When splitting, avoid making a copy of the string if the split doesn't find

anything (issue 1538).

diff --git a/Misc/NEWS b/Misc/NEWS
index efcb6c0f57107e63b5cce615af586c78dcb6e367..4db5c3d4ca6ecc801bd2427a987b8c07ca4a0150 100644 (file)
--- a/Misc/NEWS
+++ b/Misc/NEWS
@@ -12,6 +12,9 @@ What's New in Python 2.6 alpha 1?
 Core and builtins
 -----------------
 
+- Issue #1538: Avoid copying string in split/rsplit if the split
+  char is not found.
+
 - Issue #1553: An erroneous __length_hint__ can make list() raise a
   SystemError.
 

diff --git a/Objects/stringobject.c b/Objects/stringobject.c
index a0e2837a9478fb2224a94105f053896d39d7d4ff..2c8c98995320a65cae5cf14974c7b5706dc16398 100644 (file)
--- a/Objects/stringobject.c
+++ b/Objects/stringobject.c
@@ -1403,8 +1403,9 @@ static const char *stripformat[] = {"|O:lstrip", "|O:rstrip", "|O:strip"};
 #define RSKIP_NONSPACE(s, i)     { while (i>=0  && !isspace(Py_CHARMASK(s[i]))) i--; }
 
 Py_LOCAL_INLINE(PyObject *)
-split_whitespace(const char *s, Py_ssize_t len, Py_ssize_t maxsplit)
+split_whitespace(PyStringObject *self, Py_ssize_t len, Py_ssize_t maxsplit)
 {
+       const char *s = PyString_AS_STRING(self);
        Py_ssize_t i, j, count=0;
        PyObject *str;
        PyObject *list = PyList_New(PREALLOC_SIZE(maxsplit));
@@ -1419,6 +1420,13 @@ split_whitespace(const char *s, Py_ssize_t len, Py_ssize_t maxsplit)
                if (i==len) break;
                j = i; i++;
                SKIP_NONSPACE(s, i, len);
+               if (j == 0 && i == len && PyString_CheckExact(self)) {
+                       /* No whitespace in self, so just use it as list[0] */
+                       Py_INCREF(self);
+                       PyList_SET_ITEM(list, 0, (PyObject *)self);
+                       count++;
+                       break;
+               }
                SPLIT_ADD(s, j, i);
        }
 
@@ -1437,8 +1445,9 @@ split_whitespace(const char *s, Py_ssize_t len, Py_ssize_t maxsplit)
 }
 
 Py_LOCAL_INLINE(PyObject *)
-split_char(const char *s, Py_ssize_t len, char ch, Py_ssize_t maxcount)
+split_char(PyStringObject *self, Py_ssize_t len, char ch, Py_ssize_t maxcount)
 {
+       const char *s = PyString_AS_STRING(self);
        register Py_ssize_t i, j, count=0;
        PyObject *str;
        PyObject *list = PyList_New(PREALLOC_SIZE(maxcount));
@@ -1457,7 +1466,13 @@ split_char(const char *s, Py_ssize_t len, char ch, Py_ssize_t maxcount)
                        }
                }
        }
-       if (i <= len) {
+       if (i == 0 && count == 0 && PyString_CheckExact(self)) {
+               /* ch not in self, so just use self as list[0] */
+               Py_INCREF(self);
+               PyList_SET_ITEM(list, 0, (PyObject *)self);
+               count++;
+       }
+       else if (i <= len) {
                SPLIT_ADD(s, i, len);
        }
        FIX_PREALLOC_SIZE(list);
@@ -1492,7 +1507,7 @@ string_split(PyStringObject *self, PyObject *args)
        if (maxsplit < 0)
                maxsplit = PY_SSIZE_T_MAX;
        if (subobj == Py_None)
-               return split_whitespace(s, len, maxsplit);
+               return split_whitespace(self, len, maxsplit);
        if (PyString_Check(subobj)) {
                sub = PyString_AS_STRING(subobj);
                n = PyString_GET_SIZE(subobj);
@@ -1509,7 +1524,7 @@ string_split(PyStringObject *self, PyObject *args)
                return NULL;
        }
        else if (n == 1)
-               return split_char(s, len, sub[0], maxsplit);
+               return split_char(self, len, sub[0], maxsplit);
 
        list = PyList_New(PREALLOC_SIZE(maxsplit));
        if (list == NULL)
@@ -1609,8 +1624,9 @@ string_rpartition(PyStringObject *self, PyObject *sep_obj)
 }
 
