From e11b9ed32c3bd0191a9b80439a15d370df63a639 Mon Sep 17 00:00:00 2001
From: Yorick Peterse <yorickpeterse@gmail.com>
Date: Sun, 1 Jun 2014 23:02:28 +0200
Subject: [PATCH] Basic XPath parser setup.

---
 .gitignore                    |  1 +
 Rakefile                      |  5 +--
 lib/oga.rb                    |  3 ++
 lib/oga/xpath/node.rb         | 10 +++++
 lib/oga/xpath/parser.y        | 81 +++++++++++++++++++++++++++++++++++
 spec/oga/xpath/parser_spec.rb |  5 +++
 task/parser.rake              |  2 +-
 7 files changed, 103 insertions(+), 4 deletions(-)
 create mode 100644 lib/oga/xpath/node.rb
 create mode 100644 lib/oga/xpath/parser.y
 create mode 100644 spec/oga/xpath/parser_spec.rb

diff --git a/.gitignore b/.gitignore
index 4078dde..ad4f1c8 100644
--- a/.gitignore
+++ b/.gitignore
@@ -5,6 +5,7 @@ Gemfile.lock
 
 lib/oga/xml/parser.rb
 lib/oga/xpath/lexer.rb
+lib/oga/xpath/parser.rb
 
 lib/liboga.*
 
diff --git a/Rakefile b/Rakefile
index 1313afc..23e1d0f 100644
--- a/Rakefile
+++ b/Rakefile
@@ -18,13 +18,12 @@ else
   end
 end
 
-PARSER_OUTPUT = 'lib/oga/xml/parser.rb'
-
 CLEAN.include(
   'coverage',
   'yardoc',
-  PARSER_OUTPUT,
+  'lib/oga/xml/parser.rb',
   'lib/oga/xpath/lexer.rb',
+  'lib/oga/xpath/parser.rb',
   'benchmark/fixtures/big.xml',
   'profile/samples/**/*.txt',
   'lib/liboga.*',
diff --git a/lib/oga.rb b/lib/oga.rb
index 44ae6e6..731005e 100644
--- a/lib/oga.rb
+++ b/lib/oga.rb
@@ -1,3 +1,4 @@
+require 'ast'
 require 'set'
 
 # Load these first so that the native extensions don't have to define the
@@ -23,4 +24,6 @@ require_relative 'oga/xml/doctype'
 
 require_relative 'oga/html/parser'
 
+require_relative 'oga/xpath/node'
 require_relative 'oga/xpath/lexer'
+require_relative 'oga/xpath/parser'
diff --git a/lib/oga/xpath/node.rb b/lib/oga/xpath/node.rb
new file mode 100644
index 0000000..da28e53
--- /dev/null
+++ b/lib/oga/xpath/node.rb
@@ -0,0 +1,10 @@
+module Oga
+  module XPath
+    ##
+    # AST node for XPath expressions.
+    #
+    class Node < AST::Node
+
+    end # Node
+  end # XPath
+end # Oga
diff --git a/lib/oga/xpath/parser.y b/lib/oga/xpath/parser.y
new file mode 100644
index 0000000..6631418
--- /dev/null
+++ b/lib/oga/xpath/parser.y
@@ -0,0 +1,81 @@
+##
+# Parser for XPath expressions.
+#
+class Oga::XPath::Parser
+
+token T_AXIS T_COLON T_COMMA T_FLOAT T_INT T_IDENT T_OP
+token T_LBRACK T_RBRACK T_LPAREN T_RPAREN T_SLASH T_STRING
+
+options no_result_var
+
+rule
+  expressions
+    : expressions_ { s(:xpath, *val[0]) }
+    | /* none */   { s(:xpath) }
+    ;
+
+  expressions_
+    : expressions_ expression { val[0] << val[1] }
+    | expression              { val }
+    ;
+
+  expression
+    : node_test
+    ;
+
+  node_test
+    : T_IDENT                  { s(:node, nil, val[0]) }
+    | T_IDENT T_COLON T_IDENT  { s(:node, val[0], val[2]) }
+    ;
+end
+
+---- inner
+  ##
+  # @param [String] data The input to parse.
+  #
+  # @param [Hash] options
+  #
+  def initialize(data)
+    @data  = data
+    @lexer = Lexer.new(data)
+  end
+
+  ##
+  # Creates a new XPath node.
+  #
+  # @param [Symbol] type
+  # @param [Array] children
+  # @return [Oga::XPath::Node]
+  #
+  def s(type, *children)
+    return Node.new(type, children)
+  end
+
+  ##
+  # Yields the next token from the lexer.
+  #
+  # @yieldparam [Array]
+  #
+  def yield_next_token
+    @lexer.advance do |*args|
+      yield args
+    end
+
+    yield [false, false]
+  end
+
+  ##
+  # Parses the input and returns the corresponding AST.
+  #
+  # @example
+  #  parser = Oga::XPath::Parser.new('//foo')
+  #  ast    = parser.parse
+  #
+  # @return [Oga::AST::Node]
+  #
+  def parse
+    ast = yyparse(self, :yield_next_token)
+
+    return ast
+  end
+# vim: set ft=racc:
diff --git a/spec/oga/xpath/parser_spec.rb b/spec/oga/xpath/parser_spec.rb
new file mode 100644
index 0000000..0d3567e
--- /dev/null
+++ b/spec/oga/xpath/parser_spec.rb
@@ -0,0 +1,5 @@
+require 'spec_helper'
+
+describe Oga::XPath::Parser do
+  pending 'Write me!'
+end
diff --git a/task/parser.rake b/task/parser.rake
index 10ebfca..562a025 100644
--- a/task/parser.rake
+++ b/task/parser.rake
@@ -3,4 +3,4 @@ rule '.rb' => '.y' do |task|
 end
 
 desc 'Generates the parser'
-task :parser => [PARSER_OUTPUT]
+task :parser => ['lib/oga/xml/parser.rb', 'lib/oga/xpath/parser.rb']