class RPad(ValueOp):
input_type = [
string,
integer(name='length'),
string(name='pad', optional=True)
]
output_type = rules.shape_like_arg(0, 'string')
class Substring(ValueOp):
input_type = [
string,
integer(name='start'),
integer(name='length', optional=True)
]
output_type = rules.shape_like_arg(0, 'string')
class StringFind(ValueOp):
input_type = [
string,
string(name='substr'),
integer(name='start', optional=True, default=None),
integer(name='end', optional=True, default=None)
]
output_type = rules.shape_like_arg(0, 'int64')
class Truncate(ValueOp):
input_type = [
rules.timestamp,
rules.string_options(_truncate_units,
name='unit',
validator=_truncate_unit_validate)
]
output_type = rules.shape_like_arg(0, 'timestamp')
class NotNull(UnaryOp):
"""
Returns true if values are not null
Returns
-------
notnull : boolean with dimension of caller
"""
output_type = rules.shape_like_arg(0, 'boolean')
class StringLength(UnaryOp):
"""
Compute length of strings
Returns
-------
length : int32
"""
output_type = rules.shape_like_arg(0, 'int32')
class ParseURL(ValueOp):
input_type = [
string,
rules.string_options([
'PROTOCOL', 'HOST', 'PATH', 'REF', 'AUTHORITY', 'FILE', 'USERINFO',
'QUERY'
],
name='extract'),
string(name='key', optional=True)
]
output_type = rules.shape_like_arg(0, 'string')
class StringUnaryOp(UnaryOp):
input_type = [string]
output_type = rules.shape_like_arg(0, 'string')
class RegexExtract(ValueOp):
input_type = [string, string(name='pattern'), integer(name='index')]
output_type = rules.shape_like_arg(0, 'string')
class RegexReplace(ValueOp):
input_type = [string, string(name='pattern'), string(name='replacement')]
output_type = rules.shape_like_arg(0, 'string')
class FindInSet(ValueOp):
input_type = [string(name='needle'), list_of(string, min_length=1)]
output_type = rules.shape_like_arg(0, 'int64')
class FuzzySearch(BooleanValueOp):
input_type = [string, string(name='pattern')]
output_type = rules.shape_like_arg(0, 'boolean')
class Translate(ValueOp):
input_type = [string, string(name='from_str'), string(name='to_str')]
output_type = rules.shape_like_arg(0, 'string')
class ExtractTimestampField(TimestampUnaryOp):
output_type = rules.shape_like_arg(0, 'int32')
class Repeat(ValueOp):
input_type = [string, integer(name='times')]
output_type = rules.shape_like_arg(0, 'string')
class TimestampFromUNIX(ValueOp):
input_type = [value, rules.string_options(['s', 'ms', 'us'], name='unit')]
output_type = rules.shape_like_arg(0, 'timestamp')
class DecimalPrecision(DecimalUnaryOp):
output_type = rules.shape_like_arg(0, 'int32')
class StrRight(ValueOp):
input_type = [string, integer(name='nchars')]
output_type = rules.shape_like_arg(0, 'string')
class DecimalScale(UnaryOp):
output_type = rules.shape_like_arg(0, 'int32')
class Sign(UnaryOp):
# This is the Impala output for both integers and double/float
output_type = rules.shape_like_arg(0, 'float')
class RealUnaryOp(UnaryOp):
input_type = [number]
output_type = rules.shape_like_arg(0, 'double')
class TimestampDelta(ValueOp):
input_type = [rules.timestamp, rules.timedelta(name='offset')]
output_type = rules.shape_like_arg(0, 'timestamp')
class Hash(ValueOp):
input_type = [value, rules.string_options(['fnv'], name='how')]
output_type = rules.shape_like_arg(0, 'int64')
class StringAscii(UnaryOp):
output_type = rules.shape_like_arg(0, 'int32')
class Sign(UnaryOp):
output_type = rules.shape_like_arg(0, 'int32')
