Fix #986: Preserve Wikipedia <h2> headings

Readability.js

@@ -815,10 +815,29 @@ Readability.prototype = {
 
     this._forEachNode(articleContent.children, function (topCandidate) {
       this._cleanMatchedNodes(topCandidate, function (node, matchString) {
-        return (
+        var shouldRemove =
           this.REGEXPS.shareElements.test(matchString) &&
-          node.textContent.length < shareElementThreshold
-        );
+          node.textContent.length < shareElementThreshold;
+
+        // Fix for issue #986: Don't remove divs that contain h2 headings with meaningful content,
+        // even if they match share elements criteria. This prevents Wikipedia h2 headings from
+        // being removed along with their parent divs.
+        if (shouldRemove && node.tagName === "DIV") {
+          var h2Elements = this._getAllNodesWithTag(node, ["h2"]);
+          for (var i = 0; i < h2Elements.length; i++) {
+            var h2Text = this._getInnerText(h2Elements[i], false).trim();
+            if (h2Text.length) {
+              this.log(
+                "Preserving div with h2 heading despite share elements criteria:",
+                node,
+                h2Text
+              );
+              return false;
+            }
+          }
+        }
+
+        return shouldRemove;
       });
     });
 
@@ -2513,6 +2532,23 @@ Readability.prototype = {
       var contentScore = 0;
 
       if (weight + contentScore < 0) {
+        // Fix for issue #986: Don't remove divs that contain h2 headings with meaningful content,
+        // even if the div has negative class weight (e.g., contains "share" in class name).
+        // This prevents Wikipedia h2 headings from being removed along with their parent divs.
+        if (tag === "DIV") {
+          var h2Elements = this._getAllNodesWithTag(node, ["h2"]);
+          for (var i = 0; i < h2Elements.length; i++) {
+            var h2Text = this._getInnerText(h2Elements[i], false).trim();
+            if (h2Text.length) {
+              this.log(
+                "Preserving div with h2 heading despite negative class weight:",
+                node,
+                h2Text
+              );
+              return false;
+            }
+          }
+        }
         return true;
       }
 
@@ -2540,18 +2576,18 @@ Readability.prototype = {
           "iframe",
         ]);
 
-        for (var i = 0; i < embeds.length; i++) {
+        for (var k = 0; k < embeds.length; k++) {
           // If this embed has attribute that matches video regex, don't delete it.
-          for (var j = 0; j < embeds[i].attributes.length; j++) {
-            if (this._allowedVideoRegex.test(embeds[i].attributes[j].value)) {
+          for (var j = 0; j < embeds[k].attributes.length; j++) {
+            if (this._allowedVideoRegex.test(embeds[k].attributes[j].value)) {
               return false;
             }
           }
 
           // For embed with <object> tag, check inner HTML as well.
           if (
-            embeds[i].tagName === "object" &&
-            this._allowedVideoRegex.test(embeds[i].innerHTML)
+            embeds[k].tagName === "object" &&
+            this._allowedVideoRegex.test(embeds[k].innerHTML)
           ) {
             return false;
           }
@@ -2636,6 +2672,24 @@ Readability.prototype = {
 
         var haveToRemove = shouldRemoveNode();
 
+        // Fix for issue #986: Don't remove divs that contain h2 headings with meaningful content,
+        // even if they fail other cleaning criteria. This prevents Wikipedia h2 headings from
+        // being removed along with their parent divs.
+        if (haveToRemove && tag === "DIV") {
+          var h2Elements2 = this._getAllNodesWithTag(node, ["h2"]);
+          for (var m = 0; m < h2Elements2.length; m++) {
+            var h2Text2 = this._getInnerText(h2Elements2[m], false).trim();
+            if (h2Text2.length) {
+              this.log(
+                "Preserving div with h2 heading despite cleaning criteria:",
+                node,
+                h2Text2
+              );
+              return false;
+            }
+          }
+        }
+
         // Allow simple lists of images to remain in pages
         if (isList && haveToRemove) {
           for (var x = 0; x < node.children.length; x++) {
@@ -2686,6 +2740,24 @@ Readability.prototype = {
     let headingNodes = this._getAllNodesWithTag(e, ["h1", "h2"]);
     this._removeNodes(headingNodes, function (node) {
       let shouldRemove = this._getClassWeight(node) < 0;
+
+      // Fix for issue #986: Preserve h2 headings that contain meaningful content,
+      // even if their parent div contains extra links like "edit" or "talk".
+      // This prevents Wikipedia h2 headings from being incorrectly removed.
+      if (shouldRemove && node.tagName === "H2") {
+        // Check if the h2 has meaningful text content (more than just whitespace)
+        let headingText = this._getInnerText(node, false).trim();
+        if (headingText.length) {
+          // Preserve h2 headings with actual content, regardless of parent container class weights
+          shouldRemove = false;
+          this.log(
+            "Preserving h2 heading with content despite negative class weight:",
+            node,
+            headingText
+          );
+        }
+      }
+
       if (shouldRemove) {
         this.log("Removing header with low class weight:", node);
       }

