don't create string scanner for liquid tag

fix parsing quirky incomplete expressions

Gemfile

@@ -28,3 +28,5 @@ group :test do
     gem 'liquid-c', github: 'Shopify/liquid-c', ref: 'main'
   end
 end
+
+gem "strscan", ">= 3.1"

Rakefile

@@ -73,7 +73,7 @@ end
 
 namespace :benchmark do
   desc "Run the liquid benchmark with lax parsing"
-  task :run do
+  task :lax do
     ruby "./performance/benchmark.rb lax"
   end
 
@@ -82,6 +82,9 @@ namespace :benchmark do
     ruby "./performance/benchmark.rb strict"
   end
 
+  desc "Run the liquid benchmark with both lax and strict parsing"
+  task run: [:lax, :strict]
+
   desc "Run unit benchmarks"
   task :unit do
     Dir["./performance/unit/*_benchmark.rb"].each do |file|

lib/liquid/block_body.rb

@@ -246,10 +246,17 @@ module Liquid
     end
 
     def create_variable(token, parse_context)
-      if token =~ ContentOfVariable
-        markup = Regexp.last_match(1)
+      if token.end_with?("}}")
+        i = 2
+        i = 3 if token[i] == "-"
+        parse_end = token.length - 3
+        parse_end -= 1 if token[parse_end] == "-"
+        markup_end = parse_end - i + 1
+        markup = markup_end <= 0 ? "" : token.slice(i, markup_end)
+
         return Variable.new(markup, parse_context)
       end
+
       BlockBody.raise_missing_variable_terminator(token, parse_context)
     end
 

lib/liquid/lexer.rb

@@ -73,7 +73,6 @@ module Liquid
     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
@@ -90,7 +89,13 @@ module Liquid
     SINGLE_STRING_LITERAL = /'[^\']*'/
     WHITESPACE_OR_NOTHING = /\s*/
 
-    COMPARISON_JUMP_TABLE = [].tap do |table|
+    SINGLE_COMPARISON_TOKENS = [].tap do |table|
+      table["<".ord] = COMPARISON_LESS_THAN
+      table[">".ord] = COMPARISON_GREATER_THAN
+      table.freeze
+    end
+
+    TWO_CHARS_COMPARISON_JUMP_TABLE = [].tap do |table|
       table["=".ord] = [].tap do |sub_table|
         sub_table["=".ord] = COMPARISON_EQUAL
         sub_table.freeze
@@ -99,15 +104,17 @@ module Liquid
         sub_table["=".ord] = COMPARISION_NOT_EQUAL
         sub_table.freeze
       end
+      table.freeze
+    end
+
+    COMPARISON_JUMP_TABLE = [].tap do |table|
       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
@@ -163,6 +170,7 @@ module Liquid
 
         break if @ss.eos?
 
+        start_pos = @ss.pos
         peeked = @ss.peek_byte
 
         if (special = SPECIAL_TABLE[peeked])
@@ -173,7 +181,7 @@ module Liquid
             @output << DOTDOT
           elsif special == DASH
             # Special case for negative numbers
-            if NUMBER_TABLE[@ss.peek_byte]
+            if (peeked_byte = @ss.peek_byte) && NUMBER_TABLE[peeked_byte]
               @ss.pos -= 1
               @output << [:number, @ss.scan(NUMBER_LITERAL)]
             else
@@ -182,26 +190,34 @@ module Liquid
           else
             @output << special
           end
-        elsif (sub_table = COMPARISON_JUMP_TABLE[peeked])
+        elsif (sub_table = TWO_CHARS_COMPARISON_JUMP_TABLE[peeked])
           @ss.scan_byte
-          if (found = sub_table[@ss.peek_byte])
+          if (peeked_byte = @ss.peek_byte) && (found = sub_table[peeked_byte])
             @output << found
             @ss.scan_byte
           else
-            raise SyntaxError, "Unexpected character #{peeked.chr}"
+            raise_syntax_error(start_pos)
+          end
+        elsif (sub_table = COMPARISON_JUMP_TABLE[peeked])
+          @ss.scan_byte
+          if (peeked_byte = @ss.peek_byte) && (found = sub_table[peeked_byte])
+            @output << found
+            @ss.scan_byte
+          else
+            @output << SINGLE_COMPARISON_TOKENS[peeked]
           end
         else
           type, pattern = NEXT_MATCHER_JUMP_TABLE[peeked]
 
           if type && (t = @ss.scan(pattern))
             # Special case for "contains"
-            @output << if type == :id && t == "contains"
+            @output << if type == :id && t == "contains" && @output.last&.first != :dot
               COMPARISON_CONTAINS
             else
               [type, t]
             end
           else
-            raise SyntaxError, "Unexpected character #{peeked.chr}"
+            raise_syntax_error(start_pos)
           end
         end
       end
@@ -209,7 +225,14 @@ module Liquid
 
