connect collapse whitespace func to tag type map

collapse/collapse.go

@@ -50,16 +50,8 @@ import (
 	"golang.org/x/net/html"
 )
 
-// Note: Originally in the javascript version, this just checked for "pre".
-// I changed it, to also return true for "code"
-func isPreOrCode(node *html.Node) bool {
-	name := dom.NodeName(node)
-
-	return name == "pre" || name == "code"
-}
-
-func next(prev *html.Node, current *html.Node) *html.Node {
-	if (prev != nil && prev.Parent == current) || isPreOrCode(current) {
+func nextNode(prev *html.Node, current *html.Node, domFuncs *DomFuncs) *html.Node {
+	if (prev != nil && prev.Parent == current) || domFuncs.IsPreformattedNode(current) {
 		if current.NextSibling != nil {
 			return current.NextSibling
 		}
@@ -77,37 +69,7 @@ func next(prev *html.Node, current *html.Node) *html.Node {
 	return current.Parent
 }
 
-var blockElements = []string{
-	"address", "article", "aside", "audio", "blockquote", "body", "canvas", "center", "dd", "dir", "div", "dl", "dt", "fieldset", "figcaption", "figure", "footer", "form", "frameset", "h1", "h2", "h3", "h4", "h5", "h6", "header", "hgroup", "hr", "html", "isindex", "li", "main", "menu", "nav", "noframes", "noscript", "ol", "output", "p", "pre", "section", "table", "tbody", "td", "tfoot", "th", "thead", "tr", "ul",
-}
-
-var voidElements = []string{
-	// Note: Compared to the javascript implementation, I removed "source"
-	"area", "base", "br", "col", "command", "embed", "hr", "img", "input", "keygen", "link", "meta", "param" /* "source, "*/, "track", "wbr",
-}
-
-func isBlock(node *html.Node) bool {
-	name := dom.NodeName(node)
-
-	for _, elem := range blockElements {
-		if elem == name {
-			return true
-		}
-	}
-	return false
-}
-func isVoid(node *html.Node) bool {
-	name := dom.NodeName(node)
-
-	for _, elem := range voidElements {
-		if elem == name {
-			return true
-		}
-	}
-	return false
-}
-
-func remove(node *html.Node) *html.Node {
+func removeNode(node *html.Node) *html.Node {
 	next := node.NextSibling
 	if next == nil {
 		next = node.Parent
@@ -119,16 +81,41 @@ func remove(node *html.Node) *html.Node {
 
 }
 
-func Collapse(element *html.Node) {
-	if element.FirstChild == nil || isPreOrCode(element) {
+type DomFuncs struct {
+	IsBlockNode        func(node *html.Node) bool
+	IsVoidNode         func(node *html.Node) bool
+	IsPreformattedNode func(node *html.Node) bool
+}
+
+func fillDefaultDomFuncs(domFuncs *DomFuncs) *DomFuncs {
+	if domFuncs == nil {
+		domFuncs = &DomFuncs{}
+	}
+	if domFuncs.IsBlockNode == nil {
+		domFuncs.IsBlockNode = defaultIsBlockNode
+	}
+	if domFuncs.IsVoidNode == nil {
+		domFuncs.IsVoidNode = defaultIsVoidNode
+	}
+	if domFuncs.IsPreformattedNode == nil {
+		domFuncs.IsPreformattedNode = defaultIsPreformattedNode
+	}
+	return domFuncs
+
+}
+func Collapse(element *html.Node, domFuncs *DomFuncs) {
+	domFuncs = fillDefaultDomFuncs(domFuncs)
+	// - - - - - - - - - - - - - - - - - - //
+
+	if element.FirstChild == nil || domFuncs.IsPreformattedNode(element) {
 		return
 	}
 
 	var prevText *html.Node = nil
 	var keepLeadingWs = false
 
 	var prev *html.Node = nil
-	var node = next(prev, element)
+	var node = nextNode(prev, element, domFuncs)
 
 	for node != element {
 		if node.Type == html.TextNode /* node.nodeType == 4 */ { // Node.TEXT_NODE or Node.CDATA_SECTION_NODE
@@ -141,22 +128,22 @@ func Collapse(element *html.Node) {
 
 			// `text` might be empty at this point.
 			if text == "" {
-				node = remove(node)
+				node = removeNode(node)
 				continue
 			}
 
 			node.Data = text
 
 			prevText = node
 		} else if node.Type == html.ElementNode { // Node.ELEMENT_NODE
-			if isBlock(node) || dom.NodeName(node) == "br" {
+			if domFuncs.IsBlockNode(node) || dom.NodeName(node) == "br" {
 				if prevText != nil {
 					prevText.Data = strings.TrimSuffix(prevText.Data, " ")
 				}
 
 				prevText = nil
 				keepLeadingWs = false
-			} else if isVoid(node) || isPreOrCode(node) || node.Data == "code" {
+			} else if domFuncs.IsVoidNode(node) || domFuncs.IsPreformattedNode(node) || dom.NodeName(node) == "code" {
 				// Avoid trimming space around non-block, non-BR void elements and inline PRE.
 				prevText = nil
 				keepLeadingWs = true
@@ -169,19 +156,19 @@ func Collapse(element *html.Node) {
 		} else {
 			// E.g. DoctypeNode
 
