@@ -0,0 +1,247 @@
+package readability
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/PuerkitoBio/goquery"
+	xhtml "golang.org/x/net/html"
+)
+
+// TestInitialNodeScoreTagBuckets exercises every switch arm in
+// initialNodeScoreWithOptions, including the WeightClasses path.
+func TestInitialNodeScoreTagBuckets(t *testing.T) {
+	cases := []struct {
+		tag  string
+		want float64
+	}{
+		{"div", 5},
+		{"pre", 3},
+		{"td", 3},
+		{"blockquote", 3},
+		{"address", -3},
+		{"ol", -3},
+		{"ul", -3},
+		{"dl", -3},
+		{"dd", -3},
+		{"dt", -3},
+		{"li", -3},
+		{"form", -3},
+		{"h1", -5},
+		{"h2", -5},
+		{"h3", -5},
+		{"h4", -5},
+		{"h5", -5},
+		{"h6", -5},
+		{"th", -5},
+		{"span", 0}, // default arm
+	}
+	for _, c := range cases {
+		got := initialNodeScoreWithOptions(newElem(c.tag, ""), articleScoringOptions{})
+		if got != c.want {
+			t.Errorf("tag=%s: got %v, want %v", c.tag, got, c.want)
+		}
+	}
+}
+
+func TestInitialNodeScoreWeightClasses(t *testing.T) {
+	node := &xhtml.Node{Type: xhtml.ElementNode, Data: "div",
+		Attr: []xhtml.Attribute{{Key: "class", Val: "article"}}}
+	got := initialNodeScoreWithOptions(node, articleScoringOptions{WeightClasses: true})
+	// div (+5) + positive class match (+25) = 30
+	if got != 30 {
+		t.Fatalf("WeightClasses score = %v, want 30", got)
+	}
+}
+
+// TestBuildArticleContentBodyTopCandidate hits the body branch where all
+// child nodes are appended directly into the wrapper.
+func TestBuildArticleContentBodyTopCandidate(t *testing.T) {
+	body := newElem("body", "")
+	body.AppendChild(newElem("p", ""))
+	body.AppendChild(newElem("p", ""))
+
+	sel := buildArticleContent(body, map[*xhtml.Node]float64{}, 0)
+	if sel.Length() == 0 {
+		t.Fatal("expected non-empty selection")
+	}
+	if sel.Find("p").Length() != 2 {
+		t.Fatalf("expected 2 paragraphs, got %d", sel.Find("p").Length())
+	}
+}
+
+// TestBuildArticleContentSiblingBranches exercises the hr/p/long-p inclusion
+// rules and the "rewrite to div" fallback.
+func TestBuildArticleContentSiblingBranches(t *testing.T) {
+	body := newElem("body", "")
+	parent := newElem("div", "")
+	body.AppendChild(parent)
+
+	top := newElem("article", "")
+	hr := newElem("hr", "")
+
+	longP := newElem("p", "")
+	longText := strings.Repeat("paragraph text with content. ", 6) // > 80 chars, has period
+	longP.AppendChild(&xhtml.Node{Type: xhtml.TextNode, Data: longText})
+
+	shortP := newElem("p", "")
+	shortP.AppendChild(&xhtml.Node{Type: xhtml.TextNode, Data: "Short sentence."})
+
+	rejectedP := newElem("p", "")
+	// Empty p — no text, density 0, contains no period — should be rejected.
+
+	li := newElem("li", "")
+	// li scores high enough to pass the threshold but is not a valid article child.
+
+	for _, n := range []*xhtml.Node{top, hr, longP, shortP, rejectedP, li} {
+		parent.AppendChild(n)
+	}
+
+	scores := map[*xhtml.Node]float64{
+		top: 50,
+		li:  100, // forces threshold inclusion
+	}
+
+	sel := buildArticleContent(top, scores, 50)
+	if sel.Find("hr").Length() != 1 {
+		t.Errorf("expected hr included, got %d", sel.Find("hr").Length())
+	}
+	if sel.Find("p").Length() < 2 {
+		t.Errorf("expected long+short paragraphs included, got %d", sel.Find("p").Length())
+	}
+	// li should have been rewritten to div.
+	if sel.Find("div").Length() == 0 {
+		t.Errorf("expected li rewritten to div")
+	}
+	if sel.Find("li").Length() != 0 {
+		t.Errorf("expected no remaining li elements")
+	}
+}
+
+func TestBuildArticleContentClassMatchBonus(t *testing.T) {
+	body := newElem("body", "")
+	parent := newElem("div", "")
+	body.AppendChild(parent)
+
+	top := &xhtml.Node{Type: xhtml.ElementNode, Data: "article",
+		Attr: []xhtml.Attribute{{Key: "class", Val: "story"}}}
+	sib := &xhtml.Node{Type: xhtml.ElementNode, Data: "div",
+		Attr: []xhtml.Attribute{{Key: "class", Val: "story"}}}
+	parent.AppendChild(top)
+	parent.AppendChild(sib)
+
+	// scores[sib]=11; threshold = max(10, 50*0.2) = 10. With a class-match
+	// bonus of 50*0.2 = 10, the sibling sails over the threshold.
