# 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: infix_op('~'),
ops.RegexReplace: _regex_replace,
ops.Translate: fixed_arity('translate', 3),
ops.RegexExtract: _regex_extract,
ops.StringSplit: fixed_arity(sa.func.string_to_array, 2),
ops.StringJoin: _string_join,
ops.FindInSet: _find_in_set,
# math
ops.Log: _log,
ops.Log2: unary(lambda x: sa.func.log(2, x)),
ops.Log10: unary(sa.func.log),
ops.Round: _round,
ops.Modulus: _mod,
# dates and times
return sa.literal(value)
_operation_registry.update({
ir.Literal:
_literal,
# strings
ops.Substring:
_substr,
ops.StringFind:
_string_find,
ops.Capitalize:
_capitalize,
ops.RegexSearch:
infix_op('REGEXP'),
# math
ops.Log:
_log,
ops.Log2:
unary(sa.func.log2),
ops.Log10:
unary(sa.func.log10),
ops.Round:
_round,
# dates and times
ops.Date:
unary(sa.func.date),
ops.DateAdd:
# 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: infix_op('~'),
ops.RegexReplace: _regex_replace,
ops.Translate: fixed_arity('translate', 3),
ops.RegexExtract: _regex_extract,
ops.StringSplit: fixed_arity(sa.func.string_to_array, 2),
ops.StringJoin: _string_join,
ops.FindInSet: _find_in_set,
# math
ops.Log: _log,
ops.Log2: unary(lambda x: sa.func.log(2, x)),
ops.Log10: unary(sa.func.log),
ops.Round: _round,
ops.Modulus: _mod,
# dates and times
# 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: infix_op('~'),
ops.RegexReplace: _regex_replace,
ops.Translate: fixed_arity('translate', 3),
ops.RegexExtract: _regex_extract,
ops.StringSplit: fixed_arity(sa.func.string_to_array, 2),
ops.StringJoin: _string_join,
ops.FindInSet: _find_in_set,
# math
ops.Log: _log,
ops.Log2: unary(lambda x: sa.func.log(2, x)),
ops.Log10: unary(sa.func.log),
ops.Round: _round,
ops.Modulus: _mod,
# dates and times
ops.Date: unary(lambda x: sa.cast(x, sa.Date)),
ops.DateTruncate: _timestamp_truncate,
value = value.to_pydatetime()
return sa.literal(value)
def _random(t, expr):
return sa.func.random()
_operation_registry.update(
{
ops.Literal: _literal,
# strings
ops.Substring: _substr,
ops.StringFind: _string_find,
ops.Capitalize: _capitalize,
ops.RegexSearch: infix_op('REGEXP'),
# math
ops.Log: _log,
ops.Log2: unary(sa.func.log2),
ops.Log10: unary(sa.func.log10),
ops.Round: _round,
ops.RandomScalar: _random,
# dates and times
ops.Date: unary(sa.func.date),
ops.DateAdd: infix_op('+'),
ops.DateSub: infix_op('-'),
ops.DateDiff: fixed_arity(sa.func.datediff, 2),
ops.TimestampAdd: infix_op('+'),
ops.TimestampSub: infix_op('-'),
ops.TimestampDiff: _timestamp_diff,
ops.DateTruncate: _truncate,
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:
infix_op('~'),
ops.RegexReplace:
_regex_replace,
ops.Translate:
fixed_arity('translate', 3),
ops.RegexExtract:
_regex_extract,
ops.StringSplit:
fixed_arity(sa.func.string_to_array, 2),
ops.StringJoin:
_string_join,
ops.FindInSet:
_find_in_set,
# math
ops.Log:
def _string_join(t, expr):
sep, elements = expr.op().args
return sa.func.concat_ws(t.translate(sep), *map(t.translate, elements))
_operation_registry.update({
# strings
ops.Substring:
_substr,
ops.StringFind:
_string_find,
ops.Capitalize:
_capitalize,
ops.RegexSearch:
infix_op('REGEXP'),
# math
ops.Log:
_log,
ops.Log2:
unary(sa.func.log2),
ops.Log10:
unary(sa.func.log10),
ops.Round:
_round,
# dates and times
ops.DateTruncate:
_truncate,
ops.TimestampTruncate:
return list(map(sa.literal, expr.op().value))
else:
value = expr.op().value
if isinstance(value, pd.Timestamp):
value = value.to_pydatetime()
return sa.literal(value)
_operation_registry.update(
{
ops.Literal: _literal,
# strings
ops.Substring: _substr,
ops.StringFind: _string_find,
ops.Capitalize: _capitalize,
ops.RegexSearch: infix_op('REGEXP'),
# math
ops.Log: _log,
ops.Log2: unary(sa.func.log2),
ops.Log10: unary(sa.func.log10),
ops.Round: _round,
# dates and times
ops.Date: unary(sa.func.date),
ops.DateAdd: infix_op('+'),
ops.DateSub: infix_op('-'),
ops.DateDiff: fixed_arity(sa.func.datediff, 2),
ops.TimestampAdd: infix_op('+'),
ops.TimestampSub: infix_op('-'),
ops.TimestampDiff: _timestamp_diff,
ops.DateTruncate: _truncate,
ops.TimestampTruncate: _truncate,
