polars-df 0.15.0 → 0.26.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +4 -4
- data/CHANGELOG.md +274 -0
- data/Cargo.lock +1465 -867
- data/Cargo.toml +3 -0
- data/LICENSE.txt +2 -2
- data/README.md +87 -37
- data/ext/polars/Cargo.toml +47 -16
- data/ext/polars/src/c_api/allocator.rs +7 -0
- data/ext/polars/src/c_api/mod.rs +1 -0
- data/ext/polars/src/catalog/mod.rs +1 -0
- data/ext/polars/src/catalog/unity.rs +470 -0
- data/ext/polars/src/conversion/any_value.rs +99 -84
- data/ext/polars/src/conversion/categorical.rs +30 -0
- data/ext/polars/src/conversion/chunked_array.rs +71 -62
- data/ext/polars/src/conversion/datetime.rs +63 -0
- data/ext/polars/src/conversion/mod.rs +796 -312
- data/ext/polars/src/dataframe/construction.rs +6 -18
- data/ext/polars/src/dataframe/export.rs +30 -39
- data/ext/polars/src/dataframe/general.rs +294 -362
- data/ext/polars/src/dataframe/io.rs +33 -150
- data/ext/polars/src/dataframe/map.rs +175 -0
- data/ext/polars/src/dataframe/mod.rs +37 -5
- data/ext/polars/src/dataframe/serde.rs +23 -8
- data/ext/polars/src/error.rs +44 -7
- data/ext/polars/src/exceptions.rs +21 -8
- data/ext/polars/src/expr/array.rs +86 -22
- data/ext/polars/src/expr/binary.rs +50 -1
- data/ext/polars/src/expr/bitwise.rs +39 -0
- data/ext/polars/src/expr/categorical.rs +20 -0
- data/ext/polars/src/expr/datatype.rs +51 -0
- data/ext/polars/src/expr/datetime.rs +99 -41
- data/ext/polars/src/expr/extension.rs +12 -0
- data/ext/polars/src/expr/general.rs +252 -128
- data/ext/polars/src/expr/list.rs +56 -60
- data/ext/polars/src/expr/meta.rs +30 -35
- data/ext/polars/src/expr/mod.rs +28 -6
- data/ext/polars/src/expr/name.rs +29 -14
- data/ext/polars/src/expr/rolling.rs +111 -3
- data/ext/polars/src/expr/selector.rs +219 -0
- data/ext/polars/src/expr/serde.rs +28 -0
- data/ext/polars/src/expr/string.rs +118 -20
- data/ext/polars/src/expr/struct.rs +14 -1
- data/ext/polars/src/file.rs +194 -86
- data/ext/polars/src/functions/aggregation.rs +13 -12
- data/ext/polars/src/functions/business.rs +2 -3
- data/ext/polars/src/functions/eager.rs +3 -2
- data/ext/polars/src/functions/io.rs +90 -18
- data/ext/polars/src/functions/lazy.rs +267 -118
- data/ext/polars/src/functions/meta.rs +8 -7
- data/ext/polars/src/functions/misc.rs +1 -1
- data/ext/polars/src/functions/mod.rs +2 -1
- data/ext/polars/src/functions/range.rs +88 -31
- data/ext/polars/src/functions/strings.rs +6 -0
- data/ext/polars/src/functions/utils.rs +8 -0
- data/ext/polars/src/interop/arrow/mod.rs +52 -1
- data/ext/polars/src/interop/arrow/{to_ruby.rs → to_rb.rs} +37 -7
- data/ext/polars/src/interop/arrow/to_rust.rs +43 -0
- data/ext/polars/src/interop/numo/to_numo_df.rs +1 -1
- data/ext/polars/src/interop/numo/to_numo_series.rs +72 -50
- data/ext/polars/src/io/cloud_options.rs +107 -0
- data/ext/polars/src/io/mod.rs +4 -0
- data/ext/polars/src/io/scan_options.rs +113 -0
- data/ext/polars/src/io/sink_options.rs +46 -0
- data/ext/polars/src/io/sink_output.rs +21 -0
- data/ext/polars/src/lazyframe/exitable.rs +39 -0
- data/ext/polars/src/lazyframe/general.rs +846 -368
- data/ext/polars/src/lazyframe/mod.rs +58 -5
- data/ext/polars/src/lazyframe/optflags.rs +59 -0
- data/ext/polars/src/lazyframe/serde.rs +36 -4
- data/ext/polars/src/lazyframe/sink.rs +46 -0
- data/ext/polars/src/lazygroupby.rs +38 -9
- data/ext/polars/src/lib.rs +574 -165
- data/ext/polars/src/map/lazy.rs +44 -74
- data/ext/polars/src/map/mod.rs +18 -254
- data/ext/polars/src/map/series.rs +241 -1087
- data/ext/polars/src/on_startup.rs +192 -9
- data/ext/polars/src/prelude.rs +1 -0
- data/ext/polars/src/rb_modules.rs +10 -57
- data/ext/polars/src/ruby/exceptions.rs +26 -0
- data/ext/polars/src/ruby/gvl.rs +104 -0
- data/ext/polars/src/ruby/lazy.rs +46 -0
- data/ext/polars/src/ruby/mod.rs +11 -0
- data/ext/polars/src/ruby/numo.rs +52 -0
- data/ext/polars/src/ruby/plan_callback.rs +198 -0
- data/ext/polars/src/ruby/rb_modules.rs +16 -0
- data/ext/polars/src/ruby/ruby_convert_registry.rs +51 -0
- data/ext/polars/src/ruby/ruby_function.rs +11 -0
- data/ext/polars/src/ruby/ruby_udf.rs +164 -0
- data/ext/polars/src/ruby/thread.rs +65 -0
- data/ext/polars/src/ruby/utils.rs +39 -0
- data/ext/polars/src/series/aggregation.rs +116 -91
- data/ext/polars/src/series/arithmetic.rs +16 -22
- data/ext/polars/src/series/comparison.rs +101 -222
- data/ext/polars/src/series/construction.rs +80 -70
- data/ext/polars/src/series/export.rs +98 -56
- data/ext/polars/src/series/general.rs +323 -440
- data/ext/polars/src/series/import.rs +22 -5
- data/ext/polars/src/series/map.rs +103 -0
- data/ext/polars/src/series/mod.rs +57 -15
- data/ext/polars/src/series/scatter.rs +139 -82
- data/ext/polars/src/sql.rs +16 -9
- data/ext/polars/src/testing/frame.rs +31 -0
- data/ext/polars/src/testing/mod.rs +5 -0
- data/ext/polars/src/testing/series.rs +31 -0
- data/ext/polars/src/timeout.rs +105 -0
- data/ext/polars/src/utils.rs +105 -4
- data/lib/polars/array_expr.rb +500 -22
- data/lib/polars/array_name_space.rb +384 -10
- data/lib/polars/batched_csv_reader.rb +48 -66
- data/lib/polars/binary_expr.rb +217 -0
- data/lib/polars/binary_name_space.rb +155 -1
- data/lib/polars/cat_expr.rb +224 -0
- data/lib/polars/cat_name_space.rb +132 -32
- data/lib/polars/catalog/unity/catalog_info.rb +20 -0
- data/lib/polars/catalog/unity/column_info.rb +31 -0
- data/lib/polars/catalog/unity/namespace_info.rb +21 -0
- data/lib/polars/catalog/unity/table_info.rb +50 -0
- data/lib/polars/catalog.rb +448 -0
- data/lib/polars/collect_batches.rb +22 -0
- data/lib/polars/config.rb +3 -3
- data/lib/polars/convert.rb +201 -36
- data/lib/polars/data_frame.rb +2851 -1017
- data/lib/polars/data_frame_plot.rb +173 -0
- data/lib/polars/data_type_expr.rb +52 -0
- data/lib/polars/data_type_group.rb +6 -0
- data/lib/polars/data_types.rb +118 -18
- data/lib/polars/date_time_expr.rb +426 -84
- data/lib/polars/date_time_name_space.rb +384 -111
- data/lib/polars/dynamic_group_by.rb +102 -10
- data/lib/polars/exceptions.rb +50 -5
- data/lib/polars/expr.rb +2159 -915
- data/lib/polars/extension_expr.rb +39 -0
- data/lib/polars/extension_name_space.rb +39 -0
- data/lib/polars/functions/aggregation/horizontal.rb +11 -6
- data/lib/polars/functions/aggregation/vertical.rb +2 -3
- data/lib/polars/functions/as_datatype.rb +290 -8
- data/lib/polars/functions/business.rb +95 -0
- data/lib/polars/functions/col.rb +6 -5
- data/lib/polars/functions/datatype.rb +62 -0
- data/lib/polars/functions/eager.rb +426 -24
- data/lib/polars/functions/escape_regex.rb +21 -0
- data/lib/polars/functions/lazy.rb +813 -195
- data/lib/polars/functions/lit.rb +21 -10
- data/lib/polars/functions/range/int_range.rb +74 -2
- data/lib/polars/functions/range/linear_space.rb +195 -0
- data/lib/polars/functions/range/time_range.rb +1 -1
- data/lib/polars/functions/repeat.rb +7 -12
- data/lib/polars/functions/whenthen.rb +2 -2
- data/lib/polars/group_by.rb +188 -58
- data/lib/polars/iceberg_dataset.rb +108 -0
- data/lib/polars/in_process_query.rb +37 -0
- data/lib/polars/io/cloud.rb +18 -0
- data/lib/polars/io/csv.rb +336 -128
- data/lib/polars/io/database.rb +19 -4
- data/lib/polars/io/delta.rb +134 -0
- data/lib/polars/io/iceberg.rb +34 -0
- data/lib/polars/io/ipc.rb +63 -63
- data/lib/polars/io/json.rb +16 -0
- data/lib/polars/io/lines.rb +172 -0
- data/lib/polars/io/ndjson.rb +176 -20
- data/lib/polars/io/parquet.rb +173 -95
- data/lib/polars/io/scan_options.rb +55 -0
- data/lib/polars/io/sink_options.rb +27 -0
- data/lib/polars/io/utils.rb +17 -0
- data/lib/polars/lazy_frame.rb +3017 -622
- data/lib/polars/lazy_group_by.rb +436 -2
- data/lib/polars/list_expr.rb +551 -59
- data/lib/polars/list_name_space.rb +465 -51
- data/lib/polars/meta_expr.rb +146 -24
- data/lib/polars/name_expr.rb +87 -2
- data/lib/polars/query_opt_flags.rb +264 -0
- data/lib/polars/rolling_group_by.rb +90 -5
