changeset:   100635:62e3b7af0697
user:        Victor Stinner <victor.stinner@gmail.com>
date:        Mon Mar 21 10:38:58 2016 +0100
files:       Doc/whatsnew/3.6.rst Misc/ACKS Misc/NEWS Objects/bytearrayobject.c Objects/bytesobject.c
description:
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.


diff -r ea9efa06c137 -r 62e3b7af0697 Doc/whatsnew/3.6.rst
--- a/Doc/whatsnew/3.6.rst	Fri Mar 18 21:52:22 2016 +0100
+++ b/Doc/whatsnew/3.6.rst	Mon Mar 21 10:38:58 2016 +0100
@@ -339,6 +339,9 @@
 * 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
 =======================
diff -r ea9efa06c137 -r 62e3b7af0697 Misc/ACKS
--- a/Misc/ACKS	Fri Mar 18 21:52:22 2016 +0100
+++ b/Misc/ACKS	Mon Mar 21 10:38:58 2016 +0100
@@ -1376,6 +1376,7 @@
 Roy Smith
 Ryan Smith-Roberts
 Rafal Smotrzyk
+Josh Snider
 Eric Snow
 Dirk Soede
 Nir Soffer
diff -r ea9efa06c137 -r 62e3b7af0697 Misc/NEWS
--- a/Misc/NEWS	Fri Mar 18 21:52:22 2016 +0100
+++ b/Misc/NEWS	Mon Mar 21 10:38:58 2016 +0100
@@ -10,6 +10,9 @@
 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.
 
diff -r ea9efa06c137 -r 62e3b7af0697 Objects/bytearrayobject.c
--- a/Objects/bytearrayobject.c	Fri Mar 18 21:52:22 2016 +0100
+++ b/Objects/bytearrayobject.c	Mon Mar 21 10:38:58 2016 +0100
@@ -1705,17 +1705,27 @@
     self_s = PyByteArray_AS_STRING(self);
     result_s = PyByteArray_AS_STRING(result);
 
-    /* TODO: special case single character, which doesn't need memcpy */
-
-    /* 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++) {
-        *result_s++ = *self_s++;
+    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++) {
+            *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 */
diff -r ea9efa06c137 -r 62e3b7af0697 Objects/bytesobject.c
--- a/Objects/bytesobject.c	Fri Mar 18 21:52:22 2016 +0100
+++ b/Objects/bytesobject.c	Mon Mar 21 10:38:58 2016 +0100
@@ -2464,17 +2464,27 @@
     self_s = PyBytes_AS_STRING(self);
     result_s = PyBytes_AS_STRING(result);
 
-    /* TODO: special case single character, which doesn't need memcpy */
-
-    /* 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++) {
-        *result_s++ = *self_s++;
+    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++) {
+            *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 */