Merge branch 'master' into dfat

AlexanderKalistratov · Apr 23, 2020 · Mar 19, 2020 · Mar 20, 2020 · Mar 20, 2020 · Mar 26, 2020
commit 928dae88ea4f7f3c7dd53ec0dd777d88786c6d2f
diff --git a/sdc/datatypes/hpat_pandas_dataframe_functions.py b/sdc/datatypes/hpat_pandas_dataframe_functions.py
@@ -1919,6 +1919,140 @@ def _df_getitem_tuple_at_impl(self, idx):
     return func_text, global_vars
 
 
+def df_getitem_int_iloc_codegen(self, idx):
+    """
+    Example of generated implementation:
+        def _df_getitem_int_iloc_impl(self, idx):
+        if -1 < idx < len(self._dataframe.index):
+            data_0 = pandas.Series(self._dataframe._data[0])
+            result_0 = data_0.iat[idx]
+            data_1 = pandas.Series(self._dataframe._data[1])
+            result_1 = data_1.iat[idx]
+            return pandas.Series(data=[result_0, result_1], index=['A', 'B'], name=str(idx))
+        raise IndexingError('Index is out of bounds for axis')
+    """
+    func_lines = ['def _df_getitem_int_iloc_impl(self, idx):',
+                  '  if -1 < idx < len(self._dataframe.index):']
+    results = []
+    index = []
+    name = 'self._dataframe._index[idx]'
+    if isinstance(self.index, types.NoneType):
+        name = 'idx'
+    for i, c in enumerate(self.columns):
+        result_c = f"result_{i}"
+        func_lines += [f"    data_{i} = pandas.Series(self._dataframe._data[{i}])",
+                       f"    {result_c} = data_{i}.iat[idx]"]
+        results.append(result_c)
+        index.append(c)
+    data = ', '.join(col for col in results)
+    func_lines += [f"    return pandas.Series(data=[{data}], index={index}, name=str({name}))",
+                   f"  raise IndexingError('Index is out of bounds for axis')"]
+
+    func_text = '\n'.join(func_lines)
+    global_vars = {'pandas': pandas, 'numpy': numpy, 'IndexingError': IndexingError}
+
+    return func_text, global_vars
+
+
+def df_getitem_slice_iloc_codegen(self, idx):
+    """
+    Example of generated implementation:
+        def _df_getitem_slice_iloc_impl(self, idx):
+            data_0 = pandas.Series(self._dataframe._data[0])
+            result_0 = data_0.iloc[idx]
+            data_1 = pandas.Series(self._dataframe._data[1])
+            result_1 = data_1.iloc[idx]
+            return pandas.DataFrame(data={"A": result_0, "B": result_1}, index=self._dataframe.index[idx])
+    """
+    func_lines = ['def _df_getitem_slice_iloc_impl(self, idx):']
+    results = []
+    for i, c in enumerate(self.columns):
+        result_c = f"result_{i}"
+        func_lines += [f"  data_{i} = pandas.Series(self._dataframe._data[{i}])",
+                       f"  {result_c} = data_{i}.iloc[idx]"]
+        results.append((c, result_c))
+    data = ', '.join(f'"{col}": {data}' for col, data in results)
+    func_lines += [f"  return pandas.DataFrame(data={{{data}}}, index=self._dataframe.index[idx])"]
+
+    func_text = '\n'.join(func_lines)
+    global_vars = {'pandas': pandas, 'numpy': numpy}
+
+    return func_text, global_vars
+
+
+def df_getitem_list_iloc_codegen(self, idx):
+    """
+    Example of generated implementation:
+        def _df_getitem_list_iloc_impl(self, idx):
+            check_idx = False
+            for i in idx:
+                if -1 < i < len(self._dataframe.index):
+                    check_idx = True
+            if check_idx == True:
+                data_0 = pandas.Series(self._dataframe._data[0])
+                result_0 = data_0.iloc[numpy.array(idx)]
+                data_1 = pandas.Series(self._dataframe._data[1])
+                result_1 = data_1.iloc[numpy.array(idx)]
