diff --git a/codebeaver.yml b/codebeaver.yml
new file mode 100644
index 0000000..419e243
--- /dev/null
+++ b/codebeaver.yml
@@ -0,0 +1,2 @@
+from: pytest
+# This file was generated automatically by CodeBeaver based on your repository. Learn how to customize it here: https://docs.codebeaver.ai/configuration/
\ No newline at end of file
diff --git a/test/test_database.py b/test/test_database.py
new file mode 100644
index 0000000..2390037
--- /dev/null
+++ b/test/test_database.py
@@ -0,0 +1,221 @@
+import os
+import pytest
+from datetime import datetime
+from collections import OrderedDict
+from sqlalchemy.exc import IntegrityError, SQLAlchemyError
+
+from dataset import connect
+
+from .conftest import TEST_DATA
+
+
+def test_valid_database_url(db):
+    assert db.url, os.environ["DATABASE_URL"]
+
+
+def test_database_url_query_string(db):
+    db = connect("sqlite:///:memory:/?cached_statements=1")
+    assert "cached_statements" in db.url, db.url
+
+
+def test_tables(db, table):
+    assert db.tables == ["weather"], db.tables
+
+
+def test_contains(db, table):
+    assert "weather" in db, db.tables
+
+
+def test_create_table(db):
+    table = db["foo"]
+    assert db.has_table(table.table.name)
+    assert len(table.table.columns) == 1, table.table.columns
+    assert "id" in table.table.c, table.table.c
+
+
+def test_create_table_no_ids(db):
+    if db.is_mysql or db.is_sqlite:
+        return
+    table = db.create_table("foo_no_id", primary_id=False)
+    assert table.table.name == "foo_no_id"
+    assert len(table.table.columns) == 0, table.table.columns
+
+
+def test_create_table_custom_id1(db):
+    pid = "string_id"
+    table = db.create_table("foo2", pid, db.types.string(255))
+    assert db.has_table(table.table.name)
+    assert len(table.table.columns) == 1, table.table.columns
+    assert pid in table.table.c, table.table.c
+    table.insert({pid: "foobar"})
+    assert table.find_one(string_id="foobar")[pid] == "foobar"
+
+
+def test_create_table_custom_id2(db):
+    pid = "string_id"
+    table = db.create_table("foo3", pid, db.types.string(50))
+    assert db.has_table(table.table.name)
+    assert len(table.table.columns) == 1, table.table.columns
+    assert pid in table.table.c, table.table.c
+
+    table.insert({pid: "foobar"})
+    assert table.find_one(string_id="foobar")[pid] == "foobar"
+
+
+def test_create_table_custom_id3(db):
+    pid = "int_id"
+    table = db.create_table("foo4", primary_id=pid)
+    assert db.has_table(table.table.name)
+    assert len(table.table.columns) == 1, table.table.columns
+    assert pid in table.table.c, table.table.c
+
+    table.insert({pid: 123})
+    table.insert({pid: 124})
+    assert table.find_one(int_id=123)[pid] == 123
+    assert table.find_one(int_id=124)[pid] == 124
+    with pytest.raises(IntegrityError):
+        table.insert({pid: 123})
+    db.rollback()
+
+
+def test_create_table_shorthand1(db):
+    pid = "int_id"
+    table = db.get_table("foo5", pid)
+    assert len(table.table.columns) == 1, table.table.columns
+    assert pid in table.table.c, table.table.c
+
+    table.insert({"int_id": 123})
+    table.insert({"int_id": 124})
+    assert table.find_one(int_id=123)["int_id"] == 123
+    assert table.find_one(int_id=124)["int_id"] == 124
+    with pytest.raises(IntegrityError):
+        table.insert({"int_id": 123})
+
+
+def test_create_table_shorthand2(db):
+    pid = "string_id"
+    table = db.get_table("foo6", primary_id=pid, primary_type=db.types.string(255))
+    assert len(table.table.columns) == 1, table.table.columns
+    assert pid in table.table.c, table.table.c
+
+    table.insert({"string_id": "foobar"})
+    assert table.find_one(string_id="foobar")["string_id"] == "foobar"
+
+
+def test_with(db, table):
+    init_length = len(table)
+    with pytest.raises(ValueError):
+        with db:
+            table.insert(
+                {
+                    "date": datetime(2011, 1, 1),
+                    "temperature": 1,
+                    "place": "tmp_place",
+                }
+            )
+            raise ValueError()
+    db.rollback()
+    assert len(table) == init_length
+
+
+def test_invalid_values(db, table):
+    if db.is_mysql:
+        # WARNING: mysql seems to be doing some weird type casting
+        # upon insert. The mysql-python driver is not affected but
+        # it isn't compatible with Python 3
+        # Conclusion: use postgresql.
+        return
+    with pytest.raises(SQLAlchemyError):
+        table.insert({"date": True, "temperature": "wrong_value", "place": "tmp_place"})
+
+
+def test_load_table(db, table):
+    tbl = db.load_table("weather")
+    assert tbl.table.name == table.table.name
+
+
+def test_query(db, table):
+    r = db.query("SELECT COUNT(*) AS num FROM weather").next()
+    assert r["num"] == len(TEST_DATA), r
+
+
+def test_table_cache_updates(db):
+    tbl1 = db.get_table("people")
+    data = OrderedDict([("first_name", "John"), ("last_name", "Smith")])
+    tbl1.insert(data)
+    data["id"] = 1
+    tbl2 = db.get_table("people")
+    assert dict(tbl2.all().next()) == dict(data), (tbl2.all().next(), data)
+
+    
+    def test_repr(db):
+        """Test __repr__ returns safe URL representation."""
+        rep = repr(db)
+        from dataset.util import safe_url
+        assert safe_url(db.url) in rep
+    
+    def test_in_transaction(db):
+        """Test that in_transaction returns True as soon as a connection is accessed."""
+        # accessing the connection starts a transaction so this should be True.
+        assert db.in_transaction is True
+    
+    def test_close_database(db):
+        """Test that closing the database resets engine and table cache."""
+        _ = db.get_table("dummy_close")
+        db.close()
+        assert db._engine is None
+        assert db._tables == {}
+    
+    def test_ipython_key_completions(db):
+        """Test that _ipython_key_completions returns a list of table names."""
+        _ = db.get_table("ipython_test")
+        completions = db._ipython_key_completions_()
+        # normalize table names are used so 'ipython_test' should be there.
+        assert "ipython_test" in completions
+    
+    def test_flush_tables(db):
+        """Test that _flush_tables resets internal table metadata for cached tables."""
+        table_obj = db.get_table("flush_test")
+        # Set a dummy _table value
+        table_obj._table = "dummy"
+        db._flush_tables()
+        assert table_obj._table is None
+    
+    def test_query_with_text_object(db):
+        """Test that the query method accepts SQLAlchemy text objects."""
+        table_obj = db.create_table("query_text")
+        table_obj.insert({'id': 1})
+        from sqlalchemy.sql import text
+        result = list(db.query(text("SELECT * FROM query_text")))
+        assert len(result) >= 1
+        assert result[0]['id'] == 1
+    
+    def test_query_without_step(db):
+        """Test that the query method returns all rows when _step is disabled."""
+        table_obj = db.create_table("query_no_step")
+        table_obj.insert({'id': 2})
+        result = list(db.query("SELECT * FROM query_no_step", _step=False))
+        assert len(result) >= 1
+        assert result[0]['id'] == 2
+    
+    def test_context_manager_no_exception(db):
+        """Test that the context manager commits the transaction if no error occurs."""
+        weather_table = db.get_table("weather")
+        initial_count = len(list(weather_table.all()))
+        with db:
+            weather_table.insert({"date": "2023-10-11", "temperature": 25, "place": "test"})
+        new_count = len(list(weather_table.all()))
+        assert new_count == initial_count + 1
+    
+    def test_context_manager_with_exception(db):
+        """Test that the context manager rolls back the transaction if an error is raised."""
