Add new selectors. Need manual testing. Closest is temparorly commented for devolopment of selectors and combinators. Once selectors and combinatore are done testing closest will work again.

Author: udan-jayanith

querying.go

@@ -1,7 +1,7 @@
 package GoHtml
 
 import (
-	//"iter"
+	"iter"
 	"strings"
 )
 
@@ -103,23 +103,14 @@ func (node *Node) GetElementsById(idName string) NodeList {
 /*
 QuerySearch tokenizes the query string and search for nodes that matches with the right most query token. After matching right most query it proceeds to match nodes parents nodes for left over tokens and then passed that node to (yield/range). QuerySearch search the whole node tree for matches unless yield get canceled or range iterator get cancel.
 */
-/*
-func QuerySearch(node *Node, query string) iter.Seq[*Node] {
+
+func QuerySearch(node *Node, selector string) iter.Seq[*Node] {
 	traverser := NewTraverser(node)
 	return func(yield func(node *Node) bool) {
-		queryTokens := TokenizeQuery(query)
+		selectorTokens := TokenizeSelectorsAndCombinators(selector)
 		iter := traverser.Walkthrough
 		for node := range iter {
-			i := matchFromRightMostQueryToken(node, queryTokens, len(queryTokens)-1)
-			if i == len(queryTokens)-1{
-				continue
-			}
-			parentNode := node.GetParent()
-			for parentNode != nil && i>=0 {
-				i = matchFromRightMostQueryToken(parentNode, queryTokens, i)
-				parentNode = parentNode.GetParent()
-			}
-			if i < 0 && !yield(node){
+			if matchFromRightMostSelectors(node, selectorTokens) && !yield(node) {
 				return
 			}
 		}
@@ -128,40 +119,17 @@ func QuerySearch(node *Node, query string) iter.Seq[*Node] {
 }
 
 // matchFromRightMostQueryToken tries to match query tokens from right to left and return the index at which point query token last matched.
-func matchFromRightMostQueryToken(node *Node, queryTokens []QueryToken, i int) int {
-	classList := NewClassList()
-	classList.DecodeFrom(node)
-	checked := make(map[string]struct{})
-outer:
-	for i >= 0 {
-		token := queryTokens[i]
-		_, ok := checked[token.Selector]
-		if ok {
+func matchFromRightMostSelectors(node *Node, selectorTokens []CombinatorEl) bool {
+	for i := len(selectorTokens) - 1; i >= 0; i-- {
+		if node == nil {
 			break
-		} else {
-			checked[token.Selector] = struct{}{}
-		}
-
-		switch token.Type {
-		case Id:
-			idName, _ := node.GetAttribute("id")
-			if token.SelectorName != idName {
-				break outer
-			}
-		case Class:
-			if !classList.Contains(token.SelectorName) {
-				break outer
-			}
-		case Tag:
-			if node.GetTagName() != token.SelectorName {
-				break outer
-			}
 		}
-		i--
+		node = selectorTokens[i].getMatchingNode(node)
 	}
-	return i
+	return node != nil
 }
 
+
 // QuerySelector only returns the first node that matches with the QuerySearch.
 func (node *Node) QuerySelector(query string) *Node {
 	iter := QuerySearch(node, query)
@@ -182,4 +150,3 @@ func (node *Node) QuerySelectorAll(query string) NodeList {
 	return nodeList
 }
 
-*/

querying_test.go

@@ -138,45 +138,17 @@ 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
 	}
-	node = node.QuerySelector("html .ordered-list ol li .ordered-item")
+	node = node.QuerySelector("html ol li")
 	if node == nil {
 		t.Fatal("Node is nill after QuerySelector")
 	} else if node.GetInnerText() != "Apple" {
+		t.Log(node)
 		t.Fatal("Unexpected text")
 	}
 }
@@ -211,4 +183,3 @@ func TestQuerySelectorAll(t *testing.T) {
 	}
 }
 
-*/

selectors.go

@@ -19,9 +19,9 @@ type Selector struct {
 }
 
 func matchNode(node *Node, basicSelectorName string, basicSelectorType BasicSelector) bool {
-	if basicSelectorName == ""{
+	if basicSelectorName == "" {
 		return true
-	}else if node == nil {
+	} else if node == nil {
 		return false
 	}
 
@@ -86,6 +86,7 @@ func TokenizeSelectorsAndCombinators(selector string) []CombinatorEl {
 	slice := strings.SplitSeq(selector, " ")
 	currentCombinator := *new(CombinatorEl)
 	currentCombinator.Selector1 = NewSelector("")
+	currentCombinator.Type = NoneCombinator
 	for str := range slice {
 		if strings.TrimSpace(str) == "" {
 			continue
@@ -115,61 +116,77 @@ func TokenizeSelectorsAndCombinators(selector string) []CombinatorEl {
 	return list
 }
 