- data/lib/polars/scan_cast_options.rb +86 -0
- data/lib/polars/schema.rb +128 -0
- data/lib/polars/selector.rb +245 -0
- data/lib/polars/selectors.rb +1048 -201
- data/lib/polars/series.rb +2522 -774
- data/lib/polars/series_plot.rb +72 -0
- data/lib/polars/slice.rb +1 -1
- data/lib/polars/sql_context.rb +13 -6
- data/lib/polars/string_cache.rb +19 -72
- data/lib/polars/string_expr.rb +561 -107
- data/lib/polars/string_name_space.rb +781 -109
- data/lib/polars/struct_expr.rb +139 -18
- data/lib/polars/struct_name_space.rb +19 -1
- data/lib/polars/testing.rb +24 -273
- data/lib/polars/utils/constants.rb +2 -0
- data/lib/polars/utils/construction/data_frame.rb +410 -0
- data/lib/polars/utils/construction/series.rb +350 -0
- data/lib/polars/utils/construction/utils.rb +9 -0
- data/lib/polars/utils/convert.rb +18 -8
- data/lib/polars/utils/deprecation.rb +11 -0
- data/lib/polars/utils/parse.rb +62 -9
- data/lib/polars/utils/reduce_balanced.rb +43 -0
- data/lib/polars/utils/serde.rb +22 -0
- data/lib/polars/utils/unstable.rb +19 -0
- data/lib/polars/utils/various.rb +86 -1
- data/lib/polars/utils.rb +63 -48
- data/lib/polars/version.rb +1 -1
- data/lib/polars.rb +85 -2
- metadata +80 -28
- data/ext/polars/src/allocator.rs +0 -13
- data/ext/polars/src/batched_csv.rs +0 -138
- data/ext/polars/src/functions/string_cache.rs +0 -25
- data/ext/polars/src/map/dataframe.rs +0 -338
- data/lib/polars/plot.rb +0 -109
|
@@ -0,0 +1,350 @@
|
|
|
1
|
+
module Polars
|
|
2
|
+
module Utils
|
|
3
|
+
def self.sequence_to_rbseries(name, values, dtype: nil, strict: true, nan_to_null: false)
|
|
4
|
+
ruby_dtype = nil
|
|
5
|
+
|
|
6
|
+
if values.is_a?(Range)
|
|
7
|
+
if values.begin.is_a?(::String)
|
|
8
|
+
values = values.to_a
|
|
9
|
+
else
|
|
10
|
+
return range_to_series(name, values, dtype: dtype)._s
|
|
11
|
+
end
|
|
12
|
+
end
|
|
13
|
+
|
|
14
|
+
if values.length == 0 && dtype.nil?
|
|
15
|
+
dtype = Null
|
|
16
|
+
elsif [List, Array].include?(dtype)
|
|
17
|
+
ruby_dtype = ::Array
|
|
18
|
+
end
|
|
19
|
+
|
|
20
|
+
rb_temporal_types = [::Date, ::DateTime, ::Time]
|
|
21
|
+
rb_temporal_types << ActiveSupport::TimeWithZone if defined?(ActiveSupport::TimeWithZone)
|
|
22
|
+
|
|
23
|
+
value = get_first_non_none(values)
|
|
24
|
+
|
|
25
|
+
if !dtype.nil? && is_polars_dtype(dtype) && !dtype.nested? && dtype != Unknown && ruby_dtype.nil?
|
|
26
|
+
constructor = polars_type_to_constructor(dtype)
|
|
27
|
+
rbseries = _construct_series_with_fallbacks(
|
|
28
|
+
constructor, name, values, dtype, strict: strict
|
|
29
|
+
)
|
|
30
|
+
|
|
31
|
+
if [Date, Datetime, Duration, Time, Boolean, Categorical, Enum].include?(dtype) || dtype.is_a?(Decimal) || dtype.is_a?(Categorical)
|
|
32
|
+
if rbseries.dtype != dtype
|
|
33
|
+
rbseries = rbseries.cast(dtype, true, false)
|
|
34
|
+
end
|
|
35
|
+
end
|
|
36
|
+
|
|
37
|
+
# Uninstanced Decimal is a bit special and has various inference paths
|
|
38
|
+
if dtype == Decimal
|
|
39
|
+
if rbseries.dtype == String
|
|
40
|
+
rbseries = rbseries.str_to_decimal_infer(0)
|
|
41
|
+
elsif rbseries.dtype.float?
|
|
42
|
+
# Go through string so we infer an appropriate scale.
|
|
43
|
+
rbseries = rbseries.cast(
|
|
44
|
+
String, strict, false
|
|
45
|
+
).str_to_decimal_infer(0)
|
|
46
|
+
elsif rbseries.dtype.integer? || rbseries.dtype == Null
|
|
47
|
+
rbseries = rbseries.cast(
|
|
48
|
+
Decimal.new(nil, 0), strict, false
|
|
49
|
+
)
|
|
50
|
+
elsif !rbseries.dtype.is_a?(Decimal)
|
|
51
|
+
msg = "can't convert #{rbseries.dtype} to Decimal"
|
|
52
|
+
raise TypeError, msg
|
|
53
|
+
end
|
|
54
|
+
end
|
|
55
|
+
|
|
56
|
+
return rbseries
|
|
57
|
+
|
|
58
|
+
elsif dtype == Struct
|
|
59
|
+
struct_schema = dtype.is_a?(Struct) ? dtype.to_schema : nil
|
|
60
|
+
empty = {}
|
|
61
|
+
|
|
62
|
+
data = []
|
|
63
|
+
invalid = []
|
|
64
|
+
values.each_with_index do |v, i|
|
|
65
|
+
if v.nil?
|
|
66
|
+
invalid << i
|
|
67
|
+
data << empty
|
|
68
|
+
else
|
|
69
|
+
data << v
|
|
70
|
+
end
|
|
71
|
+
end
|
|
72
|
+
|
|
73
|
+
return sequence_to_rbdf(
|
|
74
|
+
data,
|
|
75
|
+
schema: struct_schema,
|
|
76
|
+
orient: "row",
|
|
77
|
+
).to_struct(name, invalid)
|
|
78
|
+
end
|
|
79
|
+
|
|
80
|
+
if ruby_dtype.nil?
|
|
81
|
+
if value.nil?
|
|
82
|
+
constructor = polars_type_to_constructor(Null)
|
|
83
|
+
return constructor.(name, values, strict)
|
|
84
|
+
end
|
|
85
|
+
|
|
86
|
+
ruby_dtype = value.class
|
|
87
|
+
end
|
|
88
|
+
|
|
89
|
+
# temporal branch
|
|
90
|
+
if rb_temporal_types.include?(ruby_dtype)
|
|
91
|
+
if dtype.nil?
|
|
92
|
+
dtype = parse_into_dtype(ruby_dtype)
|
|
93
|
+
elsif rb_temporal_types.include?(dtype)
|
|
94
|
+
dtype = parse_into_dtype(dtype)
|
|
95
|
+
end
|
|
96
|
+
|
|
97
|
+
values_dtype = value.nil? ? nil : try_parse_into_dtype(value.class)
|
|
98
|
+
if !values_dtype.nil? && values_dtype.float?
|
|
99
|
+
msg = "'float' object cannot be interpreted as a #{ruby_dtype.name.inspect}"
|
|
100
|
+
raise TypeError, msg
|
|
101
|
+
end
|
|
102
|
+
|
|
103
|
+
rb_series = RbSeries.new_from_any_values(name, values, strict)
|
|
104
|
+
|
|
105
|
+
time_unit = dtype.respond_to?(:time_unit) ? dtype.time_unit : nil
|
|
106
|
+
time_zone = dtype.respond_to?(:time_zone) ? dtype.time_zone : nil
|
|
107
|
+
|
|
108
|
+
if dtype.temporal? && values_dtype == String && dtype != Duration
|
|
109
|
+
s = wrap_s(rb_series).str.strptime(dtype, strict: strict)
|
|
110
|
+
elsif !time_unit.nil? && values_dtype != Date
|
|
111
|
+
s = wrap_s(rb_series).dt.cast_time_unit(time_unit)
|
|
112
|
+
else
|
|
113
|
+
s = wrap_s(rb_series)
|
|
114
|
+
end
|
|
115
|
+
|
|
116
|
+
if dtype == Datetime && !time_zone.nil?
|
|
117
|
+
return s.dt.convert_time_zone(time_zone)._s
|
|
118
|
+
end
|
|
119
|
+
s._s
|
|
120
|
+
|
|
121
|
+
elsif defined?(Numo::NArray) && value.is_a?(Numo::NArray) && value.shape.length == 1
|
|
122
|
+
raise Todo
|
|
123
|
+
|
|
124
|
+
elsif ruby_dtype == ::Array
|
|
125
|
+
if dtype.nil?
|
|
126
|
+
RbSeries.new_from_any_values(name, values, strict)
|
|
127
|
+
elsif dtype.is_a?(Object)
|
|
128
|
+
RbSeries.new_object(name, values, strict)
|
|
129
|
+
else
|
|
130
|
+
inner_dtype = dtype.respond_to?(:inner) ? dtype.inner : nil
|
|
131
|
+
if !inner_dtype.nil?
|
|
132
|
+
rbseries_list =
|
|
133
|
+
values.map do |value|
|
|
134
|
+
if value.nil?
|
|
135
|
+
nil
|
|
136
|
+
else
|
|
137
|
+
sequence_to_rbseries(
|
|
138
|
+
"",
|
|
139
|
+
value,
|
|
140
|
+
dtype: inner_dtype,
|
|
141
|
+
strict: strict,
|
|
142
|
+
nan_to_null: nan_to_null,
|
|
143
|
+
)
|
|
144
|
+
end
|
|
145
|
+
end
|
|
146
|
+
rbseries = RbSeries.new_series_list(name, rbseries_list, strict)
|
|
147
|
+
else
|
|
148
|
+
# panics in Python
|
|
149
|
+
raise Todo if dtype.eql?(Array)
|
|
150
|
+
|
|
151
|
+
rbseries = RbSeries.new_from_any_values_and_dtype(
|
|
152
|
+
name, values, dtype, strict
|
|
153
|
+
)
|
|
154
|
+
end
|
|
155
|
+
if dtype != rbseries.dtype
|
|
156
|
+
rbseries = rbseries.cast(dtype, false, false)
|
|
157
|
+
end
|
|
158
|
+
rbseries
|
|
159
|
+
end
|
|
160
|
+
|
|
161
|
+
elsif ruby_dtype == Series
|
|
162
|
+
RbSeries.new_series_list(
|
|
163
|
+
name, values.map { |v| !v.nil? ? v._s : v }, strict
|
|
164
|
+
)
|
|
165
|
+
|
|
166
|
+
elsif ruby_dtype == RbSeries
|
|
167
|
+
RbSeries.new_series_list(name, values, strict)
|
|
168
|
+
else
|
|
169
|
+
constructor =
|
|
170
|
+
if value.is_a?(::String)
|
|
171
|
+
if value.encoding == Encoding::UTF_8
|
|
172
|
+
RbSeries.method(:new_str)
|
|
173
|
+
else
|
|
174
|
+
RbSeries.method(:new_binary)
|
|
175
|
+
end
|
|
176
|
+
elsif value.is_a?(Integer) && values.any? { |v| v.is_a?(Float) }
|
|
177
|
+
# TODO improve performance
|
|
178
|
+
RbSeries.method(:new_opt_f64)
|
|
179
|
+
else
|
|
180
|
+
rb_type_to_constructor(value.class)
|
|
181
|
+
end
|
|
182
|
+
|
|
183
|
+
if constructor == RbSeries.method(:new_object)
|
|
184
|
+
srs = RbSeries.new_from_any_values(name, values, strict)
|
|
185
|
+
return srs
|
|
186
|
+
end
|
|
187
|
+
|
|
188
|
+
_construct_series_with_fallbacks(constructor, name, values, dtype, strict: strict)
|
|
189
|
+
end
|
|
190
|
+
end
|
|
191
|
+
|
|
192
|
+
def self._construct_series_with_fallbacks(constructor, name, values, dtype, strict:)
|
|
193
|
+
begin
|
|
194
|
+
constructor.(name, values, strict)
|
|
195
|
+
rescue
|
|
196
|
+
if dtype.nil?
|
|
197
|
+
RbSeries.new_from_any_values(name, values, strict)
|
|
198
|
+
else
|
|
199
|
+
RbSeries.new_from_any_values_and_dtype(name, values, dtype, strict)
|
|
200
|
+
end
|
|
201
|
+
end
|
|
202
|
+
end
|
|
203
|
+
|
|
204
|
+
def self.numo_to_rbseries(name, values, strict: true, nan_to_null: false)
|
|
205
|
+
# not needed yet
|
|
206
|
+
# if !values.contiguous?
|
|
207
|
+
# end
|
|
208
|
+
|
|
209
|
+
if values.shape.length == 1
|
|
210
|
+
values, dtype = numo_values_and_dtype(values)
|
|
211
|
+
constructor = numo_type_to_constructor(dtype)
|
|
212
|
+
constructor.(
|
|
213
|
+
name, values.to_a, [Numo::SFloat, Numo::DFloat].include?(dtype) ? nan_to_null : strict
|
|
214
|
+
)
|
|
215
|
+
else
|
|
216
|
+
original_shape = values.shape
|
|
217
|
+
values_1d = values.reshape(original_shape.inject(&:*))
|
|
218
|
+
|
|
219
|
+
rb_s = numo_to_rbseries(
|
|
220
|
+
name,
|
|
221
|
+
values_1d,
|
|
222
|
+
strict: strict,
|
|
223
|
+
nan_to_null: nan_to_null
|
|
224
|
+
)
|
|
225
|
+
Utils.wrap_s(rb_s).reshape(original_shape)._s
|
|
226
|
+
end
|
|
227
|
+
end
|
|
228
|
+
|
|
229
|
+
def self.series_to_rbseries(name, values, dtype: nil, strict: true)
|
|
230
|
+
s = values.clone
|
|
231
|
+
if !dtype.nil? && dtype != s.dtype
|
|
232
|
+
s = s.cast(dtype, strict: strict)
|
|
233
|
+
end
|
|
234
|
+
if !name.nil?
|
|
235
|
+
s = s.alias(name)
|
|
236
|
+
end
|
|
237
|
+
s._s
|
|
238
|
+
end
|
|
239
|
+
|
|
240
|
+
def self.dataframe_to_rbseries(
|
|
241
|
+
name,
|
|
242
|
+
values,
|
|
243
|
+
dtype: nil,
|
|
244
|
+
strict: true
|
|
245
|
+
)
|
|
246
|
+
if values.width > 1
|
|
247
|
+
name ||= ""
|
|
248
|
+
s = values.to_struct(name)
|
|
249
|
+
elsif values.width == 1
|
|
250
|
+
s = values.to_series
|
|
251
|
+
if !name.nil?
|
|
252
|
+
s = s.alias(name)
|
|
253
|
+
end
|
|
254
|
+
else
|
|
255
|
+
msg = "cannot initialize Series from DataFrame without any columns"
|
|
256
|
+
raise TypeError, msg
|
|
257
|
+
end
|
|
258
|
+
|
|
259
|
+
if !dtype.nil? && dtype != s.dtype
|
|
260
|
+
s = s.cast(dtype, strict: strict)
|
|
261
|
+
end
|
|
262
|
+
|
|
263
|
+
s._s
|
|
264
|
+
end
|
|
265
|
+
|
|
266
|
+
# TODO move rest
|
|
267
|
+
|
|
268
|
+
POLARS_TYPE_TO_CONSTRUCTOR = {
|
|
269
|
+
Float16 => RbSeries.method(:new_opt_f16),
|
|
270
|
+
Float32 => RbSeries.method(:new_opt_f32),
|
|
271
|
+
Float64 => RbSeries.method(:new_opt_f64),
|
|
272
|
+
Int8 => RbSeries.method(:new_opt_i8),
|
|
273
|
+
Int16 => RbSeries.method(:new_opt_i16),
|
|
274
|
+
Int32 => RbSeries.method(:new_opt_i32),
|
|
275
|
+
Int64 => RbSeries.method(:new_opt_i64),
|
|
276
|
+
Int128 => RbSeries.method(:new_opt_i128),
|
|
277
|
+
UInt8 => RbSeries.method(:new_opt_u8),
|
|
278
|
+
UInt16 => RbSeries.method(:new_opt_u16),
|
|
279
|
+
UInt32 => RbSeries.method(:new_opt_u32),
|
|
280
|
+
UInt64 => RbSeries.method(:new_opt_u64),
|
|
281
|
+
UInt128 => RbSeries.method(:new_opt_u128),
|
|
282
|
+
Decimal => RbSeries.method(:new_decimal),
|
|
283
|
+
Date => RbSeries.method(:new_from_any_values),
|
|
284
|
+
Datetime => RbSeries.method(:new_from_any_values),
|
|
285
|
+
Duration => RbSeries.method(:new_from_any_values),
|
|
286
|
+
Time => RbSeries.method(:new_from_any_values),
|
|
287
|
+
Boolean => RbSeries.method(:new_opt_bool),