-			node = remove(node)
+			node = removeNode(node)
 			continue
 		}
 
-		var nextNode = next(prev, node)
+		var nextNode = nextNode(prev, node, domFuncs)
 		prev = node
 		node = nextNode
 	}
 
 	if prevText != nil {
 		prevText.Data = strings.TrimSuffix(prevText.Data, " ")
 		if prevText.Data == "" {
-			remove(prevText)
+			removeNode(prevText)
 		}
 	}
 }

collapse/collapse_test.go

@@ -35,7 +35,7 @@ func TestCollapse_DocType(t *testing.T) {
 		t.Error(err)
 	}
 
-	Collapse(doc)
+	Collapse(doc, nil)
 
 	var buf bytes.Buffer
 	err = html.Render(&buf, doc)
@@ -55,7 +55,7 @@ func TestCollapse_NoFirstChild(t *testing.T) {
 		Data: "strong",
 	}
 
-	Collapse(boldNode)
+	Collapse(boldNode, nil)
 
 	var buf bytes.Buffer
 	err := html.Render(&buf, boldNode)
@@ -80,7 +80,7 @@ func TestCollapse_StartWithCode(t *testing.T) {
 	}
 	codeNode.AppendChild(textNode)
 
-	Collapse(codeNode)
+	Collapse(codeNode, nil)
 
 	var buf bytes.Buffer
 	err := html.Render(&buf, codeNode)
@@ -111,7 +111,7 @@ func TestCollapse_TwoTextNodes(t *testing.T) {
 	node1.AppendChild(node2)
 	node1.AppendChild(node3)
 
-	Collapse(node1)
+	Collapse(node1, nil)
 
 	var buf bytes.Buffer
 	err := html.Render(&buf, node1)
@@ -142,7 +142,7 @@ func TestCollapse_LastTextIsEmpty(t *testing.T) {
 	node1.AppendChild(node2)
 	node1.AppendChild(node3)
 
-	Collapse(node1)
+	Collapse(node1, nil)
 
 	var buf bytes.Buffer
 	err := html.Render(&buf, node1)
@@ -273,7 +273,7 @@ func TestCollapse_Table(t *testing.T) {
 				t.Error(err)
 			}
 
-			Collapse(doc)
+			Collapse(doc, nil)
 
 			var buf bytes.Buffer
 			err = html.Render(&buf, getBody(doc))

collapse/is_node.go

@@ -0,0 +1,43 @@
+package collapse
+
+import (
+	"github.com/JohannesKaufmann/dom"
+	"golang.org/x/net/html"
+)
+
+var blockElements = []string{
+	"address", "article", "aside", "audio", "blockquote", "body", "canvas", "center", "dd", "dir", "div", "dl", "dt", "fieldset", "figcaption", "figure", "footer", "form", "frameset", "h1", "h2", "h3", "h4", "h5", "h6", "header", "hgroup", "hr", "html", "isindex", "li", "main", "menu", "nav", "noframes", "noscript", "ol", "output", "p", "pre", "section", "table", "tbody", "td", "tfoot", "th", "thead", "tr", "ul",
+}
+
+var voidElements = []string{
+	// Note: Compared to the javascript implementation, I removed "source"
+	"area", "base", "br", "col", "command", "embed", "hr", "img", "input", "keygen", "link", "meta", "param" /* "source, "*/, "track", "wbr",
+}
+
+var defaultIsBlockNode = func(node *html.Node) bool {
+	name := dom.NodeName(node)
+
+	for _, elem := range blockElements {
+		if elem == name {
+			return true
+		}
+	}
+	return false
+}
+var defaultIsVoidNode = func(node *html.Node) bool {
+	name := dom.NodeName(node)
+
+	for _, elem := range voidElements {
+		if elem == name {
+			return true
+		}
+	}
+	return false
+}
+var defaultIsPreformattedNode = func(node *html.Node) bool {
+	// Note: Originally in the javascript version, this just checked for "pre".
+	// I changed it, to also return true for "code"
+	name := dom.NodeName(node)
+
+	return name == "pre" || name == "code"
+}

converter/register_test.go

@@ -26,7 +26,7 @@ func TestTagType_Priority(t *testing.T) {
 			if err != nil {
 				t.Fatal(err)
 			}
-			expected := "This \n\n<strong>bold</strong>\n\n and *italic* text"
+			expected := "This\n\n<strong>bold</strong>\n\nand *italic* text"
 			if output != expected {
 				t.Errorf("expected %q but got %q", expected, output)
 			}

plugin/base/base.go

@@ -83,7 +83,17 @@ func (b *base) preRenderRemove(ctx converter.Context, doc *html.Node) {
 }
 
 func (b *base) preRenderCollapse(ctx converter.Context, doc *html.Node) {
-	collapse.Collapse(doc)
+	collapse.Collapse(doc, &collapse.DomFuncs{
+		IsBlockNode: func(node *html.Node) bool {
+			tagName := dom.NodeName(node)
+			tagType, ok := ctx.GetTagType(tagName)
+			if ok {
+				return tagType == converter.TagTypeBlock
+			}
+
+			return dom.NameIsBlockNode(tagName)
+		},
+	})
 }
 
 var characterEntityReplacer = strings.NewReplacer(

plugin/base/renderers_test.go

@@ -50,8 +50,7 @@ func TestRenderAsX(t *testing.T) {
 			desc:       "RenderAsHTML Inline",
 			isInline:   true,
 			renderFunc: base.RenderAsHTML,
-			// TODO: Connect the collapse function to the tagType map. Then we can also use "TagTypeInline" for a node that is actually a block node
-			expected: "# heading\n\n<footer><strong>bold text</strong></footer>",
+			expected:   "# heading\n\n<footer><strong>bold text</strong></footer>",
 		},
 
 		{