-func (ce *CombinatorEl) IsMatchingNode(node *Node) bool {
+func (ce *CombinatorEl) getMatchingNode(node *Node) *Node {
 	switch ce.Type {
 	case Descendant:
-		return ce.isDescended(node)
+		return ce.getDescended(node)
 	case Child:
-		return ce.isDirectChild(node)
+		return ce.getDirectChild(node)
 	case NextSibling:
-		return ce.isNextSibling(node)
+		return ce.getNextSibling(node)
 	case SubsequentSibling:
-		return ce.isSubsequentSibling(node)
+		return ce.getSubsequentSibling(node)
 	case NoneCombinator:
-		return matchNode(node, ce.Selector2.selectorName, ce.Selector2.selectorType)
+		if matchNode(node, ce.Selector2.selectorName, ce.Selector2.selectorType) {
+			return node
+		}
 	}
-	return false
+	return nil
 }
 
 // isDescended returns wether the given node is a ce.Selector2 and descended of ce.Selector1.
-func (ce *CombinatorEl) isDescended(node *Node) bool {
+func (ce *CombinatorEl) getDescended(node *Node) *Node {
 	if !matchNode(node, ce.Selector2.selectorName, ce.Selector2.selectorType) {
-		return false
+		return nil
 	}
 
 	parentNode := node.GetParent()
-	for parentNode != nil && !matchNode(parentNode, ce.Selector1.selectorName, ce.Selector1.selectorType) {
+	for parentNode != nil {
+		if matchNode(parentNode, ce.Selector1.selectorName, ce.Selector1.selectorType) {
+			return parentNode
+		}
 		parentNode = parentNode.GetParent()
 	}
-	return parentNode != nil
+	return nil
 }
 
 // isDirectChild returns whether the given node is a direct child of ce.Selector1 and node is of ce.Selector2
-func (ce *CombinatorEl) isDirectChild(node *Node) bool {
+func (ce *CombinatorEl) getDirectChild(node *Node) *Node {
 	if node == nil {
-		return false
+		return nil
 	}
 
-	return matchNode(node, ce.Selector2.selectorName, ce.Selector2.selectorType) && matchNode(node.GetParent(), ce.Selector1.selectorName, ce.Selector1.selectorType)
+	if matchNode(node, ce.Selector2.selectorName, ce.Selector2.selectorType) &&
+		matchNode(node.GetParent(), ce.Selector1.selectorName, ce.Selector1.selectorType) {
+		return node.GetParent()
+	}
+	return nil
 }
 
 // isNextSibling return whether the given node is of ce.Selector2 and next sibling of ce.Selector1
-func (ce *CombinatorEl) isNextSibling(node *Node) bool {
+func (ce *CombinatorEl)   getNextSibling(node *Node) *Node {
 	if node == nil {
-		return false
+		return nil
 	}
 
-	return matchNode(node, ce.Selector2.selectorName, ce.Selector2.selectorType) && matchNode(node.GetPreviousNode(), ce.Selector1.selectorName, ce.Selector1.selectorType)
+	if matchNode(node, ce.Selector2.selectorName, ce.Selector2.selectorType) &&
+		matchNode(node.GetPreviousNode(), ce.Selector1.selectorName, ce.Selector1.selectorType) {
+		return node.GetPreviousNode()
+	}
+	return nil
 }
 
-func (ce *CombinatorEl) isSubsequentSibling(node *Node) bool {
-	if !matchNode(node, ce.Selector2.selector, ce.Selector2.selectorType) {
-		return false
+func (ce *CombinatorEl) getSubsequentSibling(node *Node) *Node {
+	if node == nil || !matchNode(node, ce.Selector2.selector, ce.Selector2.selectorType) {
+		return nil
 	}
 
 	traverser := NewTraverser(node)
-	for traverser.GetCurrentNode() != nil && !matchNode(traverser.GetCurrentNode(), ce.Selector1.selector, ce.Selector1.selectorType) {
+	for traverser.GetCurrentNode() != nil {
+		if matchNode(traverser.GetCurrentNode(), ce.Selector1.selector, ce.Selector1.selectorType){
+			return traverser.GetCurrentNode()
+		}
 		traverser.Previous()
 	}
-	return matchNode(traverser.GetCurrentNode(), ce.Selector1.selector, ce.Selector1.selectorType)
+	return nil
 }