+        weather_table = db.get_table("weather")
+        initial_count = len(list(weather_table.all()))
+        try:
+            with db:
+                weather_table.insert({"date": "2023-10-12", "temperature": 30, "place": "rollback_test"})
+                raise RuntimeError("Triggering rollback")
+        except RuntimeError:
+            pass
+        new_count = len(list(weather_table.all()))
+        assert new_count == initial_count
diff --git a/test/test_table.py b/test/test_table.py
new file mode 100644
index 0000000..f5c9d89
--- /dev/null
+++ b/test/test_table.py
@@ -0,0 +1,526 @@
+import os
+import pytest
+from datetime import datetime
+from collections import OrderedDict
+from sqlalchemy.types import BIGINT, INTEGER, VARCHAR, TEXT
+from sqlalchemy.exc import IntegrityError, SQLAlchemyError, ArgumentError
+
+from dataset import chunked
+
+from .conftest import TEST_DATA, TEST_CITY_1
+
+
+def test_insert(table):
+    assert len(table) == len(TEST_DATA), len(table)
+    last_id = table.insert(
+        {"date": datetime(2011, 1, 2), "temperature": -10, "place": "Berlin"}
+    )
+    assert len(table) == len(TEST_DATA) + 1, len(table)
+    assert table.find_one(id=last_id)["place"] == "Berlin"
+
+
+def test_insert_ignore(table):
+    table.insert_ignore(
+        {"date": datetime(2011, 1, 2), "temperature": -10, "place": "Berlin"},
+        ["place"],
+    )
+    assert len(table) == len(TEST_DATA) + 1, len(table)
+    table.insert_ignore(
+        {"date": datetime(2011, 1, 2), "temperature": -10, "place": "Berlin"},
+        ["place"],
+    )
+    assert len(table) == len(TEST_DATA) + 1, len(table)
+
+
+def test_insert_ignore_all_key(table):
+    for i in range(0, 4):
+        table.insert_ignore(
+            {"date": datetime(2011, 1, 2), "temperature": -10, "place": "Berlin"},
+            ["date", "temperature", "place"],
+        )
+    assert len(table) == len(TEST_DATA) + 1, len(table)
+
+
+def test_insert_json(table):
+    last_id = table.insert(
+        {
+            "date": datetime(2011, 1, 2),
+            "temperature": -10,
+            "place": "Berlin",
+            "info": {
+                "currency": "EUR",
+                "language": "German",
+                "population": 3292365,
+            },
+        }
+    )
+    assert len(table) == len(TEST_DATA) + 1, len(table)
+    assert table.find_one(id=last_id)["place"] == "Berlin"
+
+
+def test_upsert(table):
+    table.upsert(
+        {"date": datetime(2011, 1, 2), "temperature": -10, "place": "Berlin"},
+        ["place"],
+    )
+    assert len(table) == len(TEST_DATA) + 1, len(table)
+    table.upsert(
+        {"date": datetime(2011, 1, 2), "temperature": -10, "place": "Berlin"},
+        ["place"],
+    )
+    assert len(table) == len(TEST_DATA) + 1, len(table)
+
+
+def test_upsert_single_column(db):
+    table = db["banana_single_col"]
+    table.upsert({"color": "Yellow"}, ["color"])
+    assert len(table) == 1, len(table)
+    table.upsert({"color": "Yellow"}, ["color"])
+    assert len(table) == 1, len(table)
+
+
+def test_upsert_all_key(table):
+    assert len(table) == len(TEST_DATA), len(table)
+    for i in range(0, 2):
+        table.upsert(
+            {"date": datetime(2011, 1, 2), "temperature": -10, "place": "Berlin"},
+            ["date", "temperature", "place"],
+        )
+    assert len(table) == len(TEST_DATA) + 1, len(table)
+
+
+def test_upsert_id(db):
+    table = db["banana_with_id"]
+    data = dict(id=10, title="I am a banana!")
+    table.upsert(data, ["id"])
+    assert len(table) == 1, len(table)
+
+
+def test_update_while_iter(table):
+    for row in table:
+        row["foo"] = "bar"
+        table.update(row, ["place", "date"])
+    assert len(table) == len(TEST_DATA), len(table)
+
+
+def test_weird_column_names(table):
+    with pytest.raises(ValueError):
+        table.insert(
+            {
+                "date": datetime(2011, 1, 2),
+                "temperature": -10,
+                "foo.bar": "Berlin",
+                "qux.bar": "Huhu",
+            }
+        )
+
+
+def test_cased_column_names(db):
+    tbl = db["cased_column_names"]
+    tbl.insert({"place": "Berlin"})
+    tbl.insert({"Place": "Berlin"})
+    tbl.insert({"PLACE ": "Berlin"})
+    assert len(tbl.columns) == 2, tbl.columns
+    assert len(list(tbl.find(Place="Berlin"))) == 3
+    assert len(list(tbl.find(place="Berlin"))) == 3
+    assert len(list(tbl.find(PLACE="Berlin"))) == 3
+
+
+def test_invalid_column_names(db):
+    tbl = db["weather"]
+    with pytest.raises(ValueError):
+        tbl.insert({None: "banana"})
+
+    with pytest.raises(ValueError):
+        tbl.insert({"": "banana"})
+
+    with pytest.raises(ValueError):
+        tbl.insert({"-": "banana"})
+
+
+def test_delete(table):
+    table.insert({"date": datetime(2011, 1, 2), "temperature": -10, "place": "Berlin"})
+    original_count = len(table)
+    assert len(table) == len(TEST_DATA) + 1, len(table)
+    # Test bad use of API
+    with pytest.raises(ArgumentError):
+        table.delete({"place": "Berlin"})
+    assert len(table) == original_count, len(table)
+
+    assert table.delete(place="Berlin") is True, "should return 1"
+    assert len(table) == len(TEST_DATA), len(table)
+    assert table.delete() is True, "should return non zero"
+    assert len(table) == 0, len(table)
+
+
+def test_repr(table):
+    assert (
+        repr(table) == "<Table(weather)>"
+    ), "the representation should be <Table(weather)>"
+
+
+def test_delete_nonexist_entry(table):
+    assert (
+        table.delete(place="Berlin") is False
+    ), "entry not exist, should fail to delete"
+
+
+def test_find_one(table):
+    table.insert({"date": datetime(2011, 1, 2), "temperature": -10, "place": "Berlin"})
+    d = table.find_one(place="Berlin")
+    assert d["temperature"] == -10, d
+    d = table.find_one(place="Atlantis")
+    assert d is None, d
+
+
+def test_count(table):
+    assert len(table) == 6, len(table)
+    length = table.count(place=TEST_CITY_1)
+    assert length == 3, length
+
+
+def test_find(table):
+    ds = list(table.find(place=TEST_CITY_1))
+    assert len(ds) == 3, ds
+    ds = list(table.find(place=TEST_CITY_1, _limit=2))
+    assert len(ds) == 2, ds
+    ds = list(table.find(place=TEST_CITY_1, _limit=2, _step=1))
+    assert len(ds) == 2, ds
+    ds = list(table.find(place=TEST_CITY_1, _limit=1, _step=2))
+    assert len(ds) == 1, ds
+    ds = list(table.find(_step=2))
+    assert len(ds) == len(TEST_DATA), ds
+    ds = list(table.find(order_by=["temperature"]))
+    assert ds[0]["temperature"] == -1, ds
+    ds = list(table.find(order_by=["-temperature"]))
+    assert ds[0]["temperature"] == 8, ds
+    ds = list(table.find(table.table.columns.temperature > 4))
+    assert len(ds) == 3, ds
+
+
+def test_find_dsl(table):
+    ds = list(table.find(place={"like": "%lw%"}))
+    assert len(ds) == 3, ds
+    ds = list(table.find(temperature={">": 5}))
+    assert len(ds) == 2, ds
+    ds = list(table.find(temperature={">=": 5}))
+    assert len(ds) == 3, ds
+    ds = list(table.find(temperature={"<": 0}))
+    assert len(ds) == 1, ds
+    ds = list(table.find(temperature={"<=": 0}))
+    assert len(ds) == 2, ds
+    ds = list(table.find(temperature={"!=": -1}))
+    assert len(ds) == 5, ds
+    ds = list(table.find(temperature={"between": [5, 8]}))
+    assert len(ds) == 3, ds
+    ds = list(table.find(place={"=": "G€lway"}))
+    assert len(ds) == 3, ds
+    ds = list(table.find(place={"ilike": "%LwAy"}))
+    assert len(ds) == 3, ds
+
+
+def test_offset(table):
+    ds = list(table.find(place=TEST_CITY_1, _offset=1))
+    assert len(ds) == 2, ds
+    ds = list(table.find(place=TEST_CITY_1, _limit=2, _offset=2))
+    assert len(ds) == 1, ds
+
+
+def test_streamed_update(table):
+    ds = list(table.find(place=TEST_CITY_1, _streamed=True, _step=1))
+    assert len(ds) == 3, len(ds)
+    for row in table.find(place=TEST_CITY_1, _streamed=True, _step=1):
+        row["temperature"] = -1
+        table.update(row, ["id"])
+
+
+def test_distinct(table):
+    x = list(table.distinct("place"))
+    assert len(x) == 2, x
+    x = list(table.distinct("place", "date"))
+    assert len(x) == 6, x
+    x = list(
+        table.distinct(
+            "place",
+            "date",
+            table.table.columns.date >= datetime(2011, 1, 2, 0, 0),
+        )
+    )
+    assert len(x) == 4, x
+
+    x = list(table.distinct("temperature", place="B€rkeley"))
+    assert len(x) == 3, x
+    x = list(table.distinct("temperature", place=["B€rkeley", "G€lway"]))
+    assert len(x) == 6, x
+
+
+def test_insert_many(table):
+    data = TEST_DATA * 100
+    table.insert_many(data, chunk_size=13)
+    assert len(table) == len(data) + 6, (len(table), len(data))
+
+
+def test_chunked_insert(table):
+    data = TEST_DATA * 100
+    with chunked.ChunkedInsert(table) as chunk_tbl:
+        for item in data:
+            chunk_tbl.insert(item)
+    assert len(table) == len(data) + 6, (len(table), len(data))
+
+
+def test_chunked_insert_callback(table):
+    data = TEST_DATA * 100
+    N = 0
+
+    def callback(queue):
+        nonlocal N
+        N += len(queue)
+
+    with chunked.ChunkedInsert(table, callback=callback) as chunk_tbl:
+        for item in data:
+            chunk_tbl.insert(item)
+    assert len(data) == N
+    assert len(table) == len(data) + 6
+
+
+def test_update_many(db):
+    tbl = db["update_many_test"]
+    tbl.insert_many([dict(temp=10), dict(temp=20), dict(temp=30)])
+    tbl.update_many([dict(id=1, temp=50), dict(id=3, temp=50)], "id")
+
+    # Ensure data has been updated.
