Merge pull request #2855 from jreback/pytables_bug

BUG: proper conversion of string to bool HDFStore selection

Author: jreback

pandas/io/pytables.py

@@ -250,10 +250,14 @@ def __repr__(self):
             values = []
 
             for k in self.keys():
-                s = self.get_storer(k)
-                if s is not None:
-                    keys.append(str(s.pathname or k))
-                    values.append(str(s or 'invalid_HDFStore node'))
+                try:
+                    s = self.get_storer(k)
+                    if s is not None:
+                        keys.append(str(s.pathname or k))
+                        values.append(str(s or 'invalid_HDFStore node'))
+                except (Exception), detail:
+                    keys.append(k)
+                    values.append("[invalid_HDFStore node: %s]" % str(detail))
 
             output += adjoin(12, keys, values)
         else:
@@ -3060,7 +3064,10 @@ def convert_value(self, v):
             v = float(v)
             return [v, v]
         elif self.kind == 'bool':
-            v = bool(v)
+            if isinstance(v, basestring):
+                v = not str(v).strip().lower() in ["false", "f", "no", "n", "none", "0", "[]", "{}", ""]
+            else:
+                v = bool(v)
             return [v, v]
         elif not isinstance(v, basestring):
             return [str(v), None]

pandas/io/tests/test_pytables.py

@@ -1649,15 +1649,20 @@ def test_select_dtypes(self):
             df['bool'] = df['A'] > 0
             store.remove('df')
             store.append('df', df, data_columns = True)
-            result = store.select('df', Term('bool == True'), columns = ['A','bool'])
+
             expected = df[df.bool == True].reindex(columns=['A','bool'])
-            tm.assert_frame_equal(expected, result)
+            for v in [True,'true',1]:
+                result = store.select('df', Term('bool == %s' % str(v)), columns = ['A','bool'])
+                tm.assert_frame_equal(expected, result)
 
-            result = store.select('df', Term('bool == 1'), columns = ['A','bool'])
-            tm.assert_frame_equal(expected, result)
+            expected = df[df.bool == False ].reindex(columns=['A','bool'])
+            for v in [False,'false',0]:
+                result = store.select('df', Term('bool == %s' % str(v)), columns = ['A','bool'])
+                tm.assert_frame_equal(expected, result)
 
             # integer index
             df = DataFrame(dict(A=np.random.rand(20), B=np.random.rand(20)))
+            store.remove('df_int')
             store.append('df_int', df)
             result = store.select(
                 'df_int', [Term("index<10"), Term("columns", "=", ["A"])])
@@ -1667,6 +1672,7 @@ def test_select_dtypes(self):
             # float index
             df = DataFrame(dict(A=np.random.rand(
                         20), B=np.random.rand(20), index=np.arange(20, dtype='f8')))
+            store.remove('df_float')
             store.append('df_float', df)
             result = store.select(
                 'df_float', [Term("index<10.0"), Term("columns", "=", ["A"])])