feat: parse novel tags and add a single-request metadata fetch

Novel gains Tags, read from the anchors carrying schema.org itemprop="genre". Matching that microdata rather than the /the-loai/ href shape is what keeps the site-wide genre navigation out: a sampled page links 69 categories against the novel's own 6. NovelInfo fetches just the landing page and parses it, so a caller that only wants metadata pays one request instead of the two Novel needs for its chapter list cross-check.
2026-08-05 23:20:28 +00:00 · 2026-07-30 00:45:20 +07:00
parent f400a83678
commit d88f2a482d
3 changed files with 103 additions and 3 deletions
@@ -77,6 +77,21 @@ func (c *Crawler) Novel(ctx context.Context, novelURL string) (*Novel, error) {
 	return novel, nil
 }

+// NovelInfo fetches only the landing page and returns what it carries: title,
+// slug, tags, and the chapter list as that page shows it.
+//
+// Unlike Novel it does not also fetch a chapter page to cross-check the chapter
+// list, so it costs a single request. Callers that only want metadata should
+// prefer it; callers about to export every chapter want Novel, whose
+// reconciliation guards against a silently truncated list.
+func (c *Crawler) NovelInfo(ctx context.Context, novelURL string) (*Novel, error) {
+	page, err := c.page(ctx, novelURL)
+	if err != nil {
+		return nil, err
+	}
+	return ParseNovelPage(page, novelURL)
+}
+
 // Chapters fetches every chapter concurrently and returns them in reading
 // order. Any chapter that cannot be fetched or parsed fails the whole run
 // rather than yielding a book with a hole in it.
@@ -17,9 +17,13 @@ type ChapterRef struct {

 // Novel is a novel's landing page: its title and its chapters in reading order.
 type Novel struct {
-	Title    string
-	Slug     string
-	URL      string
+	Title string
+	Slug  string
+	URL   string
+
+	// Tags are the novel's own genres, as labelled on the site, in the order
+	// they appear. Empty when the page lists none.
+	Tags     []string
 	Chapters []ChapterRef
 }

@@ -43,6 +47,7 @@ func ParseNovelPage(page []byte, pageURL string) (*Novel, error) {
 		Title:    novelTitle(doc),
 		Slug:     slugFromNovelURL(base),
 		URL:      pageURL,
+		Tags:     tagsFromInfo(doc),
 		Chapters: chapterRefsFromList(doc, base),
 	}
 	if novel.Title == "" {
@@ -64,6 +69,35 @@ func novelTitle(doc *html.Node) string {
 	return nodeText(elementByTag(doc, "title"))
 }

+// tagsFromInfo reads the novel's own genres out of the info block.
+//
+// The anchors are matched on the schema.org itemprop="genre" microdata rather
+// than on their href. The page also carries a site-wide genre menu linking every
+// category — 69 distinct ones against this novel's 6 on a sampled page — so
+// matching the /the-loai/ URL shape would pull in the whole navigation.
+func tagsFromInfo(doc *html.Node) []string {
+	links := findAllNodes(doc, func(n *html.Node) bool {
+		return n.Type == html.ElementNode && n.Data == "a" && attr(n, "itemprop") == "genre"
+	})
+
+	var tags []string
+	seen := make(map[string]bool, len(links))
+	for _, link := range links {
+		// The label appears both as the link text and as its title attribute;
+		// prefer the text and fall back to the attribute.
+		name := nodeText(link)
+		if name == "" {
+			name = collapseSpaces(attr(link, "title"))
+		}
+		if name == "" || seen[name] {
+			continue
+		}
+		seen[name] = true
+		tags = append(tags, name)
+	}
+	return tags
+}
+
 // slugFromNovelURL turns https://host/tro-lai-nam-thang-cu.html into
 // "tro-lai-nam-thang-cu".
 func slugFromNovelURL(u *url.URL) string {
@@ -44,6 +44,57 @@ func TestParseNovelPageOrdersChaptersForReading(t *testing.T) {
 	}
 }

+// tagsFixture mirrors the real page: the novel's own genres carry
+// itemprop="genre", while a site-wide menu links every category without it. A
+// parser matching on the /the-loai/ href shape would swallow the whole menu.
+const tagsFixture = `<html><head><title>TEN TRUYEN</title></head><body>
+<h1>TEN TRUYEN</h1>
+<dl class="row">
+  <dt class="col-sm-3">Thể loại</dt>
+  <dd class="col-sm-9">
+    <a class='cate-item' itemprop='genre' title='Trọng Sinh' href='https://monkeydd.com/the-loai/trong-sinh.html'>Trọng Sinh</a>
+    <a class='cate-item' itemprop='genre' title='Cổ Đại' href='https://monkeydd.com/the-loai/co-dai.html'>Cổ  Đại</a>
+    <a class='cate-item' itemprop='genre' title='Gia Đình' href='https://monkeydd.com/the-loai/gia-dinh.html'></a>
+    <a class='cate-item' itemprop='genre' title='Cổ Đại' href='https://monkeydd.com/the-loai/co-dai.html'>Cổ Đại</a>
+  </dd>
+</dl>
+<nav class="site-menu">
+  <a href='https://monkeydd.com/the-loai/dam-my.html'>Đam Mỹ</a>
+  <a href='https://monkeydd.com/the-loai/bach-hop.html'>Bách Hợp</a>
+</nav>
+<div class="list-chapters">
+  <div class="item"><div class="episode-title"><a href="/n/1.html">1</a></div></div>
+</div></body></html>`
+
+func TestParseNovelPageReadsTags(t *testing.T) {
+	novel, err := ParseNovelPage([]byte(tagsFixture), "https://monkeydd.com/n.html")
+	if err != nil {
+		t.Fatalf("ParseNovelPage: %v", err)
+	}
+
+	// "Cổ  Đại" has its inner whitespace collapsed, the empty anchor falls back
+	// to its title attribute, and the repeated genre appears once.
+	want := []string{"Trọng Sinh", "Cổ Đại", "Gia Đình"}
+	if len(novel.Tags) != len(want) {
+		t.Fatalf("Tags = %q, want %q", novel.Tags, want)
+	}
+	for i := range want {
+		if novel.Tags[i] != want[i] {
+			t.Errorf("Tags[%d] = %q, want %q", i, novel.Tags[i], want[i])
+		}
+	}
+}
+
+func TestParseNovelPageWithoutTags(t *testing.T) {
+	novel, err := ParseNovelPage([]byte(novelFixture), "https://monkeydd.com/n.html")
+	if err != nil {
+		t.Fatalf("ParseNovelPage: %v", err)
+	}
+	if len(novel.Tags) != 0 {
+		t.Errorf("Tags = %q, want none", novel.Tags)
+	}
+}
+
 func TestParseNovelPageNoChapters(t *testing.T) {
 	if _, err := ParseNovelPage([]byte(`<html><title>x</title><body></body></html>`),
 		"https://monkeydd.com/n.html"); err == nil {