+    assert tbl.find_one(id=1)["temp"] == tbl.find_one(id=3)["temp"]
+
+
+def test_chunked_update(db):
+    tbl = db["update_many_test"]
+    tbl.insert_many(
+        [
+            dict(temp=10, location="asdf"),
+            dict(temp=20, location="qwer"),
+            dict(temp=30, location="asdf"),
+        ]
+    )
+    db.commit()
+
+    chunked_tbl = chunked.ChunkedUpdate(tbl, ["id"])
+    chunked_tbl.update(dict(id=1, temp=50))
+    chunked_tbl.update(dict(id=2, location="asdf"))
+    chunked_tbl.update(dict(id=3, temp=50))
+    chunked_tbl.flush()
+    db.commit()
+
+    # Ensure data has been updated.
+    assert tbl.find_one(id=1)["temp"] == tbl.find_one(id=3)["temp"] == 50
+    assert tbl.find_one(id=2)["location"] == tbl.find_one(id=3)["location"] == "asdf"
+
+
+def test_upsert_many(db):
+    # Also tests updating on records with different attributes
+    tbl = db["upsert_many_test"]
+
+    W = 100
+    tbl.upsert_many([dict(age=10), dict(weight=W)], "id")
+    assert tbl.find_one(id=1)["age"] == 10
+
+    tbl.upsert_many([dict(id=1, age=70), dict(id=2, weight=W / 2)], "id")
+    assert tbl.find_one(id=2)["weight"] == W / 2
+
+
+def test_drop_operations(table):
+    assert table._table is not None, "table shouldn't be dropped yet"
+    table.drop()
+    assert table._table is None, "table should be dropped now"
+    assert list(table.all()) == [], table.all()
+    assert table.count() == 0, table.count()
+
+
+def test_table_drop(db, table):
+    assert "weather" in db
+    db["weather"].drop()
+    assert "weather" not in db
+
+
+def test_table_drop_then_create(db, table):
+    assert "weather" in db
+    db["weather"].drop()
+    assert "weather" not in db
+    db["weather"].insert({"foo": "bar"})
+
+
+def test_columns(table):
+    cols = table.columns
+    assert len(list(cols)) == 4, "column count mismatch"
+    assert "date" in cols and "temperature" in cols and "place" in cols
+
+
+def test_drop_column(table):
+    try:
+        table.drop_column("date")
+        assert "date" not in table.columns
+    except RuntimeError:
+        pass
+
+
+def test_iter(table):
+    c = 0
+    for row in table:
+        c += 1
+    assert c == len(table)
+
+
+def test_update(table):
+    date = datetime(2011, 1, 2)
+    res = table.update(
+        {"date": date, "temperature": -10, "place": TEST_CITY_1}, ["place", "date"]
+    )
+    assert res, "update should return True"
+    m = table.find_one(place=TEST_CITY_1, date=date)
+    assert m["temperature"] == -10, (
+        "new temp. should be -10 but is %d" % m["temperature"]
+    )
+
+
+def test_create_column(db, table):
+    flt = db.types.float
+    table.create_column("foo", flt)
+    assert "foo" in table.table.c, table.table.c
+    assert isinstance(table.table.c["foo"].type, flt), table.table.c["foo"].type
+    assert "foo" in table.columns, table.columns
+
+
+def test_ensure_column(db, table):
+    flt = db.types.float
+    table.create_column_by_example("foo", 0.1)
+    assert "foo" in table.table.c, table.table.c
+    assert isinstance(table.table.c["foo"].type, flt), table.table.c["bar"].type
+    table.create_column_by_example("bar", 1)
+    assert "bar" in table.table.c, table.table.c
+    assert isinstance(table.table.c["bar"].type, BIGINT), table.table.c["bar"].type
+    table.create_column_by_example("pippo", "test")
+    assert "pippo" in table.table.c, table.table.c
+    assert isinstance(table.table.c["pippo"].type, TEXT), table.table.c["pippo"].type
+    table.create_column_by_example("bigbar", 11111111111)
+    assert "bigbar" in table.table.c, table.table.c
+    assert isinstance(table.table.c["bigbar"].type, BIGINT), table.table.c[
+        "bigbar"
+    ].type
+    table.create_column_by_example("littlebar", -11111111111)
+    assert "littlebar" in table.table.c, table.table.c
+    assert isinstance(table.table.c["littlebar"].type, BIGINT), table.table.c[
+        "littlebar"
+    ].type
+
+
+def test_key_order(db, table):
+    res = db.query("SELECT temperature, place FROM weather LIMIT 1")
+    keys = list(res.next().keys())
+    assert keys[0] == "temperature"
+    assert keys[1] == "place"
+
+
+def test_empty_query(table):
+    empty = list(table.find(place="not in data"))
+    assert len(empty) == 0, empty
+
+# New tests to increase test coverage
+def test_find_invalid_operator(table):
+    """Test find with an unsupported operator returns no rows using an unknown operator."""
+    # Passing an operator that is not recognized should result in a false() clause, yielding no results.
+    results = list(table.find(temperature={"unknown": 5}))
+    assert results == [], "Using an unsupported operator should return an empty result set"
+
+def test_create_index_custom_name(db):
+    """Test creating an index with a custom name on a table column."""
+    tbl = db["index_test"]
+    # Create the table by inserting a row
+    tbl.insert({"temperature": 22, "place": "TestCity"})
+    tbl.create_index(["temperature"], name="custom_index")
+    assert tbl.has_index(["temperature"]), "Custom index should be recognized by has_index"
+
+def test_insert_many_empty(table):
+    """Test that insert_many with an empty list leaves the table unchanged."""
+    initial_count = len(table)
+    table.insert_many([], chunk_size=10)
+    assert len(table) == initial_count, "Inserting an empty list should not change the table count"
+
+def test_update_many_empty(table):
+    """Test that update_many with an empty list leaves the table unchanged."""
+    initial_count = len(table)
+    table.update_many([], "id")
+    assert len(table) == initial_count, "Updating an empty list should not change the table count"
+
+def test_insert_no_ensure(table):
+    """Test that extra keys are dropped when ensure is False."""
+    # 'place' is an existing column, while 'not_exist' is not
+    row_data = {"place": "TestCity", "not_exist": "should_be_dropped"}
+    new_id = table.insert(row_data, ensure=False)
+    ret = table.find_one(id=new_id)
+    assert "place" in ret, "Existing column 'place' should be present"
+    assert "not_exist" not in ret, "Non-existing column 'not_exist' should be dropped"
+
+def test_find_operator_filters(table):
+    """Test find DSL operators: startswith, endswith, notlike, and notilike."""
