polars-df 0.10.0 → 0.12.0

data/lib/polars/utils/various.rb

@@ -0,0 +1,76 @@
+module Polars
+  module Utils
+    def self._process_null_values(null_values)
+      if null_values.is_a?(Hash)
+        null_values.to_a
+      else
+        null_values
+      end
+    end
+
+    def self._is_iterable_of(val, eltype)
+      val.all? { |x| x.is_a?(eltype) }
+    end
+
+    def self.is_bool_sequence(val)
+      val.is_a?(::Array) && val.all? { |x| x == true || x == false }
+    end
+
+    def self.is_int_sequence(val)
+      val.is_a?(::Array) && _is_iterable_of(val, Integer)
+    end
+
+    def self.is_str_sequence(val, allow_str: false)
+      if allow_str == false && val.is_a?(::String)
+        false
+      else
+        val.is_a?(::Array) && _is_iterable_of(val, ::String)
+      end
+    end
+
+    def self.arrlen(obj)
+      if obj.is_a?(Range)
+        # size only works for numeric ranges
+        obj.to_a.length
+      elsif obj.is_a?(::String)
+        nil
+      else
+        obj.length
+      end
+    rescue
+      nil
+    end
+
+    def self.normalize_filepath(path, check_not_directory: true)
+      path = File.expand_path(path)
+      if check_not_directory && File.exist?(path) && Dir.exist?(path)
+        raise ArgumentError, "Expected a file path; #{path} is a directory"
+      end
+      path
+    end
+
+    def self.scale_bytes(sz, to:)
+      scaling_factor = {
+        "b" => 1,
+        "k" => 1024,
+        "m" => 1024 ** 2,
+        "g" => 1024 ** 3,
+        "t" => 1024 ** 4
+      }[to[0]]
+      if scaling_factor > 1
+        sz / scaling_factor.to_f
+      else
+        sz
+      end
+    end
+
+    def self.extend_bool(value, n_match, value_name, match_name)
+      values = bool?(value) ? [value] * n_match : value
+      if n_match != values.length
+        msg = "the length of `#{value_name}` (#{values.length}) does not match the length of `#{match_name}` (#{n_match})"
+        raise ValueError, msg
+      end
+      values
+    end
+  end
+end

data/lib/polars/utils/wrap.rb

@@ -0,0 +1,19 @@
+module Polars
+  module Utils
+    def self.wrap_df(df)
+      DataFrame._from_rbdf(df)
+    end
+
+    def self.wrap_ldf(ldf)
+      LazyFrame._from_rbldf(ldf)
+    end
+
+    def self.wrap_s(s)
+      Series._from_rbseries(s)
+    end
+
+    def self.wrap_expr(rbexpr)
+      Expr._from_rbexpr(rbexpr)
+    end
+  end
+end

data/lib/polars/utils.rb

@@ -3,150 +3,6 @@ module Polars
   module Utils
     DTYPE_TEMPORAL_UNITS = ["ns", "us", "ms"]
 
