From 8a82cc3593f4bb7b195dc5d33e59954a147165f1 Mon Sep 17 00:00:00 2001
From: Yorick Peterse <yorickpeterse@gmail.com>
Date: Mon, 27 Jul 2015 01:00:14 +0200
Subject: [PATCH] XPath compiler support for the "=" operator

---
 lib/oga.rb                                   |   1 +
 lib/oga/xpath/compiler.rb                    | 166 +++++++++++++++--
 lib/oga/xpath/conversion.rb                  |  83 +++++++++
 spec/oga/xpath/compiler/operators/eq_spec.rb |  22 +++
 spec/oga/xpath/conversion_spec.rb            | 180 +++++++++++++++++++
 5 files changed, 442 insertions(+), 10 deletions(-)
 create mode 100644 lib/oga/xpath/conversion.rb
 create mode 100644 spec/oga/xpath/conversion_spec.rb

diff --git a/lib/oga.rb b/lib/oga.rb
index 7dbfa8f..a82b723 100644
--- a/lib/oga.rb
+++ b/lib/oga.rb
@@ -56,6 +56,7 @@ require 'oga/xpath/lexer'
 require 'oga/xpath/parser'
 require 'oga/xpath/evaluator'
 require 'oga/xpath/compiler'
+require 'oga/xpath/conversion'
 
 require 'oga/css/lexer'
 require 'oga/css/parser'
diff --git a/lib/oga/xpath/compiler.rb b/lib/oga/xpath/compiler.rb
index c20f310..577ea44 100644
--- a/lib/oga/xpath/compiler.rb
+++ b/lib/oga/xpath/compiler.rb
@@ -29,6 +29,15 @@ module Oga
         CACHE.get_or_set(ast) { new.compile(ast) }
       end
 
+      def initialize
+        reset
+      end
+
+      # Resets the internal state.
+      def reset
+        @literal_id = 0
+      end
+
       ##
       # Compiles an XPath AST into a Ruby Proc.
       #
@@ -61,6 +70,8 @@ module Oga
         source    = generator.process(proc_ast)
 
         eval(source)
+      ensure
+        reset
       end
 
       ##
@@ -246,17 +257,11 @@ module Oga
       # @return [Oga::Ruby::Node]
       #
       def on_expression_predicate(test, predicate, input)
-        catch_arg = symbol(:predicate_matched)
-
         process(test, input) do |matched_test_node|
-          catch_block = send_message('catch', catch_arg).add_block do
-            inner = process(predicate, matched_test_node) do
-              send_message('throw', catch_arg, literal('true'))
+          catch_block = catch_message(:predicate_matched) do
+            process(predicate, matched_test_node) do
+              throw_message(:predicate_matched, literal('true'))
             end
-
-            # Ensure that the "catch" only returns a value when "throw" is
-            # actually invoked.
-            inner.followed_by(literal('nil'))
           end
 
           catch_block.if_true { yield matched_test_node }
@@ -273,13 +278,99 @@ module Oga
         name_match ? condition.and(name_match) : condition
       end
 
+      ##
+      # Processes the `=` operator.
+      #
+      # The generated code is optimized so that expressions such as `a/b = c`
+      # only match the first node in both arms instead of matching all available
+      # nodes first. Because the `=` only ever operates on the first node in a
+      # set we can simply ditch the rest, possibly speeding things up quite a
+      # bit. This only works if one of the arms is:
+      #
+      # * a path
+      # * an axis
+      # * a predicate
+      #
+      # Everything else is processed the usual (and possibly slower) way.
+      #
+      # The variables used by this operator are assigned a "begin" block
+      # containing the actual result. This ensures that each variable is
+      # assigned the result of the entire block instead of the first expression
+      # that occurs.
+      #
+      # For example, take the following expression:
+      #
+      #     10 = 10 = 20
+      #
+      # Without a "begin" we'd end up with the following code (trimmed for
+      # readability):
+      #
+      #     eq_left3 = eq_left1 = ...
+      #
+      #     eq_left2 = ...
+      #
+      #     eq_left1, eq_left2 = to_compatible_types(eq_left1, eq_left2)
+      #
+      #     eq_left1 == eq_left2
+      #
+      #     eq_left4 = ...
+      #
+      #     eq_left3 == eq_left4
+      #
+      # This would be incorrect as the first boolean expression (`10 = 10`)
+      # would be ignored. By using a "begin" we instead get the following:
+      #
+      #     eq_left3 = begin
+      #       eq_left1 = ...
+      #
+      #       eq_left2 = ...
+      #
+      #       eq_left1, eq_left2 = to_compatible_types(eq_left1, eq_left2)
+      #
+      #       eq_left1 == eq_left2
+      #     end
+      #
+      #     eq_left4 = begin
+      #       ...
+      #     end
+      #
+      #     eq_left3 == eq_left4
+      #
       # @param [AST::Node] ast
       # @param [Oga::Ruby::Node] input
       # @return [Oga::Ruby::Node]
+      #
       def on_eq(ast, input)
         left, right = *ast
 
