polars-df 0.20.0 → 0.21.1

data/lib/polars/struct_expr.rb

@@ -57,6 +57,36 @@ module Polars
       Utils.wrap_expr(_rbexpr.struct_field_by_name(name))
     end
 
+    # Expand the struct into its individual fields.
+    #
+    # Alias for `Expr.struct.field("*")`.
+    #
+    # @return [Expr]
+    #
+    # @example
+    #   df = Polars::DataFrame.new(
+    #     {
+    #       "aaa" => [1, 2],
+    #       "bbb" => ["ab", "cd"],
+    #       "ccc" => [true, nil],
+    #       "ddd" => [[1, 2], [3]]
+    #     }
+    #   ).select(Polars.struct("aaa", "bbb", "ccc", "ddd").alias("struct_col"))
+    #   df.select(Polars.col("struct_col").struct.unnest)
+    #   # =>
+    #   # shape: (2, 4)
+    #   # ┌─────┬─────┬──────┬───────────┐
+    #   # │ aaa ┆ bbb ┆ ccc  ┆ ddd       │
+    #   # │ --- ┆ --- ┆ ---  ┆ ---       │
+    #   # │ i64 ┆ str ┆ bool ┆ list[i64] │
+    #   # ╞═════╪═════╪══════╪═══════════╡
+    #   # │ 1   ┆ ab  ┆ true ┆ [1, 2]    │
+    #   # │ 2   ┆ cd  ┆ null ┆ [3]       │
+    #   # └─────┴─────┴──────┴───────────┘
+    def unnest
+      field("*")
+    end
+
     # Rename the fields of the struct.
     #
     # @param names [Array]
@@ -94,5 +124,78 @@ module Polars
     def rename_fields(names)
       Utils.wrap_expr(_rbexpr.struct_rename_fields(names))
     end
+
+    # Convert this struct to a string column with json values.
+    #
+    # @return [Expr]
+    #
+    # @example
+    #   Polars::DataFrame.new(
+    #     {"a" => [{"a" => [1, 2], "b" => [45]}, {"a" => [9, 1, 3], "b" => nil}]}
+    #   ).with_columns(Polars.col("a").struct.json_encode.alias("encoded"))
+    #   # =>
+    #   # shape: (2, 2)
+    #   # ┌──────────────────┬────────────────────────┐
+    #   # │ a                ┆ encoded                │
+    #   # │ ---              ┆ ---                    │
+    #   # │ struct[2]        ┆ str                    │
+    #   # ╞══════════════════╪════════════════════════╡
+    #   # │ {[1, 2],[45]}    ┆ {"a":[1,2],"b":[45]}   │
+    #   # │ {[9, 1, 3],null} ┆ {"a":[9,1,3],"b":null} │
+    #   # └──────────────────┴────────────────────────┘
+    def json_encode
+      Utils.wrap_expr(_rbexpr.struct_json_encode)
+    end
+
+    # Add or overwrite fields of this struct.
+    #
+    # This is similar to `with_columns` on `DataFrame`.
+    #
+    # @param exprs [Array]
+    #   Field(s) to add, specified as positional arguments.
+    #   Accepts expression input. Strings are parsed as column names, other
+    #   non-expression inputs are parsed as literals.
+    # @param named_exprs [Hash]
+    #   Additional fields to add, specified as keyword arguments.
+    #   The columns will be renamed to the keyword used.
+    #
+    # @return [Expr]
+    #
+    # @example
+    #   df = Polars::DataFrame.new(
+    #     {
+    #       "coords" => [{"x" => 1, "y" => 4}, {"x" => 4, "y" => 9}, {"x" => 9, "y" => 16}],
+    #       "multiply" => [10, 2, 3]
+    #     }
+    #   )
+    #   df.with_columns(
+    #     Polars.col("coords").struct.with_fields(
+    #       Polars.field("x").sqrt,
+    #       y_mul: Polars.field("y") * Polars.col("multiply")
+    #     )
+    #   )
+    #   # =>
+    #   # shape: (3, 2)
+    #   # ┌─────────────┬──────────┐
+    #   # │ coords      ┆ multiply │
+    #   # │ ---         ┆ ---      │
+    #   # │ struct[3]   ┆ i64      │
+    #   # ╞═════════════╪══════════╡
+    #   # │ {1.0,4,40}  ┆ 10       │
+    #   # │ {2.0,9,18}  ┆ 2        │
+    #   # │ {3.0,16,48} ┆ 3        │
+    #   # └─────────────┴──────────┘
+    def with_fields(
+      *exprs,
+      **named_exprs
+    )
+      structify = ENV.fetch("POLARS_AUTO_STRUCTIFY", 0).to_i != 0
+
+      rbexprs = Utils.parse_into_list_of_expressions(
+        *exprs, **named_exprs, __structify: structify
+      )
+
+      Utils.wrap_expr(_rbexpr.struct_with_fields(rbexprs))
+    end
   end
 end