+    from datetime import datetime
+    # Insert a row with a new 'text' column (auto-created when ensure=True)
+    new_id = table.insert({
+        "date": datetime(2021, 1, 1),
+        "temperature": 0,
+        "place": "TestPlace",
+        "text": "HelloWorld"
+    }, ensure=True)
+    # Test 'startswith'
+    rows = list(table.find(text={"startswith": "Hell"}))
+    assert any(r.get("text") == "HelloWorld" for r in rows), "Row with text starting 'Hell' should be found"
+    # Test 'endswith'
+    rows = list(table.find(text={"endswith": "World"}))
+    assert any(r.get("text") == "HelloWorld" for r in rows), "Row with text ending 'World' should be found"
+    # Test 'notlike' (should exclude rows that match the pattern)
+    rows = list(table.find(text={"notlike": "%Bye%"}))
+    assert any(r.get("text") == "HelloWorld" for r in rows), "Row with text not matching '%Bye%' should be found"
+    # Test 'notilike' (case-insensitive match)
+    rows = list(table.find(text={"notilike": "%bye%"}))
+    assert any(r.get("text") == "HelloWorld" for r in rows), "Row with text not matching case-insensitive '%bye%' should be found"
+
+def test_update_empty_data(table):
+    """Test update behavior when the update dict becomes empty (only keys are used for filtering)."""
+    # Grab an existing row from the table based on a known column value
+    sample = table.find_one()
+    # Use keys present in the sample so that the update dict (non-key fields) becomes empty.
+    # The update method should then return the count of matched records rather than false.
+    count = table.update(sample, ["place", "date"])
+    assert count > 0, "Update with empty update dict should return the matched row count"
+
+def test_create_index_duplicate_call(db, table):
+    """Test that creating an index twice does not error and the index is recognized."""
+    from datetime import datetime
+    # Insert a new row to ensure the table exists
+    table.insert({"date": datetime(2022, 1, 1), "temperature": 15, "place": "IndexCity"})
+    # Create an index on the 'temperature' column twice
+    table.create_index("temperature")
+    table.create_index("temperature")
+    assert table.has_index("temperature"), "Index on 'temperature' should exist after duplicate calls"
+
+def test_find_with_in_operator(table):
+    """Test filtering using the 'in' operator for the 'place' column."""
+    from datetime import datetime
+    # Insert rows with distinct place values
+    table.insert({"date": datetime(2023, 1, 1), "temperature": 100, "place": "CityA"})
+    table.insert({"date": datetime(2023, 1, 2), "temperature": 101, "place": "CityB"})
+    table.insert({"date": datetime(2023, 1, 3), "temperature": 102, "place": "CityC"})
+    rows = list(table.find(place={"in": ["CityA", "CityC"]}))
+    places = [r.get("place", "") for r in rows]
+    assert "CityA" in places and "CityC" in places, "Rows with CityA and CityC places should be found using 'in' operator"
+
+def test_update_return_count_for_no_update(table):
+    """Test that update returns a count when there is no non-key field to update."""
+    from datetime import datetime
+    row_data = {"date": datetime(2020, 1, 1), "temperature": 20, "place": "CountCity"}
+    new_id = table.insert(row_data)
+    # Update with keys only so that the update dict becomes empty
+    count = table.update({
+        "date": row_data["date"],
+        "temperature": row_data["temperature"],
+        "place": row_data["place"]
+    }, ["place", "date"])
+    assert count >= 1, "Update should return count of matched rows when update dict is empty"
\ No newline at end of file
diff --git a/tests/test_chunked.py b/tests/test_chunked.py
new file mode 100644
index 0000000..5db0748
--- /dev/null
+++ b/tests/test_chunked.py
@@ -0,0 +1,183 @@
+import pytest
+import itertools
+
+from dataset.chunked import ChunkedInsert, ChunkedUpdate, InvalidCallback
+
+class DummyTable:
+    """A dummy table class to record operations on insert and update operations."""
+    def __init__(self):
+        self.inserts = []
+        self.updates = []
+
+    def insert_many(self, items):
+        self.inserts.extend(items)
+
+    def update_many(self, items, keys):
+        # Record the update call as a tuple: (items, keys)
+        self.updates.append((items, keys))
+
+# Test class for Chunked operations
+class TestChunked:
+    """Test suite for ChunkedInsert and ChunkedUpdate functionality."""
+
+    def test_invalid_callback_insert(self):
+        """Test that passing a non-callable callback to ChunkedInsert raises InvalidCallback."""
+        dummy = DummyTable()
+        with pytest.raises(InvalidCallback):
+            ChunkedInsert(dummy, chunksize=1, callback="I am not callable")
+
+    def test_invalid_callback_update(self):
+        """Test that passing a non-callable callback to ChunkedUpdate raises InvalidCallback."""
+        dummy = DummyTable()
+        with pytest.raises(InvalidCallback):
+            ChunkedUpdate(dummy, keys=['id'], chunksize=1, callback=123)
+
+    def test_chunked_insert_auto_flush(self):
+        """Test that ChunkedInsert flushes automatically when the chunksize is reached."""
+        dummy = DummyTable()
+        # Use a small chunksize to trigger flush automatically
+        inserter = ChunkedInsert(dummy, chunksize=2)
+        # Insert rows with different keys; fields should be merged.
+        inserter.insert({'a': 1})
+        inserter.insert({'b': 2})
+        # At this point, flush should have been called automatically
+        assert len(dummy.inserts) == 2
+        # Check that for each inserted row, the missing field has a None value
+        expected = [{'a': 1, 'b': None}, {'a': None, 'b': 2}]
+        assert dummy.inserts == expected
+
+    def test_chunked_insert_context_manager(self):
+        """Test that ChunkedInsert flushes remaining data when exiting the context manager."""
+        dummy = DummyTable()
+        with ChunkedInsert(dummy, chunksize=3) as inserter:
+            inserter.insert({'x': 10})
+            inserter.insert({'y': 20})
+            # Do not reach chunksize so flush happens on __exit__
+        # Now flush should have been called on exit and both inserted.
+        expected = [{'x': 10, 'y': None}, {'x': None, 'y': 20}]
+        assert dummy.inserts == expected
+
+    def test_chunked_update_grouping(self):
+        """Test that ChunkedUpdate groups rows correctly and calls update_many accordingly."""
+        dummy = DummyTable()
+        updater = ChunkedUpdate(dummy, keys=['id'], chunksize=2)
+        # Insert two items with same keys (so they belong to the same group)
+        updater.update({'id': 1, 'value': 'a'})
+        updater.update({'id': 2, 'value': 'b'})
+        # Flush should be called manually
+        updater.flush()
+        # Since both items have the same set of keys, they will be grouped in one update_many call
+        assert len(dummy.updates) == 1
+        updated_items, update_keys = dummy.updates[0]
+        # The update keys passed in should be ['id']
+        assert update_keys == ['id']
+        # Confirm that the updated group contains the two items
+        assert {'id': 1, 'value': 'a'} in updated_items
+        assert {'id': 2, 'value': 'b'} in updated_items
+
+    def test_chunked_update_context_manager_with_callback(self):
+        """Test that ChunkedUpdate flushes on context manager exit and that a callback is invoked."""
+        dummy = DummyTable()
+        callback_called = []
+
+        def callback(queue):
+            callback_called.append(len(queue))
+
+        with ChunkedUpdate(dummy, keys=['id'], chunksize=3, callback=callback) as updater:
+            updater.update({'id': 1, 'name': 'Alice'})
+            updater.update({'id': 2, 'name': 'Bob'})
+            # Do not add a third row so flush happens on __exit__
+        # Ensure callback was called once during flush
+        assert callback_called == [2]
+        # Check that an update_many call was made (even if with a single group)
+        assert len(dummy.updates) == 1
+
+    def test_empty_flush(self):
+        """Test that calling flush on an empty queue works without error for both insert and update."""
+        dummy = DummyTable()
+        inserter = ChunkedInsert(dummy, chunksize=2)
+        # Call flush on an empty queue via context manager
+        with inserter:
+            pass
+        # Also test update flush on empty queue
+        updater = ChunkedUpdate(dummy, keys=['id'], chunksize=2)
+        with updater:
+            pass
+        # No assertions needed; just ensuring no exceptions are raised.
+
+    def test_chunked_insert_callback_invocation(self):
+        """Test that ChunkedInsert invokes its callback and merges fields correctly during flush."""
+        dummy = DummyTable()
+        callback_called = []
+        def callback(queue):
+            callback_called.append(len(queue))
+        # Use a chunksize of 3 to delay flush until 3 items have been added.
+        inserter = ChunkedInsert(dummy, chunksize=3, callback=callback)
+        inserter.insert({'a': 1})
+        inserter.insert({'b': 2})
+        # At this point, flush hasn't been triggered so callback should not have been called.