-        process(left, input).eq(process(right, input))
+        left_var  = unique_literal('eq_left')
+        right_var = unique_literal('eq_right')
+
+        text_sym   = symbol(:text)
+        conversion = literal('Conversion')
+
+        if return_nodeset?(left)
+          left_ast = match_first_node(left, input)
+        else
+          left_ast = process(left, input)
+        end
+
+        if return_nodeset?(right)
+          right_ast = match_first_node(right, input)
+        else
+          right_ast = process(right, input)
+        end
+
+        initial_assign = left_var.assign(left_ast.wrap)
+          .followed_by(right_var.assign(right_ast.wrap))
+
+        compatible_assign = mass_assign(
+          [left_var, right_var],
+          conversion.to_compatible_types(left_var, right_var)
+        )
+
+        initial_assign.followed_by(compatible_assign)
+          .followed_by(left_var.eq(right_var))
       end
 
       # @param [AST::Node] ast
@@ -322,6 +413,14 @@ module Oga
         Ruby::Node.new(:lit, [value.to_s])
       end
 
+      # @param [String] name
+      # @return [Oga::Ruby::Node]
+      def unique_literal(name)
+        new_id = @literal_id += 1
+
+        literal("#{name}#{new_id}")
+      end
+
       # @param [#to_s] value
       # @return [Oga::Ruby::Node]
       def string(value)
@@ -367,6 +466,21 @@ module Oga
         condition
       end
 
+      ##
+      # Returns an AST matching the first node of a node set.
+      #
+      # @param [Oga::Ruby::Node] ast
+      # @param [Oga::Ruby::Node] input
+      # @return [Oga::Ruby::Node]
+      #
+      def match_first_node(ast, input)
+        catch_message(:value) do
+          process(ast, input) do |node|
+            throw_message(:value, literal('Conversion').to_string(node))
+          end
+        end
+      end
+
       # @return [Oga::Ruby::Node]
       def matched_literal
         literal('matched')
@@ -388,12 +502,44 @@ module Oga
         literal(ast.children[0].to_i.to_s)
       end
 
+      ##
+      # @param [Array] vars The variables to assign.
+      # @param [Oga::Ruby::Node] value
+      # @return [Oga::Ruby::Node]
+      #
+      def mass_assign(vars, value)
+        Ruby::Node.new(:massign, [vars, value])
+      end
+
       # @param [AST::Node] ast
       # @return [TrueClass|FalseClass]
       def number?(ast)
         ast.type == :int || ast.type == :float
       end
 