data/lib/polars/struct_name_space.rb

@@ -80,7 +80,7 @@ module Polars
       super
     end
 
-    # Get the struct definition as a name/dtype schema dict.
+    # Get the struct definition as a name/dtype schema hash.
     #
     # @return [Object]
     #
@@ -116,5 +116,23 @@ module Polars
     def unnest
       Utils.wrap_df(_s.struct_unnest)
     end
+
+    # Convert this struct to a string column with json values.
+    #
+    # @return [Series]
+    #
+    # @example
+    #   s = Polars::Series.new("a", [{"a" => [1, 2], "b" => [45]}, {"a" => [9, 1, 3], "b" => nil}])
+    #   s.struct.json_encode
+    #   # =>
+    #   # shape: (2,)
+    #   # Series: 'a' [str]
+    #   # [
+    #   #         "{"a":[1,2],"b":[45]}"
+    #   #         "{"a":[9,1,3],"b":null}"
+    #   # ]
+    def json_encode
+      super
+    end
   end
 end

data/lib/polars/utils/parse.rb

@@ -33,6 +33,46 @@ module Polars
       exprs
     end
 
+    def self.parse_into_selector(i, strict: true)
+      if i.is_a?(::String)
+        cs = Selectors
+
+        cs.by_name([i], require_all: strict)
+      elsif i.is_a?(Selector)
+        i
+      elsif i.is_a?(Expr)
+        i.meta.as_selector
+      else
+        msg = "cannot turn #{i.inspect} into selector"
+        raise TypeError, msg
+      end
+    end
+
+    def self.parse_list_into_selector(inputs, strict: true)
+      if inputs.is_a?(::Array)
+        cs = Selectors
+
+        columns = inputs.select { |i| i.is_a?(::String) }
+        selector = cs.by_name(columns, require_all: strict)
+
+        if columns.length == inputs.length
+          return selector
+        end
+
+        # A bit cleaner
+        if columns.length == 0
+          selector = cs.empty
+        end
+
+        inputs.each do |i|
+          selector |= parse_into_selector(i, strict: strict)
+        end
+        selector
+      else
+        parse_into_selector(inputs, strict: strict)
+      end
+    end
+
     def self._parse_positional_inputs(inputs, structify: false)
       inputs_iter = _parse_inputs_as_iterable(inputs)
       inputs_iter.map { |e| parse_into_expression(e, structify: structify) }

data/lib/polars/utils/various.rb

@@ -16,7 +16,10 @@ module Polars
       val.is_a?(::Array) && val.all? { |x| pathlike?(x) }
     end
 
-    def self.is_bool_sequence(val)
+    def self.is_bool_sequence(val, include_series: false)
+      if include_series && val.is_a?(Series)
+        return val.dtype == Boolean
+      end
       val.is_a?(::Array) && val.all? { |x| x == true || x == false }
     end
 
@@ -24,6 +27,10 @@ module Polars
       val.is_a?(::Array) && _is_iterable_of(val, Integer)
     end
 
