From d77442228d769ba20d7006bf4befea3a4c2f06d9 Mon Sep 17 00:00:00 2001
From: hhang <17309584214@163.com>
Date: Fri, 6 Nov 2026 14:52:32 +0800
Subject: [PATCH] fix(desktop): align exported XML formatting whitespace
 (#9020)

* fix(desktop): align exported XML formatting whitespace

* test(process): align symlink comparison snapshots

* test(boundary): align inherited CLI and MCP fixtures

---------

Co-authored-by: hhang <169027046+xu91102@users.noreply.github.com>
---
 docs/desktop-application-state.md             |  3 +
 src/domain/desktop/appExportPythonXml.ts      |  5 +-
 .../desktopXmlWhitespacePython.test.ts        | 59 +++++++++++++++++++
 3 files changed, 66 insertions(+), 1 deletion(-)
 create mode 100644 tests/boundary/providers/desktopXmlWhitespacePython.test.ts

diff --git a/docs/desktop-application-state.md b/docs/desktop-application-state.md
index 605d43c728..7ab24697a0 100644
--- a/docs/desktop-application-state.md
+++ b/docs/desktop-application-state.md
@@ -86,6 +86,9 @@ element that holds elements is dropped unless `xml:space="preserve"` applies.
 The view keeps up to 5,000 nodes, comments and processing instructions
 counted, nested up to 64 levels, and sets `truncated` when it leaves some out.
 
+Exported XML views use REA's same formatting-whitespace classification, so
+retained XML text and the task checks that address it stay aligned.
+
 SQLite databases are read through REA's SQLite worker from a private copy,
 with any retained `-wal` or `-journal` file staged beside it; the worker
 accepts an inactive rollback journal and refuses a hot one. A version is not
diff --git a/src/domain/desktop/appExportPythonXml.ts b/src/domain/desktop/appExportPythonXml.ts
index fadd5d40d3..30c7571c48 100644
--- a/src/domain/desktop/appExportPythonXml.ts
+++ b/src/domain/desktop/appExportPythonXml.ts
@@ -15,6 +15,9 @@ class _XmlReader:
     """REA's XML view built as expat reads: names as spelled, declarations as attributes, text split where REA's DOM
     splits it (an empty CDATA section joins the text around it), and each node counted against REA's limits."""
 
+    #: Match REA String.trim(); default strip also drops U+0085 and keeps U+FEFF.
+    FORMATTING_SPACE = "\t\n\v\f\r \u00a0\u1680\u2000\u2001\u2002\u2003\u2004\u2005\u2006\u2007\u2008\u2009\u200a\u2028\u2029\u202f\u205f\u3000\ufeff"
+
     def __init__(self) -> None:
         self.remaining, self.truncated, self.root = MAX_XML_NODES, False, None
         #: Per open element: its view or None when left out, whether xml:space="preserve" applies inside it, and
@@ -83,7 +86,7 @@ class _XmlReader:
         self._flush()
         view, preserve, holds_element = self.open.pop()
         if view is not None and holds_element and not preserve:
-            view["children"] = [child for child in view["children"] if not isinstance(child, str) or child.strip()]
+            view["children"] = [child for child in view["children"] if not isinstance(child, str) or child.strip(self.FORMATTING_SPACE)]
         if not self.open:
             self.root = view
 
diff --git a/tests/boundary/providers/desktopXmlWhitespacePython.test.ts b/tests/boundary/providers/desktopXmlWhitespacePython.test.ts
new file mode 100644
index 0000000000..6ea9560acd
--- /dev/null
+++ b/tests/boundary/providers/desktopXmlWhitespacePython.test.ts
@@ -0,0 +1,59 @@
+import { execFileSync } from "node:child_process";
+import { expect, it } from "vitest";
+import {
+  PYTHON_HELPERS,
+  PYTHON_IMPORTS,
+} from "../../../src/domain/desktop/appExportPython.js";
+import { xmlView } from "../../../src/domain/desktop/stateFormats.js";
+
+/** Execute the unmodified generated XML reader with its standard-library imports. */
+const pythonViews = (texts: readonly string[]) => {
+  const imports = PYTHON_IMPORTS.split("\n")
+    .filter((line) => !line.includes("agentenv_protocol"))
+    .join("\n");
+  return JSON.parse(
+    execFileSync(
+      "python3",
+      [
+        "-I",
+        "-B",
+        "-c",
+        `${imports}${PYTHON_HELPERS}\nprint(json.dumps([_xml_view(text) for text in json.load(sys.stdin)]))`,
+      ],
+      { input: JSON.stringify(texts), encoding: "utf8" },
+    ),
+  );
+};
+
+it.each(["\u0085", "\ufeff"])(
+  "matches REA's formatting decision for %j between elements",
+  (text) => {
+    const xml = `<prefs><entry/>${text}</prefs>`;
+    expect(pythonViews([xml])).toEqual([xmlView(xml)]);
+  },
+);
+
+it("matches the ECMAScript whitespace set for XML-valid characters and preserves other text", () => {
+  const characters =
+    "\t\n\r \u00a0\u1680\u2000\u2001\u2002\u2003\u2004\u2005\u2006\u2007\u2008\u2009\u200a\u2028\u2029\u202f\u205f\u3000\ufeff\u0085\u180e\u200b";
+  const texts = [...characters].map((text) => `<prefs><entry/>${text}</prefs>`);
+  expect(pythonViews(texts)).toEqual(texts.map(xmlView));
+});
+
+it("keeps leaf text and xml:space-preserved text, with the original parsing refusals", () => {
+  const texts = [
+    "<prefs><entry>\ufeff</entry></prefs>",
+    "<prefs><entry>\u0085</entry></prefs>",
+    '<prefs xml:space="preserve"><entry/>\ufeff</prefs>',
+    '<prefs xml:space="preserve"><entry/>\u0085</prefs>',
+  ];
+  const actual = pythonViews(texts) as { root: { children: unknown[] } }[];
+  const expected = texts.map(xmlView) as { root: { children: unknown[] } }[];
+  expect(actual.map((view) => view.root.children)).toEqual(
+    expected.map((view) => view.root.children),
+  );
+  expect(pythonViews(["<!DOCTYPE prefs><prefs/>", "<p:prefs/>"])).toEqual([
+    null,
+    null,
+  ]);
+});