From e63f1eb7e02b3c920f8f24cedc462ca3485e88aa Mon Sep 17 00:00:00 2001
From: venti <1308199824@qq.com>
Date: Sat, 30 May 2026 16:26:48 +0800
Subject: [PATCH] fix: decode exiftool JSON output as UTF-8 instead of locale
 encoding

ExifTool always outputs JSON in UTF-8 (per RFC 8259), but the
exiftool_metadata() function decoded the output using
locale.getpreferredencoding(). On systems with non-UTF-8 locale
encoding (e.g. Windows with Chinese locale cp936), this caused
UnicodeDecodeError when ExifTool returned non-ASCII metadata.

Fix:
- Use encoding='utf-8' instead of text=True in version-check
  subprocess.run() so stdout is decoded as UTF-8.
- Decode JSON output bytes with 'utf-8' instead of
  locale.getpreferredencoding().
- Remove unused import of locale.

Fixes #1972
---
 .../src/markitdown/converters/_exiftool.py    |  5 ++---
 packages/markitdown/tests/test_module_misc.py | 22 +++++++++++++++++++
 2 files changed, 24 insertions(+), 3 deletions(-)

diff --git a/packages/markitdown/src/markitdown/converters/_exiftool.py b/packages/markitdown/src/markitdown/converters/_exiftool.py
index f605024fd..8766fcc4a 100644
--- a/packages/markitdown/src/markitdown/converters/_exiftool.py
+++ b/packages/markitdown/src/markitdown/converters/_exiftool.py
@@ -1,5 +1,4 @@
 import json
-import locale
 import subprocess
 from typing import Any, BinaryIO, Union
 
@@ -22,7 +21,7 @@ def exiftool_metadata(
         version_output = subprocess.run(
             [exiftool_path, "-ver"],
             capture_output=True,
-            text=True,
+            encoding="utf-8",
             check=True,
         ).stdout.strip()
         version = _parse_version(version_output)
@@ -46,7 +45,7 @@ def exiftool_metadata(
         ).stdout
 
         return json.loads(
-            output.decode(locale.getpreferredencoding(False)),
+            output.decode("utf-8"),
         )[0]
     finally:
         file_stream.seek(cur_pos)
diff --git a/packages/markitdown/tests/test_module_misc.py b/packages/markitdown/tests/test_module_misc.py
index 4d62e4919..4fe916e2a 100644
--- a/packages/markitdown/tests/test_module_misc.py
+++ b/packages/markitdown/tests/test_module_misc.py
@@ -462,6 +462,27 @@ def test_markitdown_exiftool() -> None:
         assert target in result.text_content
 
 
+def test_exiftool_metadata_decodes_utf8() -> None:
+    """ExifTool JSON output must be decoded as UTF-8 per RFC 8259."""
+    from unittest.mock import patch, MagicMock
+    from markitdown.converters._exiftool import exiftool_metadata
+
+    version_mock = MagicMock(stdout="13.0\n")
+    # JSON with non-ASCII characters encoded in UTF-8 (C3 89 = É, C3 A9 = é)
+    metadata_mock = MagicMock(
+        stdout=b'[{"Author":"\xc3\x89lena","Title":"Caf\xc3\xa9 R\xc3\xa9sum\xc3\xa9"}]'
+    )
+
+    with patch("subprocess.run") as mock_run:
+        mock_run.side_effect = [version_mock, metadata_mock]
+        result = exiftool_metadata(
+            io.BytesIO(b"fake-data"), exiftool_path="/fake/exiftool"
+        )
+
+    assert result["Author"] == "Élena"
+    assert result["Title"] == "Café Résumé"
+
+
 def test_markitdown_llm_parameters() -> None:
     """Test that LLM parameters are correctly passed to the client."""
     mock_client = MagicMock()
@@ -545,6 +566,7 @@ def test_markitdown_llm() -> None:
         test_exceptions,
         test_doc_rlink,
         test_markitdown_exiftool,
+        test_exiftool_metadata_decodes_utf8,
         test_markitdown_llm_parameters,
         test_markitdown_llm,
     ]: