FIX-#2408: Fix read_csv and read_table args when used inside a decora… (#2486)

Signed-off-by: Weiwen Gu <[email protected]>

Author: staftermath

modin/pandas/io.py

@@ -103,9 +103,15 @@ def parser_func(
         memory_map=False,
         float_precision=None,
     ):
-        _, _, _, kwargs = inspect.getargvalues(inspect.currentframe())
-        if kwargs.get("sep", sep) is False:
-            kwargs["sep"] = "\t"
+        # ISSUE #2408: parse parameter shared with pandas read_csv and read_table and update with provided args
+        _pd_read_csv_signature = {
+            val.name for val in inspect.signature(pandas.read_csv).parameters.values()
+        }
+        _, _, _, f_locals = inspect.getargvalues(inspect.currentframe())
+        if f_locals.get("sep", sep) is False:
+            f_locals["sep"] = "\t"
+
+        kwargs = {k: v for k, v in f_locals.items() if k in _pd_read_csv_signature}
         return _read(**kwargs)
 
     return parser_func

modin/pandas/test/test_io.py

@@ -43,6 +43,7 @@
     IO_OPS_DATA_DIR,
     io_ops_bad_exc,
     eval_io_from_str,
+    dummy_decorator,
 )
 
 from modin.config import Engine, Backend, IsExperimental
@@ -1460,6 +1461,28 @@ def test_from_sas():
     df_equals(modin_df, pandas_df)
 
 
+def test_from_csv_within_decorator(make_csv_file):
+    make_csv_file()
+
+    @dummy_decorator()
+    def wrapped_read_csv(file, method):
+        if method == "pandas":
+            return pandas.read_csv(file)
+
+        if method == "modin":
+            return pd.read_csv(file)
+
+    pandas_df = wrapped_read_csv(TEST_CSV_FILENAME, method="pandas")
+    modin_df = wrapped_read_csv(TEST_CSV_FILENAME, method="modin")
+
+    df_equals(modin_df, pandas_df)
+
+    pandas_df = wrapped_read_csv(Path(TEST_CSV_FILENAME), method="pandas")
+    modin_df = wrapped_read_csv(Path(TEST_CSV_FILENAME), method="modin")
+
+    df_equals(modin_df, pandas_df)
+
+
 @pytest.mark.parametrize("nrows", [35, None])
 def test_from_csv_sep_none(make_csv_file, nrows):
     make_csv_file()
@@ -1642,6 +1665,28 @@ def test_from_table(make_csv_file):
     df_equals(modin_df, pandas_df)
 
 
+def test_from_table_within_decorator(make_csv_file):
+    make_csv_file(delimiter="\t")
+
+    @dummy_decorator()
+    def wrapped_read_table(file, method):
+        if method == "pandas":
+            return pandas.read_table(file)
+
+        if method == "modin":
+            return pd.read_table(file)
+
+    pandas_df = wrapped_read_table(TEST_CSV_FILENAME, method="pandas")
+    modin_df = wrapped_read_table(TEST_CSV_FILENAME, method="modin")
+
+    df_equals(modin_df, pandas_df)
+
+    pandas_df = wrapped_read_table(Path(TEST_CSV_FILENAME), method="pandas")
+    modin_df = wrapped_read_table(Path(TEST_CSV_FILENAME), method="modin")
+
+    df_equals(modin_df, pandas_df)
+
+
 @pytest.mark.skipif(Engine.get() == "Python", reason="Using pandas implementation")
 def test_from_csv_s3(make_csv_file):
     dataset_url = "s3://noaa-ghcn-pds/csv/1788.csv"

modin/pandas/test/utils.py

@@ -1045,3 +1045,18 @@ def check(*a, **kw):
             ), f"Unexpected open handles left for: {', '.join(item[0] for item in leaks)}"
 
     return check
+
+
+def dummy_decorator():
+    """A problematic decorator that does not use `functools.wraps`. This introduces unwanted local variables for
+    inspect.currentframe. This decorator is used in test_io to test `read_csv` and `read_table`
+    """
+
+    def wrapper(method):
+        def wrapped_function(self, *args, **kwargs):
+            result = method(self, *args, **kwargs)
+            return result
+
+        return wrapped_function
+
+    return wrapper