Speed up lexing

.gitignore

@@ -6,3 +6,4 @@ pkg
 .rvmrc
 .bundle
 .byebug_history
+Gemfile.lock

.ruby-version

@@ -1 +1 @@
-3.3.0
+3.3.4

Gemfile.lock

@@ -1,75 +0,0 @@
-GIT
-  remote: https://github.com/Shopify/liquid-c.git
-  revision: 5a786af7284df55e013ea20551c4b688d02e8326
-  ref: main
-  specs:
-    liquid-c (4.2.0)
-      liquid (>= 5.0.1)
-
-PATH
-  remote: .
-  specs:
-    liquid (5.6.0.alpha)
-
-GEM
-  remote: https://rubygems.org/
-  specs:
-    ast (2.4.2)
-    base64 (0.2.0)
-    benchmark-ips (2.13.0)
-    json (2.7.2)
-    language_server-protocol (3.17.0.3)
-    memory_profiler (1.0.1)
-    minitest (5.22.3)
-    parallel (1.24.0)
-    parser (3.3.0.5)
-      ast (~> 2.4.1)
-      racc
-    racc (1.7.3)
-    rainbow (3.1.1)
-    rake (13.2.1)
-    regexp_parser (2.9.0)
-    rexml (3.2.6)
-    rubocop (1.61.0)
-      json (~> 2.3)
-      language_server-protocol (>= 3.17.0)
-      parallel (~> 1.10)
-      parser (>= 3.3.0.2)
-      rainbow (>= 2.2.2, < 4.0)
-      regexp_parser (>= 1.8, < 3.0)
-      rexml (>= 3.2.5, < 4.0)
-      rubocop-ast (>= 1.30.0, < 2.0)
-      ruby-progressbar (~> 1.7)
-      unicode-display_width (>= 2.4.0, < 3.0)
-    rubocop-ast (1.31.2)
-      parser (>= 3.3.0.4)
-    rubocop-performance (1.19.1)
-      rubocop (>= 1.7.0, < 2.0)
-      rubocop-ast (>= 0.4.0)
-    rubocop-shopify (2.12.0)
-      rubocop (~> 1.44)
-    ruby-progressbar (1.13.0)
-    stackprof (0.2.26)
-    terminal-table (3.0.2)
-      unicode-display_width (>= 1.1.1, < 3)
-    unicode-display_width (2.5.0)
-
-PLATFORMS
-  ruby
-
-DEPENDENCIES
-  base64
-  benchmark-ips
-  liquid!
-  liquid-c!
-  memory_profiler
-  minitest
-  rake (~> 13.0)
-  rubocop (~> 1.61.0)
-  rubocop-performance
-  rubocop-shopify (~> 2.12.0)
-  stackprof
-  terminal-table
-
-BUNDLED WITH
-   2.5.7

Rakefile

@@ -81,6 +81,14 @@ namespace :benchmark do
   task :strict do
     ruby "./performance/benchmark.rb strict"
   end
+
+  desc "Run unit benchmarks"
+  task :unit do
+    Dir["./performance/unit/*_benchmark.rb"].each do |file|
+      puts "🧪 Running #{file}"
+      ruby file
+    end
+  end
 end
 