-    def self.wrap_s(s)
-      Series._from_rbseries(s)
-    end
-
-    def self.wrap_df(df)
-      DataFrame._from_rbdf(df)
-    end
-
-    def self.wrap_ldf(ldf)
-      LazyFrame._from_rbldf(ldf)
-    end
-
-    def self.wrap_expr(rbexpr)
-      Expr._from_rbexpr(rbexpr)
-    end
-
-    def self.col(name)
-      Polars.col(name)
-    end
-
-    def self.arrlen(obj)
-      if obj.is_a?(Range)
-        # size only works for numeric ranges
-        obj.to_a.length
-      elsif obj.is_a?(::String)
-        nil
-      else
-        obj.length
-      end
-    rescue
-      nil
-    end
-
-    def self._timedelta_to_pl_duration(td)
-      td
-    end
-
-    def self._datetime_to_pl_timestamp(dt, time_unit)
-      dt = dt.to_datetime.to_time
-      if time_unit == "ns"
-        nanos = dt.nsec
-        dt.to_i * 1_000_000_000 + nanos
-      elsif time_unit == "us"
-        micros = dt.usec
-        dt.to_i * 1_000_000 + micros
-      elsif time_unit == "ms"
-        millis = dt.usec / 1000
-        dt.to_i * 1_000 + millis
-      elsif time_unit.nil?
-        # Ruby has ns precision
-        nanos = dt.nsec
-        dt.to_i * 1_000_000_000 + nanos
-      else
-        raise ArgumentError, "time_unit must be one of {{'ns', 'us', 'ms'}}, got #{tu}"
-      end
-    end
-
-    def self._date_to_pl_date(d)
-      dt = d.to_datetime.to_time
-      dt.to_i / (3600 * 24)
-    end
-
-    def self._to_ruby_time(value)
-      if value == 0
-        ::Time.utc(2000, 1, 1)
-      else
-        seconds, nanoseconds = value.divmod(1_000_000_000)
-        minutes, seconds = seconds.divmod(60)
-        hours, minutes = minutes.divmod(60)
-        ::Time.utc(2000, 1, 1, hours, minutes, seconds, nanoseconds / 1000.0)
-      end
-    end
-
-    def self._to_ruby_duration(value, time_unit = "ns")
-      if time_unit == "ns"
-        value / 1e9
-      elsif time_unit == "us"
-        value / 1e6
-      elsif time_unit == "ms"
-        value / 1e3
-      else
-        raise ArgumentError, "time_unit must be one of {{'ns', 'us', 'ms'}}, got #{time_unit}"
-      end
-    end
-
-    def self._to_ruby_date(value)
-      # days to seconds
-      # important to create from utc. Not doing this leads
-      # to inconsistencies dependent on the timezone you are in.
-      ::Time.at(value * 86400).utc.to_date
-    end
-
-    def self._to_ruby_datetime(value, time_unit = "ns", time_zone = nil)
-      if time_zone.nil? || time_zone == ""
-        if time_unit == "ns"
-          return ::Time.at(value / 1000000000, value % 1000000000, :nsec).utc
-        elsif time_unit == "us"
-          return ::Time.at(value / 1000000, value % 1000000, :usec).utc
-        elsif time_unit == "ms"
-          return ::Time.at(value / 1000, value % 1000, :millisecond).utc
-        else
-          raise ArgumentError, "time_unit must be one of {{'ns', 'us', 'ms'}}, got #{time_unit}"
-        end
-      else
-        raise Todo
-      end
-    end
-
-    def self._to_ruby_decimal(digits, scale)
-      BigDecimal("#{digits}e#{scale}")
-    end
-
-    def self.selection_to_rbexpr_list(exprs)
-      if exprs.is_a?(::String) || exprs.is_a?(Symbol) || exprs.is_a?(Expr) || exprs.is_a?(Series)
-        exprs = [exprs]
-      end
-
-      exprs.map { |e| expr_to_lit_or_expr(e, str_to_lit: false)._rbexpr }
-    end
-
-    def self.expr_to_lit_or_expr(expr, str_to_lit: true)
-      if (expr.is_a?(::String) || expr.is_a?(Symbol)) && !str_to_lit
-        col(expr)
-      elsif expr.is_a?(Integer) || expr.is_a?(Float) || expr.is_a?(::String) || expr.is_a?(Symbol) || expr.is_a?(Series) || expr.nil?
-        lit(expr)
-      elsif expr.is_a?(Expr)
-        expr
-      else
-        raise ArgumentError, "did not expect value #{expr} of type #{expr.class.name}, maybe disambiguate with Polars.lit or Polars.col"
-      end
-    end
-
-    def self.lit(value)
-      Polars.lit(value)
-    end
-
-    def self.normalise_filepath(path, check_not_directory: true)
-      path = File.expand_path(path)
-      if check_not_directory && File.exist?(path) && Dir.exist?(path)
-        raise ArgumentError, "Expected a file path; #{path} is a directory"
-      end
-      path
-    end
-
     # TODO fix
     def self.is_polars_dtype(data_type, include_unknown: false)
       if data_type == Unknown
@@ -191,19 +47,11 @@ module Polars
       end
     end
 