+	scores := map[*xhtml.Node]float64{
+		top: 50,
+		sib: 11,
+	}
+	sel := buildArticleContent(top, scores, 50)
+	// One class="story" article + one class="story" div should both appear.
+	if sel.Find("[class='story']").Length() != 2 {
+		t.Fatalf("expected class-match bonus to include sibling, got %d",
+			sel.Find("[class='story']").Length())
+	}
+}
+
+func TestHasDataLoadPlaylistSiblingShortCircuit(t *testing.T) {
+	parent := newElem("div", "")
+	target := newElem("svg", "")
+	playlist := &xhtml.Node{Type: xhtml.ElementNode, Data: "div",
+		Attr: []xhtml.Attribute{{Key: "data-load-playlist", Val: "1"}}}
+	parent.AppendChild(target)
+	parent.AppendChild(playlist)
+	if !hasDataLoadPlaylistSibling(target) {
+		t.Fatal("expected playlist sibling to be detected")
+	}
+
+	// p sibling between svg and playlist short-circuits the search.
+	parent2 := newElem("div", "")
+	target2 := newElem("svg", "")
+	parent2.AppendChild(target2)
+	parent2.AppendChild(newElem("p", ""))
+	parent2.AppendChild(&xhtml.Node{Type: xhtml.ElementNode, Data: "div",
+		Attr: []xhtml.Attribute{{Key: "data-load-playlist", Val: "1"}}})
+	if hasDataLoadPlaylistSibling(target2) {
+		t.Fatal("expected p sibling to short-circuit detection")
+	}
+}
+
+func TestDirectionForCandidateBodyAndAncestor(t *testing.T) {
+	html := &xhtml.Node{Type: xhtml.ElementNode, Data: "html",
+		Attr: []xhtml.Attribute{{Key: "dir", Val: "rtl"}}}
+	body := newElem("body", "")
+	html.AppendChild(body)
+	if got := directionForCandidate(body); got != "rtl" {
+		t.Errorf("body candidate dir = %q, want rtl", got)
+	}
+
+	// nil → empty string
+	if got := directionForCandidate(nil); got != "" {
+		t.Errorf("nil dir = %q", got)
+	}
+
+	// ancestor walk for non-body candidate
+	root := &xhtml.Node{Type: xhtml.ElementNode, Data: "div",
+		Attr: []xhtml.Attribute{{Key: "dir", Val: "ltr"}}}
+	mid := newElem("div", "")
+	leaf := newElem("article", "")
+	root.AppendChild(mid)
+	mid.AppendChild(leaf)
+	if got := directionForCandidate(leaf); got != "ltr" {
+		t.Errorf("ancestor dir = %q, want ltr", got)
+	}
+}
+
+// TestPrepareArticleScoringSkipAndDialog verifies prepareArticleScoring
+// removes skip-links, aria-modal dialogs, and unlikely-role elements while
+// keeping #comments / continuation markers intact.
+func TestPrepareArticleScoringSkipAndDialog(t *testing.T) {
+	html := `<html><body>
+<a class="skip-link" href="#main">Skip to content</a>
+<div role="dialog" aria-modal="true">popup</div>
+<div role="navigation">menu</div>
+<div id="comments">comments stay</div>
+<main id="content"><p>The article body has enough useful text to be picked as the top candidate during scoring.</p></main>
+</body></html>`
+
+	doc, err := goquery.NewDocumentFromReader(strings.NewReader(html))
+	if err != nil {
+		t.Fatal(err)
+	}
+	prepareArticleScoring(doc, "", articleScoringOptions{StripUnlikely: true})
+
+	if doc.Find(".skip-link").Length() != 0 {
+		t.Error("skip-link should have been removed")
+	}
+	if doc.Find("[aria-modal='true']").Length() != 0 {
+		t.Error("aria-modal dialog should have been removed")
+	}
+	if doc.Find("[role='navigation']").Length() != 0 {
+		t.Error("unlikely-role navigation should have been removed")
+	}
+	if doc.Find("#comments").Length() != 1 {
+		t.Error("comments wrapper should be retained")
+	}
+	if doc.Find("main#content").Length() != 0 {
+		t.Error("main#content should have been renamed to div")
+	}
+	if doc.Find("div#content").Length() != 1 {
+		t.Error("expected main rewritten to div with id=content")
+	}
+}
+
+func TestPrepareArticleScoringRemovesDuplicateHeader(t *testing.T) {
+	html := `<html><body>
+<h1>The Quick Brown Fox</h1>
+<p>The article body has enough useful text to be picked as the top candidate during scoring.</p>
+</body></html>`
+	doc, err := goquery.NewDocumentFromReader(strings.NewReader(html))
+	if err != nil {
+		t.Fatal(err)
+	}
+	prepareArticleScoring(doc, "The Quick Brown Fox", articleScoringOptions{StripUnlikely: true})
+	if doc.Find("h1").Length() != 0 {
+		t.Fatal("expected duplicate-title h1 to be removed")
+	}
+}