+                return pandas.DataFrame(data={"A": result_0, "B": result_1}, index=idx)
+            raise IndexingError('Index is out of bounds for axis')
+    """
+    func_lines = ['def _df_getitem_list_iloc_impl(self, idx):',
+                  '  check_idx = False',
+                  '  for i in idx:',
+                  '    if -1 < i < len(self._dataframe.index):',
+                  '      check_idx = True',
+                  '  if check_idx == True:']
+    results = []
+    index = '[self._dataframe._index[i] for i in idx]'
+    if isinstance(self.index, types.NoneType):
+        index = 'idx'
+    for i, c in enumerate(self.columns):
+        result_c = f"result_{i}"
+        func_lines += [f"    data_{i} = pandas.Series(self._dataframe._data[{i}])",
+                       f"    {result_c} = data_{i}.iloc[numpy.array(idx)]"]
+        results.append((c, result_c))
+    data = ', '.join(f'"{col}": {data}' for col, data in results)
+    func_lines += [f"    return pandas.DataFrame(data={{{data}}}, index={index})",
+                   f"  raise IndexingError('Index is out of bounds for axis')"]
+
+    func_text = '\n'.join(func_lines)
+    global_vars = {'pandas': pandas, 'numpy': numpy, 'IndexingError': IndexingError}
+
+    return func_text, global_vars
+
+
+def df_getitem_list_bool_iloc_codegen(self, idx):
+    """
+    Example of generated implementation:
+        def _df_getitem_list_bool_iloc_impl(self, idx):
+            if len(self._dataframe.index) == len(idx):
+                data_0 = self._dataframe._data[0]
+                result_0 = pandas.Series(data_0[numpy.array(idx)])
+                data_1 = self._dataframe._data[1]
+                result_1 = pandas.Series(data_1[numpy.array(idx)])
+                return pandas.DataFrame(data={"A": result_0, "B": result_1},
+                    index=self._dataframe.index[numpy.array(idx)])
+            raise IndexingError('Item wrong length')
+    """
+    func_lines = ['def _df_getitem_list_bool_iloc_impl(self, idx):']
+    results = []
+    index = 'self._dataframe.index[numpy.array(idx)]'
+    func_lines += ['  if len(self._dataframe.index) == len(idx):']
+    for i, c in enumerate(self.columns):
+        result_c = f"result_{i}"
+        func_lines += [f"    data_{i} = self._dataframe._data[{i}]",
+                       f"    {result_c} = pandas.Series(data_{i}[numpy.array(idx)])"]
+        results.append((c, result_c))
+    data = ', '.join(f'"{col}": {data}' for col, data in results)
+    func_lines += [f"    return pandas.DataFrame(data={{{data}}}, index={index})",
+                   f"  raise IndexingError('Item wrong length')"]
+
+    func_text = '\n'.join(func_lines)
+    global_vars = {'pandas': pandas, 'numpy': numpy, 'IndexingError': IndexingError}
+
+    return func_text, global_vars
+
+
 def gen_df_getitem_tuple_at_impl(self, row, col):
     func_text, global_vars = df_getitem_tuple_at_codegen(self, row, col)
     loc_vars = {}
@@ -1928,6 +2062,19 @@ def gen_df_getitem_tuple_at_impl(self, row, col):
     return _reduce_impl
 
 
+gen_df_getitem_iloc_int_impl = gen_impl_generator(
+    df_getitem_int_iloc_codegen, '_df_getitem_int_iloc_impl')
+
+gen_df_getitem_iloc_slice_impl = gen_impl_generator(
+    df_getitem_slice_iloc_codegen, '_df_getitem_slice_iloc_impl')
+
+gen_df_getitem_iloc_list_impl = gen_impl_generator(
+    df_getitem_list_iloc_codegen, '_df_getitem_list_iloc_impl')
+
+gen_df_getitem_iloc_list_bool_impl = gen_impl_generator(
+    df_getitem_list_bool_iloc_codegen, '_df_getitem_list_bool_iloc_impl')
+
+
 @sdc_overload(operator.getitem)
 def sdc_pandas_dataframe_accessor_getitem(self, idx):
     if not isinstance(self, DataFrameGetitemAccessorType):