-    def self._process_null_values(null_values)
-      if null_values.is_a?(Hash)
-        null_values.to_a
-      else
-        null_values
-      end
-    end
-
-    def self._prepare_row_count_args(row_count_name = nil, row_count_offset = 0)
-      if !row_count_name.nil?
-        [row_count_name, row_count_offset]
-      else
+    def self.parse_row_index_args(row_index_name = nil, row_index_offset = 0)
+      if row_index_name.nil?
         nil
+      else
+        [row_index_name, row_index_offset]
       end
     end
 
@@ -223,21 +71,6 @@ module Polars
       [projection, columns]
     end
 
-    def self.scale_bytes(sz, to:)
-      scaling_factor = {
-        "b" => 1,
-        "k" => 1024,
-        "m" => 1024 ** 2,
-        "g" => 1024 ** 3,
-        "t" => 1024 ** 4
-      }[to[0]]
-      if scaling_factor > 1
-        sz / scaling_factor.to_f
-      else
-        sz
-      end
-    end
-
     def self.bool?(value)
       value.is_a?(TrueClass) || value.is_a?(FalseClass)
     end
@@ -250,108 +83,10 @@ module Polars
       value.is_a?(::String) || (defined?(Pathname) && value.is_a?(Pathname))
     end
 
-    def self._is_iterable_of(val, eltype)
-      val.all? { |x| x.is_a?(eltype) }
-    end
-
-    def self.is_bool_sequence(val)
-      val.is_a?(::Array) && val.all? { |x| x == true || x == false }
-    end
-
-    def self.is_dtype_sequence(val)
-      val.is_a?(::Array) && val.all? { |x| is_polars_dtype(x) }
-    end
-
-    def self.is_int_sequence(val)
-      val.is_a?(::Array) && _is_iterable_of(val, Integer)
-    end
-
-    def self.is_expr_sequence(val)
-      val.is_a?(::Array) && _is_iterable_of(val, Expr)
-    end
-
-    def self.is_rbexpr_sequence(val)
-      val.is_a?(::Array) && _is_iterable_of(val, RbExpr)
-    end
-
-    def self.is_str_sequence(val, allow_str: false)
-      if allow_str == false && val.is_a?(::String)
-        false
-      else
-        val.is_a?(::Array) && _is_iterable_of(val, ::String)
-      end
-    end
-
     def self.local_file?(file)
       Dir.glob(file).any?
     end
 
-    def self.parse_as_list_of_expressions(*inputs, __structify: false, **named_inputs)
-      exprs = _parse_positional_inputs(inputs, structify: __structify)
-      if named_inputs.any?
-        named_exprs = _parse_named_inputs(named_inputs, structify: __structify)
-        exprs.concat(named_exprs)
-      end
-
-      exprs
-    end
-
-    def self._parse_positional_inputs(inputs, structify: false)
-      inputs_iter = _parse_inputs_as_iterable(inputs)
-      inputs_iter.map { |e| parse_as_expression(e, structify: structify) }
-    end
-
-    def self._parse_inputs_as_iterable(inputs)
-      if inputs.empty?
-        return []
-      end
-
-      if inputs.length == 1 && inputs[0].is_a?(::Array)
-        return inputs[0]
-      end
-
-      inputs
-    end
-
-    def self._parse_named_inputs(named_inputs, structify: false)
-      named_inputs.map do |name, input|
-        parse_as_expression(input, structify: structify)._alias(name.to_s)
-      end
-    end
-
-    def self.parse_as_expression(input, str_as_lit: false, list_as_lit: true, structify: false, dtype: nil)
-      if input.is_a?(Expr)
-        expr = input
-      elsif input.is_a?(::String) && !str_as_lit
-        expr = Polars.col(input)
-        structify = false
-      elsif input.is_a?(::Array) && !list_as_lit
-        expr = Polars.lit(Series.new(input), dtype: dtype)
-        structify = false
-      else
-        expr = Polars.lit(input, dtype: dtype)
-        structify = false
-      end
-
-      if structify
-        raise Todo
-      end
-
-      expr._rbexpr
-    end
-
-    USE_EARLIEST_TO_AMBIGUOUS = {
-      true => "earliest",
-      false => "latest"
-    }
-
-    def self.rename_use_earliest_to_ambiguous(use_earliest, ambiguous)
-      unless use_earliest.nil?
-        ambiguous = USE_EARLIEST_TO_AMBIGUOUS.fetch(use_earliest)
-      end
-      ambiguous
-    end
-
     def self._check_arg_is_1byte(arg_name, arg, can_be_empty = false)
       if arg.is_a?(::String)
         arg_byte_length = arg.bytesize
@@ -385,38 +120,11 @@ module Polars
       false
     end
 
