Test for GH43540

Author: michal-gh

pandas/tests/io/test_common.py

@@ -497,6 +497,23 @@ def test_warning_missing_utf_bom(self, encoding, compression_):
             with pytest.raises(UnicodeError, match=msg):
                 pd.read_csv(path, compression=compression_, encoding=encoding)
 
+    def test_chunk_splits_multibyte_char(self):
+        """
+        Chunk splits a multibyte character with memory_map=True
+
+        GH 43540
+        """
+        # DEFAULT_CHUNKSIZE = 262144, defined in parsers.pyx
+        df = pd.DataFrame(data=["a" * 127] * 2048)
+
+        # Put two-bytes utf-8 encoded character "ą" at the end of chunk
+        # utf-8 encoding of "ą" is b'\xc4\x85'
+        df.iloc[2047] = "a" * 127 + "ą"
+        df.to_csv("data/bug-gh43540.csv", index=False, header=False, encoding="utf-8")
+        dfr = pd.read_csv("data/bug-gh43540.csv", header=None, memory_map=True)
+        tm.ensure_clean("data/bug-gh43540.csv")
+        assert (dfr == df).all()[0]
+
 
 def test_is_fsspec_url():
     assert icom.is_fsspec_url("gcs://pandas/somethingelse.com")