fix: support DOCTYPE internal subsets

CHANGELOG.md

@@ -6,6 +6,10 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/) a
 
 ## [Unreleased]
 
+### Changed
+
+- Format files whose `<!DOCTYPE>` has an internal subset (`[ ... ]`) instead of failing to parse; the subset is printed as written.
+
 ## [3.4.2] - 2025-07-08
 
 ### Changed

src/parser.js

@@ -16,6 +16,56 @@ function createError(message, options) {
   return Object.assign(error, options);
 }
 
+// The @xml-tools/parser grammar has no rule for the internal subset of a
+// DOCTYPE (`<!DOCTYPE x [ ... ]>`). Find its `[ ... ]` range so it can be
+// blanked out before parsing and printed back verbatim.
+function findInternalSubset(text) {
+  let idx = 0;
+
+  // Skip whitespace, processing instructions and comments before the DOCTYPE.
+  for (;;) {
+    while (/\s/.test(text[idx] ?? "")) idx += 1;
+
+    const close = text.startsWith("<?", idx)
+      ? "?>"
+      : text.startsWith("<!--", idx)
+        ? "-->"
+        : null;
+    const end = close ? text.indexOf(close, idx) : -1;
+
+    if (end === -1) break;
+    idx = end + close.length;
+  }
+
+  if (!text.startsWith("<!DOCTYPE", idx)) {
+    return null;
+  }
+
+  let start = -1;
+  let quote = null;
+
+  for (idx += "<!DOCTYPE".length; idx < text.length; idx += 1) {
+    const char = text[idx];
+
+    if (quote) {
+      if (char === quote) quote = null;
+    } else if (char === '"' || char === "'") {
+      quote = char;
+    } else if (start === -1) {
+      if (char === "[") start = idx;
+      else if (char === ">") return null;
+    } else if (text.startsWith("<!--", idx)) {
+      const commentEnd = text.indexOf("-->", idx);
+      if (commentEnd === -1) return null;
+      idx = commentEnd + 2;
+    } else if (char === "]") {
+      return { start, end: idx };
+    }
+  }
+
+  return null;
+}
+
 function simplifyCST(node) {
   switch (node.name) {
     case "attribute": {
@@ -68,6 +118,7 @@ function simplifyCST(node) {
         DocType: DocType[0].image,
         Name: Name[0].image,
         externalID: externalID ? simplifyCST(externalID[0]) : null,
+        internalSubset: null,
         CLOSE: CLOSE[0].image,
         location: node.location
       };
@@ -166,7 +217,14 @@ function simplifyCST(node) {
 
 const parser = {
   parse(text) {
-    const { lexErrors, parseErrors, cst } = xmlToolsParse(text);
+    const subset = findInternalSubset(text);
+    const { lexErrors, parseErrors, cst } = xmlToolsParse(
+      subset
+        ? text.slice(0, subset.start) +
+            text.slice(subset.start, subset.end + 1).replace(/[^\r\n]/g, " ") +
+            text.slice(subset.end + 1)
+        : text
+    );
 
     // If there are any lexical errors, throw the first of them as an error.
     if (lexErrors.length > 0) {
@@ -200,7 +258,16 @@ const parser = {
     }
 
     // Otherwise return the CST.
-    return simplifyCST(cst);
+    const document = simplifyCST(cst);
+
+    if (subset) {
+      document.docTypeDecl.internalSubset = text.slice(
+        subset.start + 1,
+        subset.end
+      );
+    }
+
+    return document;
   },
   astFormat: "xml",
   locStart(node) {

src/printer.js

@@ -184,13 +184,17 @@ function printContent(path, opts, print) {
 }
 
 function printDocTypeDecl(path, opts, print) {
-  const { DocType, Name, externalID, CLOSE } = path.node;
+  const { DocType, Name, externalID, internalSubset, CLOSE } = path.node;
   const parts = [DocType, " ", Name];
 
   if (externalID) {
     parts.push(" ", print("externalID"));
   }
 
+  if (internalSubset !== null) {
+    parts.push(" [", doc.utils.replaceEndOfLine(internalSubset), "]");
+  }
+
   return group([...parts, CLOSE]);
 }
 

test/__snapshots__/format.test.js.snap

@@ -1,5 +1,15 @@
 // Jest Snapshot v1, https://jestjs.io/docs/snapshot-testing
 
+exports[`DOCTYPE with an internal subset 1`] = `
+"<?xml version="1.0" ?>
+<!DOCTYPE x SYSTEM "x.dtd" [
+  <!ENTITY a "]>">
+  <!-- ] -->
+]>
+<x>&a;</x>
+"
+`;
+
 exports[`bracketSameLine => true 1`] = `
 "<?xml version="1.0" encoding="UTF-8" ?>
 <!DOCTYPE module PUBLIC "-//Puppy Crawl//DTD Check Configuration 1.3//EN"

test/format.test.js

@@ -104,3 +104,17 @@ test("xmlQuoteAttributes => double", async () => {
 
   expect(formatted).toMatchSnapshot();
 });
+
+test("DOCTYPE with an internal subset", async () => {
+  const content = `<?xml version="1.0"?>
+<!DOCTYPE x SYSTEM "x.dtd" [
+  <!ENTITY a "]>">
+  <!-- ] -->
+]>
+<x>&a;</x>
+`;
+
+  const formatted = await format(content);
+  expect(formatted).toMatchSnapshot();
+  expect(await format(formatted)).toBe(formatted);
+});

test/parser.test.js

@@ -10,3 +10,11 @@ test("parseError", () => {
 
   expect(() => parser.parse("<foo /")).toThrow(expected);
 });
+
+test("DOCTYPE internal subset", () => {
+  const { docTypeDecl } = parser.parse(
+    '<!DOCTYPE x [<!ENTITY a "]">]>\n<x>&a;</x>'
+  );
+
+  expect(docTypeDecl.internalSubset).toBe('<!ENTITY a "]">');
+});