops.Sign:
unary('SIGN'),
ops.Modulus:
fixed_arity('MOD', 2),
ops.Date:
ops.StringAscii: _string_ascii,
ops.StringFind: _string_find,
ops.StrRight: _string_right,
ops.Repeat: fixed_arity('REPEAT', 2),
ops.RegexSearch: _regex_search,
ops.RegexExtract: _regex_extract,
ops.RegexReplace: _regex_replace,
ops.GroupConcat: fixed_arity('STRING_AGG', 2),
ops.IfNull: fixed_arity('IFNULL', 2),
ops.Cast: _cast,
ops.StructField: _struct_field,
ops.ArrayCollect: unary('ARRAY_AGG'),
ops.ArrayConcat: _array_concat,
ops.ArrayIndex: _array_index,
ops.ArrayLength: unary('ARRAY_LENGTH'),
# BigQuery doesn't have these operations built in.
# ops.ArrayRepeat: _array_repeat,
# ops.ArraySlice: _array_slice,
ir.Literal: _literal,
})
_invalid_operations = {
ops.Translate,
ops.FindInSet,
ops.Capitalize,
ops.IsNan,
ops.StringAscii: _string_ascii,
ops.StringFind: _string_find,
ops.StrRight: _string_right,
ops.Repeat: fixed_arity('REPEAT', 2),
ops.RegexSearch: _regex_search,
ops.RegexExtract: _regex_extract,
ops.RegexReplace: _regex_replace,
ops.GroupConcat: fixed_arity('STRING_AGG', 2),
ops.IfNull: fixed_arity('IFNULL', 2),
ops.Cast: _cast,
ops.StructField: _struct_field,
ops.ArrayCollect: unary('ARRAY_AGG'),
ops.ArrayConcat: _array_concat,
ops.ArrayIndex: _array_index,
ops.ArrayLength: unary('ARRAY_LENGTH'),
ops.HLLCardinality: _reduction('APPROX_COUNT_DISTINCT'),
ops.Log: _log,
ops.Sign: unary('SIGN'),
ops.Modulus: fixed_arity('MOD', 2),
ops.Date: unary('DATE'),
# BigQuery doesn't have these operations built in.
# ops.ArrayRepeat: _array_repeat,
# ops.ArraySlice: _array_slice,
ops.Literal: _literal,
def _round(translator, expr):
op = expr.op()
arg, digits = op.args
arg_formatted = translator.translate(arg)
if digits is not None:
digits_formatted = translator.translate(digits)
return 'bround({}, {})'.format(arg_formatted, digits_formatted)
return 'bround({})'.format(arg_formatted)
_operation_registry = impala_compiler._operation_registry.copy()
_operation_registry.update({
ops.IsNan:
unary('isnan'),
ops.IfNull:
fixed_arity('ifnull', 2),
ops.StructField:
_struct_field,
ops.MapValueForKey:
_map_value_for_key,
ops.ArrayLength:
unary('size'),
ops.Round:
_round,
ops.HLLCardinality:
_reduction('approx_count_distinct'),
ops.StrRight:
fixed_arity('right', 2),
ops.StringSplit:
ops.RegexSearch:
_regex_search,
ops.RegexExtract:
_regex_extract,
ops.RegexReplace:
_regex_replace,
ops.GroupConcat:
fixed_arity('STRING_AGG', 2),
ops.IfNull:
fixed_arity('IFNULL', 2),
ops.Cast:
_cast,
ops.StructField:
_struct_field,
ops.ArrayCollect:
unary('ARRAY_AGG'),
ops.ArrayConcat:
_array_concat,
ops.ArrayIndex:
_array_index,
ops.ArrayLength:
unary('ARRAY_LENGTH'),
ops.Log:
_log,
ops.Modulus:
fixed_arity('MOD', 2),
ops.Date:
unary('DATE'),
# BigQuery doesn't have these operations built in.
# ops.ArrayRepeat: _array_repeat,
def _extract_epoch_seconds(translator, expr):
return 'CAST({} AS BIGINT)'.format(
unary('unix_timestamp')(translator, expr))
ops.StringReplace: fixed_arity('REPLACE', 3),
ops.StringSplit: fixed_arity('SPLIT', 2),
ops.StringConcat: _string_concat,
ops.StringJoin: _string_join,
ops.StringAscii: _string_ascii,
ops.StringFind: _string_find,
ops.StrRight: _string_right,
ops.Repeat: fixed_arity('REPEAT', 2),
ops.RegexSearch: _regex_search,
ops.RegexExtract: _regex_extract,
ops.RegexReplace: _regex_replace,
ops.GroupConcat: fixed_arity('STRING_AGG', 2),
ops.IfNull: fixed_arity('IFNULL', 2),
ops.Cast: _cast,
ops.StructField: _struct_field,
ops.ArrayCollect: unary('ARRAY_AGG'),
ops.ArrayConcat: _array_concat,
ops.ArrayIndex: _array_index,
ops.ArrayLength: unary('ARRAY_LENGTH'),
ops.HLLCardinality: _reduction('APPROX_COUNT_DISTINCT'),
ops.Log: _log,
ops.Sign: unary('SIGN'),
ops.Modulus: fixed_arity('MOD', 2),
ops.Date: unary('DATE'),
# BigQuery doesn't have these operations built in.
# ops.ArrayRepeat: _array_repeat,
# ops.ArraySlice: _array_slice,
ops.Literal: _literal,
ops.Arbitrary: _arbitrary,
ops.TimestampTruncate: _truncate('TIMESTAMP', _timestamp_units),
ops.DateTruncate: _truncate('DATE', _date_units),
