copy an optimization from CPython: when the search string of str.replace and

str.split doesn't occur in the string, don't create a copy but just reuse self.

4 files changed, 30 insertions, 2 deletions

diff --git a/pypy/objspace/std/bytesobject.py b/pypy/objspace/std/bytesobject.py
index 6315c5d6cf..2316f6e513 100644
--- a/pypy/objspace/std/bytesobject.py
+++ b/pypy/objspace/std/bytesobject.py
@@ -690,15 +690,33 @@ class W_BytesObject(W_AbstractBytesObject):
                 self_as_unicode._utf8.find(w_sub._utf8) >= 0)
         return self._StringMethods_descr_contains(space, w_sub)
 
-    _StringMethods_descr_replace = descr_replace
     @unwrap_spec(count=int)
     def descr_replace(self, space, w_old, w_new, count=-1):
+        from rpython.rlib.rstring import replace
         old_is_unicode = space.isinstance_w(w_old, space.w_unicode)
         new_is_unicode = space.isinstance_w(w_new, space.w_unicode)
         if old_is_unicode or new_is_unicode:
             self_as_uni = unicode_from_encoded_object(space, self, None, None)
             return self_as_uni.descr_replace(space, w_old, w_new, count)
-        return self._StringMethods_descr_replace(space, w_old, w_new, count)
+
+        # almost copy of StringMethods.descr_replace :-(
+        input = self._value
+
+        sub = self._op_val(space, w_old)
+        by = self._op_val(space, w_new)
+        # the following two lines are for being bug-to-bug compatible
+        # with CPython: see issue #2448
+        if count >= 0 and len(input) == 0:
+            return self._empty()
+        try:
+            res = replace(input, sub, by, count)
+        except OverflowError:
+            raise oefmt(space.w_OverflowError, "replace string is too long")
+        # difference: reuse self if no replacement was done
+        if type(self) is W_BytesObject and res is input:
+            return self
+
+        return self._new(res)
 
     _StringMethods_descr_join = descr_join
     def descr_join(self, space, w_list):
diff --git a/pypy/objspace/std/test/test_bytesobject.py b/pypy/objspace/std/test/test_bytesobject.py
index cc15f97d54..2feca7ab5e 100644
--- a/pypy/objspace/std/test/test_bytesobject.py
+++ b/pypy/objspace/std/test/test_bytesobject.py
@@ -342,6 +342,10 @@ class AppTestBytesObject:
         assert 'one'.replace(buffer('o'), buffer('n'), 1) == 'nne'
         assert 'one'.replace(buffer('o'), buffer('n')) == 'nne'
 
+    def test_replace_no_occurrence(self):
+        x = b"xyz"
+        assert x.replace(b"a", b"b") is x
+
     def test_strip(self):
         s = " a b "
         assert s.strip() == "a b"
diff --git a/pypy/objspace/std/test/test_unicodeobject.py b/pypy/objspace/std/test/test_unicodeobject.py
index 51faff763d..6b1c7315da 100644
--- a/pypy/objspace/std/test/test_unicodeobject.py
+++ b/pypy/objspace/std/test/test_unicodeobject.py
@@ -1303,3 +1303,7 @@ class AppTestUnicodeString:
 
     def test_newlist_utf8_non_ascii(self):
         'ä'.split("\n")[0] # does not crash
+
+    def test_replace_no_occurrence(self):
+        x = u"xyz"
+        assert x.replace(u"a", u"b") is x
diff --git a/pypy/objspace/std/unicodeobject.py b/pypy/objspace/std/unicodeobject.py
index 1dcd415912..1b7a8d07b4 100644
--- a/pypy/objspace/std/unicodeobject.py
+++ b/pypy/objspace/std/unicodeobject.py
@@ -880,6 +880,8 @@ class W_UnicodeObject(W_Root):
                                               count, isutf8=True)
         except OverflowError:
             raise oefmt(space.w_OverflowError, "replace string is too long")
+        if type(self) is W_UnicodeObject and replacements == 0:
+            return self
 
         newlength = self._length + replacements * (w_by._length - w_sub._length)
         return W_UnicodeObject(res, newlength)