Changeset 95

Timestamp:

01/27/2007 12:03:07 (22 months ago)

Author:

why

Message:

• lib/hpricot/parse.rb: no raw strings for text, cdata, comment, procins. redundant waste of cycles.
  
• ext/hpricot_scan/hpricot_scan.rl: ditto. and a small bug in checking for text elements which was causing two doctypes (a real one and a text one). see #36.
  
• test/test_preserved.rb: check that Hpricot preserves all the HTML of our test pages when to_original_html is used.
  

Location:

trunk

Files:

• ext/hpricot_scan/hpricot_scan.rl (modified) (2 diffs)
• lib/hpricot/parse.rb (modified) (5 diffs)
• test/test_parser.rb (modified) (1 diff)
• test/test_preserved.rb (modified) (2 diffs)

trunk/ext/hpricot_scan/hpricot_scan.rl

@@ -18 +18 @@
 #define ELE(N) \
   if (tokend > tokstart || text == 1) { \
-    ele_open = 0; \
-    rb_yield_tokens(sym_##N, tag, attr, tokstart == 0 ? Qnil : rb_str_new(tokstart, tokend-tokstart), taint); \
+    VALUE raw_string = Qnil; \
+    ele_open = 0; text = 0; \
+    if (tokstart != 0 && sym_##N != sym_cdata && sym_##N != sym_text && sym_##N != sym_procins && sym_##N != sym_comment) { \
+      raw_string = rb_str_new(tokstart, tokend-tokstart); \
+    } \
+    rb_yield_tokens(sym_##N, tag, attr, raw_string, taint); \
   }
 
@@ -54 +58 @@
     }
 
-#define EBLK(N, T) CAT(tag, p - T + 1); ELE(N); text = 0;
+#define EBLK(N, T) CAT(tag, p - T + 1); ELE(N);
 
 %%{

trunk/lib/hpricot/parse.rb

@@ -206 +206 @@
   def Text.parse_pcdata(raw_string)
     result = Text.new(raw_string)
-    result.raw_string = raw_string
     result
   end
@@ -212 +211 @@
   def Text.parse_cdata_content(raw_string)
     result = CData.new(raw_string)
-    result.raw_string = raw_string
     result
   end
@@ -218 +216 @@
   def Text.parse_cdata_section(content)
     result = CData.new(content)
-    result.raw_string = content
     result
   end
@@ -256 +253 @@
     _, target, content = *raw_string.match(/\A<\?(\S+)\s+(.+)/m)
     result = ProcIns.new(target, content)
-    result.raw_string = "#{raw_string}?>"
     result
   end
@@ -262 +258 @@
   def Comment.parse(content)
     result = Comment.new(content)
-    result.raw_string = "<!--" + content.to_s + "-->"
     result
   end

trunk/test/test_parser.rb

@@ -48 +48 @@
 
   def scan_basic doc
+    assert_kind_of Hpricot::XMLDecl, doc.children.first 
+    assert_not_equal doc.children.first.to_s, doc.children[1].to_s 
     assert_equal 'link1', doc.at('#link1')['id']
     assert_equal 'link1', doc.at("p a")['id']

trunk/test/test_preserved.rb

@@ -3 +3 @@
 require 'test/unit'
 require 'hpricot'
+require 'load_files'
 
 class TestPreserved < Test::Unit::TestCase
+  def assert_roundtrip str
+    doc = Hpricot(str)
+    yield doc if block_given?
+    str2 = doc.to_original_html
+    [*str].zip([*str2]).each do |s1, s2|
+      assert_equal s1, s2
+    end
+  end
+
   def assert_html str1, str2
     doc = Hpricot(str2)
@@ -29 +39 @@
   end
 
+  def test_files
+    assert_roundtrip TestFiles::BASIC
+    assert_roundtrip TestFiles::BOINGBOING
+  end
 end