test/wikipedia-h2.js

@@ -0,0 +1,116 @@
+/* eslint-env node, mocha */
+
+/**
+ * Test case for issue #986: Wikipedia h2 headings being skipped when parent div contains extra links
+ *
+ * This test verifies that h2 headings are preserved even when their parent div contains
+ * links like "edit" or "talk" that might trigger negative class weight filtering.
+ */
+
+const { JSDOM } = require("jsdom");
+const chai = require("chai");
+const expect = chai.expect;
+const Readability = require("../index").Readability;
+
+// Sample HTML that mimics Wikipedia structure with h2 headings in divs containing edit/talk links
+const testHTML = `
+<!DOCTYPE html>
+<html>
+<head>
+    <title>Test Wikipedia Article</title>
+</head>
+<body>
+    <div id="content">
+        <h1>Main Article Title</h1>
+
+        <p>This is the introduction paragraph.</p>
+
+        <!-- This h2 should be preserved despite the parent div having edit/talk links -->
+        <div class="mw-headline" id="History">
+            <h2>History</h2>
+            <span class="mw-editsection">
+                <span class="mw-editsection-bracket">[</span>
+                <a href="/w/index.php?title=Test&action=edit&section=1" title="Edit section: History">edit</a>
+                <span class="mw-editsection-bracket">]</span>
+            </span>
+        </div>
+
+        <p>This is content under the History section.</p>
+
+        <!-- Another h2 that should be preserved -->
+        <div class="mw-headline" id="Development">
+            <h2>Development</h2>
+            <span class="mw-editsection">
+                <span class="mw-editsection-bracket">[</span>
+                <a href="/w/index.php?title=Test&action=edit&section=2" title="Edit section: Development">edit</a>
+                <span class="mw-editsection-bracket">]</span>
+            </span>
+        </div>
+
+        <p>This is content under the Development section.</p>
+
+        <!-- h2 with negative class that should still be preserved if it has content -->
+        <div class="edit-talk-share">
+            <h2>Important Section</h2>
+            <a href="/talk">talk</a>
+            <a href="/share">share</a>
+        </div>
+
+        <p>This is content under the Important Section.</p>
+
+        <!-- Empty h2 that should be removed -->
+        <div class="edit-talk-share">
+            <h2></h2>
+            <a href="/talk">talk</a>
+        </div>
+
+        <p>This paragraph should remain.</p>
+    </div>
+</body>
+</html>
+`;
+
+describe("Wikipedia H2 Heading Preservation", function () {
+  it("should preserve h2 headings even when parent div contains edit/talk links", function () {
+    const dom = new JSDOM(testHTML, {
+      url: "https://example.com",
+      contentType: "text/html",
+    });
+
+    const reader = new Readability(dom.window.document, {
+      debug: false,
+    });
+
+    const article = reader.parse();
+
+    expect(article).to.not.be.null;
+    expect(article.content).to.not.be.undefined;
+
+    // Check that h2 headings are preserved
+    const h2Count = (article.content.match(/<h2>/g) || []).length;
+
+    // We expect 4 h2 headings: "Main Article Title", "History", "Development", and "Important Section"
+    // The empty h2 should be removed
+    expect(h2Count).to.equal(4, `Expected 4 h2 headings, but found ${h2Count}`);
+
+    // Check for specific headings
+    expect(article.content).to.include(
+      "<h2>History</h2>",
+      "History h2 heading not found"
+    );
+    expect(article.content).to.include(
+      "<h2>Development</h2>",
+      "Development h2 heading not found"
+    );
+    expect(article.content).to.include(
+      "<h2>Important Section</h2>",
+      "Important Section h2 heading not found"
+    );
+
+    // Check that empty h2 was removed
+    expect(article.content).to.not.include(
+      "<h2></h2>",
+      "Empty h2 heading should have been removed"
+    );
+  });
+});