Optimize bytes.replace(b'', b'.')

Issue #26574: Optimize bytes.replace(b'', b'.') and
bytearray.replace(b'', b'.'): up to 80% faster. Patch written by Josh Snider.

Doc/whatsnew/3.6.rst

@@ -339,6 +339,9 @@ Optimizations
 * Optimize :meth:`bytes.fromhex` and :meth:`bytearray.fromhex`: they are now
   between 2x and 3.5x faster. (Contributed by Victor Stinner in :issue:`25401`).
 
+* Optimize ``bytes.replace(b'', b'.')`` and ``bytearray.replace(b'', b'.')``:
+  up to 80% faster. (Contributed by Josh Snider in :issue:`26574`).
+
 
 Build and C API Changes
 =======================

Misc/ACKS

@@ -1376,6 +1376,7 @@ Mark Smith
 Roy Smith
 Ryan Smith-Roberts
 Rafal Smotrzyk
+Josh Snider
 Eric Snow
 Dirk Soede
 Nir Soffer

Misc/NEWS

@@ -10,6 +10,9 @@ Release date: tba
 Core and Builtins
 -----------------
 
+- Issue #26574: Optimize ``bytes.replace(b'', b'.')`` and
+  ``bytearray.replace(b'', b'.')``. Patch written by Josh Snider.
+
 - Issue #26581: If coding cookie is specified multiple times on a line in
   Python source code file, only the first one is taken to account.
 

Objects/bytearrayobject.c

@@ -1705,18 +1705,28 @@ replace_interleave(PyByteArrayObject *self,
     self_s = PyByteArray_AS_STRING(self);
     result_s = PyByteArray_AS_STRING(result);
 
-    /* TODO: special case single character, which doesn't need memcpy */
-
+    if (to_len > 1) {
         /* Lay the first one down (guaranteed this will occur) */
         Py_MEMCPY(result_s, to_s, to_len);
         result_s += to_len;
         count -= 1;
 
-    for (i=0; i<count; i++) {
+        for (i = 0; i < count; i++) {
             *result_s++ = *self_s++;
             Py_MEMCPY(result_s, to_s, to_len);
             result_s += to_len;
         }
+    }
+    else {
+        result_s[0] = to_s[0];
+        result_s += to_len;
+        count -= 1;
+        for (i = 0; i < count; i++) {
+            *result_s++ = *self_s++;
+            result_s[0] = to_s[0];
+            result_s += to_len;
+        }
+    }
 
     /* Copy the rest of the original string */
     Py_MEMCPY(result_s, self_s, self_len-i);

Objects/bytesobject.c

@@ -2464,18 +2464,28 @@ replace_interleave(PyBytesObject *self,
     self_s = PyBytes_AS_STRING(self);
     result_s = PyBytes_AS_STRING(result);
 
-    /* TODO: special case single character, which doesn't need memcpy */
-
+    if (to_len > 1) {
         /* Lay the first one down (guaranteed this will occur) */
         Py_MEMCPY(result_s, to_s, to_len);
         result_s += to_len;
         count -= 1;
 
-    for (i=0; i<count; i++) {
+        for (i = 0; i < count; i++) {
             *result_s++ = *self_s++;
             Py_MEMCPY(result_s, to_s, to_len);
             result_s += to_len;
         }
+    }
+    else {
+        result_s[0] = to_s[0];
+        result_s += to_len;
+        count -= 1;
+        for (i = 0; i < count; i++) {
+            *result_s++ = *self_s++;
+            result_s[0] = to_s[0];
+            result_s += to_len;
+        }
+    }
 
     /* Copy the rest of the original string */
     Py_MEMCPY(result_s, self_s, self_len-i);