+      # @param [AST::Node] ast
+      # @return [TrueClass|FalseClass]
+      def string?(ast)
+        ast.type == :string
+      end
+
+      # @param [Symbol] name
+      # @return [Oga::Ruby::Node]
+      def catch_message(name)
+        send_message('catch', symbol(name)).add_block do
+          # Ensure that the "catch" only returns a value when "throw" is
+          # actually invoked.
+          yield.followed_by(literal('nil'))
+        end
+      end
+
+      # @param [Symbol] name
+      # @param [Array] args
+      # @return [Oga::Ruby::Node]
+      def throw_message(name, *args)
+        send_message('throw', symbol(name), *args)
+      end
+
       # @param [AST::Node] ast
       # @return [TrueClass|FalseClass]
       def return_nodeset?(ast)
diff --git a/lib/oga/xpath/conversion.rb b/lib/oga/xpath/conversion.rb
new file mode 100644
index 0000000..b66168f
--- /dev/null
+++ b/lib/oga/xpath/conversion.rb
@@ -0,0 +1,83 @@
+module Oga
+  module XPath
+    ##
+    # Module for converting XPath objects such as NodeSets.
+    #
+    module Conversion
+      ##
+      # Converts both arguments to a type that can be compared using ==.
+      #
+      # @return [Array]
+      #
+      def self.to_compatible_types(left, right)
+        if left.is_a?(XML::NodeSet)
+          left = to_string(left)
+        end
+
+        if right.is_a?(XML::NodeSet)
+          right = to_string(right)
+        end
+
+        if left.is_a?(Numeric) and !right.is_a?(Numeric)
+          right = to_float(right)
+        end
+
+        if left.is_a?(String) and !right.is_a?(String)
+          right = to_string(right)
+        end
+
+        if boolean?(left) and !boolean?(right)
+          right = to_boolean(right)
+        end
+
+        [left, right]
+      end
+
+      # @return [String]
+      def self.to_string(value)
+        # If we have a number that has a zero decimal (e.g. 10.0) we want to
+        # get rid of that decimal. For this we'll first convert the number to
+        # an integer.
+        if value.is_a?(Float) and value.modulo(1).zero?
+          value = value.to_i
+        end
+
+        if value.is_a?(XML::NodeSet)
+          first = value.first
+          value = first.respond_to?(:text) ? first.text : ''
+        end
+
+        if value.respond_to?(:text)
+          value = value.text
+        end
+
+        value.to_s
+      end
+
+      # @return [Float]
+      def self.to_float(value)
+        Float(value) rescue Float::NAN
+      end
+
+      # @return [TrueClass|FalseClass]
+      def self.to_boolean(value)
+        bool = false
+
+        if value.is_a?(Float)
+          bool = !value.nan? && !value.zero?
+        elsif value.is_a?(Fixnum)
+          bool = !value.zero?
+        elsif value.respond_to?(:empty?)
+          bool = !value.empty?
+        end
+
+        bool
+      end
+
+      # @return [TrueClass|FalseClass]
+      def self.boolean?(value)
+        value.is_a?(TrueClass) || value.is_a?(FalseClass)
+      end
+    end # Conversion
+  end # XPath
+end # Oga
diff --git a/spec/oga/xpath/compiler/operators/eq_spec.rb b/spec/oga/xpath/compiler/operators/eq_spec.rb
index 63ce4d3..157036e 100644
--- a/spec/oga/xpath/compiler/operators/eq_spec.rb
+++ b/spec/oga/xpath/compiler/operators/eq_spec.rb
@@ -10,6 +10,14 @@ describe Oga::XPath::Compiler do
       evaluate_xpath(@document, '10 = 10').should == true
     end
 
+    it 'returns true if two numbers and 1 are equal' do
+      evaluate_xpath(@document, '10 = 10 = 1').should == true
+    end
+
+    it 'returns false if two numbers and 0 are not equal' do
+      evaluate_xpath(@document, '10 = 10 = 0').should == false
+    end
+
     it 'returns false if two numbers are not equal' do
       evaluate_xpath(@document, '10 = 15').should == false
     end
@@ -34,6 +42,14 @@ describe Oga::XPath::Compiler do
       evaluate_xpath(@document, 'root/a = root/b').should == true
     end
 
