Changeset 106

Timestamp:

01/31/2007 19:47:13 (22 months ago)

Author:

why

Message:

• lib/hpricot/elements.rb: here we go. now support for a[text()="Click Me!"] and h3[text()*="space"] and the like.
  

Location:

trunk

Files:

• lib/hpricot/elements.rb (modified) (3 diffs)
• test/test_parser.rb (modified) (1 diff)

trunk/lib/hpricot/elements.rb

@@ -194 +194 @@
     end
 
-    ATTR_RE = %r!\[ *(@)([a-zA-Z0-9\(\)_-]+) *([~\!\|\*$\^=]*) *'?"?([^'"]*)'?"? *\]!i
+    ATTR_RE = %r!\[ *(?:(@)([\w\(\)-]+)|([\w\(\)-]+\(\))) *([~\!\|\*$\^=]*) *'?"?([^'"]*)'?"? *\]!i
     BRACK_RE = %r!(\[) *([^\]]*) *\]!i
     FUNC_RE = %r!(:)?([a-zA-Z0-9\*_-]*)\( *[\"']?([^ \)]*?)['\"]? *\)!
@@ -217 +217 @@
                 nodes, = Elements.filter(nodes, m[2], false)
             else
-                meth = "filter[#{m[0]}]"
+                meth = "filter[#{m[0]}#{m[1]}]"
                 if Traverse.method_defined? meth
-                    args = m[1..-1]
+                    args = m[2..-1]
                 else
-                    meth = "filter[#{m[0]}#{m[1]}]"
+                    meth = "filter[#{m[0]}]"
                     if Traverse.method_defined? meth
-                        args = m[2..-1]
+                        args = m[1..-1]
                     end
                 end
@@ -362 +362 @@
     end
 
-    filter '@=' do |attr,val,i|
-      self.elem? and get_attribute(attr).to_s == val
-    end
-
-    filter '@!=' do |attr,val,i|
-      self.elem? and get_attribute(attr).to_s != val
-    end
-
-    filter '@~=' do |attr,val,i|
-      self.elem? and get_attribute(attr).to_s.split(/\s+/).include? val
-    end
-
-    filter '@|=' do |attr,val,i|
-      self.elem? and get_attribute(attr).to_s =~ /^#{Regexp::quote val}(-|$)/
-    end
-
-    filter '@^=' do |attr,val,i|
-      self.elem? and get_attribute(attr).to_s.index(val) == 0
-    end
-
-    filter '@$=' do |attr,val,i|
-      self.elem? and get_attribute(attr).to_s =~ /#{Regexp::quote val}$/
-    end
-
-    filter '@*=' do |attr,val,i|
-      if self.elem?
-        idx = get_attribute(attr).to_s.index(val)
-        idx >= 0 if idx
-      end
+    pred_procs =
+      {'text()' => proc { |ele, *_| ele.inner_text.strip },
+       '@'      => proc { |ele, attr, *_| ele.get_attribute(attr).to_s if ele.elem? }}
+    
+    oper_procs =
+      {'='      => proc { |a,b| a == b },
+       '!='     => proc { |a,b| a != b },
+       '~='     => proc { |a,b| a.split(/\s+/).include?(b) },
+       '|='     => proc { |a,b| a =~ /^#{Regexp::quote b}(-|$)/ },
+       '^='     => proc { |a,b| a.index(b) == 0 },
+       '$='     => proc { |a,b| a =~ /#{Regexp::quote b}$/ },
+       '*='     => proc { |a,b| idx = a.index(b) }}
+
+    pred_procs.each do |pred_n, pred_f|
+      oper_procs.each do |oper_n, oper_f|
+        filter "#{pred_n}#{oper_n}" do |*a|
+          qual = pred_f[self, *a]
+          oper_f[qual, a[-2]] if qual
+        end
+      end
+    end
+
+    filter 'text()' do |val,i|
+      !self.inner_text.strip.empty?
     end
 

trunk/test/test_parser.rb

@@ -90 +90 @@
     assert_equal 1, @boingboing.search("//a[@name='027906']").length
     assert_equal 10, @boingboing.search("script comment()").length
+    assert_equal 3, @boingboing.search("a[text()*='Boing']").length
+    assert_equal 1, @boingboing.search("h3[text()='College kids reportedly taking more smart drugs']").length
+    assert_equal 0, @boingboing.search("h3[text()='College']").length
+    assert_equal 60, @boingboing.search("h3").length
+    assert_equal 59, @boingboing.search("h3[text()!='College kids reportedly taking more smart drugs']").length
+    assert_equal 17, @boingboing.search("h3[text()$='s']").length
+    assert_equal 128, @boingboing.search("p[text()]").length
+    assert_equal 211, @boingboing.search("p").length
   end