reimplemented ucs4_get_end_p to go from end

Author: flexatone

arraymap.c

@@ -123,16 +123,37 @@ typedef enum ViewKind{
 
 
 // NOTE: would like to use strchr(str, '\0') instead of this routine, but some buffers might not have a null terminator and stread by full to the the dt_size.
+// static inline Py_UCS4*
+// ucs4_get_end_p(Py_UCS4* p, Py_ssize_t dt_size) {
+//     Py_UCS4* p_end = p + dt_size; // invalid
+//     while (p < p_end && *p != '\0') {
+//         p++;
+//     }
+//     return p; // this might point to null, or point to one after end
+// }
+
+// 5 start, length of 4, valid values are (5, 6, 7, 8)
+// last valid is start + len - 1
+
+// Return a pointer to the last valid character in the field.
 static inline Py_UCS4*
-ucs4_get_end_p(Py_UCS4* p, Py_ssize_t dt_size) {
-    Py_UCS4* p_end = p + dt_size;
-    while (p < p_end && *p != '\0') {
-        p++;
+ucs4_get_end_p(Py_UCS4* p_start, Py_ssize_t dt_size) {
+    Py_UCS4* p = p_start + dt_size - 1;
+    while (p > p_start) {
+        if (*p != '\0') {
+            return p + 1;
+        }
+        p--;
+    }
+    // p is start; it may or may not be null; if its not null, return 1 more
+    if (*p != '\0') {
+        return p + 1;
     }
     return p;
 }
 
 
+
 static inline char*
 char_get_end_p(char* p, Py_ssize_t dt_size) {
     char* p_end = p + dt_size;

test/test_unit.py

@@ -145,6 +145,17 @@ def test_fam_constructor_array_unicode_d():
     assert list(fam) == ["", "\x000"]
 
 
+# >>> u = "\x000\x00"
+# >>> len(u)
+# 3
+# >>> a1 = np.array(['', ''], dtype='U4')
+# >>> a1[0] = u
+# >>> a1
+# array(['\x000', ''], dtype='<U4')
+# >>> len(a1[0])
+# 2
+
+
 def test_fam_copy_array_unicode_a():
     a1 = np.array(("a", "ccc", "bb"))
     a1.flags.writeable = False