|
|
288
|
+
Utf8 => RbSeries.method(:new_str),
|
|
289
|
+
Object => RbSeries.method(:new_object),
|
|
290
|
+
Categorical => RbSeries.method(:new_str),
|
|
291
|
+
Enum => RbSeries.method(:new_str),
|
|
292
|
+
Binary => RbSeries.method(:new_binary),
|
|
293
|
+
Null => RbSeries.method(:new_null)
|
|
294
|
+
}
|
|
295
|
+
|
|
296
|
+
def self.polars_type_to_constructor(dtype)
|
|
297
|
+
if dtype.is_a?(Array)
|
|
298
|
+
return lambda do |name, values, strict|
|
|
299
|
+
RbSeries.new_array(name, values, strict, dtype)
|
|
300
|
+
end
|
|
301
|
+
end
|
|
302
|
+
|
|
303
|
+
begin
|
|
304
|
+
base_type = dtype.base_type
|
|
305
|
+
POLARS_TYPE_TO_CONSTRUCTOR.fetch(base_type)
|
|
306
|
+
rescue KeyError
|
|
307
|
+
raise ArgumentError, "Cannot construct RbSeries for type #{dtype}."
|
|
308
|
+
end
|
|
309
|
+
end
|
|
310
|
+
|
|
311
|
+
RB_TYPE_TO_CONSTRUCTOR = {
|
|
312
|
+
Float => RbSeries.method(:new_opt_f64),
|
|
313
|
+
TrueClass => RbSeries.method(:new_opt_bool),
|
|
314
|
+
FalseClass => RbSeries.method(:new_opt_bool),
|
|
315
|
+
Integer => RbSeries.method(:new_opt_i64),
|
|
316
|
+
::String => RbSeries.method(:new_str)
|
|
317
|
+
}
|
|
318
|
+
|
|
319
|
+
def self.rb_type_to_constructor(dtype)
|
|
320
|
+
RB_TYPE_TO_CONSTRUCTOR.fetch(dtype) do
|
|
321
|
+
if defined?(::BigDecimal) && dtype == ::BigDecimal
|
|
322
|
+
RbSeries.method(:new_decimal)
|
|
323
|
+
else
|
|
324
|
+
RbSeries.method(:new_object)
|
|
325
|
+
end
|
|
326
|
+
end
|
|
327
|
+
end
|
|
328
|
+
|
|
329
|
+
def self.numo_values_and_dtype(values)
|
|
330
|
+
[values, values.class]
|
|
331
|
+
end
|
|
332
|
+
|
|
333
|
+
def self.numo_type_to_constructor(dtype)
|
|
334
|
+
{
|
|
335
|
+
Numo::Float32 => RbSeries.method(:new_opt_f32),
|
|
336
|
+
Numo::Float64 => RbSeries.method(:new_opt_f64),
|
|
337
|
+
Numo::Int8 => RbSeries.method(:new_opt_i8),
|
|
338
|
+
Numo::Int16 => RbSeries.method(:new_opt_i16),
|
|
339
|
+
Numo::Int32 => RbSeries.method(:new_opt_i32),
|
|
340
|
+
Numo::Int64 => RbSeries.method(:new_opt_i64),
|
|
341
|
+
Numo::UInt8 => RbSeries.method(:new_opt_u8),
|
|
342
|
+
Numo::UInt16 => RbSeries.method(:new_opt_u16),
|
|
343
|
+
Numo::UInt32 => RbSeries.method(:new_opt_u32),
|
|
344
|
+
Numo::UInt64 => RbSeries.method(:new_opt_u64)
|
|
345
|
+
}.fetch(dtype)
|
|
346
|
+
rescue KeyError
|
|
347
|
+
RbSeries.method(:new_object)
|
|
348
|
+
end
|
|
349
|
+
end
|
|
350
|
+
end
|
data/lib/polars/utils/convert.rb
CHANGED
|
@@ -63,19 +63,24 @@ module Polars
|
|
|
63
63
|
end
|
|
64
64
|
|
|
65
65
|
def self._to_ruby_datetime(value, time_unit = "ns", time_zone = nil)
|
|
66
|
-
|
|
66
|
+
utc_time =
|
|
67
67
|
if time_unit == "ns"
|
|
68
|
-
::Time.at(value /
|
|
68
|
+
::Time.at(value / 1_000_000_000, value % 1_000_000_000, :nsec).utc
|
|
69
69
|
elsif time_unit == "us"
|
|
70
|
-
::Time.at(value /
|
|
70
|
+
::Time.at(value / 1_000_000, value % 1_000_000, :usec).utc
|
|
71
71
|
elsif time_unit == "ms"
|
|
72
|
-
::Time.at(value /
|
|
72
|
+
::Time.at(value / 1_000, value % 1_000, :millisecond).utc
|
|
73
73
|
else
|
|
74
74
|
raise ArgumentError, "time_unit must be one of {{'ns', 'us', 'ms'}}, got #{time_unit}"
|
|
75
75
|
end
|
|
76
|
-
|
|
77
|
-
|
|
76
|
+
return utc_time if time_zone.nil?