 namespace :profile do

lib/liquid/lexer.rb

@@ -1,8 +1,9 @@
 # frozen_string_literal: true
 
 require "strscan"
+
 module Liquid
-  class Lexer
+  class Lexer1
     SPECIALS = {
       '|' => :pipe,
       '.' => :dot,
@@ -58,4 +59,141 @@ module Liquid
       @output << [:end_of_string]
     end
   end
+
+  class Lexer2
+    CLOSE_ROUND = [:close_round, ")"].freeze
+    CLOSE_SQUARE = [:close_square, "]"].freeze
+    COLON = [:colon, ":"].freeze
+    COMMA = [:comma, ","].freeze
+    COMPARISION_NOT_EQUAL = [:comparison, "!="].freeze
+    COMPARISON_CONTAINS = [:comparison, "contains"].freeze
+    COMPARISON_EQUAL = [:comparison, "=="].freeze
+    COMPARISON_GREATER_THAN = [:comparison, ">"].freeze
+    COMPARISON_GREATER_THAN_OR_EQUAL = [:comparison, ">="].freeze
+    COMPARISON_LESS_THAN = [:comparison, "<"].freeze
+    COMPARISON_LESS_THAN_OR_EQUAL = [:comparison, "<="].freeze
+    COMPARISON_NOT_EQUAL_ALT = [:comparison, "<>"].freeze
+    CONTAINS = /contains(?=\s)/
+    DASH = [:dash, "-"].freeze
+    DOT = [:dot, "."].freeze
+    DOTDOT = [:dotdot, ".."].freeze
+    DOT_ORD = ".".ord
+    DOUBLE_STRING_LITERAL = /"[^\"]*"/
+    EOS = [:end_of_string].freeze
+    IDENTIFIER            = /[a-zA-Z_][\w-]*\??/
+    NUMBER_LITERAL        = /-?\d+(\.\d+)?/
+    OPEN_ROUND = [:open_round, "("].freeze
+    OPEN_SQUARE = [:open_square, "["].freeze
+    PIPE = [:pipe, "|"].freeze
+    QUESTION = [:question, "?"].freeze
+    RUBY_WHITESPACE = [" ", "\t", "\r", "\n", "\f"].freeze
+    SINGLE_STRING_LITERAL = /'[^\']*'/
+    WHITESPACE_OR_NOTHING = /\s*/
+
+    COMPARISON_JUMP_TABLE = [].tap do |table|
+      table["=".ord] = [].tap do |sub_table|
+        sub_table["=".ord] = COMPARISON_EQUAL
+        sub_table.freeze
+      end
+      table["!".ord] = [].tap do |sub_table|
+        sub_table["=".ord] = COMPARISION_NOT_EQUAL
+        sub_table.freeze
+      end
+      table["<".ord] = [].tap do |sub_table|
+        sub_table["=".ord] = COMPARISON_LESS_THAN_OR_EQUAL
+        sub_table[">".ord] = COMPARISON_NOT_EQUAL_ALT
+        RUBY_WHITESPACE.each { |c| sub_table[c.ord] = COMPARISON_LESS_THAN }
+        sub_table.freeze
+      end
+      table[">".ord] = [].tap do |sub_table|
+        sub_table["=".ord] = COMPARISON_GREATER_THAN_OR_EQUAL
+        RUBY_WHITESPACE.each { |c| sub_table[c.ord] = COMPARISON_GREATER_THAN }
+        sub_table.freeze
+      end
+      table.freeze
+    end
+
+    NEXT_MATCHER_JUMP_TABLE = [].tap do |table|
+      "a".upto("z") do |c|
+        table[c.ord] = [:id, IDENTIFIER].freeze
+        table[c.upcase.ord] = [:id, IDENTIFIER].freeze
+      end
+      table["_".ord] = [:id, IDENTIFIER].freeze
+
+      "0".upto("9") do |c|
+        table[c.ord] = [:number, NUMBER_LITERAL].freeze
+      end
+      table["-".ord] = [:number, NUMBER_LITERAL].freeze
+
+      table["'".ord] = [:string, SINGLE_STRING_LITERAL].freeze
+      table["\"".ord] = [:string, DOUBLE_STRING_LITERAL].freeze
+      table.freeze
+    end
+
+    SPECIAL_TABLE = [].tap do |table|
+      table["|".ord] = PIPE
+      table[".".ord] = DOT
+      table[":".ord] = COLON
+      table[",".ord] = COMMA
+      table["[".ord] = OPEN_SQUARE
+      table["]".ord] = CLOSE_SQUARE
+      table["(".ord] = OPEN_ROUND
+      table[")".ord] = CLOSE_ROUND
+      table["?".ord] = QUESTION
+      table["-".ord] = DASH
+    end
+
+    def initialize(input)
+      @ss = StringScanner.new(input)
+    end
+
+    def tokenize
+      @output = []
+
+      until @ss.eos?
+        @ss.skip(WHITESPACE_OR_NOTHING)
+
+        break if @ss.eos?
+
+        peeked = @ss.peek_byte
+
+        if (special = SPECIAL_TABLE[peeked])
+          @ss.scan_byte
+          if special == DOT && @ss.peek_byte == DOT_ORD
+            @ss.scan_byte
+            @output << DOTDOT
+          else
+            @output << special
+          end
+        elsif (sub_table = COMPARISON_JUMP_TABLE[peeked])
+          @ss.scan_byte
+          next_peeked = @ss.peek_byte
+          if (found = sub_table[next_peeked])
+            @output << found
+            @ss.scan_byte
+          else
+            raise SyntaxError, "Unexpected character #{peeked.chr}"
+          end
+        else
+          type, pattern = NEXT_MATCHER_JUMP_TABLE[peeked]
+
+          if type && (t = @ss.scan(pattern))
+            # rubocop:disable Metrics/BlockNesting
+            @output << if type == :id && t == "contains"
+              COMPARISON_CONTAINS
+            else
+              [type, t]
+            end
+            # rubocop:enable Metrics/BlockNesting
+          else
+            raise SyntaxError, "Unexpected character #{peeked.chr}"
+          end
+        end
+      end
+
+      @output << EOS
+    end
+  end
+
+  Lexer = StringScanner.instance_methods.include?(:scan_byte) ? Lexer2 : Lexer1
 end