       @output << EOS
     end
+
+    def raise_syntax_error(start_pos)
+      @ss.pos = start_pos
+      # the character could be a UTF-8 character, use getch to get all the bytes
+      raise SyntaxError, "Unexpected character #{@ss.getch}"
+    end
   end
 
+  # Remove this once we can depend on strscan >= 3.1.1
   Lexer = StringScanner.instance_methods.include?(:scan_byte) ? Lexer2 : Lexer1
 end

lib/liquid/standardfilters.rb

@@ -877,7 +877,7 @@ module Liquid
     #   - [`nil`](/docs/api/liquid/basics#nil)
     # @liquid_syntax variable | default: variable
     # @liquid_return [untyped]
-    # @liquid_optional_param allow_false [boolean] Whether to use false values instead of the default.
+    # @liquid_optional_param allow_false: [boolean] Whether to use false values instead of the default.
     def default(input, default_value = '', options = {})
       options = {} unless options.is_a?(Hash)
       false_check = options['allow_false'] ? input.nil? : !Liquid::Utils.to_liquid_value(input)

lib/liquid/tokenizer.rb

@@ -3,38 +3,23 @@
 require "strscan"
 
 module Liquid
-  class Tokenizer
+  class Tokenizer1
     attr_reader :line_number, :for_liquid_tag
 
-    TAG_END = /%\}/
-    TAG_OR_VARIABLE_START = /\{[\{\%}]/
-    NEWLINE = /\n/
-
-    OPEN_CURLEY = 123
-    CLOSE_CURLEY = 125
-    PERCENTAGE = 37
-
     def initialize(source, line_numbers = false, line_number: nil, for_liquid_tag: false)
-      @line_number = line_number || (line_numbers ? 1 : nil)
-
+      @source         = source
+      @line_number    = line_number || (line_numbers ? 1 : nil)
       @for_liquid_tag = for_liquid_tag
-      @tokens = []
-      tokenize(source)
-    end
-
-    private def tokenize(source)
-      ss = StringScanner.new(source)
-
-      while token = t_shift(ss)
-        @tokens.push(token)
-      end
+      @offset         = 0
+      @tokens         = tokenize
     end
 
     def shift
-      token = @tokens.shift
-
+      token = @tokens[@offset]
       return nil unless token
 
+      @offset += 1
+
       if @line_number
         @line_number += @for_liquid_tag ? 1 : token.count("\n")
       end
@@ -42,105 +27,164 @@ module Liquid
       token
     end
 
-    private def t_shift(ss)
-      return nil if ss.eos?
+    private
 
-      @for_liquid_tag ? next_liquid_token(ss) : next_token(ss)
+    def tokenize
+      return [] if @source.empty?
+
+      return @source.split("\n") if @for_liquid_tag
+
+      tokens = @source.split(TemplateParser)
+
+      # removes the rogue empty element at the beginning of the array
+      if tokens[0]&.empty?
+        @offset += 1
+      end
+
+      tokens
+    end
+  end
+
+  class Tokenizer2
+    attr_reader :line_number, :for_liquid_tag
+
+    TAG_END = /%\}/
+    TAG_OR_VARIABLE_START = /\{[\{\%]/
+    NEWLINE = /\n/
+
+    OPEN_CURLEY = "{".ord
+    CLOSE_CURLEY = "}".ord
+    PERCENTAGE = "%".ord
+
+    def initialize(source, line_numbers = false, line_number: nil, for_liquid_tag: false)
+      @line_number    = line_number || (line_numbers ? 1 : nil)
+      @for_liquid_tag = for_liquid_tag
+      @source         = source
+      @offset         = 0
+      @tokens         = []
+      tokenize
+    end
+
+    def shift
+      token = @tokens[@offset]
+
+      return unless token
+
+      @offset += 1
+
+      if @line_number
+        @line_number += @for_liquid_tag ? 1 : token.count("\n")
+      end
+
+      token
     end
 
     private
 
-    def next_liquid_token(ss)
-      # read until we find a \n
-      start = ss.pos
-      if ss.scan_until(NEWLINE).nil?
-        token = ss.rest
-        ss.terminate
-        return token
+    def tokenize
+      if @for_liquid_tag
+        @tokens = @source.split("\n")
+      else
+        @ss = StringScanner.new(@source)
+        @tokens << shift_normal until @ss.eos?
       end
 