+        assert callback_called == []
+        # Inserting the third item triggers flush.
+        inserter.insert({'c': 3})
+        # After flush, callback should have been invoked with a queue of length 3.
+        assert callback_called == [3]
+        # All inserted rows should have merged fields from all items: keys 'a', 'b', and 'c'.
+        expected = [
+            {'a': 1, 'b': None, 'c': None},
+            {'a': None, 'b': 2, 'c': None},
+            {'a': None, 'b': None, 'c': 3}
+        ]
+        assert dummy.inserts == expected
+
+    def test_chunked_update_multiple_groups(self):
+        """Test that ChunkedUpdate groups updated items by their set of keys when they differ."""
+        dummy = DummyTable()
+        updater = ChunkedUpdate(dummy, keys=['id'], chunksize=5)
+        # Two items with the same keys (group 1)
+        updater.update({'id': 1, 'value': 'a'})
+        updater.update({'id': 2, 'value': 'b'})
+        # One item with a different keys set (group 2)
+        updater.update({'id': 3})
+        # One more item matching group 1 keys
+        updater.update({'id': 4, 'value': 'd'})
+        # Manually flush to trigger update_many calls
+        updater.flush()
+        # Ensure that exactly two update_many calls were made, one for each distinct keys set.
+        assert len(dummy.updates) == 2
+        # Verify that one group contains three items (group1 with keys {'id', 'value'})
+        # and the other group contains one item (group2 with keys {'id'}).
+        group_sizes = sorted(len(items) for items, _ in dummy.updates)
+        assert group_sizes == [1, 3]
+        # Also verify that all update_many calls used the update keys ['id'].
+        for items, keys in dummy.updates:
+            assert keys == ['id']
+    def test_manual_flush_insert(self):
+        """Test that manually calling flush in ChunkedInsert works as expected (without reaching chunksize)."""
+        dummy = DummyTable()
+        inserter = ChunkedInsert(dummy, chunksize=5)
+        inserter.insert({'p': 100})
+        inserter.insert({'q': 200})
+        # Manually call flush even though the chunksize has not been reached
+        inserter.flush()
+        expected = [{'p': 100, 'q': None}, {'p': None, 'q': 200}]
+        assert dummy.inserts == expected
+
+    def test_manual_flush_update(self):
+        """Test that manually calling flush in ChunkedUpdate triggers the callback and update_many correctly."""
+        dummy = DummyTable()
+        callback_called = []
+
+        def update_callback(queue):
+            callback_called.append(len(queue))
+
+        updater = ChunkedUpdate(dummy, keys=['id'], chunksize=5, callback=update_callback)
+        updater.update({'id': 10, 'r': 'x'})
+        # Manually flush the pending update operation
+        updater.flush()
+        # Verify that the callback was invoked with a queue length of 1
+        assert callback_called == [1]
+        # Verify that update_many was called exactly once and with the correct parameters
+        assert len(dummy.updates) == 1
+        updated_items, update_keys = dummy.updates[0]
+        assert update_keys == ['id']
+        assert {'id': 10, 'r': 'x'} in updated_items
\ No newline at end of file
diff --git a/tests/test_test_row_type.py b/tests/test_test_row_type.py
new file mode 100644
index 0000000..d8002d0
--- /dev/null
+++ b/tests/test_test_row_type.py
@@ -0,0 +1,544 @@
+import pytest
+from datetime import datetime
+from test.test_row_type import Constructor
+
+def test_constructor_attribute_get():
+    """Test that attribute access on Constructor returns expected values."""
+    data = {"temperature": 20, "place": "Paris", "date": datetime(2020, 1, 1)}
+    c = Constructor(data)
+    assert c.temperature == 20
+    assert c.place == "Paris"
+    assert c["date"] == datetime(2020, 1, 1)
+
+def test_constructor_missing_attribute_raises_keyerror():
+    """Test that accessing a missing attribute raises KeyError."""
+    c = Constructor({"a": 1})
+    with pytest.raises(KeyError):
+        _ = c.nonexistent
+
+def test_constructor_update_affects_attribute_access():
+    """Test that updating the Constructor dictionary reflects in attribute access."""
+    c = Constructor({"a": 5})
+    c.update({"b": 10})
+    assert c.b == 10
+
+def test_constructor_overlapping_method_names():
+    """Test that keys overlapping with dict methods still return correct value via __getattr__."""
+    c = Constructor({"keys": "value_of_keys", "get": "value_of_get"})
+    # Direct attribute access for 'keys' returns the built-in method, so we use __getattr__ for testing.
+    assert c["keys"] == "value_of_keys"
+    assert c.__getattr__("get") == "value_of_get"
+# New tests to increase test coverage for table operations and custom row type conversion.
+class FakeDB:
+    """Fake database class for testing that holds a row_type attribute."""
+    def __init__(self):
+        self.row_type = dict
+
+class FakeTable:
+    """Fake table implementing insert, find_one, find, distinct, __iter__, and __len__ operations."""
+    def __init__(self, db):
+        self.records = []
+        self.db = db
+
+    def insert(self, record):
+        self.records.append(record)
+
+    def __apply_row_type(self, rec):
+        if rec is None:
+            return None
+        if hasattr(self.db, "row_type") and self.db.row_type:
+            return self.db.row_type(rec)
+        return rec
+
+    def find_one(self, **kwargs):
+        for rec in self.records:
+            match = True
+            for k, v in kwargs.items():
+                if k == "_limit":
+                    continue
+                if rec.get(k) != v:
+                    match = False
+                    break
+            if match:
+                return self.__apply_row_type(rec)
+        return None
+
+    def find(self, **kwargs):
+        limit = kwargs.pop('_limit', None)
+        result = []
+        for rec in self.records:
+            if all(rec.get(k) == v for k, v in kwargs.items()):
+                result.append(self.__apply_row_type(rec))
+                if limit is not None and len(result) >= limit:
+                    break
+        return result
+
+    def distinct(self, *fields):
+        seen = set()
+        result = []
+        for rec in self.records:
+            key = tuple(rec.get(f) for f in fields)
+            if key not in seen:
+                seen.add(key)
+                rec_data = {f: rec.get(f) for f in fields}
+                result.append(self.__apply_row_type(rec_data))
+        return result
+
+    def __iter__(self):
+        for rec in self.records:
+            yield self.__apply_row_type(rec)
+
+    def __len__(self):
+        return len(self.records)
+
+def test_integration_find_one_fake_table():
+    """Test integration of find_one using FakeDB and FakeTable with Constructor as row type."""
+    db = FakeDB()
+    db.row_type = Constructor  # use Constructor to allow attribute access
+    table = FakeTable(db)
+    from datetime import datetime
+    row = {"date": datetime(2011, 1, 2), "temperature": -10, "place": "Berlin"}
+    table.insert(row)
+    found = table.find_one(place="Berlin")
+    assert found is not None
+    # Test both dict and attribute access
+    assert found["temperature"] == -10
+    assert found.temperature == -10
+    missing = table.find_one(place="Atlantis")
+    assert missing is None
+
+def test_integration_find_fake_table():
+    """Test integration of the find method with limit functionality using FakeDB and FakeTable."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    # Insert multiple records
+    rows = [
+        {"place": "CityA", "value": 1},
+        {"place": "CityA", "value": 2},
+        {"place": "CityA", "value": 3},
+        {"place": "CityB", "value": 4},
+    ]
+    for r in rows:
+        table.insert(r)
+    results = table.find(place="CityA")
+    assert len(results) == 3
+    for item in results:
+        assert isinstance(item, Constructor)
+    limited = table.find(place="CityA", _limit=2)
+    assert len(limited) == 2
+
+def test_integration_distinct_fake_table():
+    """Test integration of the distinct method using FakeDB and FakeTable with multiple field criteria."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    # Insert records with duplicates
+    rows = [
+        {"cat": "A", "num": 1},
+        {"cat": "A", "num": 2},
+        {"cat": "B", "num": 3},
+        {"cat": "B", "num": 3},  # duplicate
+        {"cat": "C", "num": 4},
+    ]
+    for r in rows:
+        table.insert(r)
+    distinct_cats = table.distinct("cat")
+    # Expect three distinct cat values: A, B, and C
+    assert len(distinct_cats) == 3
+    for item in distinct_cats:
+        assert isinstance(item, Constructor)
+    distinct_cat_num = table.distinct("cat", "num")
+    # Expect four distinct records (duplicates for B,3 should be filtered)
+    assert len(distinct_cat_num) == 4
+
+def test_integration_iter_fake_table():
+    """Test the iteration functionality of FakeTable ensuring all records are converted to Constructor."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    rows = [
+        {"id": 1},
+        {"id": 2},
+        {"id": 3},
+    ]
+    for r in rows:
+        table.insert(r)
+    count = 0
+    for rec in table:
+        count += 1
+        assert isinstance(rec, Constructor)
+def test_no_row_type_returns_dict():
+    """Test that when db.row_type is None, FakeTable returns plain dicts instead of using Constructor."""