|
|
77
|
+
|
|
78
|
+
unless defined?(::TZInfo)
|
|
79
|
+
# TODO improve error message
|
|
80
|
+
require "tzinfo"
|
|
78
81
|
end
|
|
82
|
+
|
|
83
|
+
utc_time.getlocal(::TZInfo::Timezone.get(time_zone))
|
|
79
84
|
end
|
|
80
85
|
|
|
81
86
|
def self._to_ruby_duration(value, time_unit = "ns")
|
|
@@ -90,8 +95,13 @@ module Polars
|
|
|
90
95
|
end
|
|
91
96
|
end
|
|
92
97
|
|
|
93
|
-
def self._to_ruby_decimal(
|
|
94
|
-
BigDecimal
|
|
98
|
+
def self._to_ruby_decimal(prec, value)
|
|
99
|
+
unless defined?(::BigDecimal)
|
|
100
|
+
# TODO improve error message
|
|
101
|
+
require "bigdecimal"
|
|
102
|
+
end
|
|
103
|
+
|
|
104
|
+
BigDecimal(value, prec)
|
|
95
105
|
end
|
|
96
106
|
end
|
|
97
107
|
end
|
data/lib/polars/utils/parse.rb
CHANGED
|
@@ -5,7 +5,8 @@ module Polars
|
|
|
5
5
|
str_as_lit: false,
|
|
6
6
|
list_as_series: false,
|
|
7
7
|
structify: false,
|
|
8
|
-
dtype: nil
|
|
8
|
+
dtype: nil,
|
|
9
|
+
require_selector: false
|
|
9
10
|
)
|
|
10
11
|
if input.is_a?(Expr)
|
|
11
12
|
expr = input
|
|
@@ -14,17 +15,22 @@ module Polars
|
|
|
14
15
|
end
|
|
15
16
|
elsif (input.is_a?(::String) || input.is_a?(Symbol)) && !str_as_lit
|
|
16
17
|
expr = F.col(input)
|
|
17
|
-
elsif input.is_a?(::Array) && list_as_series
|
|
18
|
-
expr = F.lit(Series.new(input), dtype: dtype)
|
|
19
18
|
else
|
|
20
|
-
|
|
19
|
+
if require_selector
|
|
20
|
+
msg = "cannot turn #{input.inspect} into selector"
|
|
21
|
+
raise TypeError, msg
|
|
22
|
+
elsif input.is_a?(::Array) && list_as_series
|
|
23
|
+
expr = F.lit(Series.new(input), dtype: dtype)
|
|
24
|
+
else
|
|
25
|
+
expr = F.lit(input, dtype: dtype)
|
|
26
|
+
end
|
|
21
27
|
end
|
|
22
28
|
|
|
23
29
|
expr._rbexpr
|
|
24
30
|
end
|
|
25
31
|
|
|
26
|
-
def self.parse_into_list_of_expressions(*inputs, __structify: false, **named_inputs)
|
|
27
|
-
exprs = _parse_positional_inputs(inputs, structify: __structify)
|
|
32
|
+
def self.parse_into_list_of_expressions(*inputs, __structify: false, __require_selectors: false, **named_inputs)
|
|
33
|
+
exprs = _parse_positional_inputs(inputs, structify: __structify, require_selectors: __require_selectors)
|
|
28
34
|
if named_inputs.any?
|
|
29
35
|
named_exprs = _parse_named_inputs(named_inputs, structify: __structify)
|
|
30
36
|
exprs.concat(named_exprs)
|
|
@@ -33,9 +39,49 @@ module Polars
|
|
|
33
39
|
exprs
|
|
34
40
|
end
|
|
35
41
|
|
|
36
|
-
def self.
|
|
42
|
+
def self.parse_into_selector(i, strict: true)
|
|
43
|
+
if i.is_a?(::String)
|
|
44
|
+
cs = Selectors
|
|
45
|
+
|
|
46
|
+
cs.by_name([i], require_all: strict)
|
|
47
|
+
elsif i.is_a?(Selector)
|
|
48
|
+
i
|
|
49
|
+
elsif i.is_a?(Expr)
|
|
50
|
+
i.meta.as_selector
|
|
51
|
+
else
|
|
52
|
+
msg = "cannot turn #{i.inspect} into selector"
|
|
53
|
+
raise TypeError, msg
|
|
54
|
+
end
|
|
55
|
+
end
|
|
56
|
+
|
|
57
|
+
def self.parse_list_into_selector(inputs, strict: true)
|
|
58
|
+
if inputs.is_a?(::Array)
|
|
59
|
+
cs = Selectors
|
|
60
|
+
|
|
61
|
+
columns = inputs.select { |i| i.is_a?(::String) }
|
|
62
|
+
selector = cs.by_name(columns, require_all: strict)
|
|
63
|
+
|
|
64
|
+
if columns.length == inputs.length
|
|
65
|
+
return selector
|
|
66
|
+
end
|
|
67
|
+
|
|
68
|
+
# A bit cleaner
|
|
69
|
+
if columns.length == 0
|
|
70
|
+
selector = cs.empty
|
|
71
|
+
end
|
|
72
|
+
|
|
73
|
+
inputs.each do |i|
|
|
74
|
+
selector |= parse_into_selector(i, strict: strict)
|
|
75
|
+
end
|
|
76
|
+
selector
|
|
77
|
+
else
|
|
78
|
+
parse_into_selector(inputs, strict: strict)
|
|
79
|
+
end
|
|
80
|
+
end
|
|
81
|
+
|
|
82
|
+
def self._parse_positional_inputs(inputs, structify: false, require_selectors: false)
|
|
37
83
|
inputs_iter = _parse_inputs_as_iterable(inputs)
|
|
38
|
-
inputs_iter.map { |e| parse_into_expression(e, structify: structify) }
|
|
84
|
+
inputs_iter.map { |e| parse_into_expression(e, structify: structify, require_selector: require_selectors) }
|
|
39
85
|
end
|
|
40
86
|
|
|
41
87
|
def self._parse_inputs_as_iterable(inputs)
|
|
@@ -52,7 +98,7 @@ module Polars
|
|
|
52
98
|
|
|
53
99
|
def self._parse_named_inputs(named_inputs, structify: false)
|
|
54
100
|
named_inputs.map do |name, input|
|
|
55
|
-
parse_into_expression(input, structify: structify).