-      ss.string.byteslice(start, ss.pos - start - 1)
+      @ss = nil
+      @source = nil
     end
 
-    def next_token(ss)
-      if ss.pos == 0
-        if ss.string.start_with?("{{")
-          ss.pos = 2
-          return next_variable_token(ss)
-        elsif ss.string.start_with?("{%")
-          ss.pos = 2
-          return next_tag_token(ss)
-        else
-          return next_text_token(ss)
-        end
-      end
+    def shift_normal
+      token = next_token
 
+      return unless token
+
+      token
+    end
+
+    def next_token
       # possible states: :text, :tag, :variable
-      byte_a = ss.scan_byte
+      byte_a = @ss.peek_byte
 
       if byte_a == OPEN_CURLEY
-        byte_b = ss.scan_byte
+        @ss.scan_byte
+
+        byte_b = @ss.peek_byte
 
         if byte_b == PERCENTAGE
-          return next_tag_token(ss)
+          @ss.scan_byte
+          return next_tag_token
         elsif byte_b == OPEN_CURLEY
-          return next_variable_token(ss)
+          @ss.scan_byte
+          return next_variable_token
         end
 
-        ss.pos -= 1
+        @ss.pos -= 1
       end
 
-      ss.pos -= 1
-      next_text_token(ss)
+      next_text_token
     end
 
-    def next_text_token(ss)
-      start = ss.pos
+    def next_text_token
+      start = @ss.pos
 
-      unless ss.skip_until(TAG_OR_VARIABLE_START)
-        token = ss.rest
-        ss.terminate
+      unless @ss.skip_until(TAG_OR_VARIABLE_START)
+        token = @ss.rest
+        @ss.terminate
         return token
       end
 
-      ss.pos -= 2
-      ss.string.byteslice(start, ss.pos - start)
+      pos = @ss.pos -= 2
+      @source.byteslice(start, pos - start)
     end
 
-    def next_variable_token(ss)
-      start = ss.pos - 2
+    def next_variable_token
+      start = @ss.pos - 2
 
-      # it is possible to see a {% before a }} so we need to check for that
-      byte_a = ss.scan_byte
+      byte_a = byte_b = @ss.scan_byte
 
-      until ss.eos?
-        while ss.eos? == false && byte_a != CLOSE_CURLEY && byte_a != OPEN_CURLEY
-          byte_a = ss.scan_byte
+      while byte_b
+        byte_a = @ss.scan_byte while byte_a && (byte_a != CLOSE_CURLEY && byte_a != OPEN_CURLEY)
+
+        break unless byte_a
+
+        if @ss.eos?
+          return byte_a == CLOSE_CURLEY ? @source.byteslice(start, @ss.pos - start) : "{{"
         end
 
-        break if ss.eos?
+        byte_b = @ss.scan_byte
 
-        byte_b = ss.scan_byte
-
-        if byte_b != CLOSE_CURLEY && byte_b != PERCENTAGE
-          byte_a = byte_b
-          next
-        elsif byte_a == CLOSE_CURLEY && byte_b == CLOSE_CURLEY
-          return ss.string.byteslice(start, ss.pos - start)
+        if byte_a == CLOSE_CURLEY
+          if byte_b == CLOSE_CURLEY
+            return @source.byteslice(start, @ss.pos - start)
+          elsif byte_b != CLOSE_CURLEY
+            @ss.pos -= 1
+            return @source.byteslice(start, @ss.pos - start)
+          end
         elsif byte_a == OPEN_CURLEY && byte_b == PERCENTAGE
-          return next_tag_token(ss, start)
+          return next_tag_token_with_start(start)
         end
+
+        byte_a = byte_b
       end
 
