From 83276b057b02f125e09870efa2bae6404cecc81c Mon Sep 17 00:00:00 2001
From: Quinton Miller <nicetas.c@gmail.com>
Date: Fri, 15 Nov 2024 20:16:20 +0800
Subject: [PATCH] handle non-ascii whitespace

---
 spec/std/string_spec.cr              |  3 +++
 src/float/fast_float.cr              | 35 +++++++++++-----------------
 src/float/fast_float/parse_number.cr |  8 +------
 src/string.cr                        |  6 +++--
 4 files changed, 22 insertions(+), 30 deletions(-)

diff --git a/spec/std/string_spec.cr b/spec/std/string_spec.cr
index 595ece196bac..3c63b7b23460 100644
--- a/spec/std/string_spec.cr
+++ b/spec/std/string_spec.cr
@@ -504,6 +504,7 @@ describe "String" do
     "  1234.56  ".to_f?(whitespace: false).should be_nil
     expect_raises(ArgumentError) { "  1234.56foo".to_f }
     "  1234.56foo".to_f?.should be_nil
+    "\u{A0}\u{2028}\u{2029}1234.56\u{A0}\u{2028}\u{2029}".to_f.should eq(1234.56_f64)
     "123.45 x".to_f64(strict: false).should eq(123.45_f64)
     expect_raises(ArgumentError) { "x1.2".to_f64 }
     "x1.2".to_f64?.should be_nil
@@ -548,6 +549,7 @@ describe "String" do
     "  1234.56  ".to_f32?(whitespace: false).should be_nil
     expect_raises(ArgumentError) { "  1234.56foo".to_f32 }
     "  1234.56foo".to_f32?.should be_nil
+    "\u{A0}\u{2028}\u{2029}1234.56\u{A0}\u{2028}\u{2029}".to_f32.should eq(1234.56_f32)
     "123.45 x".to_f32(strict: false).should eq(123.45_f32)
     expect_raises(ArgumentError) { "x1.2".to_f32 }
     "x1.2".to_f32?.should be_nil
@@ -591,6 +593,7 @@ describe "String" do
     "  1234.56  ".to_f64?(whitespace: false).should be_nil
     expect_raises(ArgumentError) { "  1234.56foo".to_f64 }
     "  1234.56foo".to_f64?.should be_nil
+    "\u{A0}\u{2028}\u{2029}1234.56\u{A0}\u{2028}\u{2029}".to_f64.should eq(1234.56_f64)
     "123.45 x".to_f64(strict: false).should eq(123.45_f64)
     expect_raises(ArgumentError) { "x1.2".to_f64 }
     "x1.2".to_f64?.should be_nil
diff --git a/src/float/fast_float.cr b/src/float/fast_float.cr
index cbb3305657d1..010476db4bca 100644
--- a/src/float/fast_float.cr
+++ b/src/float/fast_float.cr
@@ -42,11 +42,14 @@ struct Float
       finish = start + str.bytesize
       options = ParseOptionsT(typeof(str.to_unsafe.value)).new(format: :general)
 
-      ret = BinaryFormat_Float64.new.from_chars_advanced(start, finish, pointerof(value), options, whitespace: whitespace)
-      if ret.ec == Errno::NONE
-        if trailing_chars_allowed?(ret.ptr, finish, whitespace, strict)
-          value
-        end
+      if whitespace
+        start += str.calc_excess_left
+        finish -= str.calc_excess_right
+      end
+
+      ret = BinaryFormat_Float64.new.from_chars_advanced(start, finish, pointerof(value), options)
+      if ret.ec == Errno::NONE && (!strict || ret.ptr == finish)
+        value
       end
     end
 
@@ -56,24 +59,14 @@ struct Float
       finish = start + str.bytesize
       options = ParseOptionsT(typeof(str.to_unsafe.value)).new(format: :general)
 
-      ret = BinaryFormat_Float32.new.from_chars_advanced(start, finish, pointerof(value), options, whitespace: whitespace)
-      if ret.ec == Errno::NONE
-        if trailing_chars_allowed?(ret.ptr, finish, whitespace, strict)
-          value
-        end
+      if whitespace
+        start += str.calc_excess_left
+        finish -= str.calc_excess_right
       end
-    end
 
-    private def self.trailing_chars_allowed?(ptr, finish, whitespace, strict)
-      if strict
-        if whitespace
-          while ptr < finish && ptr.value.unsafe_chr.ascii_whitespace?
-            ptr += 1
-          end
-        end
-        ptr == finish
-      else
-        true
+      ret = BinaryFormat_Float32.new.from_chars_advanced(start, finish, pointerof(value), options)
+      if ret.ec == Errno::NONE && (!strict || ret.ptr == finish)
+        value
       end
     end
   end
diff --git a/src/float/fast_float/parse_number.cr b/src/float/fast_float/parse_number.cr
index ef9663dca74e..3c1ac4c1cb24 100644
--- a/src/float/fast_float/parse_number.cr
+++ b/src/float/fast_float/parse_number.cr
@@ -172,18 +172,12 @@ module Float::FastFloat
       FromCharsResultT(UC).new(ptr, ec)
     end
 
-    # NOTE(crystal): *whitespace* is a normal argument
-    def from_chars_advanced(first : UC*, last : UC*, value : T*, options : ParseOptionsT(UC), whitespace : Bool) : FromCharsResultT(UC) forall UC
+    def from_chars_advanced(first : UC*, last : UC*, value : T*, options : ParseOptionsT(UC)) : FromCharsResultT(UC) forall UC
       {% raise "only some floating-point types are supported" unless T == Float32 || T == Float64 %}
 
       # TODO(crystal): support UInt16 and UInt32
       {% raise "only UInt8 is supported" unless UC == UInt8 %}
 
-      if whitespace
-        while first != last && first.value.unsafe_chr.ascii_whitespace?
-          first += 1
-        end
-      end
       if first == last
         return FromCharsResultT(UC).new(first, Errno::EINVAL)
       end
diff --git a/src/string.cr b/src/string.cr
index 5cfb1d1f4c54..4014cb0ff52a 100644
--- a/src/string.cr
+++ b/src/string.cr
@@ -2105,7 +2105,8 @@ class String
     remove_excess_left(excess_left)
   end
 
-  private def calc_excess_right
+  # :nodoc:
+  def calc_excess_right
     if single_byte_optimizable?
       i = bytesize - 1
       while i >= 0 && to_unsafe[i].unsafe_chr.ascii_whitespace?
@@ -2143,7 +2144,8 @@ class String
     bytesize - byte_index
   end
 
-  private def calc_excess_left
+  # :nodoc:
+  def calc_excess_left
     if single_byte_optimizable?
       excess_left = 0
       # All strings end with '\0', and it's not a whitespace