-    def self.parse_predicates_constraints_as_expression(*predicates, **constraints)
-      all_predicates = _parse_positional_inputs(predicates)
-
-      if constraints.any?
-        constraint_predicates = _parse_constraints(constraints)
-        all_predicates.concat(constraint_predicates)
-      end
-
-      _combine_predicates(all_predicates)
-    end
-
-    def self._parse_constraints(constraints)
-      constraints.map do |name, value|
-        Polars.col(name).eq(value)._rbexpr
+    def self.parse_interval_argument(interval)
+      if interval.include?(" ")
+        interval = interval.gsub(" ", "")
       end
-    end
-
-    def self._combine_predicates(predicates)
-      if !predicates.any?
-        msg = "at least one predicate or constraint must be provided"
-        raise TypeError, msg
-      end
-
-      if predicates.length == 1
-        return predicates[0]
-      end
-
-      Plr.all_horizontal(predicates)
-    end
-
-    def self.parse_when_inputs(*predicates, **constraints)
-      parse_predicates_constraints_as_expression(*predicates, **constraints)
+      interval.downcase
     end
   end
 end

data/lib/polars/version.rb

@@ -1,4 +1,4 @@
 module Polars
   # @private
-  VERSION = "0.10.0"
+  VERSION = "0.12.0"
 end

data/lib/polars/whenthen.rb

@@ -8,7 +8,7 @@ module Polars
     end
 
     def then(statement)
-      statement_rbexpr = Utils.parse_as_expression(statement)
+      statement_rbexpr = Utils.parse_into_expression(statement)
       Then.new(_when.then(statement_rbexpr))
     end
   end
@@ -30,12 +30,12 @@ module Polars
     end
 
     def when(*predicates, **constraints)
-      condition_rbexpr = Utils.parse_when_inputs(*predicates, **constraints)
+      condition_rbexpr = Utils.parse_predicates_constraints_into_expression(*predicates, **constraints)
       ChainedWhen.new(_then.when(condition_rbexpr))
     end
 
     def otherwise(statement)
-      statement_rbexpr = Utils.parse_as_expression(statement)
+      statement_rbexpr = Utils.parse_into_expression(statement)
       Utils.wrap_expr(_then.otherwise(statement_rbexpr))
     end
   end
@@ -49,7 +49,7 @@ module Polars
     end
 
     def then(statement)
-      statement_rbexpr = Utils.parse_as_expression(statement)
+      statement_rbexpr = Utils.parse_into_expression(statement)
       ChainedThen.new(_chained_when.then(statement_rbexpr))
     end
   end
@@ -71,12 +71,12 @@ module Polars
     end
 
     def when(*predicates, **constraints)
-      condition_rbexpr = Utils.parse_when_inputs(*predicates, **constraints)
+      condition_rbexpr = Utils.parse_predicates_constraints_into_expression(*predicates, **constraints)
       ChainedWhen.new(_chained_then.when(condition_rbexpr))
     end
 
     def otherwise(statement)
-      statement_rbexpr = Utils.parse_as_expression(statement)
+      statement_rbexpr = Utils.parse_into_expression(statement)
       Utils.wrap_expr(_chained_then.otherwise(statement_rbexpr))
     end
   end

data/lib/polars.rb

@@ -42,9 +42,17 @@ require_relative "polars/functions/whenthen"
 require_relative "polars/functions/aggregation/horizontal"
 require_relative "polars/functions/aggregation/vertical"
 require_relative "polars/functions/range/date_range"
+require_relative "polars/functions/range/datetime_range"
 require_relative "polars/functions/range/int_range"
+require_relative "polars/functions/range/time_range"
 require_relative "polars/group_by"
-require_relative "polars/io"
+require_relative "polars/io/avro"
+require_relative "polars/io/csv"
+require_relative "polars/io/database"
+require_relative "polars/io/ipc"
+require_relative "polars/io/json"
+require_relative "polars/io/ndjson"
+require_relative "polars/io/parquet"
 require_relative "polars/lazy_frame"
 require_relative "polars/lazy_group_by"
 require_relative "polars/list_expr"
@@ -62,6 +70,11 @@ require_relative "polars/struct_expr"
 require_relative "polars/struct_name_space"
 require_relative "polars/testing"
 require_relative "polars/utils"
+require_relative "polars/utils/constants"
+require_relative "polars/utils/convert"
+require_relative "polars/utils/parse"
+require_relative "polars/utils/various"
+require_relative "polars/utils/wrap"
 require_relative "polars/version"
 require_relative "polars/whenthen"
 
