✨ Add encoding parameter to write_file() (#511)

Co-authored-by: claell <26320273+claell@users.noreply.github.com>

Author: MiWeiss

bibtexparser/entrypoint.py

@@ -139,6 +139,7 @@ def write_file(
     unparse_stack: Optional[Iterable[Middleware]] = None,
     prepend_middleware: Optional[Iterable[Middleware]] = None,
     bibtex_format: Optional[BibtexFormat] = None,
+    encoding: str = "UTF-8",
 ) -> None:
     """Write a BibTeX database to a file.
 
@@ -148,15 +149,16 @@ def write_file(
                         If None, a default stack will be used.
     :param prepend_middleware: List of middleware to prepend to the default stack.
                         Only applicable if `unparse_stack` is None.
-    :param bibtex_format: Customized BibTeX format to use (optional)."""
+    :param bibtex_format: Customized BibTeX format to use (optional).
+    :param encoding: Encoding of the .bib file. Default encoding is ``"UTF-8"``."""
     bibtex_str = write_string(
         library=library,
         unparse_stack=unparse_stack,
         prepend_middleware=prepend_middleware,
         bibtex_format=bibtex_format,
     )
     if isinstance(file, str):
-        with open(file, "w") as f:
+        with open(file, "w", encoding=encoding) as f:
             f.write(bibtex_str)
     else:
         file.write(bibtex_str)

tests/test_entrypoint.py

@@ -1,6 +1,13 @@
-"""Testing the parse_file function."""
+"""Testing the parse_file and write_file functions."""
+
+import os
+import tempfile
 
 from bibtexparser import parse_file
+from bibtexparser import write_file
+from bibtexparser.library import Library
+from bibtexparser.model import Entry
+from bibtexparser.model import Field
 
 
 def test_gbk():
@@ -9,3 +16,77 @@ def test_gbk():
     assert library.entries[0]["title"] == "Test Title"
     assert library.entries[0]["year"] == "2013"
     assert library.entries[0]["journal"] == "测试期刊"
+
+
+def test_write_file_default_encoding():
+    """Test write_file uses UTF-8 by default."""
+    entry = Entry(
+        entry_type="article",
+        key="test2024",
+        fields=[
+            Field(key="author", value="Müller"),
+            Field(key="title", value="Ångström measurements"),
+        ],
+    )
+    library = Library([entry])
+
+    with tempfile.NamedTemporaryFile(mode="w", suffix=".bib", delete=False) as f:
+        temp_path = f.name
+
+    try:
+        write_file(temp_path, library)
+        # Read back and verify
+        with open(temp_path, encoding="UTF-8") as f:
+            content = f.read()
+        assert "Müller" in content
+        assert "Ångström" in content
+    finally:
+        os.unlink(temp_path)
+
+
+def test_write_file_gbk_encoding():
+    """Test write_file with GBK encoding for Chinese characters."""
+    entry = Entry(
+        entry_type="article",
+        key="test2024",
+        fields=[
+            Field(key="author", value="凯撒"),
+            Field(key="title", value="Test Title"),
+            Field(key="journal", value="测试期刊"),
+        ],
+    )
+    library = Library([entry])
+
+    with tempfile.NamedTemporaryFile(mode="w", suffix=".bib", delete=False) as f:
+        temp_path = f.name
+
+    try:
+        write_file(temp_path, library, encoding="gbk")
+        # Read back with GBK and verify
+        with open(temp_path, encoding="gbk") as f:
+            content = f.read()
+        assert "凯撒" in content
+        assert "测试期刊" in content
+    finally:
+        os.unlink(temp_path)
+
+
+def test_write_file_roundtrip_gbk():
+    """Test round-trip: parse GBK file, write with GBK, parse again."""
+    # Parse original GBK file
+    library = parse_file("tests/resources/gbk_test.bib", encoding="gbk")
+    original_author = library.entries[0]["author"]
+    original_journal = library.entries[0]["journal"]
+
+    with tempfile.NamedTemporaryFile(mode="w", suffix=".bib", delete=False) as f:
+        temp_path = f.name
+
+    try:
+        # Write with GBK encoding
+        write_file(temp_path, library, encoding="gbk")
+        # Parse back
+        library2 = parse_file(temp_path, encoding="gbk")
+        assert library2.entries[0]["author"] == original_author
+        assert library2.entries[0]["journal"] == original_journal
+    finally:
+        os.unlink(temp_path)