Add error handling to decode_text function

dissect/esedb/c_esedb.py

@@ -493,7 +493,7 @@ def decode_bit(buf: bytes) -> bool:
     return c_esedb.uint8(buf) == 0xFF
 
 
-def decode_text(buf: bytes, encoding: CODEPAGE) -> str:
+def decode_text(buf: bytes, encoding: CODEPAGE, errors: str | None = "backslashreplace") -> str:
     """Decode text with the appropriate encoding.
 
     Args:
@@ -504,7 +504,7 @@ def decode_text(buf: bytes, encoding: CODEPAGE) -> str:
     if encoding == CODEPAGE.UNICODE and len(buf) % 2:
         buf += b"\x00"
 
-    return buf.decode(CODEPAGE_MAP[encoding]).rstrip("\x00")
+    return buf.decode(CODEPAGE_MAP[encoding], errors=errors).rstrip("\x00")
 
 
 def decode_guid(buf: bytes) -> str:

tests/conftest.py

@@ -67,3 +67,8 @@ def sru_db() -> Iterator[BinaryIO]:
 @pytest.fixture
 def ual_db() -> Iterator[BinaryIO]:
     yield from open_file_gz("_data/Current.mdb.gz")
+
+
+@pytest.fixture
+def windows_search_db() -> Iterator[BinaryIO]:
+    yield from open_file_gz("_data/Windows.edb.gz")

tests/test_record.py

@@ -63,3 +63,19 @@ def test_comparison(basic_db: BinaryIO) -> None:
 
     assert set(records) == {records[0], records[1]}
     assert set(records) | {obj} == {records[0], records[1]}
+
+
+def test_parse_value_encoding(windows_search_db: BinaryIO) -> None:
+    """Test if we can parse invalid utf-16-le (Long)Text columns.
+
+    Resources:
+        - https://github.com/fox-it/dissect.esedb/pull/48
+    """
+
+    db = EseDB(windows_search_db)
+    table = db.table("SystemIndex_PropertyStore")
+
+    record = table.search(WorkID=1017)
+    auto_summary = record.get("4625-System_Search_AutoSummary")
+    assert auto_summary.startswith("Hong Kong SCS AdobeMingStd-Light-Acro-HKscs-B5-H ASCII")
+    assert auto_summary.endswith("\\x4c\\xd8")