feat(range): improve hashing function

tree-sitter · May 11, 2024 · ce904b9 · ce904b9
1 parent a3ce31b
commit ce904b9
Showing 1 changed file with 9 additions and 34 deletions.
diff --git a/tree_sitter/binding/range.c b/tree_sitter/binding/range.c
@@ -1,5 +1,7 @@
 #include "range.h"
 
+#define HASH_COMBINE_CONSTANT (sizeof(Py_hash_t) >= 8 ? 0x517cc1b727220a95 : 0x9e3779b9)
+
 int range_init(Range *self, PyObject *args, PyObject *kwargs) {
     uint32_t start_row, start_col, end_row, end_col, start_byte, end_byte;
     char *keywords[] = {
@@ -32,42 +34,15 @@ PyObject *range_repr(Range *self) {
 }
 
 Py_hash_t range_hash(Range *self) {
-    // FIXME: replace with an efficient integer hashing algorithm
-    PyObject *row_tuple = PyTuple_Pack(2, PyLong_FromSize_t(self->range.start_point.row),
-                                       PyLong_FromLong(self->range.end_point.row));
-    if (!row_tuple) {
-        return NULL;
-    }
-
-    PyObject *col_tuple = PyTuple_Pack(2, PyLong_FromSize_t(self->range.start_point.column),
-                                       PyLong_FromSize_t(self->range.end_point.column));
-    if (!col_tuple) {
-        Py_DECREF(row_tuple);
-        return NULL;
-    }
-
-    PyObject *bytes_tuple = PyTuple_Pack(2, PyLong_FromSize_t(self->range.start_byte),
-                                         PyLong_FromSize_t(self->range.end_byte));
-    if (!bytes_tuple) {
-        Py_DECREF(row_tuple);
-        Py_DECREF(col_tuple);
-        return NULL;
-    }
-
-    PyObject *range_tuple = PyTuple_Pack(3, row_tuple, col_tuple, bytes_tuple);
-    if (!range_tuple) {
-        Py_DECREF(row_tuple);
-        Py_DECREF(col_tuple);
-        Py_DECREF(bytes_tuple);
-        return NULL;
-    }
+    Py_hash_t hash = 6581;
 
-    Py_hash_t hash = PyObject_Hash(range_tuple);
+    hash ^= self->range.start_point.row + HASH_COMBINE_CONSTANT + (hash << 6) + (hash >> 2);
+    hash ^= self->range.start_point.column + HASH_COMBINE_CONSTANT + (hash << 6) + (hash >> 2);
+    hash ^= self->range.start_byte + HASH_COMBINE_CONSTANT + (hash << 6) + (hash >> 2);
+    hash ^= self->range.end_point.row + HASH_COMBINE_CONSTANT + (hash << 6) + (hash >> 2);
+    hash ^= self->range.end_point.column + HASH_COMBINE_CONSTANT + (hash << 6) + (hash >> 2);
+    hash ^= self->range.end_byte + HASH_COMBINE_CONSTANT + (hash << 6) + (hash >> 2);
 
-    Py_DECREF(range_tuple);
-    Py_DECREF(row_tuple);
-    Py_DECREF(col_tuple);
-    Py_DECREF(bytes_tuple);
     return hash;
 }