Merge pull request #225 from russross/issue-165

Implement support for CDATA section

block.go

@@ -320,6 +320,11 @@ func (p *parser) html(out *bytes.Buffer, data []byte, doRender bool) int {
 			return size
 		}
 
+		// check for HTML CDATA
+		if size := p.htmlCDATA(out, data, doRender); size > 0 {
+			return size
+		}
+
 		// no special case recognized
 		return 0
 	}
@@ -397,12 +402,10 @@ func (p *parser) html(out *bytes.Buffer, data []byte, doRender bool) int {
 	return i
 }
 
-// HTML comment, lax form
-func (p *parser) htmlComment(out *bytes.Buffer, data []byte, doRender bool) int {
-	i := p.inlineHtmlComment(out, data)
-	// needs to end with a blank line
-	if j := p.isEmpty(data[i:]); j > 0 {
-		size := i + j
+func (p *parser) renderHTMLBlock(out *bytes.Buffer, data []byte, start int, doRender bool) int {
+	// html block needs to end with a blank line
+	if i := p.isEmpty(data[start:]); i > 0 {
+		size := start + i
 		if doRender {
 			// trim trailing newlines
 			end := size
@@ -416,6 +419,35 @@ func (p *parser) htmlComment(out *bytes.Buffer, data []byte, doRender bool) int
 	return 0
 }
 
+// HTML comment, lax form
+func (p *parser) htmlComment(out *bytes.Buffer, data []byte, doRender bool) int {
+	i := p.inlineHTMLComment(out, data)
+	return p.renderHTMLBlock(out, data, i, doRender)
+}
+
+// HTML CDATA section
+func (p *parser) htmlCDATA(out *bytes.Buffer, data []byte, doRender bool) int {
+	const cdataTag = "<![cdata["
+	const cdataTagLen = len(cdataTag)
+	if len(data) < cdataTagLen+1 {
+		return 0
+	}
+	if !bytes.Equal(bytes.ToLower(data[:cdataTagLen]), []byte(cdataTag)) {
+		return 0
+	}
+	i := cdataTagLen
+	// scan for an end-of-comment marker, across lines if necessary
+	for i < len(data) && !(data[i-2] == ']' && data[i-1] == ']' && data[i] == '>') {
+		i++
+	}
+	i++
+	// no end-of-comment marker
+	if i >= len(data) {
+		return 0
+	}
+	return p.renderHTMLBlock(out, data, i, doRender)
+}
+
 // HR, which is the only self-closing block tag considered
 func (p *parser) htmlHr(out *bytes.Buffer, data []byte, doRender bool) int {
 	if data[0] != '<' || (data[1] != 'h' && data[1] != 'H') || (data[2] != 'r' && data[2] != 'R') {
@@ -432,19 +464,7 @@ func (p *parser) htmlHr(out *bytes.Buffer, data []byte, doRender bool) int {
 	}
 
 	if data[i] == '>' {
-		i++
-		if j := p.isEmpty(data[i:]); j > 0 {
-			size := i + j
-			if doRender {
-				// trim newlines
-				end := size
-				for end > 0 && data[end-1] == '\n' {
-					end--
-				}
-				p.r.BlockHtml(out, data[:end])
-			}
-			return size
-		}
+		return p.renderHTMLBlock(out, data, i+1, doRender)
 	}
 
 	return 0

block_test.go

@@ -1530,3 +1530,41 @@ func TestBlockComments(t *testing.T) {
 	}
 	doTestsBlock(t, tests, 0)
 }
+
+func TestCDATA(t *testing.T) {
+	var tests = []string{
+		"Some text\n\n<![CDATA[foo]]>\n",
+		"<p>Some text</p>\n\n<![CDATA[foo]]>\n",
+
+		"CDATA ]]\n\n<![CDATA[]]]]>\n",
+		"<p>CDATA ]]</p>\n\n<![CDATA[]]]]>\n",
+
+		"CDATA >\n\n<![CDATA[>]]>\n",
+		"<p>CDATA &gt;</p>\n\n<![CDATA[>]]>\n",
+
+		"Lots of text\n\n<![CDATA[lots of te><t\non\nseveral\nlines]]>\n",
+		"<p>Lots of text</p>\n\n<![CDATA[lots of te><t\non\nseveral\nlines]]>\n",
+
+		"<![CDATA[>]]>\n",
+		"<![CDATA[>]]>\n",
+	}
+	doTestsBlock(t, tests, 0)
+	doTestsBlock(t, []string{
+		"``` html\n<![CDATA[foo]]>\n```\n",
+		"<pre><code class=\"language-html\">&lt;![CDATA[foo]]&gt;\n</code></pre>\n",
+
+		"<![CDATA[\n``` python\ndef func():\n    pass\n```\n]]>\n",
+		"<![CDATA[\n``` python\ndef func():\n    pass\n```\n]]>\n",
+
+		`<![CDATA[
+> def func():
+>     pass
+]]>
+`,
+		`<![CDATA[
+> def func():
+>     pass
+]]>
+`,
+	}, EXTENSION_FENCED_CODE)
+}

inline.go

@@ -575,7 +575,7 @@ func link(p *parser, out *bytes.Buffer, data []byte, offset int) int {
 	return i
 }
 
-func (p *parser) inlineHtmlComment(out *bytes.Buffer, data []byte) int {
+func (p *parser) inlineHTMLComment(out *bytes.Buffer, data []byte) int {
 	if len(data) < 5 {
 		return 0
 	}
@@ -599,7 +599,7 @@ func leftAngle(p *parser, out *bytes.Buffer, data []byte, offset int) int {
 	data = data[offset:]
 	altype := LINK_TYPE_NOT_AUTOLINK
 	end := tagLength(data, &altype)
-	if size := p.inlineHtmlComment(out, data); size > 0 {
+	if size := p.inlineHTMLComment(out, data); size > 0 {
 		end = size
 	}
 	if end > 2 {