change get_string_data signature to more pythonic

Author: anmyachev

pandas/_libs/hashtable_class_helper.pxi.in

@@ -597,7 +597,7 @@ cdef class StringHashTable(HashTable):
         cdef:
             khiter_t k
             const char *v
-        get_string_data_checked(val, &v, NULL)
+        v = get_string_data_checked(val, NULL)
 
         k = kh_get_str(self.table, v)
         if k != self.table.n_buckets:
@@ -611,7 +611,7 @@ cdef class StringHashTable(HashTable):
             int ret = 0
             const char *v
 
-        get_string_data_checked(val, &v, NULL)
+        v = get_string_data_checked(val, NULL)
 
         k = kh_put_str(self.table, v, &ret)
         self.table.keys[k] = key
@@ -634,7 +634,7 @@ cdef class StringHashTable(HashTable):
         vecs = <const char **>malloc(n * sizeof(char *))
         for i in range(n):
             val = values[i]
-            get_string_data_checked(val, &v, NULL)
+            v = get_string_data_checked(val, NULL)
             vecs[i] = v
 
         with nogil:
@@ -664,9 +664,9 @@ cdef class StringHashTable(HashTable):
             val = values[i]
 
             if isinstance(val, (str, unicode)):
-                get_string_data(val, &v, NULL)
+                v = get_string_data(val, NULL)
             else:
-                get_string_data(self.na_string_sentinel, &v, NULL)
+                v = get_string_data(self.na_string_sentinel, NULL)
             vecs[i] = v
 
         with nogil:
@@ -697,9 +697,9 @@ cdef class StringHashTable(HashTable):
             val = values[i]
 
             if isinstance(val, (str, unicode)):
-                get_string_data(val, &v, NULL)
+                v = get_string_data(val, NULL)
             else:
-                get_string_data(self.na_string_sentinel, &v, NULL)
+                v = get_string_data(self.na_string_sentinel, NULL)
             vecs[i] = v
 
         with nogil:
@@ -778,7 +778,7 @@ cdef class StringHashTable(HashTable):
                 labels[i] = na_sentinel
             else:
                 # if ignore_na is False, we also stringify NaN/None/etc.
-                get_string_data_checked(val, &v, NULL)
+                v = get_string_data_checked(val, NULL)
                 vecs[i] = v
 
         # compute

pandas/_libs/tslibs/np_datetime.pyx

@@ -177,6 +177,6 @@ cdef inline int _string_to_dts(object val, npy_datetimestruct* dts,
         Py_ssize_t length
         const char* tmp
 
-    get_string_data_checked(val, &tmp, &length)
+    tmp = get_string_data_checked(val, &length)
     return parse_iso_8601_datetime(tmp, length,
                                    dts, out_local, out_tzoffset)

pandas/_libs/tslibs/util.pxd

@@ -238,8 +238,7 @@ cdef inline bint is_nan(object val):
     return (is_float_object(val) or is_complex_object(val)) and val != val
 
 
-cdef inline bint get_string_data(object s, const char **buf,
-                                 Py_ssize_t *length):
+cdef inline const char* get_string_data(object s, Py_ssize_t *length):
     """
     Extract internal char * buffer of unicode or bytes object `s` to `buf` with
     getting length of this internal buffer saved in `length`.
@@ -255,25 +254,31 @@ cdef inline bint get_string_data(object s, const char **buf,
     Parameters
     ----------
     s      : object
-    buf    : const char**
     length : Py_ssize_t*
 
     Returns
     -------
-    bint
+    buf : const char*
     """
+    cdef:
+        const char *buf
+
     if PyUnicode_Check(s):
-        buf[0] = PyUnicode_AsUTF8AndSize(s, length)
-        return buf[0] != NULL
+        buf = PyUnicode_AsUTF8AndSize(s, length)
     if PyBytes_Check(s):
-        return PyBytes_AsStringAndSize(s, <char**>buf, length) == 0
-    return False
+        if PyBytes_AsStringAndSize(s, <char**>&buf, length) != 0:
+            return NULL
+    return buf
+
 
-cdef inline void get_string_data_checked(object s, const char **buf,
-                                         Py_ssize_t *length):
+cdef inline const char* get_string_data_checked(object s, Py_ssize_t *length):
     """
     This is a wrapper for get_string_data() that raises TypeError
     when supplied with neither unicode nor bytes object
     """
-    if not get_string_data(s, buf, length):
+    cdef:
+        const char *buf = get_string_data(s, length)
+
+    if not buf:
         PyErr_BadArgument()
+    return buf