+    it 'returns true if two node sets and 1 are equal' do
+      evaluate_xpath(@document, 'root/a = root/b = 1').should == true
+    end
+
+    it 'returns false if two node sets and 0 are not equal' do
+      evaluate_xpath(@document, 'root/a = root/b = 0').should == false
+    end
+
     it 'returns false if two node sets are not equal' do
       evaluate_xpath(@document, 'root/a = root/c').should == false
     end
@@ -57,5 +73,11 @@ describe Oga::XPath::Compiler do
     it 'returns true if an attribute and string are equal' do
       evaluate_xpath(@document, 'root/b/@class = "foo"').should == true
     end
+
+    it 'returns true if an axis and a string are equal' do
+      element = @document.at_xpath('root/b')
+
+      evaluate_xpath(element, '@class = "foo"').should == true
+    end
   end
 end
diff --git a/spec/oga/xpath/conversion_spec.rb b/spec/oga/xpath/conversion_spec.rb
new file mode 100644
index 0000000..c8d79c8
--- /dev/null
+++ b/spec/oga/xpath/conversion_spec.rb
@@ -0,0 +1,180 @@
+require 'spec_helper'
+
+describe Oga::XPath::Conversion do
+  describe 'to_compatible_types' do
+    it 'returns two Strings when using two NodeSets' do
+      set1 = node_set(Oga::XML::Text.new(:text => 'foo'))
+      set2 = node_set(Oga::XML::Text.new(:text => 'bar'))
+
+      left, right = described_class.to_compatible_types(set1, set2)
+
+      left.should  == 'foo'
+      right.should == 'bar'
+    end
+
+    it 'returns two Strings when using a NodeSet and Float' do
+      set = node_set(Oga::XML::Text.new(:text => 'foo'))
+
+      left, right = described_class.to_compatible_types(set, 10.5)
+
+      left.should  == 'foo'
+      right.should == '10.5'
+    end
+
+    it 'returns two Floats when using a Float and NodeSet' do
+      set = node_set(Oga::XML::Text.new(:text => '20'))
+
+      left, right = described_class.to_compatible_types(10.5, set)
+
+      left.should  == 10.5
+      right.should == 20.0
+    end
+
+    it 'returns two Strings when using a String and a Float' do
+      left, right = described_class.to_compatible_types('foo', 10.5)
+
+      left.should  == 'foo'
+      right.should == '10.5'
+    end
+
+    it 'returns two booleans when using a boolean and a non-zero Fixnum' do
+      left, right = described_class.to_compatible_types(true, 10)
+
+      left.should  == true
+      right.should == true
+    end
+
+    it 'returns two booleans when using a boolean and 0' do
+      left, right = described_class.to_compatible_types(true, 0)
+
+      left.should  == true
+      right.should == false
+    end
+
+    it 'returns two booleans when using a boolean and a negative Fixnum' do
+      left, right = described_class.to_compatible_types(true, -5)
+
+      left.should  == true
+      right.should == true
+    end
+
+    it 'returns two booleans when using a boolean and a non-empty NodeSet' do
+      set = node_set(Oga::XML::Text.new(:text => '20'))
+
+      left, right = described_class.to_compatible_types(true, set)
+
+      left.should  == true
+      right.should == true
+    end
+
+    it 'returns two booleans when using a boolean and an empty NodeSet' do
+      set = node_set
+
+      left, right = described_class.to_compatible_types(true, set)
+
+      left.should  == true
+      right.should == false
+    end
+  end
+
+  describe 'to_string' do
+    describe 'using a Float' do
+      it 'converts 10.0 to a String' do
+        described_class.to_string(10.0).should == '10'
+      end
+
+      it 'converts 10.5 to a String' do
+        described_class.to_string(10.5).should == '10.5'
+      end
+    end
+
+    describe 'using a Node' do
+      it 'converts an Element to a String' do
+        node = Oga::XML::Element.new(:name => 'p')
+        node.inner_text = 'foo'
+
+        described_class.to_string(node).should == 'foo'
+      end
+
+      it 'converts a Text to a String' do
+        node = Oga::XML::Text.new(:text => 'foo')
+
+        described_class.to_string(node).should == 'foo'
+      end
+    end
+
+    describe 'using a NodeSet' do
+      it 'returns the text of the first node' do
+        node1 = Oga::XML::Text.new(:text => 'foo')
+        node2 = Oga::XML::Text.new(:text => 'bar')
+        set   = node_set(node1, node2)
+
+        described_class.to_string(set).should == 'foo'
+      end
+
+      it 'returns an empty String for an empty NodeSet' do
+        described_class.to_string(node_set).should == ''
+      end
+    end
+
+    describe 'using a Fixnum' do
+      it 'converts 10 to a String' do
+        described_class.to_string(10).should == '10'
+      end
+    end
+  end
+
+  describe 'to_float' do
+    it 'returns a Float for a valid value' do
+      described_class.to_float('10.5').should == 10.5
+    end
+
+    it 'returns Float::NAN for an invalid value' do
+      described_class.to_float('foo').nan?.should == true
+    end
+  end
+
+  describe 'to_boolean' do
+    it 'returns true for a non-empty String' do
+      described_class.to_boolean('foo').should == true
+    end
+
+    it 'returns false for an empty String' do
+      described_class.to_boolean('').should == false
+    end
+
+    it 'returns true for a positive Fixnum' do
+      described_class.to_boolean(10).should == true
+    end
+
+    it 'returns true for a positive Float' do
+      described_class.to_boolean(10.0).should == true
+    end
+
+    it 'returns true for a negative Fixnum' do
+      described_class.to_boolean(-10).should == true
+    end
+
+    it 'returns true for a negative Float' do
+      described_class.to_boolean(-10.0).should == true
+    end
+
+    it 'returns false for 0' do
+      described_class.to_boolean(0).should == false
+    end
+
+    it 'returns false for 0.0' do
+      described_class.to_boolean(0.0).should == false
+    end
+
+    it 'returns true for a non-empty NodeSet' do
+      set = node_set(Oga::XML::Node.new)
+
+      described_class.to_boolean(set).should == true
+    end
+
+    it 'returns false for an empty NodeSet' do
+      described_class.to_boolean(node_set).should == false
+    end
+  end
+end