+    def self.is_sequence(val, include_series: false)
+      val.is_a?(::Array) || (include_series && val.is_a?(Series))
+    end
+
     def self.is_str_sequence(val, allow_str: false)
       if allow_str == false && val.is_a?(::String)
         false
@@ -76,5 +83,15 @@ module Polars
       end
       values
     end
+
+    def self.require_same_type(current, other)
+      if !other.is_a?(current.class) && !current.is_a?(other.class)
+        msg = (
+          "expected `other` to be a #{current.inspect}, " +
+          "not #{other.inspect}"
+        )
+        raise TypeError, msg
+      end
+    end
   end
 end

data/lib/polars/utils.rb

@@ -107,7 +107,7 @@ module Polars
     end
 
     def self.is_selector(obj)
-      obj.is_a?(Selectors::SelectorProxy)
+      obj.is_a?(Selector)
     end
 
     def self.expand_selector(target, selector, strict: true)
@@ -161,5 +161,13 @@ module Polars
       # escapes _only_ those metachars with meaning to the rust regex crate
       Plr.re_escape(s)
     end
+
+    def self.parse_into_datatype_expr(input)
+      if input.is_a?(DataTypeExpr)
+        input
+      else
+        parse_into_dtype(input).to_dtype_expr
+      end
+    end
   end
 end

data/lib/polars/version.rb

@@ -1,4 +1,4 @@
 module Polars
   # @private
-  VERSION = "0.20.0"
+  VERSION = "0.21.1"
 end

data/lib/polars.rb

@@ -19,11 +19,17 @@ require_relative "polars/binary_expr"
 require_relative "polars/binary_name_space"
 require_relative "polars/cat_expr"
 require_relative "polars/cat_name_space"
+require_relative "polars/catalog"
+require_relative "polars/catalog/unity/catalog_info"
+require_relative "polars/catalog/unity/column_info"
+require_relative "polars/catalog/unity/namespace_info"
+require_relative "polars/catalog/unity/table_info"
 require_relative "polars/config"
 require_relative "polars/convert"
 require_relative "polars/plot"
 require_relative "polars/data_frame"
 require_relative "polars/data_types"
+require_relative "polars/data_type_expr"
 require_relative "polars/data_type_group"
 require_relative "polars/date_time_expr"
 require_relative "polars/date_time_name_space"
@@ -32,6 +38,7 @@ require_relative "polars/exceptions"
 require_relative "polars/expr"
 require_relative "polars/functions/as_datatype"
 require_relative "polars/functions/col"
+require_relative "polars/functions/datatype"
 require_relative "polars/functions/eager"
 require_relative "polars/functions/lazy"
 require_relative "polars/functions/len"
@@ -54,6 +61,7 @@ require_relative "polars/io/ipc"
 require_relative "polars/io/json"
 require_relative "polars/io/ndjson"
 require_relative "polars/io/parquet"
+require_relative "polars/io/scan_options"
 require_relative "polars/lazy_frame"
 require_relative "polars/lazy_group_by"
 require_relative "polars/list_expr"
@@ -61,7 +69,9 @@ require_relative "polars/list_name_space"
 require_relative "polars/meta_expr"
 require_relative "polars/name_expr"
 require_relative "polars/rolling_group_by"
+require_relative "polars/scan_cast_options"
 require_relative "polars/schema"
+require_relative "polars/selector"
 require_relative "polars/selectors"
 require_relative "polars/series"
 require_relative "polars/slice"

