Merge pull request #10 from udan-jayanith/newQueryings

V0.0.2

Author: udan-jayanith

FUTURE-CHANGELOG.md

@@ -8,6 +8,5 @@
 * GetElementsByClassName
 * GetElementsByTagName
 * GetElementsById
-* QueryList
 * QuerySelector
 * QuerySelectorAll

go.mod

@@ -1,6 +1,6 @@
 module github.com/udan-jayanith/GoHTML
 
-go 1.24.1
+go 1.25.0
 
 require github.com/emirpasic/gods v1.18.1
 

node-tree.go

@@ -68,13 +68,14 @@ func (node *Node) SetTagName(tagName string) {
 
 // GetAttribute returns the specified attribute value form the node. If the specified attribute doesn't exists GetAttribute returns a empty string and false.
 func (node *Node) GetAttribute(attributeName string) (string, bool) {
-	v, ok := node.attributes[attributeName]
+	v, ok := node.attributes[strings.TrimSpace(strings.ToLower(attributeName))]
 	return v, ok
 }
 
 // RemoveAttribute remove or delete the specified attribute.
 func (node *Node) RemoveAttribute(attributeName string) {
-	delete(node.attributes, attributeName)
+	delete(node.attributes, strings.TrimSpace(strings.ToLower(attributeName)))
+	
 }
 
 // IterateAttributes calls callback at every attribute in the node by passing attribute and value of the node.
@@ -87,7 +88,7 @@ func (node *Node) IterateAttributes(callback func(attribute, value string)) {
 
 // SetAttribute add a attribute to the node.
 func (node *Node) SetAttribute(attribute, value string) {
-	node.attributes[strings.TrimSpace(attribute)] = strings.TrimSpace(value)
+	node.attributes[strings.ToLower(strings.TrimSpace(attribute))] = strings.TrimSpace(value)
 }
 
 // GetText returns text on the node. This does not returns text on it's child nodes. If you also wants child nodes text use GetInnerText method on the node.

querying.go

@@ -39,12 +39,12 @@ func (node *Node) GetElementByClassName(className string) *Node {
 }
 
 // GetElementByID returns the first node that match with the given idName by advancing from the node.
-func (node *Node) GetElementByID(idName string) *Node{
+func (node *Node) GetElementByID(idName string) *Node {
 	traverser := NewTraverser(node)
 	var returnNode *Node
 	traverser.Walkthrough(func(node *Node) TraverseCondition {
-		id, _ := node.GetAttribute("id") 
-		if id == idName{
+		id, _ := node.GetAttribute("id")
+		if id == idName {
 			returnNode = node
 			return StopWalkthrough
 		}
@@ -54,15 +54,15 @@ func (node *Node) GetElementByID(idName string) *Node{
 }
 
 // GetElementsByClassName returns a NodeList containing nodes that have the given className from the node.
-func (node *Node) GetElementsByClassName(className string) NodeList{
+func (node *Node) GetElementsByClassName(className string) NodeList {
 	traverser := NewTraverser(node)
 	nodeList := NewNodeList()
 
 	traverser.Walkthrough(func(node *Node) TraverseCondition {
 		classList := NewClassList()
 		classList.DecodeFrom(node)
 
-		if classList.Contains(className){
+		if classList.Contains(className) {
 			nodeList.Append(node)
 		}
 		return ContinueWalkthrough
@@ -71,12 +71,12 @@ func (node *Node) GetElementsByClassName(className string) NodeList{
 }
 
 // GetElementsByTagName returns a NodeList containing nodes that have the given tagName from the node.
-func (node *Node) GetElementsByTagName(tagName string) NodeList{
+func (node *Node) GetElementsByTagName(tagName string) NodeList {
 	traverser := NewTraverser(node)
 	nodeList := NewNodeList()
 
 	traverser.Walkthrough(func(node *Node) TraverseCondition {
-		if node.GetTagName() == tagName{
+		if node.GetTagName() == tagName {
 			nodeList.Append(node)
 		}
 		return ContinueWalkthrough
@@ -85,16 +85,119 @@ func (node *Node) GetElementsByTagName(tagName string) NodeList{
 }
 
