Fix paywall false positive on Anime News Network

ANN wraps article bodies in <div class="KonaBody"> with no semantic
<article>/<main> tags, so body-length extraction fell below the 500-char
threshold and the poller flagged stories as gated. Broaden the container
fallback to also try itemprop="articleBody", common content-container
classes, and ANN's KonaBody.

internal/ingestion/article.go

@@ -375,9 +375,11 @@ func extractBodyText(doc *goquery.Document) string {
 		}
 	}
 	// Sites like Phoronix use <br>-separated text inside <article> rather
-	// than <p> tags. Fall back to the container's raw text in that case.
+	// than <p> tags, and sites like Anime News Network skip semantic
+	// article/main tags entirely and put the body in a custom div. Fall
+	// back to the raw text of the longest known container.
 	if len(out) < PaywallBodyThreshold {
-		if alt := containerText(doc.Find("article, main").First()); len(alt) > len(out) {
+		if alt := bestContainerText(doc); len(alt) > len(out) {
 			out = alt
 		}
 	}
@@ -387,6 +389,34 @@ func extractBodyText(doc *goquery.Document) string {
 	return out
 }
 
+// bodyContainerSelectors is the list of selectors we try when <article>/<main>
+// aren't present or are too thin. Ordered loosely from semantic to generic;
+// we take the longest match across all of them.
+var bodyContainerSelectors = []string{
+	"article",
+	"main",
+	`[itemprop="articleBody"]`,
+	`[role="main"]`,
+	".article-body",
+	".article-content",
+	".entry-content",
+	".post-content",
+	".story-body",
+	".KonaBody", // Anime News Network
+	"#maincontent",
+	"#content",
+}
+
+func bestContainerText(doc *goquery.Document) string {
+	var best string
+	for _, sel := range bodyContainerSelectors {
+		if t := containerText(doc.Find(sel).First()); len(t) > len(best) {
+			best = t
+		}
+	}
+	return best
+}
+
 func joinParagraphText(sel *goquery.Selection) string {
 	var b strings.Builder
 	sel.Each(func(_ int, s *goquery.Selection) {
@@ -514,7 +544,7 @@ func extractBodyChars(doc *goquery.Document) int {
 		}
 	}
 	if best < PaywallBodyThreshold {
-		if n := len(containerText(doc.Find("article, main").First())); n > best {
+		if n := len(bestContainerText(doc)); n > best {
 			best = n
 		}
 	}