Support for lexing/parsing XPath type tests.

Unlike what I thought before syntax such as "node()" is not a function call.
Instead this is a special node test that tests the *types* of nodes, not their
names.

lib/oga/xpath/lexer.rl

@@ -249,9 +249,7 @@ module Oga
           # added on lexer level to make it easier to handle these cases on
           # parser/evaluator level.
           if AXIS_EMIT_NODE.include?(value)
-            add_token(:T_IDENT, 'node')
+            add_token(:T_NODE_TYPE, 'node')
-            add_token(:T_LPAREN)
-            add_token(:T_RPAREN)
 
             if AXIS_EMIT_EXTRA_SLASH.include?(value) and te != eof
               add_token(:T_SLASH)
@@ -305,10 +303,25 @@ module Oga
           | op_sub
           ;
 
+        # Node types
+        #
+        # While these look like functions they are actually node tests. For
+        # example, comment() matches all comment nodes.
+        #
+        # See http://www.w3.org/TR/xpath/#NT-NodeType for more information.
+
+        node_type = 'comment' | 'text' | 'processing-instruction' | 'node';
+
+        action emit_node_type {
+          emit(:T_NODE_TYPE, ts, te - 2)
+        }
+
         main := |*
           operator;
           whitespace | slash | lparen | rparen | comma | colon;
 
+          node_type '()' => emit_node_type;
+
           '[' => { add_token(:T_LBRACK) };
           ']' => { add_token(:T_RBRACK) };
 

lib/oga/xpath/parser.y

@@ -3,7 +3,7 @@
 #
 class Oga::XPath::Parser
 
-token T_AXIS T_COLON T_COMMA T_FLOAT T_INT T_IDENT
+token T_AXIS T_COLON T_COMMA T_FLOAT T_INT T_IDENT T_NODE_TYPE
 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
 token T_SUB T_MUL
@@ -76,6 +76,11 @@ rule
   node_test
     : node_name           { s(:test, *val[0]) }
     | node_name predicate { s(:test, *val[0], val[1]) }
+    | node_type           { val[0] }
+    ;
+
+  node_type
+    : T_NODE_TYPE { s(:node_type, val[0]) }
     ;
 
   node_name

spec/oga/xpath/lexer/axes_spec.rb

@@ -120,9 +120,7 @@ describe Oga::XPath::Lexer do
       lex_xpath('//A').should == [
         [:T_SLASH, nil],
         [:T_AXIS, 'descendant-or-self'],
-        [:T_IDENT, 'node'],
+        [:T_NODE_TYPE, 'node'],
-        [:T_LPAREN, nil],
-        [:T_RPAREN, nil],
         [:T_SLASH, nil],
         [:T_IDENT, 'A']
       ]
@@ -132,9 +130,7 @@ describe Oga::XPath::Lexer do
       lex_xpath('/..').should == [
         [:T_SLASH, nil],
         [:T_AXIS, 'parent'],
-        [:T_IDENT, 'node'],
+        [:T_NODE_TYPE, 'node']
-        [:T_LPAREN, nil],
-        [:T_RPAREN, nil],
       ]
     end
 
@@ -142,18 +138,14 @@ describe Oga::XPath::Lexer do
       lex_xpath('/.').should == [
         [:T_SLASH, nil],
         [:T_AXIS, 'self'],
-        [:T_IDENT, 'node'],
+        [:T_NODE_TYPE, 'node']
-        [:T_LPAREN, nil],
-        [:T_RPAREN, nil],
       ]
     end
 
     example 'lex the . axis followed by a path' do
       lex_xpath('./foo').should == [
         [:T_AXIS, 'self'],
-        [:T_IDENT, 'node'],
+        [:T_NODE_TYPE, 'node'],
-        [:T_LPAREN, nil],
-        [:T_RPAREN, nil],
         [:T_SLASH, nil],
         [:T_IDENT, 'foo']
       ]

spec/oga/xpath/lexer/general_spec.rb

@@ -37,9 +37,7 @@ describe Oga::XPath::Lexer do
         [:T_IDENT, 'wikimedia'],
         [:T_SLASH, nil],
         [:T_AXIS, 'descendant-or-self'],
-        [:T_IDENT, 'node'],
+        [:T_NODE_TYPE, 'node'],
-        [:T_LPAREN, nil],
-        [:T_RPAREN, nil],
         [:T_SLASH, nil],
         [:T_IDENT, 'editions']
       ]
@@ -66,9 +64,7 @@ describe Oga::XPath::Lexer do
         [:T_SLASH, nil],
         [:T_IDENT, 'edition'],
         [:T_SLASH, nil],
-        [:T_IDENT, 'text'],
+        [:T_NODE_TYPE, 'text']
-        [:T_LPAREN, nil],
-        [:T_RPAREN, nil]
       ]
     end
   end

spec/oga/xpath/lexer/node_type_spec.rb

@@ -0,0 +1,23 @@
+require 'spec_helper'
+
+describe Oga::XPath::Lexer do
+  context 'node types' do
+    example 'lex the "node" type' do
+      lex_xpath('node()').should == [[:T_NODE_TYPE, 'node']]
+    end
+
+    example 'lex the "comment" type' do
+      lex_xpath('comment()').should == [[:T_NODE_TYPE, 'comment']]
+    end
+
+    example 'lex the "text" type' do
+      lex_xpath('text()').should == [[:T_NODE_TYPE, 'text']]
+    end
+
+    example 'lex the "processing-instruction" type' do
+      lex_xpath('processing-instruction()').should == [
+        [:T_NODE_TYPE, 'processing-instruction']
+      ]
+    end
+  end
+end

spec/oga/xpath/parser/axes_spec.rb

@@ -105,7 +105,7 @@ describe Oga::XPath::Parser do
     example 'parse the // axis' do
       parse_xpath('//A').should == s(
         :absolute_path,
-        s(:axis, 'descendant-or-self', s(:call, 'node')),
+        s(:axis, 'descendant-or-self', s(:node_type, 'node')),
         s(:axis, 'child', s(:test, nil, 'A'))
       )
     end
@@ -113,14 +113,14 @@ describe Oga::XPath::Parser do
     example 'parse the .. axis' do
       parse_xpath('/..').should == s(
         :absolute_path,
-        s(:axis, 'parent', s(:call, 'node'))
+        s(:axis, 'parent', s(:node_type, 'node'))
       )
     end
 
     example 'parse the . axis' do
       parse_xpath('/.').should == s(
         :absolute_path,
-        s(:axis, 'self', s(:call, 'node'))
+        s(:axis, 'self', s(:node_type, 'node'))
       )
     end
   end

spec/oga/xpath/parser/node_type_spec.rb

@@ -0,0 +1,23 @@
+require 'spec_helper'
+
+describe Oga::XPath::Parser do
+  context 'node types' do
+    example 'parse the "node" type' do
+      parse_xpath('node()').should == s(:axis, 'child', s(:node_type, 'node'))
+    end
+
+    example 'parse the "comment" type' do
+      parse_xpath('comment()')
+        .should == s(:axis, 'child', s(:node_type, 'comment'))
+    end
+
+    example 'parse the "text" type' do
+      parse_xpath('text()').should == s(:axis, 'child', s(:node_type, 'text'))
+    end
+
+    example 'parse the "processing-instruction" type' do
+      parse_xpath('processing-instruction()')
+        .should == s(:axis, 'child', s(:node_type, 'processing-instruction'))
+    end
+  end
+end