 // GetElementsByClassName returns a NodeList containing nodes that have the given idName from the node.
-func (node *Node) GetElementsById(idName string) NodeList{
+func (node *Node) GetElementsById(idName string) NodeList {
 	traverser := NewTraverser(node)
 	nodeList := NewNodeList()
 
 	traverser.Walkthrough(func(node *Node) TraverseCondition {
 		id, _ := node.GetAttribute("id")
-		if id == idName{
+		if id == idName {
 			nodeList.Append(node)
 		}
 		return ContinueWalkthrough
 	})
 	return nodeList
+}
+
+// Selector types
+const (
+	Id int = iota
+	Tag
+	Class
+)
+
+// QueryToken store data about basic css selectors(ids, classes, tags).
+type QueryToken struct {
+	Type         int
+	SelectorName string
+	Selector     string
+}
+
+// TokenizeQuery tokenizes the query and returns a list of QueryToken.
+func TokenizeQuery(query string) []QueryToken {
+	slice := make([]QueryToken, 0, 1)
+	if strings.TrimSpace(query) == "" {
+		return slice
+	}
+
+	iter := strings.SplitSeq(query, " ")
+	for sec := range iter {
+		token := QueryToken{}
+		switch sec {
+		case "", " ", ".", "#":
+			continue
+		}
+
+		switch string(sec[0]) {
+		case ".":
+			token.Type = Class
+			token.SelectorName = sec[1:]
+		case "#":
+			token.Type = Id
+			token.SelectorName = sec[1:]
+		default:
+			token.Type = Tag
+			token.SelectorName = sec
+		}
+		token.Selector = sec
+		slice = append(slice, token)
+	}
+
+	return slice
+}
+
+func matchQueryTokens(node *Node, queryTokens []QueryToken) bool {
+	if len(queryTokens) == 0 {
+		return false
+	}
+	classList := NewClassList()
+	classList.DecodeFrom(node)
+	for _, token := range queryTokens {
+		switch token.Type {
+		case Id:
+			idName, _ := node.GetAttribute("id")
+			if token.SelectorName != idName {
+				return false
+			}
+		case Tag:
+			if node.GetTagName() != token.SelectorName {
+				return false
+			}
+		case Class:
+			if !classList.Contains(token.SelectorName) {
+				return false
+			}
+		}
+	}
+	return true
+}
+
+// QuerySelector returns the first node that matches with the give node.
+func (node *Node) QuerySelector(query string) *Node {
+	queryTokens := TokenizeQuery(query)
+
+	traverser := NewTraverser(node)
+	var res *Node
+	traverser.Walkthrough(func(node *Node) TraverseCondition {
+		if matchQueryTokens(node, queryTokens) {
+			res = node
+			return StopWalkthrough
+		}
+		return ContinueWalkthrough
+	})
+	return res
+}
+
+// QuerySelectorAll returns a NodeList containing nodes that matched with the given query.
+func (node *Node) QuerySelectorAll(query string) NodeList{
+	nodeList := NewNodeList()
+	queryTokens := TokenizeQuery(query)
+	traverser := NewTraverser(node)
+
+	for node := range traverser.Walkthrough{
+		if matchQueryTokens(node, queryTokens) {
+			nodeList.Append(node)
+		}
+	}
+	return nodeList
 }