-      return "{{"
+      "{{"
     end
 
-    def next_tag_token(ss, start = nil)
-      start ||= ss.pos - 2
+    def next_tag_token
+      start = @ss.pos - 2
+      if (len = @ss.skip_until(TAG_END))
+        @source.byteslice(start, len + 2)
+      else
+        "{%"
+      end
+    end
 
-      ss.scan_until(TAG_END)
-
-      ss.string.byteslice(start, ss.pos - start)
+    def next_tag_token_with_start(start)
+      @ss.skip_until(TAG_END)
+      @source.byteslice(start, @ss.pos - start)
     end
   end
+
+  # Remove this once we can depend on strscan >= 3.1.1
+  Tokenizer = StringScanner.instance_methods.include?(:scan_byte) ? Tokenizer2 : Tokenizer1
 end

performance/benchmark.rb

@@ -8,14 +8,17 @@ Liquid::Template.error_mode = ARGV.first.to_sym if ARGV.first
 profiler = ThemeRunner.new
 
 Benchmark.ips do |x|
-  x.time   = 10
-  x.warmup = 5
+  x.time   = 20
+  x.warmup = 10
 
   puts
   puts "Running benchmark for #{x.time} seconds (with #{x.warmup} seconds warmup)."
   puts
 
-  x.report("parse:") { profiler.compile }
-  x.report("render:") { profiler.render }
-  x.report("parse & render:") { profiler.run }
+  phase = ENV["PHASE"] || "all"
+
+  x.report("tokenize:") { profiler.tokenize } if phase == "all" || phase == "tokenize"
+  x.report("parse:") { profiler.compile } if phase == "all" || phase == "parse"
+  x.report("render:") { profiler.render } if phase == "all" || phase == "render"
+  x.report("parse & render:") { profiler.run } if phase == "all" || phase == "run"
 end

performance/theme_runner.rb

@@ -48,6 +48,14 @@ class ThemeRunner
     end
   end
 
+  # `tokenize` will just test the tokenizen portion of liquid without any templates
+  def tokenize
+    @tests.each do |test_hash|
+      tokenizer = Liquid::Tokenizer.new(test_hash[:liquid], true)
+      while tokenizer.shift; end
+    end
+  end
+
   # `run` is called to benchmark rendering and compiling at the same time
   def run
     each_test do |liquid, layout, assigns, page_template, template_name|

test/integration/parsing_quirks_test.rb

@@ -131,4 +131,24 @@ class ParsingQuirksTest < Minitest::Test
   def test_contains_in_id
     assert_template_result(' YES ', '{% if containsallshipments == true %} YES {% endif %}', { 'containsallshipments' => true })
   end
+
+  def test_incomplete_expression
+    with_error_mode(:lax) do
+      assert_template_result("false", "{{ false - }}")
+      assert_template_result("false", "{{ false > }}")
+      assert_template_result("false", "{{ false < }}")
+      assert_template_result("false", "{{ false = }}")
+      assert_template_result("false", "{{ false ! }}")
+      assert_template_result("false", "{{ false 1 }}")
+      assert_template_result("false", "{{ false a }}")
+
+      assert_template_result("false", "{% liquid assign foo = false -\n%}{{ foo }}")
+      assert_template_result("false", "{% liquid assign foo = false >\n%}{{ foo }}")
+      assert_template_result("false", "{% liquid assign foo = false <\n%}{{ foo }}")
+      assert_template_result("false", "{% liquid assign foo = false =\n%}{{ foo }}")
+      assert_template_result("false", "{% liquid assign foo = false !\n%}{{ foo }}")
+      assert_template_result("false", "{% liquid assign foo = false 1\n%}{{ foo }}")
+      assert_template_result("false", "{% liquid assign foo = false a\n%}{{ foo }}")
+    end
+  end
 end # ParsingQuirksTest

test/integration/tags/raw_tag_test.rb

@@ -16,6 +16,7 @@ class RawTagTest < Minitest::Test
     assert_template_result('>{{ test }}<', '> {%- raw -%}{{ test }}{%- endraw -%} <')
     assert_template_result("> inner  <", "> {%- raw -%} inner {%- endraw %} <")
     assert_template_result("> inner <", "> {%- raw -%} inner {%- endraw -%} <")
+    assert_template_result("{Hello}", "{% raw %}{{% endraw %}Hello{% raw %}}{% endraw %}")
   end
 
   def test_open_tag_in_raw

test/test_helper.rb

@@ -13,7 +13,7 @@ if (env_mode = ENV['LIQUID_PARSER_MODE'])
   puts "-- #{env_mode.upcase} ERROR MODE"
   mode = env_mode.to_sym
 end
-Liquid::Template.error_mode = mode
+Liquid::Environment.default.error_mode = mode
 
 if ENV['LIQUID_C'] == '1'
   puts "-- LIQUID C"

test/unit/block_unit_test.rb

@@ -32,6 +32,12 @@ class BlockUnitTest < Minitest::Test
     assert_equal(String, template.root.nodelist[2].class)
   end
 