 Py_LOCAL_INLINE(PyObject *)
-rsplit_whitespace(const char *s, Py_ssize_t len, Py_ssize_t maxsplit)
+rsplit_whitespace(PyStringObject *self, Py_ssize_t len, Py_ssize_t maxsplit)
 {
+       const char *s = PyString_AS_STRING(self);
        Py_ssize_t i, j, count=0;
        PyObject *str;
        PyObject *list = PyList_New(PREALLOC_SIZE(maxsplit));
@@ -1625,6 +1641,13 @@ rsplit_whitespace(const char *s, Py_ssize_t len, Py_ssize_t maxsplit)
                if (i<0) break;
                j = i; i--;
                RSKIP_NONSPACE(s, i);
+               if (j == len-1 && i < 0 && PyString_CheckExact(self)) {
+                       /* No whitespace in self, so just use it as list[0] */
+                       Py_INCREF(self);
+                       PyList_SET_ITEM(list, 0, (PyObject *)self);
+                       count++;
+                       break;
+               }
                SPLIT_ADD(s, i + 1, j + 1);
        }
        if (i >= 0) {
@@ -1645,8 +1668,9 @@ rsplit_whitespace(const char *s, Py_ssize_t len, Py_ssize_t maxsplit)
 }
 
 Py_LOCAL_INLINE(PyObject *)
-rsplit_char(const char *s, Py_ssize_t len, char ch, Py_ssize_t maxcount)
+rsplit_char(PyStringObject *self, Py_ssize_t len, char ch, Py_ssize_t maxcount)
 {
+       const char *s = PyString_AS_STRING(self);
        register Py_ssize_t i, j, count=0;
        PyObject *str;
        PyObject *list = PyList_New(PREALLOC_SIZE(maxcount));
@@ -1664,7 +1688,13 @@ rsplit_char(const char *s, Py_ssize_t len, char ch, Py_ssize_t maxcount)
                        }
                }
        }
-       if (j >= -1) {
+       if (i < 0 && count == 0 && PyString_CheckExact(self)) {
+               /* ch not in self, so just use self as list[0] */
+               Py_INCREF(self);
+               PyList_SET_ITEM(list, 0, (PyObject *)self);
+               count++;
+       }
+       else if (j >= -1) {
                SPLIT_ADD(s, 0, j + 1);
        }
        FIX_PREALLOC_SIZE(list);
@@ -1691,7 +1721,7 @@ string_rsplit(PyStringObject *self, PyObject *args)
 {
        Py_ssize_t len = PyString_GET_SIZE(self), n, i, j;
        Py_ssize_t maxsplit = -1, count=0;
-       const char *s = PyString_AS_STRING(self), *sub;
+       const char *s, *sub;
        PyObject *list, *str, *subobj = Py_None;
 
        if (!PyArg_ParseTuple(args, "|On:rsplit", &subobj, &maxsplit))
@@ -1699,7 +1729,7 @@ string_rsplit(PyStringObject *self, PyObject *args)
        if (maxsplit < 0)
                maxsplit = PY_SSIZE_T_MAX;
        if (subobj == Py_None)
-               return rsplit_whitespace(s, len, maxsplit);
+               return rsplit_whitespace(self, len, maxsplit);
        if (PyString_Check(subobj)) {
                sub = PyString_AS_STRING(subobj);
                n = PyString_GET_SIZE(subobj);
@@ -1716,7 +1746,7 @@ string_rsplit(PyStringObject *self, PyObject *args)
                return NULL;
        }
        else if (n == 1)
-               return rsplit_char(s, len, sub[0], maxsplit);
+               return rsplit_char(self, len, sub[0], maxsplit);
 
        list = PyList_New(PREALLOC_SIZE(maxsplit));
        if (list == NULL)
@@ -1725,6 +1755,7 @@ string_rsplit(PyStringObject *self, PyObject *args)
        j = len;
        i = j - n;
 
+       s = PyString_AS_STRING(self);
        while ( (i >= 0) && (maxsplit-- > 0) ) {
                for (; i>=0; i--) {
                        if (Py_STRING_MATCH(s, i, sub, n)) {