From 6535421997f9b06ac48a9b4569a77b1f7754db64 Mon Sep 17 00:00:00 2001 From: Herwin Date: Fri, 28 Feb 2025 13:21:34 +0100 Subject: [PATCH] Add specs for Kernel.Float with hex strings They are very much driven by the current (inconsistent) behaviour of MRI. I'm not sure if these should be seen as a specification or as MRI specific implementation quirks. --- core/kernel/Float_spec.rb | 125 +++++++++++++++++++++++--------------- 1 file changed, 77 insertions(+), 48 deletions(-) diff --git a/core/kernel/Float_spec.rb b/core/kernel/Float_spec.rb index 6cedfe061..5aeca7642 100644 --- a/core/kernel/Float_spec.rb +++ b/core/kernel/Float_spec.rb @@ -222,59 +222,88 @@ def to_f() 1.2 end end end - describe "for hexadecimal literals with binary exponent" do - %w(p P).each do |p| - it "interprets the fractional part (on the left side of '#{p}') in hexadecimal" do - @object.send(:Float, "0x10#{p}0").should == 16.0 - end - - it "interprets the exponent (on the right of '#{p}') in decimal" do - @object.send(:Float, "0x1#{p}10").should == 1024.0 - end - - it "raises an ArgumentError if #{p} is the trailing character" do - -> { @object.send(:Float, "0x1#{p}") }.should raise_error(ArgumentError) - end - - it "raises an ArgumentError if #{p} is the leading character" do - -> { @object.send(:Float, "0x#{p}1") }.should raise_error(ArgumentError) - end - - it "returns Infinity for '0x1#{p}10000'" do - @object.send(:Float, "0x1#{p}10000").should == Float::INFINITY - end - - it "returns 0 for '0x1#{p}-10000'" do - @object.send(:Float, "0x1#{p}-10000").should == 0 - end - - it "allows embedded _ in a number on either side of the #{p}" do - @object.send(:Float, "0x1_0#{p}10").should == 16384.0 - @object.send(:Float, "0x10#{p}1_0").should == 16384.0 - @object.send(:Float, "0x1_0#{p}1_0").should == 16384.0 - end - - it "raises an exception if a space is embedded on either side of the '#{p}'" do - -> { @object.send(:Float, "0x1 0#{p}10") }.should raise_error(ArgumentError) - -> { @object.send(:Float, "0x10#{p}1 0") }.should raise_error(ArgumentError) - end + context "for hexadecimal literals" do + it "interprets the 0x prefix as hexadecimal" do + @object.send(:Float, "0x10").should == 16.0 + @object.send(:Float, "0x0F").should == 15.0 + @object.send(:Float, "0x0f").should == 15.0 + end - it "raises an exception if there's a leading _ on either side of the '#{p}'" do - -> { @object.send(:Float, "0x_10#{p}10") }.should raise_error(ArgumentError) - -> { @object.send(:Float, "0x10#{p}_10") }.should raise_error(ArgumentError) - end + it "accepts embedded _ if the number does not contain a-f" do + @object.send(:Float, "0x1_0").should == 16.0 + -> { @object.send(:Float, "0x1_0a") }.should raise_error(ArgumentError) + @object.send(:Float, "0x1_0a", exception: false).should be_nil + end - it "raises an exception if there's a trailing _ on either side of the '#{p}'" do - -> { @object.send(:Float, "0x10_#{p}10") }.should raise_error(ArgumentError) - -> { @object.send(:Float, "0x10#{p}10_") }.should raise_error(ArgumentError) - end + it "does not accept _ before, after or inside the 0x prefix" do + -> { @object.send(:Float, "_0x10") }.should raise_error(ArgumentError) + -> { @object.send(:Float, "0_x10") }.should raise_error(ArgumentError) + -> { @object.send(:Float, "0x_10") }.should raise_error(ArgumentError) + @object.send(:Float, "_0x10", exception: false).should be_nil + @object.send(:Float, "0_x10", exception: false).should be_nil + @object.send(:Float, "0x_10", exception: false).should be_nil + end - it "allows hexadecimal points on the left side of the '#{p}'" do - @object.send(:Float, "0x1.8#{p}0").should == 1.5 + ruby_version_is "3.4"..."" do + it "accepts a fractional part" do + @object.send(:Float, "0x0.8").should == 0.5 end + end - it "raises an ArgumentError if there's a decimal point on the right side of the '#{p}'" do - -> { @object.send(:Float, "0x1#{p}1.0") }.should raise_error(ArgumentError) + describe "with binary exponent" do + %w(p P).each do |p| + it "interprets the fractional part (on the left side of '#{p}') in hexadecimal" do + @object.send(:Float, "0x10#{p}0").should == 16.0 + end + + it "interprets the exponent (on the right of '#{p}') in decimal" do + @object.send(:Float, "0x1#{p}10").should == 1024.0 + end + + it "raises an ArgumentError if #{p} is the trailing character" do + -> { @object.send(:Float, "0x1#{p}") }.should raise_error(ArgumentError) + end + + it "raises an ArgumentError if #{p} is the leading character" do + -> { @object.send(:Float, "0x#{p}1") }.should raise_error(ArgumentError) + end + + it "returns Infinity for '0x1#{p}10000'" do + @object.send(:Float, "0x1#{p}10000").should == Float::INFINITY + end + + it "returns 0 for '0x1#{p}-10000'" do + @object.send(:Float, "0x1#{p}-10000").should == 0 + end + + it "allows embedded _ in a number on either side of the #{p}" do + @object.send(:Float, "0x1_0#{p}10").should == 16384.0 + @object.send(:Float, "0x10#{p}1_0").should == 16384.0 + @object.send(:Float, "0x1_0#{p}1_0").should == 16384.0 + end + + it "raises an exception if a space is embedded on either side of the '#{p}'" do + -> { @object.send(:Float, "0x1 0#{p}10") }.should raise_error(ArgumentError) + -> { @object.send(:Float, "0x10#{p}1 0") }.should raise_error(ArgumentError) + end + + it "raises an exception if there's a leading _ on either side of the '#{p}'" do + -> { @object.send(:Float, "0x_10#{p}10") }.should raise_error(ArgumentError) + -> { @object.send(:Float, "0x10#{p}_10") }.should raise_error(ArgumentError) + end + + it "raises an exception if there's a trailing _ on either side of the '#{p}'" do + -> { @object.send(:Float, "0x10_#{p}10") }.should raise_error(ArgumentError) + -> { @object.send(:Float, "0x10#{p}10_") }.should raise_error(ArgumentError) + end + + it "allows hexadecimal points on the left side of the '#{p}'" do + @object.send(:Float, "0x1.8#{p}0").should == 1.5 + end + + it "raises an ArgumentError if there's a decimal point on the right side of the '#{p}'" do + -> { @object.send(:Float, "0x1#{p}1.0") }.should raise_error(ArgumentError) + end end end end