Parse in quirksmode if no doctype html

Fixes #2197
jhy · Sep 10, 2024 · 8601e85 · 8601e85
1 parent d3104a0
commit 8601e85
Show file tree

Hide file tree

Showing 3 changed files with 25 additions and 2 deletions.
diff --git a/CHANGES.md b/CHANGES.md
@@ -22,6 +22,8 @@
   character. [2169](https://github.com/jhy/jsoup/issues/2169)
 * When tracking source ranges, a text node following an invalid self-closing element may be left
   untracked.[2175](https://github.com/jhy/jsoup/issues/2175)
+* When a document has no doctype, or a doctype not named `html`, it should be parsed in Quirks
+  Mode. [2197](https://github.com/jhy/jsoup/issues/2197)
 
 ## 1.18.1 (2024-Jul-10)
 

diff --git a/src/main/java/org/jsoup/parser/HtmlTreeBuilderState.java b/src/main/java/org/jsoup/parser/HtmlTreeBuilderState.java
@@ -25,18 +25,19 @@ enum HtmlTreeBuilderState {
                 tb.insertCommentNode(t.asComment());
             } else if (t.isDoctype()) {
                 // todo: parse error check on expected doctypes
-                // todo: quirk state check on doctype ids
                 Token.Doctype d = t.asDoctype();
                 DocumentType doctype = new DocumentType(
                     tb.settings.normalizeTag(d.getName()), d.getPublicIdentifier(), d.getSystemIdentifier());
                 doctype.setPubSysKey(d.getPubSysKey());
                 tb.getDocument().appendChild(doctype);
                 tb.onNodeInserted(doctype);
-                if (d.isForceQuirks())
+                // todo: quirk state check on more doctype ids, if deemed useful (most are ancient legacy and presumably irrelevant)
+                if (d.isForceQuirks() || !doctype.name().equals("html") || doctype.publicId().equalsIgnoreCase("HTML"))
                     tb.getDocument().quirksMode(Document.QuirksMode.quirks);
                 tb.transition(BeforeHtml);
             } else {
                 // todo: check not iframe srcdoc
+                tb.getDocument().quirksMode(Document.QuirksMode.quirks); // missing doctype
                 tb.transition(BeforeHtml);
                 return tb.process(t); // re-process token
             }

diff --git a/src/test/java/org/jsoup/parser/HtmlParserTest.java b/src/test/java/org/jsoup/parser/HtmlParserTest.java
@@ -1888,4 +1888,24 @@ private static void assertMathNamespace(Element el) {
         img.ownerDocument().outputSettings().charset("ascii");
         assertEquals("<img multi=\"&#x1f4af;\" single=\"&#x1f4af;\" hexsingle=\"&#x1f4af;\">", img.outerHtml());
     }
+
+    @Test void tableInPInQuirksMode() {
+        // https://github.com/jhy/jsoup/issues/2197
+        String html = "<p><span><table><tbody><tr><td><span>Hello table data</span></td></tr></tbody></table></span></p>";
+        Document doc = Jsoup.parse(html);
+        assertEquals(Document.QuirksMode.quirks, doc.quirksMode());
+        assertEquals(
+            "<p><span><table><tbody><tr><td><span>Hello table data</span></td></tr></tbody></table></span></p>", // quirks, allows table in p
+            TextUtil.normalizeSpaces(doc.body().html())
+        );
+
+        // doctype set, no quirks
+        html ="<!DOCTYPE html><p><span><table><tbody><tr><td><span>Hello table data</span></td></tr></tbody></table></span></p>";
+        doc = Jsoup.parse(html);
+        assertEquals(Document.QuirksMode.noQuirks, doc.quirksMode());
+        assertEquals(
+            "<p><span></span></p><table><tbody><tr><td><span>Hello table data</span></td></tr></tbody></table><p></p>", // no quirks, p gets closed
+            TextUtil.normalizeSpaces(doc.body().html())
+        );
+    }
 }