MAINT/BUG: Default inplace to False in pd.eval

Deprecated in 0.18.0.

xref gh-11149.

Also patches bug where we were improperly handling
the inplace=False condition, as we were assuming
that target input was non-None when that wasn't
necessarily enforced.

Author: gfyoung

doc/source/whatsnew/v0.21.0.txt

@@ -76,6 +76,7 @@ Removal of prior version deprecations/changes
 
 - ``pd.read_excel()`` has dropped the ``has_index_names`` parameter (:issue:`10967`)
 - ``Categorical`` has dropped the ``.order()`` and ``.sort()`` methods in favor of ``.sort_values()`` (:issue:`12882`)
+- ``pd.eval`` and ``DataFrame.eval`` have changed the default of ``inplace`` from ``None`` to ``False`` (:issue:`11149`)
 
 
 .. _whatsnew_0210.performance:
@@ -138,3 +139,4 @@ Categorical
 
 Other
 ^^^^^
+- Bug in ``pd.eval()`` where the ``inplace`` parameter was being incorrectly handled (:issue:`16732`)

pandas/core/computation/eval.py

@@ -3,7 +3,6 @@
 """Top level ``eval`` module.
 """
 
-import warnings
 import tokenize
 from pandas.io.formats.printing import pprint_thing
 from pandas.core.computation import _NUMEXPR_INSTALLED
@@ -148,7 +147,7 @@ def _check_for_locals(expr, stack_level, parser):
 
 def eval(expr, parser='pandas', engine=None, truediv=True,
          local_dict=None, global_dict=None, resolvers=(), level=0,
-         target=None, inplace=None):
+         target=None, inplace=False):
     """Evaluate a Python expression as a string using various backends.
 
     The following arithmetic operations are supported: ``+``, ``-``, ``*``,
@@ -207,18 +206,24 @@ def eval(expr, parser='pandas', engine=None, truediv=True,
         scope. Most users will **not** need to change this parameter.
     target : a target object for assignment, optional, default is None
         essentially this is a passed in resolver
-    inplace : bool, default True
-        If expression mutates, whether to modify object inplace or return
-        copy with mutation.
+    inplace : bool, default False
+        If `target` is provided, and the expression mutates `target`, whether
+        to modify `target` inplace. Otherwise, return a copy of `target` with
+        the mutation.
 
-        WARNING: inplace=None currently falls back to to True, but
-        in a future version, will default to False.  Use inplace=True
-        explicitly rather than relying on the default.
+        If `inplace=True`, but `target` cannot be modified inplace, a
+        ValueError will be raised. Examples of targets that cannot be
+        modified inplace are integers and strings. Examples of targets
+        that can be modified inplace are lists and class instances.
 
     Returns
     -------
     ndarray, numeric scalar, DataFrame, Series
 
+    Raises
+    ------
+    ValueError : `inplace=True`, but the provided `target` could not be
+                 modified inplace.
     Notes
     -----
     The ``dtype`` of any objects involved in an arithmetic ``%`` operation are
@@ -232,8 +237,20 @@ def eval(expr, parser='pandas', engine=None, truediv=True,
     pandas.DataFrame.query
     pandas.DataFrame.eval
     """
+
     inplace = validate_bool_kwarg(inplace, 'inplace')
-    first_expr = True
+
+    # "Primitives" cannot be modified inplace because
+    # they have no attributes that we can modify.
+    #
+    # The only exceptions are lists and dictionaries,
+    # whose elements can be modified inplace.
+    target_type = type(target).__name__
+    modifiable = hasattr(target, "__dict__") or target_type in ('list', 'dict')
+
+    if inplace and not modifiable:
+        raise ValueError("Cannot modify the provided target inplace")
+
     if isinstance(expr, string_types):
         _check_expression(expr)
         exprs = [e.strip() for e in expr.splitlines() if e.strip() != '']
@@ -245,7 +262,9 @@ def eval(expr, parser='pandas', engine=None, truediv=True,
         raise ValueError("multi-line expressions are only valid in the "
                          "context of data, use DataFrame.eval")
 
+    ret = None
     first_expr = True
+
     for expr in exprs:
         expr = _convert_expression(expr)
         engine = _check_engine(engine)
@@ -272,14 +291,6 @@ def eval(expr, parser='pandas', engine=None, truediv=True,
 
         # assign if needed
         if env.target is not None and parsed_expr.assigner is not None:
-            if inplace is None:
-                warnings.warn(
-                    "eval expressions containing an assignment currently"
-                    "default to operating inplace.\nThis will change in "
-                    "a future version of pandas, use inplace=True to "
-                    "avoid this warning.",
-                    FutureWarning, stacklevel=3)
-                inplace = True
 
             # if returning a copy, copy only on the first assignment
             if not inplace and first_expr:
@@ -304,7 +315,6 @@ def eval(expr, parser='pandas', engine=None, truediv=True,
             ret = None
             first_expr = False
 
-    if not inplace and inplace is not None:
-        return target
-
-    return ret
+    # We want to exclude `inplace=None` as being False.
+    if inplace is False:
+        return target or ret

pandas/core/frame.py

@@ -2224,21 +2224,18 @@ def query(self, expr, inplace=False, **kwargs):
         else:
             return new_data
 
-    def eval(self, expr, inplace=None, **kwargs):
+    def eval(self, expr, inplace=False, **kwargs):
         """Evaluate an expression in the context of the calling DataFrame
         instance.
 
         Parameters
         ----------
         expr : string
             The expression string to evaluate.
-        inplace : bool
-            If the expression contains an assignment, whether to return a new
-            DataFrame or mutate the existing.
-
-            WARNING: inplace=None currently falls back to to True, but
-            in a future version, will default to False.  Use inplace=True
-            explicitly rather than relying on the default.
+        inplace : bool, default False
+            If the expression contains an assignment, whether to perform the
+            operation inplace and mutate the existing DataFrame. Otherwise,
+            a new DataFrame is returned.
 
             .. versionadded:: 0.18.0
 

pandas/tests/computation/test_eval.py

@@ -1311,14 +1311,6 @@ def assignment_not_inplace(self):
         expected['c'] = expected['a'] + expected['b']
         tm.assert_frame_equal(df, expected)
 
-        # Default for inplace will change
-        with tm.assert_produces_warnings(FutureWarning):
-            df.eval('c = a + b')
-
-        # but don't warn without assignment
-        with tm.assert_produces_warnings(None):
-            df.eval('a + b')
-
     def test_multi_line_expression(self):
         # GH 11149
         df = pd.DataFrame({'a': [1, 2, 3], 'b': [4, 5, 6]})