fix: Decode function panics when r io.Reader is empty.

udan-jayanith · udan-jayanith · commit 702a4374927c · 2025-12-03T18:58:03.000+05:30
diff --git a/parser.go b/parser.go
@@ -1,13 +1,20 @@
 package GoHtml
 
 import (
+	"errors"
 	"io"
 	"strings"
 
 	"golang.org/x/net/html"
 )
 
+var (
+	NoNodesFound error = errors.New("No nodes found in the node tree")
+)
+
 // Decode reads from rd and create a node-tree. Then returns the root node and nil.
+// If no nodes are found Decode returns a error.
+// r must no be nil.
 func Decode(r io.Reader) (*Node, error) {
 	t := NewTokenizer(r)
 	nodeTreeBuilder := NewNodeTreeBuilder()
@@ -19,7 +26,11 @@ func Decode(r io.Reader) (*Node, error) {
 
 		nodeTreeBuilder.WriteNodeTree(t.GetCurrentNode(), tt)
 	}
-	return nodeTreeBuilder.GetRootNode(), nil
+	rootNode := nodeTreeBuilder.GetRootNode()
+	if rootNode == nil {
+		return nil, NoNodesFound
+	}
+	return rootNode, nil
 }
 
 // HTMLToNodeTree return html code as a node-tree. If error were to occur it would be SyntaxError.
@@ -28,4 +39,3 @@ func HTMLToNodeTree(html string) (*Node, error) {
 	node, err := Decode(rd)
 	return node, err
 }
-
diff --git a/parser_test.go b/parser_test.go
@@ -27,6 +27,15 @@ func TestDecode(t *testing.T) {
 	GoHtml.Encode(&builder, node)
 }
 
+func TestDecodeWithAEmptyReader(t *testing.T) {
+	node, err := GoHtml.Decode(strings.NewReader(""))
+	if err == nil {
+		t.Fatal("Expected a error but got no error")
+	} else if node != nil {
+		t.Fatal("Expected node to be nil but got a Node")
+	}
+}
+
 func ExampleDecode() {
 	r := strings.NewReader(`
 	<!DOCTYPE html>
@@ -52,6 +61,6 @@ func ExampleDecode() {
 		title = titleNode.GetInnerText()
 	}
 	fmt.Println(title)
-	//Output: 
+	//Output:
 	//User Profile
 }
diff --git a/serializer.go b/serializer.go
@@ -23,9 +23,10 @@ func encodeListAttributes(node *Node) string {
 }
 
 // Encode writes to w encoding of the node tree from rootNode.
-func Encode(w io.Writer, rootNode *Node) {
+// w must not be nil
+func Encode(w io.Writer, rootNode *Node) error {
 	if rootNode == nil {
-		return
+		return NoNodesFound
 	}
 
 	type stackFrame struct {
@@ -43,27 +44,36 @@ func Encode(w io.Writer, rootNode *Node) {
 		currentStackFrame := v.(stackFrame)
 
 		if currentStackFrame.isClosingTag {
-			fmt.Fprintf(w, "</%s>", currentStackFrame.node.GetTagName())
+			_, err := fmt.Fprintf(w, "</%s>", currentStackFrame.node.GetTagName())
+			if err != nil {
+				return err
+			}
 			continue
 		} else if currentStackFrame.node.IsTextNode() {
-			fmt.Fprint(w, html.EscapeString(currentStackFrame.node.GetText()))
+			_, err := fmt.Fprint(w, html.EscapeString(currentStackFrame.node.GetText()))
+			if err != nil {
+				return err
+			}
 		} else {
-			fmt.Fprintf(w, "<%s%s>", func() string {
+			_, err := fmt.Fprintf(w, "<%s%s>", func() string {
 				tagName := currentStackFrame.node.GetTagName()
 				tagNameUpperCased := strings.ToUpper(tagName)
 				if tagNameUpperCased == DOCTYPEDTD {
 					tagName = tagNameUpperCased
 				}
 				return tagName
 			}(), encodeListAttributes(currentStackFrame.node))
+			if err != nil {
+				return err
+			}
 		}
 
 		if currentStackFrame.node.GetNextNode() != nil {
 			stack.Push(stackFrame{
 				node: currentStackFrame.node.GetNextNode(),
 			})
 		}
-		if !IsVoidTag(currentStackFrame.node.GetTagName()) && !currentStackFrame.node.IsTextNode(){
+		if !IsVoidTag(currentStackFrame.node.GetTagName()) && !currentStackFrame.node.IsTextNode() {
 			stack.Push(stackFrame{
 				node:         currentStackFrame.node,
 				isClosingTag: true,
@@ -75,6 +85,7 @@ func Encode(w io.Writer, rootNode *Node) {
 			})
 		}
 	}
+	return nil
 }
 
 // NodeTreeToHTML returns encoding of node-tree as a string.
diff --git a/tokenizer.go b/tokenizer.go
@@ -14,6 +14,7 @@ type Tokenizer struct {
 }
 
 // NewTokenizer returns a new Tokenizer.
+// r must not be nil
 func NewTokenizer(r io.Reader) Tokenizer {
 	return Tokenizer{
 		z: html.NewTokenizer(r),

Original file line number	Diff line number	Diff line change
`@@ -14,6 +14,7 @@ type Tokenizer struct {`
`14`	`14`	`}`
`15`	`15`
`16`	`16`	`// NewTokenizer returns a new Tokenizer.`
	`17`	`+// r must not be nil`
`17`	`18`	`func NewTokenizer(r io.Reader) Tokenizer {`
`18`	`19`	`return Tokenizer{`
`19`	`20`	`z: html.NewTokenizer(r),`