|
|
101
|
+
parse_into_expression(input, structify: structify).alias(name.to_s)
|
|
56
102
|
end
|
|
57
103
|
end
|
|
58
104
|
|
|
@@ -85,5 +131,12 @@ module Polars
|
|
|
85
131
|
|
|
86
132
|
Plr.all_horizontal(predicates)
|
|
87
133
|
end
|
|
134
|
+
|
|
135
|
+
def self._holidays_to_expr(holidays)
|
|
136
|
+
if !(holidays.is_a?(Expr) || holidays.is_a?(Series))
|
|
137
|
+
holidays = Series.new("", [holidays], strict: false, dtype: List.new(Date))
|
|
138
|
+
end
|
|
139
|
+
parse_into_expression(holidays, dtype: Date)
|
|
140
|
+
end
|
|
88
141
|
end
|
|
89
142
|
end
|
|
@@ -0,0 +1,43 @@
|
|
|
1
|
+
module Polars
|
|
2
|
+
module Utils
|
|
3
|
+
def self.reduce_balanced(function, iterable)
|
|
4
|
+
values = iterable.to_a
|
|
5
|
+
|
|
6
|
+
if values.empty?
|
|
7
|
+
msg = "reduce_balanced() of empty iterable"
|
|
8
|
+
raise TypeError, msg
|
|
9
|
+
end
|
|
10
|
+
|
|
11
|
+
if values.length == 1
|
|
12
|
+
return values.shift
|
|
13
|
+
end
|
|
14
|
+
|
|
15
|
+
stack = [[0, values.length]]
|
|
16
|
+
|
|
17
|
+
i = 0
|
|
18
|
+
|
|
19
|
+
while i < stack.length
|
|
20
|
+
offset, length = stack[i]
|
|
21
|
+
half = -(length / -2)
|
|
22
|
+
|
|
23
|
+
if length > 3
|
|
24
|
+
stack << [offset + half, length - half]
|
|
25
|
+
end
|
|
26
|
+
|
|
27
|
+
if length > 2
|
|
28
|
+
stack << [offset, half]
|
|
29
|
+
end
|
|
30
|
+
|
|
31
|
+
stack[i] = [offset, offset + half]
|
|
32
|
+
|
|
33
|
+
i += 1
|
|
34
|
+
end
|
|
35
|
+
|
|
36
|
+
stack.reverse.each do |idx_l, idx_r|
|
|
37
|
+
values[idx_l] = function.(values[idx_l], values[idx_r])
|
|
38
|
+
end
|
|
39
|
+
|
|
40
|
+
values[0]
|
|
41
|
+
end
|
|
42
|
+
end
|
|
43
|
+
end
|
|
@@ -0,0 +1,22 @@
|
|
|
1
|
+
module Polars
|
|
2
|
+
module Utils
|
|
3
|
+
def self.serialize_polars_object(serializer, file)
|
|
4
|
+
serialize_to_bytes = lambda do
|
|
5
|
+
buf = StringIO.new
|
|
6
|
+
serializer.(buf)
|
|
7
|
+
buf.string
|
|
8
|
+
end
|
|
9
|
+
|
|
10
|
+
if file.nil?
|
|
11
|
+
serialize_to_bytes.call
|
|
12
|
+
elsif pathlike?(file)
|
|
13
|
+
file = normalize_filepath(file)
|
|
14
|
+
serializer.(file)
|
|
15
|
+
nil
|
|
16
|
+
else
|
|
17
|
+
serializer.(file)
|
|
18
|
+
nil
|
|
19
|
+
end
|
|
20
|
+
end
|
|
21
|
+
end
|
|
22
|
+
end
|
|
@@ -0,0 +1,19 @@
|
|
|
1
|
+
module Polars
|
|
2
|
+
module Utils
|
|
3
|
+
def self.issue_unstable_warning(message = nil)
|
|
4
|
+
warnings_enabled = ENV.fetch("POLARS_WARN_UNSTABLE", 0).to_i != 0
|
|
5
|
+
if !warnings_enabled
|
|
6
|
+
return
|
|
7
|
+
end
|
|
8
|
+
|
|
9
|
+
if message.nil?
|
|
10
|
+
message = "this functionality is considered unstable."
|
|
11
|
+
end
|
|
12
|
+
message += (
|
|
13
|
+
" It may be changed at any point without it being considered a breaking change."
|
|
14
|
+
)
|
|
15
|
+
|
|
16
|
+
warn message
|
|
17
|
+
end
|
|
18
|
+
end
|
|
19
|
+
end
|