@@ -69,4 +82,10 @@ module Polars
   extend Convert
   extend Functions
   extend IO
+
+  # @private
+  F = self
+
+  # @private
+  N_INFER_DEFAULT = 100
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: polars-df
 version: !ruby/object:Gem::Version
-  version: 0.10.0
+  version: 0.12.0
 platform: ruby
 authors:
 - Andrew Kane
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-05-03 00:00:00.000000000 Z
+date: 2024-07-11 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bigdecimal
@@ -54,10 +54,14 @@ files:
 - ext/polars/Cargo.toml
 - ext/polars/extconf.rb
 - ext/polars/src/batched_csv.rs
-- ext/polars/src/conversion/anyvalue.rs
+- ext/polars/src/conversion/any_value.rs
 - ext/polars/src/conversion/chunked_array.rs
 - ext/polars/src/conversion/mod.rs
-- ext/polars/src/dataframe.rs
+- ext/polars/src/dataframe/construction.rs
+- ext/polars/src/dataframe/export.rs
+- ext/polars/src/dataframe/general.rs
+- ext/polars/src/dataframe/io.rs
+- ext/polars/src/dataframe/mod.rs
 - ext/polars/src/error.rs
 - ext/polars/src/expr/array.rs
 - ext/polars/src/expr/binary.rs
@@ -83,6 +87,10 @@ files:
 - ext/polars/src/functions/range.rs
 - ext/polars/src/functions/string_cache.rs
 - ext/polars/src/functions/whenthen.rs
+- ext/polars/src/interop/mod.rs
+- ext/polars/src/interop/numo/mod.rs
+- ext/polars/src/interop/numo/to_numo_df.rs
+- ext/polars/src/interop/numo/to_numo_series.rs
 - ext/polars/src/lazyframe/mod.rs
 - ext/polars/src/lazygroupby.rs
 - ext/polars/src/lib.rs
@@ -133,11 +141,19 @@ files:
 - lib/polars/functions/lit.rb
 - lib/polars/functions/random.rb
 - lib/polars/functions/range/date_range.rb
+- lib/polars/functions/range/datetime_range.rb
 - lib/polars/functions/range/int_range.rb
+- lib/polars/functions/range/time_range.rb
 - lib/polars/functions/repeat.rb
 - lib/polars/functions/whenthen.rb
 - lib/polars/group_by.rb
-- lib/polars/io.rb
+- lib/polars/io/avro.rb
+- lib/polars/io/csv.rb
+- lib/polars/io/database.rb
+- lib/polars/io/ipc.rb
+- lib/polars/io/json.rb
+- lib/polars/io/ndjson.rb
+- lib/polars/io/parquet.rb
 - lib/polars/lazy_frame.rb
 - lib/polars/lazy_group_by.rb
 - lib/polars/list_expr.rb
@@ -156,9 +172,14 @@ files:
 - lib/polars/struct_name_space.rb
 - lib/polars/testing.rb
 - lib/polars/utils.rb
+- lib/polars/utils/constants.rb
+- lib/polars/utils/convert.rb
+- lib/polars/utils/parse.rb
+- lib/polars/utils/various.rb
+- lib/polars/utils/wrap.rb
 - lib/polars/version.rb
 - lib/polars/whenthen.rb
-homepage: https://github.com/ankane/polars-ruby
+homepage: https://github.com/ankane/ruby-polars
 licenses:
 - MIT
 metadata: {}
@@ -177,7 +198,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.5.9
+rubygems_version: 3.5.11
 signing_key:
 specification_version: 4
 summary: Blazingly fast DataFrames for Ruby