+  def test_variable_with_multibyte_character
+    template = Liquid::Template.parse("{{ '❤️' }}")
+    assert_equal(1, template.root.nodelist.size)
+    assert_equal(Variable, template.root.nodelist[0].class)
+  end
+
   def test_variable_many_embedded_fragments
     template = Liquid::Template.parse("  {{funk}} {{so}} {{brother}} ")
     assert_equal(7, template.root.nodelist.size)

test/unit/lexer_unit_test.rb

@@ -6,58 +6,134 @@ class LexerUnitTest < Minitest::Test
   include Liquid
 
   def test_strings
-    tokens = Lexer.new(%( 'this is a test""' "wat 'lol'")).tokenize
-    assert_equal([[:string, %('this is a test""')], [:string, %("wat 'lol'")], [:end_of_string]], tokens)
+    assert_equal(
+      [[:string, %('this is a test""')], [:string, %("wat 'lol'")], [:end_of_string]],
+      tokenize(%( 'this is a test""' "wat 'lol'")),
+    )
   end
 
   def test_integer
-    tokens = Lexer.new('hi 50').tokenize
-    assert_equal([[:id, 'hi'], [:number, '50'], [:end_of_string]], tokens)
+    assert_equal(
+      [[:id, 'hi'], [:number, '50'], [:end_of_string]],
+      tokenize('hi 50'),
+    )
   end
 
   def test_float
-    tokens = Lexer.new('hi 5.0').tokenize
-    assert_equal([[:id, 'hi'], [:number, '5.0'], [:end_of_string]], tokens)
+    assert_equal(
+      [[:id, 'hi'], [:number, '5.0'], [:end_of_string]],
+      tokenize('hi 5.0'),
+    )
   end
 
   def test_comparison
-    tokens = Lexer.new('== <> contains ').tokenize
-    assert_equal([[:comparison, '=='], [:comparison, '<>'], [:comparison, 'contains'], [:end_of_string]], tokens)
+    assert_equal(
+      [[:comparison, '=='], [:comparison, '<>'], [:comparison, 'contains'], [:end_of_string]],
+      tokenize('== <> contains '),
+    )
+  end
+
+  def test_comparison_without_whitespace
+    assert_equal(
+      [[:number, '1'], [:comparison, '>'], [:number, '0'], [:end_of_string]],
+      tokenize('1>0'),
+    )
+  end
+
+  def test_comparison_with_negative_number
+    assert_equal(
+      [[:number, '1'], [:comparison, '>'], [:number, '-1'], [:end_of_string]],
+      tokenize('1>-1'),
+    )
+  end
+
+  def test_raise_for_invalid_comparison
+    assert_raises(SyntaxError) do
+      tokenize('1>!1')
+    end
+
+    assert_raises(SyntaxError) do
+      tokenize('1=<1')
+    end
+
+    assert_raises(SyntaxError) do
+      tokenize('1!!1')
+    end
   end
 
   def test_specials
-    tokens = Lexer.new('| .:').tokenize
-    assert_equal([[:pipe, '|'], [:dot, '.'], [:colon, ':'], [:end_of_string]], tokens)
-    tokens = Lexer.new('[,]').tokenize
-    assert_equal([[:open_square, '['], [:comma, ','], [:close_square, ']'], [:end_of_string]], tokens)
+    assert_equal(
+      [[:pipe, '|'], [:dot, '.'], [:colon, ':'], [:end_of_string]],
+      tokenize('| .:'),
+    )
+
+    assert_equal(
+      [[:open_square, '['], [:comma, ','], [:close_square, ']'], [:end_of_string]],
+      tokenize('[,]'),
+    )
   end
 
   def test_fancy_identifiers
-    tokens = Lexer.new('hi five?').tokenize
-    assert_equal([[:id, 'hi'], [:id, 'five?'], [:end_of_string]], tokens)
+    assert_equal([[:id, 'hi'], [:id, 'five?'], [:end_of_string]], tokenize('hi five?'))
 