querying_test.go

@@ -52,7 +52,7 @@ func TestGetElementByClassName(t *testing.T) {
 	node = node.GetElementByClassName("ordered-item")
 	if node == nil {
 		t.Fatal("Node is nil")
-	}else if node.GetInnerText() != "Apple" {
+	} else if node.GetInnerText() != "Apple" {
 		t.Fatal("Expected Apple but got ", node.GetInnerText())
 	}
 }
@@ -66,7 +66,7 @@ func TestGetElementByTagName(t *testing.T) {
 	node = node.GetElementByTagName("h2")
 	if node == nil {
 		t.Fatal("Node is nil")
-	}else if node.GetInnerText() != "List 1"{
+	} else if node.GetInnerText() != "List 1" {
 		t.Fatal("Expected List 1 but got ", node.GetInnerText())
 	}
 }
@@ -137,3 +137,58 @@ func TestGetElementsById(t *testing.T) {
 		}
 	}
 }
+
+func TestSelectorTokenizer(t *testing.T) {
+	stack := linkedliststack.New()
+	stack.Push("article .content")
+	stack.Push("article p h1")
+	stack.Push("article p")
+	stack.Push(".title #user")
+	stack.Push("#user title .title-1")
+
+	for stack.Size() > 0 {
+		val, _ := stack.Pop()
+		selector := val.(string)
+
+		tokens := GoHtml.TokenizeQuery(selector)
+		s := ""
+		for _, token := range tokens {
+			if s == "" {
+				s += token.Selector
+			} else {
+				s += " " + token.Selector
+			}
+		}
+
+		if s != selector {
+			t.Fatal("Expected ", selector, "but got", s)
+		}
+	}
+}
+
+func TestQuerySelector(t *testing.T) {
+	node, err := testFile4NodeTree()
+	if err != nil {
+		t.Fatal(err)
+		return
+	}
+	imgEl := node.QuerySelector("img #idElement")
+	imgSrc, _ := imgEl.GetAttribute("src")
+	imgAlt, _ := imgEl.GetAttribute("alt")
+	if imgSrc != "" || imgAlt != "" {
+		t.Fatal("")
+	}
+}
+
+func TestQuerySelectorAll(t *testing.T) {
+	node, err := testFile4NodeTree()
+	if err != nil {
+		t.Fatal(err)
+		return
+	}
+
+	nodeList := node.QuerySelectorAll("h2")
+	if nodeList.Len() != 2{
+		t.Fatal("")
+	}
+}

traverser.go

@@ -42,15 +42,16 @@ func (t *Traverser) Previous() *Node {
 	return t.GetCurrentNode()
 }
 
-type TraverseCondition bool
+type TraverseCondition = bool
 
 const (
-	StopWalkthrough     TraverseCondition = true
-	ContinueWalkthrough TraverseCondition = false
+	StopWalkthrough     TraverseCondition = false
+	ContinueWalkthrough TraverseCondition = true
 )
 
-// Walkthrough traverse the node tree from the current node to the end of the node tree by visiting every node. If callback returned StopWalkthrough walkthrough function will stop else if it returned ContinueWalkthrough it advanced to the next node.
-// Walkthrough calls callback at every node and pass that node. Walkthrough traverse the node tree similar to DFS without visiting visited nodes iteratively.
+// Walkthrough traverse the node tree from the current node to the end of the node tree by visiting every node.
+// Walkthrough traverse the node tree similar to DFS without visiting visited nodes iteratively.
+// Walkthrough can be used as a range over iterator or a function that takes a callback and pass every node one by one.
 func (t *Traverser) Walkthrough(callback func(node *Node) TraverseCondition) {
 	stack := linkedliststack.New()
 	if t.GetCurrentNode() == nil {
@@ -60,7 +61,7 @@ func (t *Traverser) Walkthrough(callback func(node *Node) TraverseCondition) {
 
 	for stack.Size() > 0 {
 		currentNode, _ := stack.Pop()
-		if callback(currentNode.(*Node)) == StopWalkthrough {
+		if !callback(currentNode.(*Node)) {
 			return
 		}
 

traverser_test.go

@@ -19,7 +19,7 @@ func TestWalkthrough(t *testing.T) {
 	traverser := GoHtml.NewTraverser(body)
 
 	resList := make([]*GoHtml.Node, 0)
-	traverser.Walkthrough(func(node *GoHtml.Node) GoHtml.TraverseCondition {
+	traverser.Walkthrough(func(node *GoHtml.Node) bool {
 		resList = append(resList, node)
 		return GoHtml.ContinueWalkthrough
 	})