Support for parsing basic XPath expressions.

lib/oga/xpath/parser.y

@@ -3,30 +3,74 @@
 #
 class Oga::XPath::Parser
 
-token T_AXIS T_COLON T_COMMA T_FLOAT T_INT T_IDENT T_OP T_STAR
+token T_AXIS T_COLON T_COMMA T_FLOAT T_INT T_IDENT T_STAR
 token T_LBRACK T_RBRACK T_LPAREN T_RPAREN T_SLASH T_STRING
+token T_PIPE T_AND T_OR T_ADD T_DIV T_MOD T_EQ T_NEQ T_LT T_GT T_LTE T_GTE
 
 options no_result_var
 
+prechigh
+  left  T_EQ
+  right T_OR
+preclow
+
 rule
-  expressions
-    : expressions_ { s(:xpath, *val[0]) }
+  xpath
+    : expressions { s(:xpath, *val[0]) }
     | /* none */  { s(:xpath) }
     ;
 
-  expressions_
-    : expressions_ expression { val[0] << val[1] }
+  expressions
+    : expressions expression { val[0] << val[1] }
     | expression             { val }
     ;
 
   expression
-    : node_test
+    : path
+    | node_test
+    | operator
+    | axis
+    | string
+    | number
+    ;
+
+  path
+    : T_SLASH node_test { s(:path, val[1]) }
     ;
 
   node_test
-    : T_IDENT                  { s(:node, nil, val[0]) }
-    | T_IDENT T_COLON T_IDENT  { s(:node, val[0], val[2]) }
+    : node_name           { s(:node_test, val[0]) }
+    | node_name predicate { s(:node_test, val[0], *val[1]) }
     ;
+
+  node_name
+    # foo
+    : T_IDENT { s(:name, nil, val[0]) }
+
+    # foo:bar
+    | T_IDENT T_COLON T_IDENT { s(:name, val[0], val[1]) }
+    ;
+
+  predicate
+    : T_LBRACK expressions T_RBRACK { val[1] }
+    ;
+
+  operator
+    : expression T_EQ expression { s(:eq, val[0], val[2]) }
+    | expression T_OR expression { s(:or, val[0], val[2]) }
+    ;
+
+  axis
+    : T_AXIS T_IDENT { s(:axis, val[0], val[1]) }
+    ;
+
+  string
+    : T_STRING { s(:string, val[0]) }
+    ;
+
+  number
+    : T_INT   { s(:int, val[0]) }
+    | T_FLOAT { s(:float, val[0]) }
 end
 
 ---- inner

spec/oga/xpath/parser/axes_spec.rb

@@ -0,0 +1,24 @@
+require 'spec_helper'
+
+describe Oga::XPath::Parser do
+  context 'axes' do
+    example 'parse an absolute path' do
+      parse_xpath('/foo').should == s(
+        :xpath,
+        s(:path, s(:node_test, s(:name, nil, 'foo')))
+      )
+    end
+
+    example 'parse a relative path' do
+      parse_xpath('foo').should == s(:xpath, s(:node_test, s(:name, nil, 'foo')))
+    end
+
+    example 'parse an expression using two paths' do
+      parse_xpath('/foo/bar').should == s(
+        :xpath,
+        s(:path, s(:node_test, s(:name, nil, 'foo'))),
+        s(:path, s(:node_test, s(:name, nil, 'bar')))
+      )
+    end
+  end
+end

spec/oga/xpath/parser/predicates_spec.rb

@@ -0,0 +1,37 @@
+require 'spec_helper'
+
+describe Oga::XPath::Parser do
+  context 'predicates' do
+    example 'parse a single predicate' do
+      parse_xpath('/foo[@class="bar"]').should == s(
+        :xpath,
+        s(
+          :path,
+          s(
+            :node_test,
+            s(:name, nil, 'foo'),
+            s(:eq, s(:axis, 'attribute', 'class'), s(:string, 'bar'))
+          )
+        )
+      )
+    end
+
+    example 'parse a predicate using the or operator' do
+      parse_xpath('/foo[@class="bar" or @class="baz"]').should == s(
+        :xpath,
+        s(
+          :path,
+          s(
+            :node_test,
+            s(:name, nil, 'foo'),
+            s(
+              :or,
+              s(:eq, s(:axis, 'attribute', 'class'), s(:string, 'bar')),
+              s(:eq, s(:axis, 'attribute', 'class'), s(:string, 'baz'))
+            )
+          )
+        )
+      )
+    end
+  end
+end

spec/oga/xpath/parser_spec.rb

@@ -1,5 +0,0 @@
-require 'spec_helper'
-
-describe Oga::XPath::Parser do
-  pending 'Write me!'
-end

spec/support/parsing.rb

@@ -5,10 +5,11 @@ module Oga
     #
     # @param [Symbol] type
     # @param [Array] cihldren
-    # @return [Oga::AST::Node]
+    # @return [AST::Node]
     #
     def s(type, *children)
-      return Oga::AST::Node.new(type, children)
+      # TODO: add support for CSS AST nodes.
+      return Oga::XPath::Node.new(type, children)
     end
 
     ##
@@ -32,6 +33,16 @@ module Oga
       return Oga::XPath::Lexer.new(input).lex
     end
 
+    ##
+    # Parses an XPath expression.
+    #
+    # @param [String] input
+    # @return [Oga::XPath::Node]
+    #
+    def parse_xpath(input)
+      return Oga::XPath::Parser.new(input).parse
+    end
+
     ##
     # Parses the given XML and returns an AST.
     #