+    db = FakeDB()
+    db.row_type = None
+    table = FakeTable(db)
+    row = {"id": 1, "value": "test"}
+    table.insert(row)
+    result = table.find_one(id=1)
+    assert result is not None
+    assert isinstance(result, dict)
+    # Ensure that it is not converted to a Constructor
+    assert not hasattr(result, "value") or result["value"] == "test"
+def test_find_no_filters_returns_all():
+    """Test that calling find with no filter kwargs returns all records."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    rows = [
+        {"id": 1},
+        {"id": 2},
+        {"id": 3},
+    ]
+    for row in rows:
+        table.insert(row)
+    results = table.find()
+    assert len(results) == len(rows)
+    for res in results:
+        assert isinstance(res, Constructor)
+def test_distinct_missing_field():
+    """Test distinct method handling of keys that are missing in some records."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    rows = [
+        {"id": 1, "value": "a"},
+        {"id": 2},  # missing 'value', so rec.get('value') will be None
+        {"id": 3, "value": "a"},
+        {"id": 4, "value": None},
+    ]
+    for row in rows:
+        table.insert(row)
+    results = table.distinct("value")
+    distinct_values = set()
+    for res in results:
+        distinct_values.add(res.value)
+    # Expect one distinct value "a" and one None
+    assert "a" in distinct_values
+    assert None in distinct_values
+    assert len(results) == 2
+def test_iter_empty_table():
+    """Test that iterating over an empty FakeTable yields no records."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    count = 0
+    for rec in table:
+        count += 1
+    assert count == 0
+    assert len(table) == 0
+def test_find_limit_zero_fake_table():
+    """Test the find method with _limit=0 edge case.
+    Due to the logic ordering in FakeTable.find, a _limit of 0 still returns 1 record."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    rows = [
+        {"id": 1, "value": "x"},
+        {"id": 2, "value": "x"},
+    ]
+    for r in rows:
+        table.insert(r)
+    results = table.find(value="x", _limit=0)
+    # The first matching record is appended then _limit is checked causing an early break
+    assert len(results) == 1
+
+def test_custom_row_type_enhanced_fake_table():
+    """Test using a custom EnhancedConstructor as the row type that adds a description method."""
+    # Define a custom row type that extends Constructor with an extra method.
+    class EnhancedConstructor(Constructor):
+        def description(self):
+            return f"{self.place} on {self.date.strftime('%Y-%m-%d')}"
+
+    db = FakeDB()
+    db.row_type = EnhancedConstructor
+    table = FakeTable(db)
+    from datetime import datetime
+    row = {"date": datetime(2021, 12, 25), "place": "New York", "temperature": 5}
+    table.insert(row)
+    result = table.find_one(place="New York")
+    assert result is not None
+    # Test that the custom method works alongside dictionary and attribute access.
+    assert result.description() == "New York on 2021-12-25"
+    assert result.temperature == 5
+
+def test_find_with_nonexistent_field_returns_empty_list():
+    """Test that searching with a filter for a nonexistent field returns an empty list."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    rows = [
+        {"id": 1, "value": "a"},
+        {"id": 2, "value": "b"},
+    ]
+    for r in rows:
+        table.insert(r)
+    results = table.find(nonexistent="anything")
+    assert results == []
+
+def test_updated_record_affects_iteration():
+    """Test that updating a record in the underlying table affects the result during iteration.
+    This confirms that the conversion via __apply_row_type happens on the fly."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    row = {"id": 1, "value": "initial"}
+    table.insert(row)
+    # Update the record directly in table.records
+    table.records[0]["value"] = "updated"
+    for rec in table:
+        assert rec.value == "updated"
+def test_find_with_multiple_conditions():
+    """Test that find_one and find work with multiple filtering conditions."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    rows = [
+        {"id": 1, "a": "foo", "b": "bar", "c": 10},
+        {"id": 2, "a": "foo", "b": "baz", "c": 20},
+        {"id": 3, "a": "qux", "b": "bar", "c": 30},
+        {"id": 4, "a": "foo", "b": "bar", "c": 40},
+    ]
+    for row in rows:
+        table.insert(row)
+    # Test find_one with multiple conditions (should match the first occurrence)
+    res = table.find_one(a="foo", b="bar")
+    assert res is not None
+    assert res.a == "foo"
+    assert res.b == "bar"
+    # Test find with the same conditions returns all matching records (id 1 and 4)
+    results = table.find(a="foo", b="bar")
+    assert len(results) == 2
+
+def test_find_invalid_limit():
+    """Test the find method with a negative _limit value to verify edge case behavior."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    rows = [
+        {"id": 1, "value": "x"},
+        {"id": 2, "value": "x"},
+        {"id": 3, "value": "x"},
+    ]
+    for r in rows:
+        table.insert(r)
+    results = table.find(value="x", _limit=-1)
+    # With a negative _limit the condition is met almost immediately so only one record is returned
+    assert len(results) == 1
+
+def test_custom_lambda_row_type():
+    """Test using a custom lambda as the row type to transform records."""
+    db = FakeDB()
+    # Custom lambda that adds an 'extra' key to the record
+    db.row_type = lambda rec: {**rec, "extra": "yes"}
+    table = FakeTable(db)
+    row = {"id": 100, "value": "lambda test"}
+    table.insert(row)
+    # Test find_one conversion
+    res_one = table.find_one(id=100)
+    assert res_one is not None
+    assert res_one["extra"] == "yes"
+    # Test find conversion
+    results = table.find(id=100)
+    for rec in results:
+        assert rec["extra"] == "yes"
+    # Test distinct conversion using custom lambda row type
+    distinct = table.distinct("value")
+    for rec in distinct:
+        assert rec["extra"] == "yes"
+    # Test iteration conversion
+    found = False
+    for rec in table:
+        if rec.get("id") == 100:
+            found = True
+            assert rec["extra"] == "yes"
+    assert found
+
+def test_non_callable_row_type():
+    """Test that setting a non-callable row_type raises an error when transforming a record."""
+    db = FakeDB()
+    db.row_type = 5  # non-callable row_type
+    table = FakeTable(db)
+    row = {"id": 1, "value": "error expected"}
+    table.insert(row)
+    import pytest
+    with pytest.raises(TypeError):
+        _ = table.find_one(id=1)
+def test_custom_namedtuple_row_type():
+    """Test using a custom row_type as collections.namedtuple returning a tuple conversion."""
+    from collections import namedtuple
+    CustomRow = namedtuple("CustomRow", ["id", "value"])
+    db = FakeDB()
+    db.row_type = lambda rec: CustomRow(**rec)
+    table = FakeTable(db)
+    row = {"id": 1, "value": "test"}
+    table.insert(row)
+    result = table.find_one(id=1)
+    assert isinstance(result, CustomRow)
+    assert result.id == 1
+    assert result.value == "test"
+
+def test_exception_in_custom_row_type():
+    """Test that exceptions in custom row_type conversion propagate as expected."""
+    def faulty_row_type(record):
+        # Attempt to access a key that might be missing so a KeyError is raised.
+        return record["mandatory"]
+    db = FakeDB()
+    db.row_type = faulty_row_type
+    table = FakeTable(db)
+    row = {"id": 1, "value": "error expected"}
+    table.insert(row)
+    import pytest
+    with pytest.raises(KeyError):
+        _ = table.find_one(id=1)
+def test_distinct_empty_fields():
+    """Test that calling distinct with no field filters returns a single entry (an empty dict) as distinct."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    rows = [
+        {"a": 1},
+        {"b": 2},
+    ]
+    for row in rows:
+        table.insert(row)
+    # When no field is provided, every record yields an empty key tuple so only one item is returned.
+    result = table.distinct()
+    assert len(result) == 1
+    # The distinct entry should be an empty dict converted by Constructor.
+    assert result[0] == {}
+
+def test_find_one_with_limit_keyword_ignored():
+    """Test that passing '_limit' to find_one is ignored so it behaves normally."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    row = {"id": 10, "value": "test"}
+    table.insert(row)
+    # Pass _limit as extra keyword; it should be skipped during filtering.