lib/liquid/parser.rb

@@ -53,7 +53,7 @@ module Liquid
         str = consume
         str << variable_lookups
       when :open_square
-        str = consume
+        str = consume.dup
         str << expression
         str << consume(:close_square)
         str << variable_lookups

liquid.gemspec

@@ -28,6 +28,9 @@ Gem::Specification.new do |s|
 
   s.require_path = "lib"
 
+  s.add_dependency("strscan")
+  s.add_dependency("bigdecimal")
+
   s.add_development_dependency('rake', '~> 13.0')
   s.add_development_dependency('minitest')
 end

performance/benchmark.rb

@@ -3,6 +3,7 @@
 require 'benchmark/ips'
 require_relative 'theme_runner'
 
+RubyVM::YJIT.enable if defined?(RubyVM::YJIT)
 Liquid::Template.error_mode = ARGV.first.to_sym if ARGV.first
 profiler = ThemeRunner.new
 

performance/unit/lexer_benchmark.rb

@@ -0,0 +1,49 @@
+# frozen_string_literal: true
+
+require "benchmark/ips"
+
+# benchmark liquid lexing
+
+require 'liquid'
+
+RubyVM::YJIT.enable
+
+EXPRESSIONS = [
+  "foo[1..2].baz",
+  "12.0",
+  "foo.bar.based",
+  "21 - 62",
+  "foo.bar.baz",
+  "foo > 12",
+  "foo < 12",
+  "foo <= 12",
+  "foo >= 12",
+  "foo <> 12",
+  "foo == 12",
+  "foo != 12",
+  "foo contains 12",
+  "foo contains 'bar'",
+  "foo != 'bar'",
+  "'foo' contains 'bar'",
+  '234089',
+]
+
+Benchmark.ips do |x|
+  x.config(time: 10, warmup: 5)
+
+  x.report("Liquid::Lexer1#tokenize") do
+    EXPRESSIONS.each do |expr|
+      l = Liquid::Lexer1.new(expr)
+      l.tokenize
+    end
+  end
+
+  x.report("Liquid::Lexer2#tokenize") do
+    EXPRESSIONS.each do |expr|
+      l = Liquid::Lexer2.new(expr)
+      l.tokenize
+    end
+  end
+
+  x.compare!
+end