Merge pull request #8978 from jreback/unicode

TST/COMPAT: tests and compat for unicode for lib.max_len_string_array

Author: jreback

pandas/lib.pyx

@@ -898,17 +898,17 @@ def clean_index_list(list obj):
 
 @cython.boundscheck(False)
 @cython.wraparound(False)
-def max_len_string_array(ndarray[object, ndim=1] arr):
+def max_len_string_array(ndarray arr):
     """ return the maximum size of elements in a 1-dim string array """
     cdef:
         int i, m, l
-        length = arr.shape[0]
+        int length = arr.shape[0]
         object v
 
     m = 0
     for i from 0 <= i < length:
         v = arr[i]
-        if PyString_Check(v) or PyBytes_Check(v):
+        if PyString_Check(v) or PyBytes_Check(v) or PyUnicode_Check(v):
             l = len(v)
 
             if l > m:

pandas/tests/test_lib.py

@@ -3,10 +3,21 @@
 import numpy as np
 
 import pandas as pd
-from pandas.lib import isscalar, item_from_zerodim
+from pandas.lib import isscalar, item_from_zerodim, max_len_string_array
 import pandas.util.testing as tm
 from pandas.compat import u
 
+class TestMisc(tm.TestCase):
+
+    def test_max_len_string_array(self):
+
+        arr = np.array(['foo','b',np.nan],dtype='object')
+        self.assertTrue(max_len_string_array(arr),3)
+
+        # unicode
+        arr = arr.astype('U')
+        self.assertTrue(max_len_string_array(arr),3)
+
 class TestIsscalar(tm.TestCase):
     def test_isscalar_builtin_scalars(self):
         self.assertTrue(isscalar(None))