Naive hash function

Author: Ezibenroc

pyroaring/abstract_bitmap.pxi

@@ -769,6 +769,7 @@ cdef class AbstractBitMap64:
     An efficient and light-weight ordered set of 64 bits integers.
     """
     cdef croaring.roaring64_bitmap_t* _c_bitmap
+    cdef int64_t _h_val
 
     def __cinit__(self, values=None, optimize=True, no_init=False):
         if no_init:
@@ -780,6 +781,7 @@ cdef class AbstractBitMap64:
             self._c_bitmap = croaring.roaring64_bitmap_create()
         elif isinstance(values, AbstractBitMap64):
             self._c_bitmap = croaring.roaring64_bitmap_copy((<AbstractBitMap64?>values)._c_bitmap)
+            self._h_val = (<AbstractBitMap64?>values)._h_val
         elif isinstance(values, range):
             _, (start, stop, step) = values.__reduce__()
             if step < 0:
@@ -806,6 +808,8 @@ cdef class AbstractBitMap64:
             if size > 0:
                 buff_vect = values
                 croaring.roaring64_bitmap_add_many(self._c_bitmap, size, &buff_vect[0])
+        if not isinstance(values, AbstractBitMap64):
+            self._h_val = 0
         if optimize:
             self.run_optimize()
 
@@ -891,6 +895,31 @@ cdef class AbstractBitMap64:
             raise AssertionError('range_end must not be lower than range_start')
         return croaring.roaring64_bitmap_range_cardinality(self._c_bitmap, range_start, range_end)
 
+    cdef compute_hash(self):
+        cdef int64_t h_val = 0
+        cdef uint32_t i, count, max_count=256
+        cdef croaring.roaring64_iterator_t *iterator = croaring.roaring64_iterator_create(self._c_bitmap)
+        cdef uint64_t *buff = <uint64_t*>malloc(max_count*8)
+        while True:
+            count = croaring.roaring64_iterator_read(iterator, buff, max_count)
+            i = 0
+            while i < count:
+                h_val += buff[i]
+                # TODO find a good hash formula
+                i += 1
+            if count != max_count:
+                break
+        croaring.roaring64_iterator_free(iterator)
+        free(buff)
+        if not self:
+            return -1
+        return h_val
+
+    def __hash__(self):
+        if self._h_val == 0:
+            self._h_val = self.compute_hash()
+        return self._h_val
+
     def iter_equal_or_larger(self, uint64_t val):
         """
         Iterate over items in the bitmap equal or larger than a given value.