Faster String#squish (#1159)

* Faster String#squish

  squish original ascii-only whitespace (600 bytes)  42.92k ( 23.30µs) (± 0.55%)  5.92kB/op   8.58× slower
 squish optimized ascii-only whitespace (600 bytes) 368.48k (  2.71µs) (± 3.49%)  1.19kB/op        fastest

  squish original w/ unicode whitespace (60 bytes) 528.12k (  1.89µs) (± 2.31%)  624B/op   3.39× slower
squish optimized w/ unicode whitespace (60 bytes)   1.79M (557.76ns) (± 2.61%)  225B/op        fastest

* Replace existing benchmark with one for String#squish per comment in

* Include original spec for String#squish

bench.cr

@@ -1,21 +1,86 @@
 require "benchmark"
 
-text = <<-TEXT
-Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua
-TEXT
+class String
+  # Original
+  def squish_regex : String
+    gsub(/[[:space:]]+/, " ").strip
+  end
 
-string_text = text * 40_000
-io_text = IO::Memory.new(string_text)
+  def squish_new : String
+    if ascii_only?
+      squish_ascii
+    else
+      squish_unicode
+    end
+  end
 
-Benchmark.ips do |x|
-  x.report("string") do
-    io = IO::Memory.new
+  def squish_simplified : String
+    squish_unicode
+  end
 
-    io.print(io_text.to_s)
+  private def squish_ascii : String
+    String.build(size) do |str|
+      print_blank = false
+      each_char do |chr|
+        if chr.ascii_whitespace?
+          if print_blank
+            str << ' '
+            print_blank = false
+          end
+        else
+          print_blank = true
+          str << chr
+        end
+      end
+    end.strip
   end
-  x.report("io") do
-    io = IO::Memory.new
 
-    io.print(io_text)
+  private def squish_unicode : String
+    String.build(size) do |str|
+      print_blank = false
+      each_char do |chr|
+        if chr.whitespace?
+          if print_blank
+            str << ' '
+            print_blank = false
+          end
+        else
+          print_blank = true
+          str << chr
+        end
+      end
+    end.strip
   end
 end
+
+puts "Sanity check the return output is consistent:"
+example = " f f\u00A0\u00A0\u00A0f f \n \t \v\v \f\f 11111  a l0* あ\u00A0\u00A0\u00A0 "
+puts "String to squish " + example.inspect
+puts "regex: " + example.squish_regex.inspect
+puts "new: " + example.squish_new.inspect
+puts "simplified: " + example.squish_simplified.inspect
+
+# Original regex doesn't seem to work correctly with trailing unicode
+if example.squish_regex != example.squish_new
+  puts "WARN: regex version does not match:"
+  puts "Regex:  #{example.squish_regex.inspect}".ljust(50)
+  puts "Ours:  #{example.squish_new.inspect}".ljust(50)
+end
+
+puts
+puts "Benchmarking String#squish ..."
+puts
+example = " f f f f \n \t\r\r   11111  a l0* " * 20
+Benchmark.ips(warmup: 5, calculation: 10) do |x|
+  x.report("squish regex ascii-only whitespace (#{example.bytesize} bytes)") { example.squish_regex }
+  x.report("squish optimized ascii-only whitespace (#{example.bytesize} bytes)") { example.squish_new }
+  x.report("squish simplified ascii-only whitespace (#{example.bytesize} bytes)") { example.squish_simplified }
+end
+
+puts
+example = "あ\u00A0あ\u00A0\u00A0 \t  z  \n \r \r \t \v \f zzzz XXX  asdf    k ; ;, \u1680\u1680 " * 10
+Benchmark.ips(warmup: 5, calculation: 10) do |x|
+  x.report("squish regex w/ unicode whitespace (#{example.bytesize} bytes)") { example.squish_regex }
+  x.report("squish optimized w/ unicode whitespace (#{example.bytesize} bytes)") { example.squish_new }
+  x.report("squish simplified w/ unicode whitespace (#{example.bytesize} bytes)") { example.squish_simplified }
+end

spec/charms/string_spec.cr

@@ -8,5 +8,11 @@ describe "String charm" do
       og_string.squish.should eq("foo bar boo")
       og_string.should eq(" foo   bar    \n   \t   boo")
     end
+
+    it "squishes the text by removing ascii/unicode whitespace" do
+      og_string = "\u1680 \v\v\v\v\v\r\r\r\r hello foo bar\n\u00A0\t\t\u00A0\u1680\u1680   "
+
+      og_string.squish.should eq("hello foo bar")
+    end
   end
 end

src/charms/string_extensions.cr

@@ -10,6 +10,45 @@ class String
   # replace newlines with a single space and convert mutiple spaces to just one
   # space.
   def squish : String
-    gsub(/[[:space:]]+/, " ").strip
+    if ascii_only?
+      squish_ascii
+    else
+      squish_unicode
+    end
+  end
+
+  # Optimized for ASCII using String#ascii_whitespace?
+  private def squish_ascii : String
+    String.build(size) do |str|
+      print_blank = false
+      each_char do |chr|
+        if chr.ascii_whitespace?
+          if print_blank
+            str << ' '
+            print_blank = false
+          end
+        else
+          print_blank = true
+          str << chr
+        end
+      end
+    end.strip
+  end
+
+  private def squish_unicode : String
+    String.build(size) do |str|
+      print_blank = false
+      each_char do |chr|
+        if chr.whitespace?
+          if print_blank
+            str << ' '
+            print_blank = false
+          end
+        else
+          print_blank = true
+          str << chr
+        end
+      end
+    end.strip
   end
 end