-    tokens = Lexer.new('2foo').tokenize
-    assert_equal([[:number, '2'], [:id, 'foo'], [:end_of_string]], tokens)
+    assert_equal([[:number, '2'], [:id, 'foo'], [:end_of_string]], tokenize('2foo'))
   end
 
   def test_whitespace
-    tokens = Lexer.new("five|\n\t ==").tokenize
-    assert_equal([[:id, 'five'], [:pipe, '|'], [:comparison, '=='], [:end_of_string]], tokens)
+    assert_equal(
+      [[:id, 'five'], [:pipe, '|'], [:comparison, '=='], [:end_of_string]],
+      tokenize("five|\n\t =="),
+    )
   end
 
   def test_unexpected_character
     assert_raises(SyntaxError) do
-      Lexer.new("%").tokenize
+      tokenize("%")
     end
   end
 
   def test_negative_numbers
-    tokens = Lexer.new("foo | default: -1").tokenize
-    assert_equal([[:id, 'foo'], [:pipe, '|'], [:id, 'default'], [:colon, ":"], [:number, '-1'], [:end_of_string]], tokens)
+    assert_equal(
+      [[:id, 'foo'], [:pipe, '|'], [:id, 'default'], [:colon, ":"], [:number, '-1'], [:end_of_string]],
+      tokenize("foo | default: -1"),
+    )
   end
 
   def test_greater_than_two_digits
-    tokens = Lexer.new("foo > 12").tokenize
-    assert_equal([[:id, 'foo'], [:comparison, '>'], [:number, '12'], [:end_of_string]], tokens)
+    assert_equal(
+      [[:id, 'foo'], [:comparison, '>'], [:number, '12'], [:end_of_string]],
+      tokenize("foo > 12"),
+    )
+  end
+
+  def test_error_with_utf8_character
+    error = assert_raises(SyntaxError) do
+      tokenize("1 < 1Ø")
+    end
+
+    assert_equal(
+      'Liquid syntax error: Unexpected character Ø',
+      error.message,
+    )
+  end
+
+  def test_contains_as_attribute_name
+    assert_equal(
+      [[:id, "a"], [:dot, "."], [:id, "contains"], [:dot, "."], [:id, "b"], [:end_of_string]],
+      tokenize("a.contains.b"),
+    )
+  end
+
+  def test_tokenize_incomplete_expression
+    assert_equal([[:id, "false"], [:dash, "-"], [:end_of_string]], tokenize("false -"))
+    assert_equal([[:id, "false"], [:comparison, "<"], [:end_of_string]], tokenize("false <"))
+    assert_equal([[:id, "false"], [:comparison, ">"], [:end_of_string]], tokenize("false >"))
+    assert_equal([[:id, "false"], [:number, "1"], [:end_of_string]], tokenize("false 1"))
+  end
+
+  private
+
+  def tokenize(input)
+    Lexer.new(input).tokenize
   end
 end

test/unit/tokenizer_unit_test.rb

@@ -6,6 +6,7 @@ class TokenizerTest < Minitest::Test
   def test_tokenize_strings
     assert_equal([' '], tokenize(' '))
     assert_equal(['hello world'], tokenize('hello world'))
+    assert_equal(['{}'], tokenize('{}'))
   end
 
   def test_tokenize_variables
@@ -30,6 +31,19 @@ class TokenizerTest < Minitest::Test
     assert_equal([1, 1, 3], tokenize_line_numbers(" {{\n funk \n}} "))
   end
 
+  def test_incomplete_curly_braces
+    assert_equal(["{{.}", " "], tokenize('{{.} '))
+    assert_equal(["{{}", "%}"], tokenize('{{}%}'))
+    assert_equal(["{{}}", "}"], tokenize('{{}}}'))
+  end
+
+  def test_unmatching_start_and_end
+    assert_equal(["{{%}"], tokenize('{{%}'))
+    assert_equal(["{{%%%}}"], tokenize('{{%%%}}'))
+    assert_equal(["{%", "}}"], tokenize('{%}}'))
+    assert_equal(["{%%}", "}"], tokenize('{%%}}'))
+  end
+
   private
 
   def new_tokenizer(source, parse_context: Liquid::ParseContext.new, start_line_number: nil)