FIX-modin-project#6549: remove usage of dfsql module (modin-project#6550)

Signed-off-by: Anatoly Myachev <anatoly.myachev@intel.com>
Co-authored-by: Vasily Litvinov <fam1ly.n4me@yandex.ru>

Author: anmyachev

.github/workflows/ci.yml

@@ -378,7 +378,7 @@ jobs:
       # these variables to test writing to the mock s3 filesystem.
       - run: mpiexec -n 1 -genv AWS_ACCESS_KEY_ID foobar_key -genv AWS_SECRET_ACCESS_KEY foobar_secret python -m pytest modin/pandas/test/test_io.py --verbose
       - run: mpiexec -n 1 python -m pytest modin/experimental/pandas/test/test_io_exp.py
-      - run: pip install "dfsql>=0.4.2" "pyparsing<=2.4.7" && mpiexec -n 1 python -m pytest modin/experimental/sql/test/test_sql.py
+      - run: mpiexec -n 1 python -m pytest modin/experimental/sql/test/test_sql.py
       - run: mpiexec -n 1 python -m pytest modin/test/interchange/dataframe_protocol/test_general.py
       - run: mpiexec -n 1 python -m pytest modin/test/interchange/dataframe_protocol/pandas/test_protocol.py
       - run: |
@@ -499,7 +499,7 @@ jobs:
         if: matrix.engine == 'python' || matrix.test_task == 'group_4'
       - run: python -m pytest modin/experimental/pandas/test/test_io_exp.py
         if: matrix.engine == 'python' || matrix.test_task == 'group_4'
-      - run: pip install "dfsql>=0.4.2" "pyparsing<=2.4.7" && python -m pytest modin/experimental/sql/test/test_sql.py
+      - run: python -m pytest modin/experimental/sql/test/test_sql.py
         if: matrix.os == 'ubuntu' && (matrix.engine == 'python' || matrix.test_task == 'group_4')
       - run: python -m pytest modin/test/interchange/dataframe_protocol/test_general.py
         if: matrix.engine == 'python' || matrix.test_task == 'group_4'

docs/usage_guide/advanced_usage/modin_sql.rst

@@ -7,10 +7,9 @@ due to the architecture of Modin. Currently, Modin has a query compiler that act
 intermediate layer between the query language (e.g. SQL, pandas) and the execution
 (See :doc:`architecture </development/architecture>` documentation for details).
 
-To execute SQL queries, Modin uses either dfsql third-party library or, in case of HDK
-engine (See :doc:`Using HDK </development/using_hdk>` documentation for details)
-the queries are executed directly by HDK. Thus, to execute SQL queries, either dfsql
-or pyhdk module must be installed.
+To execute SQL queries, Modin uses HDK engine
+(See :doc:`Using HDK </development/using_hdk>` documentation for details)
+Thus, to execute SQL queries, pyhdk module must be installed.
 
 
 A Short Example Using the Google Play Store
@@ -42,10 +41,6 @@ App, Category, and Rating, where Price is ‘0’.
     # you can also ignore the FROM part in the query string:
     sql_str = "SELECT App, Category, Rating WHERE Price = '0' "
 
-    # DataFrame.sql() can take query strings without FROM statement
-    # NOTE: this method required the dfsql module to be installed!
-    result_df = gstore_apps_df.sql(sql_str)
-
 Writing Complex Queries
 """""""""""""""""""""""
 

modin/experimental/sql/__init__.py

@@ -20,7 +20,7 @@
 
 def query(sql: str, *args, **kwargs) -> pd.DataFrame:
     """
-    Execute SQL query using either HDK engine or dfsql.
+    Execute SQL query using HDK engine.
 
     Parameters
     ----------
@@ -42,20 +42,6 @@ def query(sql: str, *args, **kwargs) -> pd.DataFrame:
         if cfg.StorageFormat.get() == "Hdk":
             from modin.experimental.sql.hdk.query import hdk_query as _query_impl
         else:
-            from dfsql import sql_query as _query_impl
+            raise NotImplementedError
 
     return _query_impl(sql, *args, **kwargs)
-
-
-# dfsql adds the sql() method to the DataFrame class.
-# This code is used for lazy dfsql extensions initialization.
-if not hasattr(pd.DataFrame, "sql"):
-
-    def dfsql_init(df, query):
-        delattr(pd.DataFrame, "sql")
-        import modin.experimental.sql.dfsql.query  # noqa: F401
-
-        df.sql = pd.DataFrame.sql(df)
-        return df.sql(query)
-
-    pd.DataFrame.sql = dfsql_init

modin/experimental/sql/dfsql/query.py

@@ -13,8 +13,8 @@
 
 import pandas
 import modin.pandas as pd
-import modin.config as cfg
 from modin.pandas.test.utils import default_to_pandas_ignore_string, df_equals
+from modin.config import StorageFormat
 
 import io
 import pytest
@@ -34,6 +34,10 @@
 """
 
 
+@pytest.mark.skipif(
+    StorageFormat.get() != "Hdk",
+    reason="Lack of implementation for other storage formats.",
+)
 def test_sql_query():
     from modin.experimental.sql import query
 
@@ -52,28 +56,10 @@ def test_sql_query():
     assert (values_left == values_right).all()
 
 
-def test_sql_extension():
-    # This test is for DataFrame.sql() method, that is injected by
-    # dfsql.extensions. In the HDK environment, there is no dfsql
-    # module and, thus, this test fails.
-    if cfg.StorageFormat.get() == "Hdk":
-        return
-
-    import modin.experimental.sql  # noqa: F401
-
-    df = pd.read_csv(io.StringIO(titanic_snippet))
-
-    expected_df = df[df["survived"] == 1][["passenger_id", "survived"]]
-
-    sql = "SELECT passenger_id, survived WHERE survived = 1"
-    query_result = df.sql(sql)
-    assert list(query_result.columns) == ["passenger_id", "survived"]
-    values_left = expected_df.values
-    values_right = query_result.values
-    assert values_left.shape == values_right.shape
-    assert (values_left == values_right).all()
-
-
+@pytest.mark.skipif(
+    StorageFormat.get() != "Hdk",
+    reason="Lack of implementation for other storage formats.",
+)
 def test_string_cast():
     from modin.experimental.sql import query
 

modin/experimental/sql/test/test_sql.py

@@ -10,7 +10,6 @@
 ray_deps = ["ray[default]>=1.13.0,!=2.5.0", "pyarrow>=7.0.0", "pydantic<2"]
 unidist_deps = ["unidist[mpi]>=0.2.1"]
 spreadsheet_deps = ["modin-spreadsheet>=0.1.0"]
-sql_deps = ["dfsql>=0.4.2", "pyparsing<=2.4.7"]
 all_deps = dask_deps + ray_deps + unidist_deps + spreadsheet_deps
 
 # Distribute 'modin-autoimport-pandas.pth' along with binary and source distributions.
@@ -60,7 +59,6 @@ def make_distribution(self):
         "ray": ray_deps,
         "unidist": unidist_deps,
         "spreadsheet": spreadsheet_deps,
-        "sql": sql_deps,
         "all": all_deps,
     },
     python_requires=">=3.9",