def add_operation(op, func_name, db):
"""
Registers the given operation within the Ibis SQL translation toolchain
Parameters
----------
op: operator class
name: used in issuing statements to SQL engine
database: database the relevant operator is registered to
"""
full_name = '{0}.{1}'.format(db, func_name)
# TODO
# if op.input_type is rlz.listof:
# translator = comp.varargs(full_name)
# else:
arity = len(op.signature)
translator = fixed_arity(full_name, arity)
comp._operation_registry[op] = translator
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 = {**operation_registry}
_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:
fixed_arity('SPLIT', 2),
ops.RegexSearch:
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 = {**operation_registry}
_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: fixed_arity('SPLIT', 2),
ops.RegexSearch: fixed_arity('rlike', 2),
ops.StringConcat: _string_concat,
ops.ArrayConcat: fixed_arity('concat', 2),
ops.GroupConcat: _group_concat,
ops.Cast: _cast,
ops.ExtractYear: unary('year'),
ops.ExtractMonth: unary('month'),
ops.ExtractDay: unary('day'),
ops.ExtractDay:
_extract_field('day'),
ops.ExtractHour:
_extract_field('hour'),
ops.ExtractMinute:
_extract_field('minute'),
ops.ExtractSecond:
_extract_field('second'),
ops.ExtractMillisecond:
_extract_field('millisecond'),
ops.ExtractEpochSeconds:
_extract_field('epochseconds'),
ops.Hash:
_hash,
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:
ops.Uppercase:
unary('upper'),
ops.Reverse:
unary('reverse'),
ops.Strip:
unary('trim'),
ops.LStrip:
unary('ltrim'),
ops.RStrip:
unary('rtrim'),
ops.Capitalize:
unary('initcap'),
ops.Substring:
_substring,
ops.StrRight:
fixed_arity('strright', 2),
ops.Repeat:
fixed_arity('repeat', 2),
ops.StringFind:
_string_find,
ops.Translate:
fixed_arity('translate', 3),
ops.FindInSet:
_find_in_set,
ops.LPad:
fixed_arity('lpad', 3),
ops.RPad:
fixed_arity('rpad', 3),
ops.StringJoin:
_string_join,
ops.StringSQLLike:
ops.ExtractYear:
_extract_field("year"),
ops.ExtractMonth:
_extract_field("month"),
ops.ExtractDay:
_extract_field("day"),
ops.ExtractHour:
_extract_field("hour"),
ops.ExtractMinute:
_extract_field("minute"),
ops.ExtractSecond:
_extract_field("second"),
ops.ExtractMillisecond:
_extract_field("millisecond"),
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:
_reduction('max'),
ops.Min:
_reduction('min'),
ops.Sum:
_reduction('sum'),
ops.Mean:
_reduction('avg', 'float64'),
# string methods
ops.LStrip:
unary(sa.func.ltrim),
ops.Lowercase:
unary(sa.func.lower),
ops.RStrip:
unary(sa.func.rtrim),
ops.Repeat:
fixed_arity(sa.func.replicate, 2),
ops.Reverse:
unary(sa.func.reverse),
ops.StringFind:
_string_find,
ops.StringLength:
unary(sa.func.length),
ops.StringReplace:
fixed_arity(sa.func.replace, 3),
ops.Strip:
unary(sa.func.trim),
ops.Substring:
_substr,
ops.Uppercase:
unary(sa.func.upper),
# math
return sa.func.regexp_substr(sa_arg, sa_expr, sa_group)
_operation_registry.update(
{
ops.Literal: _literal,
# We override this here to support time zones
ops.TableColumn: _table_column,
# types
ops.Cast: _cast,
ops.TypeOf: _typeof,
# Floating
ops.IsNan: _is_nan,
ops.IsInf: _is_inf,
# null handling
ops.IfNull: fixed_arity(sa.func.coalesce, 2),
# boolean reductions
ops.Any: unary(sa.func.bool_or),
ops.All: unary(sa.func.bool_and),
ops.NotAny: unary(lambda x: sa.not_(sa.func.bool_or(x))),
ops.NotAll: unary(lambda x: sa.not_(sa.func.bool_and(x))),
# strings
ops.Substring: _substr,
ops.StringFind: _string_find,
ops.GroupConcat: _string_agg,
ops.Capitalize: unary(sa.func.initcap),
ops.RegexSearch: _regex_search,
ops.RegexReplace: _regex_replace,
ops.RegexExtract: _regex_extract,
ops.Translate: fixed_arity('translate', 3),
ops.StringSplit: fixed_arity(sa.func.string_to_array, 2),
_operation_registry = {
**base_sql.operation_registry,
**impala_compiler._operation_registry,
}
_operation_registry.update(
{
ops.ExtractYear: _extract_field('year'),
ops.ExtractQuarter: _extract_field('quarter'),
ops.ExtractMonth: _extract_field('month'),
ops.ExtractDay: _extract_field('day'),
ops.ExtractHour: _extract_field('hour'),
ops.ExtractMinute: _extract_field('minute'),
ops.ExtractSecond: _extract_field('second'),
ops.ExtractMillisecond: _extract_field('millisecond'),
ops.ExtractEpochSeconds: _extract_field('epochseconds'),
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: reduction('STRING_AGG'),
ops.IfNull: fixed_arity('IFNULL', 2),
ops.Cast: _cast,
ops.StructField: _struct_field,
ops.ArrayCollect: unary('ARRAY_AGG'),