+    result = table.find_one(id=10, _limit=100)
+    assert result is not None
+    assert result.id == 10
+
+def test_insert_non_dict_record_exception():
+    """Test that inserting a non-dictionary record will raise an error when trying to perform get on it."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    table.insert(["not", "a", "dict"])
+    with pytest.raises(AttributeError):
+        _ = table.find_one(somekey="value")
+
+def test_find_with_none_as_filter_value():
+    """Test that find correctly matches records with a None filter value."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    rows = [
+        {"id": 1, "optional": None},
+        {"id": 2, "optional": "present"},
+    ]
+    for row in rows:
+        table.insert(row)
+    result = table.find(optional=None)
+    assert len(result) == 1
+    assert result[0].id == 1
+def test_switch_row_type():
+    """Test that switching the db.row_type dynamically affects row conversion."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    row = {"id": 99, "name": "First"}
+    table.insert(row)
+    result1 = table.find_one(id=99)
+    # Check that result1 was converted using Constructor
+    assert hasattr(result1, "name")
+    # Now switch the row type to a lambda that adds an extra key
+    db.row_type = lambda rec: {**rec, "extra": "modified"}
+    result2 = table.find_one(id=99)
+    # Check that the new conversion is applied (a plain dict with extra key)
+    assert isinstance(result2, dict)
+    assert result2.get("extra") == "modified"
+
+def test_find_invalid_limit_type():
+    """Test that using a non-integer _limit value causes a TypeError during comparison."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    rows = [
+        {"id": 1, "value": "x"},
+        {"id": 2, "value": "x"},
+    ]
+    for r in rows:
+        table.insert(r)
+    import pytest
+    with pytest.raises(TypeError):
+        _ = table.find(value="x", _limit="non_integer")
+def test_modification_of_converted_record_does_not_affect_underlying_data():
+    """Test that modifying a converted record does not change the underlying record in FakeTable."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    row = {"id": 1, "value": "original"}
+    table.insert(row)
+    converted = table.find_one(id=1)
+    # Modify the converted record
+    converted["value"] = "modified"
+    # Get a new conversion from the underlying record
+    fresh = table.find_one(id=1)
+    assert fresh["value"] == "original", "Underlying record should not be affected by modifications to the converted record"
+
+def test_find_returns_fresh_instances_on_each_call():
+    """Test that each call to find_one returns a fresh conversion instance and not a cached object."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    row = {"id": 2, "value": "constant"}
+    table.insert(row)
+    first_instance = table.find_one(id=2)
+    second_instance = table.find_one(id=2)
+    assert first_instance is not second_instance, "Each conversion should produce a new instance"
+
+def test_find_no_matching_record_returns_none():
+    """Test that find_one returns None when no record matches the provided filter."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    table.insert({"id": 3, "value": "test"})
+    result = table.find_one(id=999)
+    assert result is None, "find_one should return None if no matching record is found"
+def test_custom_constant_row_type():
+    """Test using a constant lambda as the row_type that returns a constant value regardless of record content.
+    This ensures that find_one, iteration, and distinct all return the constant conversion."""
+    db = FakeDB()
+    db.row_type = lambda rec: "constant"
+    table = FakeTable(db)
+    row = {"id": 101, "value": "dummy"}
+    table.insert(row)
+    result = table.find_one(id=101)
+    assert result == "constant", "Expected constant conversion for find_one"
+    iter_results = list(table)
+    # All results via iteration should yield the constant value
+    for item in iter_results:
+        assert item == "constant", "Iteration did not yield constant conversion"
+    distinct_results = table.distinct("value")
+    for item in distinct_results:
+        assert item == "constant", "Distinct did not yield constant conversion"
+
+def test_find_with_list_field():
+    """Test that the find method correctly matches records with list fields using equality.
+    The record should be found only when the list exactly matches."""
+    db = FakeDB()
+    db.row_type = Constructor
+    table = FakeTable(db)
+    row1 = {"id": 201, "tags": ["a", "b", "c"]}
+    row2 = {"id": 202, "tags": ["x", "y"]}
+    table.insert(row1)
+    table.insert(row2)
+    # Search for the record with an exact match on the list field.
+    result = table.find_one(tags=["a", "b", "c"])
+    assert result is not None, "Expected to find record with matching list"
+    assert result.id == 201
+    # A search for a list that does not match should return None.
+    result_none = table.find_one(tags=["non", "matching"])
+    assert result_none is None, "Expected no record with non-matching list"
\ No newline at end of file
diff --git a/tests/test_util.py b/tests/test_util.py
new file mode 100644
index 0000000..31fc981
--- /dev/null
+++ b/tests/test_util.py
@@ -0,0 +1,302 @@
+import pytest
+from collections import namedtuple, OrderedDict
+from urllib.parse import urlparse, parse_qs
+from hashlib import sha1
+from sqlalchemy.exc import ResourceClosedError
+from dataset.util import (
+    convert_row,
+    DatasetException,
+    iter_result_proxy,
+    make_sqlite_url,
+    ResultIter,
+    normalize_column_name,
+    normalize_column_key,
+    normalize_table_name,
+    safe_url,
+    index_name,
+    pad_chunk_columns,
+)
+
+# Fake ResultProxy for testing iter_result_proxy and ResultIter
+class FakeCursor:
+    def __init__(self, rows, keys=None, fetchmany_size=None, raise_on_keys=False):
+        self.rows = rows[:]  # copy list of rows
+        self.fetchmany_size = fetchmany_size
+        self.closed = False
+        self._keys = keys if keys is not None else (list(rows[0]._fields) if rows else [])
+        self.raise_on_keys = raise_on_keys
+
+    def fetchall(self):
+        result = self.rows
+        self.rows = []
+        return result
+
+    def fetchmany(self, size):
+        if self.fetchmany_size is not None:
+            size = self.fetchmany_size
+        chunk = self.rows[:size]
+        self.rows = self.rows[size:]
+        return chunk
+
+    def keys(self):
+        if self.raise_on_keys:
+            raise ResourceClosedError("Cursor is closed")
+        return self._keys
+
+    def close(self):
+        self.closed = True
+
+# Define a dummy row type to simulate namedtuple rows for testing convert_row and ResultIter.
+DummyRow = namedtuple("DummyRow", ["a", "b"])
+
+class TestUtil:
+    def test_convert_row_with_valid_row(self):
+        """Test convert_row with a valid row-like object."""
+        row = DummyRow(a=1, b=2)
+        result = convert_row(OrderedDict, row)
+        assert result == OrderedDict([("a", 1), ("b", 2)])
+
+    def test_convert_row_with_none(self):
+        """Test that convert_row returns None when given None."""
+        assert convert_row(OrderedDict, None) is None
+
+    def test_iter_result_proxy_fetchall(self):
+        """Test iter_result_proxy using fetchall method."""
+        rows = [DummyRow(a=i, b=i+1) for i in range(5)]
+        fake_cursor = FakeCursor(rows)
+        result = list(iter_result_proxy(fake_cursor))
+        assert result == rows
+
+    def test_iter_result_proxy_fetchmany(self):
+        """Test iter_result_proxy using fetchmany method."""
+        rows = [DummyRow(a=i, b=i+1) for i in range(5)]
+        fake_cursor = FakeCursor(rows, fetchmany_size=2)
+        result = list(iter_result_proxy(fake_cursor, step=2))
+        assert result == rows
+
+    def test_make_sqlite_url_basic(self):
+        """Test that make_sqlite_url returns a basic URL when no extra params are provided."""
+        url = make_sqlite_url("test.db")
+        assert url == "sqlite:///test.db"
+
+    def test_make_sqlite_url_with_params(self):
+        """Test that make_sqlite_url returns a URL with encoded query parameters."""
+        url = make_sqlite_url("test.db", cache="shared", timeout=30, mode="rw", check_same_thread=False, immutable=True, nolock=True)
+        # The URL should start with 'sqlite:///file:test.db?' due to URI scheme
+        assert url.startswith("sqlite:///file:test.db?")
+        parsed = urlparse(url)
+        qs = parse_qs(parsed.query)
+        expected = {
+            "cache": ["shared"],
+            "timeout": ["30"],
+            "mode": ["rw"],
+            "nolock": ["1"],
+            "immutable": ["1"],
+            "check_same_thread": ["false"],
+            "uri": ["true"]
+        }
+        assert qs == expected
+
+    def test_result_iter_normal_iteration(self):
+        """Test that ResultIter iterates over rows and closes the cursor upon completion."""