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: polars-df
 version: !ruby/object:Gem::Version
-  version: 0.20.0
+  version: 0.21.1
 platform: ruby
 authors:
 - Andrew Kane
@@ -53,7 +53,10 @@ files:
 - ext/polars/extconf.rb
 - ext/polars/src/allocator.rs
 - ext/polars/src/batched_csv.rs
+- ext/polars/src/catalog/mod.rs
+- ext/polars/src/catalog/unity.rs
 - ext/polars/src/conversion/any_value.rs
+- ext/polars/src/conversion/categorical.rs
 - ext/polars/src/conversion/chunked_array.rs
 - ext/polars/src/conversion/mod.rs
 - ext/polars/src/dataframe/construction.rs
@@ -66,7 +69,9 @@ files:
 - ext/polars/src/exceptions.rs
 - ext/polars/src/expr/array.rs
 - ext/polars/src/expr/binary.rs
+- ext/polars/src/expr/bitwise.rs
 - ext/polars/src/expr/categorical.rs
+- ext/polars/src/expr/datatype.rs
 - ext/polars/src/expr/datetime.rs
 - ext/polars/src/expr/general.rs
 - ext/polars/src/expr/list.rs
@@ -74,6 +79,7 @@ files:
 - ext/polars/src/expr/mod.rs
 - ext/polars/src/expr/name.rs
 - ext/polars/src/expr/rolling.rs
+- ext/polars/src/expr/selector.rs
 - ext/polars/src/expr/string.rs
 - ext/polars/src/expr/struct.rs
 - ext/polars/src/file.rs
@@ -96,6 +102,7 @@ files:
 - ext/polars/src/interop/numo/numo_rs.rs
 - ext/polars/src/interop/numo/to_numo_df.rs
 - ext/polars/src/interop/numo/to_numo_series.rs
+- ext/polars/src/io/mod.rs
 - ext/polars/src/lazyframe/general.rs
 - ext/polars/src/lazyframe/mod.rs
 - ext/polars/src/lazyframe/serde.rs
@@ -117,6 +124,7 @@ files:
 - ext/polars/src/series/export.rs
 - ext/polars/src/series/general.rs
 - ext/polars/src/series/import.rs
+- ext/polars/src/series/map.rs
 - ext/polars/src/series/mod.rs
 - ext/polars/src/series/scatter.rs
 - ext/polars/src/sql.rs
@@ -130,9 +138,15 @@ files:
 - lib/polars/binary_name_space.rb
 - lib/polars/cat_expr.rb
 - lib/polars/cat_name_space.rb
+- lib/polars/catalog.rb
+- lib/polars/catalog/unity/catalog_info.rb
+- lib/polars/catalog/unity/column_info.rb
+- lib/polars/catalog/unity/namespace_info.rb
+- lib/polars/catalog/unity/table_info.rb
 - lib/polars/config.rb
 - lib/polars/convert.rb
 - lib/polars/data_frame.rb
+- lib/polars/data_type_expr.rb
 - lib/polars/data_type_group.rb
 - lib/polars/data_types.rb
 - lib/polars/date_time_expr.rb
@@ -145,6 +159,7 @@ files:
 - lib/polars/functions/aggregation/vertical.rb
 - lib/polars/functions/as_datatype.rb
 - lib/polars/functions/col.rb
+- lib/polars/functions/datatype.rb
 - lib/polars/functions/eager.rb
 - lib/polars/functions/lazy.rb
 - lib/polars/functions/len.rb
@@ -165,6 +180,7 @@ files:
 - lib/polars/io/json.rb
 - lib/polars/io/ndjson.rb
 - lib/polars/io/parquet.rb
+- lib/polars/io/scan_options.rb
 - lib/polars/lazy_frame.rb
 - lib/polars/lazy_group_by.rb
 - lib/polars/list_expr.rb
@@ -173,7 +189,9 @@ files:
 - lib/polars/name_expr.rb
 - lib/polars/plot.rb
 - lib/polars/rolling_group_by.rb
+- lib/polars/scan_cast_options.rb
 - lib/polars/schema.rb
+- lib/polars/selector.rb
 - lib/polars/selectors.rb
 - lib/polars/series.rb
 - lib/polars/slice.rb
@@ -210,7 +228,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.6.7
+rubygems_version: 3.6.9
 specification_version: 4
 summary: Blazingly fast DataFrames for Ruby
 test_files: []