+        rows = [DummyRow(a=i, b=i+1) for i in range(3)]
+        fake_cursor = FakeCursor(rows)
+        result_iter = ResultIter(fake_cursor)
+        collected = list(result_iter)
+        expected = [OrderedDict([("a", row.a), ("b", row.b)]) for row in rows]
+        assert collected == expected
+        # After iteration, the cursor should be closed
+        assert fake_cursor.closed is True
+
+    def test_result_iter_closed_cursor(self):
+        """Test that ResultIter handles ResourceClosedError by using an empty iterator."""
+        fake_cursor = FakeCursor([], raise_on_keys=True)
+        result_iter = ResultIter(fake_cursor)
+        result = list(result_iter)
+        assert result == []
+        fake_cursor.close()
+        assert fake_cursor.closed is True
+
+    def test_normalize_column_name_valid(self):
+        """Test normalize_column_name with a valid column name."""
+        valid_name = "valid_column"
+        assert normalize_column_name(valid_name) == valid_name
+
+    def test_normalize_column_name_invalid(self):
+        """Test normalize_column_name with column names that should be rejected."""
+        with pytest.raises(ValueError):
+            normalize_column_name("invalid.column")
+        with pytest.raises(ValueError):
+            normalize_column_name("invalid-column")
+
+    def test_normalize_column_name_length(self):
+        """Test that normalize_column_name truncates column names exceeding 63 bytes."""
+        long_name = "a" * 100
+        result = normalize_column_name(long_name)
+        assert len(result.encode("utf-8")) < 64
+
+    def test_normalize_column_key(self):
+        """Test normalize_column_key converts strings to uppercase and removes spaces."""
+        result = normalize_column_key(" col Name ")
+        assert result == "COLNAME"
+        # Check that None is returned as None
+        assert normalize_column_key(None) is None
+
+    def test_normalize_table_name_valid(self):
+        """Test normalize_table_name with a valid table name."""
+        table_name = "table_name"
+        assert normalize_table_name(table_name) == table_name
+
+    def test_normalize_table_name_invalid(self):
+        """Test normalize_table_name raises ValueError for invalid names."""
+        with pytest.raises(ValueError):
+            normalize_table_name("")
+
+    def test_safe_url(self):
+        """Test that safe_url masks passwords contained in a connection URL."""
+        original_url = "postgres://user:secret@localhost/db"
+        result = safe_url(original_url)
+        assert "secret" not in result
+        assert "*****" in result
+
+    def test_index_name(self):
+        """Test that index_name generates an index name in the expected format."""
+        table = "mytable"
+        columns = ["col1", "col2"]
+        idx = index_name(table, columns)
+        assert idx.startswith("ix_mytable_")
+        key_part = idx[len("ix_mytable_"):]
+        # The key should be exactly 16 hex characters; assert its length and that it is valid hex.
+        assert len(key_part) == 16
+        int(key_part, 16)
+
+    def test_pad_chunk_columns(self):
+        """Test that pad_chunk_columns adds missing columns with None values."""
+        chunk = [{"a": 1}, {"b": 2}]
+        columns = ["a", "b", "c"]
+        result = pad_chunk_columns(chunk, columns)
+        for record in result:
+            for col in columns:
+                assert col in record
+        # Ensure that existing keys are retained
+        assert result[0]["a"] == 1
+
+    def test_dataset_exception(self):
+        """Test that DatasetException carries the correct message."""
+        exc = DatasetException("Test error")
+        assert str(exc) == "Test error"
+
+    def test_make_sqlite_url_invalid_cache(self):
+        """Test that make_sqlite_url raises AssertionError for an invalid cache value."""
+        with pytest.raises(AssertionError):
+            make_sqlite_url("test.db", cache="invalid")
+
+    def test_make_sqlite_url_invalid_mode(self):
+        """Test that make_sqlite_url raises AssertionError for an invalid mode value."""
+        with pytest.raises(AssertionError):
+            make_sqlite_url("test.db", mode="invalid")
+
+    def test_normalize_column_name_non_string(self):
+        """Test that normalize_column_name raises ValueError when given a non-string."""
+        with pytest.raises(ValueError):
+            normalize_column_name(123)
+
+    def test_normalize_table_name_non_string(self):
+        """Test that normalize_table_name raises ValueError when given a non-string."""
+        with pytest.raises(ValueError):
+            normalize_table_name(123)
+
+    def test_normalize_column_key_non_string(self):
+        """Test that normalize_column_key returns None when given a non-string."""
+        assert normalize_column_key(123) is None
+
+    def test_iter_result_proxy_with_zero(self):
+        """Test iter_result_proxy with step=0 returns an empty iterator even if there are rows."""
+        rows = [DummyRow(a=i, b=i+1) for i in range(3)]
+        fake_cursor = FakeCursor(rows, fetchmany_size=0)
+        result = list(iter_result_proxy(fake_cursor, step=0))
+        assert result == []
+    def test_safe_url_no_password(self):
+        """Test that safe_url returns the URL unchanged when there is no password."""
+        url = "postgres://user@localhost/db"
+        result = safe_url(url)
+        assert result == url
+
+    def test_convert_row_custom_row_type(self):
+        """Test convert_row with a custom row_type (dict) instead of OrderedDict."""
+        row = DummyRow(a=10, b=20)
+        result = convert_row(dict, row)
+        expected = dict([("a", 10), ("b", 20)])
+        assert result == expected
+
+    def test_pad_chunk_columns_empty_chunk(self):
+        """Test that pad_chunk_columns returns an empty list when given an empty chunk."""
+        chunk = []
+        columns = ["a", "b"]
+        result = pad_chunk_columns(chunk, columns)
+        assert result == []
+
+    def test_pad_chunk_columns_all_present(self):
+        """Test that pad_chunk_columns does not modify the chunk if all columns are present."""
+        chunk = [{"a": 1, "b": 2}]
+        columns = ["a", "b"]
+        result = pad_chunk_columns(chunk, columns)
+        assert result == chunk
+
+    def test_result_iter_next_after_exhaustion(self):
+        """Test that ResultIter raises StopIteration after all rows have been iterated."""
+        rows = [DummyRow(a=1, b=2)]
+        fake_cursor = FakeCursor(rows)
+        result_iter = ResultIter(fake_cursor)
+        # Get the only element.
+        first = next(result_iter)
+        assert first == OrderedDict([("a", 1), ("b", 2)])
+        # Next call should raise StopIteration and close the cursor.
+        with pytest.raises(StopIteration):
+            next(result_iter)
+        # Ensure cursor is closed.
+    def test_iter_result_proxy_empty(self):
+        """Test iter_result_proxy returns an empty list when given an empty cursor."""
+        fake_cursor = FakeCursor([])
+        result = list(iter_result_proxy(fake_cursor))
+        assert result == []
+
+    def test_iter_result_proxy_negative_step(self):
+        """Test iter_result_proxy with a negative step value produces expected slice behavior."""
+        # Create 3 dummy rows. With step = -1, fetchmany will return rows[:-1] on the first call.
+        rows = [DummyRow(a=i, b=i+10) for i in range(3)]
+        fake_cursor = FakeCursor(rows)
+        result = list(iter_result_proxy(fake_cursor, step=-1))
+        # When step is negative, list slicing returns all but the last element.
+        expected = rows[:-1]
+        assert result == expected
+
+    def test_normalize_column_name_trimmed(self):
+        """Test normalize_column_name trims whitespace from a valid column name."""
+        input_name = "   trimmed_column   "
+        result = normalize_column_name(input_name)
+        # The result should be trimmed of whitespace and still be valid.
+        assert result == "trimmed_column"
+
+    def test_normalize_table_name_trimmed(self):
+        """Test normalize_table_name trims whitespace from a valid table name."""
+        input_name = "   trimmed_table   "
+        result = normalize_table_name(input_name)
+        assert result == "trimmed_table"
+
+    def test_result_iter_iter_method(self):
+        """Test that the __iter__ method of ResultIter returns self."""
+        rows = [DummyRow(a=99, b=100)]
+        fake_cursor = FakeCursor(rows)
+        result_iter = ResultIter(fake_cursor)
+        # Calling iter() on result_iter should return itself.
+        # Verify that __iter__ returns self without closing the cursor
+        it = iter(result_iter)
+        assert it is result_iter
+        # Before iteration the cursor should still be open
+        assert not fake_cursor.closed
+        # Fully consume the iterator so that the cursor gets closed
+        list(result_iter)
+        # After iteration, the cursor should be closed
+        assert fake_cursor.closed
+# End of TestUtil class
\ No newline at end of file