Merged revisions 84394 via svnmerge from

svn+ssh://pythondev@svn.python.org/python/branches/py3k

........
  r84394 | antoine.pitrou | 2010-09-01 17:10:12 +0200 (mer., 01 sept. 2010) | 4 lines

  Issue #7415: PyUnicode_FromEncodedObject() now uses the new buffer API
  properly.  Patch by Stefan Behnel.
........

diff --git a/Misc/ACKS b/Misc/ACKS
index dfbbc81e133929cd37855fc384978e6d529fe0fd..b4ac3a70f816d96a72d2cbd8caff61177296abcf 100644 (file)
--- a/Misc/ACKS
+++ b/Misc/ACKS
@@ -58,6 +58,7 @@ David Beazley
 Robin Becker
 Neal Becker
 Bill Bedford
+Stefan Behnel
 Reimer Behrends
 Ben Bell
 Thomas Bellman

diff --git a/Misc/NEWS b/Misc/NEWS
index 8fc8d0838420c9349af3a3ef7a74239f36489ecb..4086378017cdf3b371daf6b4c7de88ec20800223 100644 (file)
--- a/Misc/NEWS
+++ b/Misc/NEWS
@@ -12,6 +12,9 @@ What's New in Python 3.1.3?
 Core and Builtins
 -----------------
 
+- Issue #7415: PyUnicode_FromEncodedObject() now uses the new buffer API
+  properly.  Patch by Stefan Behnel.
+
 - Restore GIL in nis_cat in case of error.
 
 - Issue #9712: Fix tokenize on identifiers that start with non-ascii names.

diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c
index 028b42d8ebf837bba959259e36815f08d2557677..b50293c89c6ad05a11f33ecaef203305b7fa6f1a 100644 (file)
--- a/Objects/unicodeobject.c
+++ b/Objects/unicodeobject.c
@@ -1162,8 +1162,7 @@ PyObject *PyUnicode_FromEncodedObject(register PyObject *obj,
                                       const char *encoding,
                                       const char *errors)
 {
-    const char *s = NULL;
-    Py_ssize_t len;
+    Py_buffer buffer;
     PyObject *v;
 
     if (obj == NULL) {
@@ -1171,44 +1170,44 @@ PyObject *PyUnicode_FromEncodedObject(register PyObject *obj,
         return NULL;
     }
 
+    /* Decoding bytes objects is the most common case and should be fast */
+    if (PyBytes_Check(obj)) {
+        if (PyBytes_GET_SIZE(obj) == 0) {
+            Py_INCREF(unicode_empty);
+            v = (PyObject *) unicode_empty;
+        }
+        else {
+            v = PyUnicode_Decode(
+                    PyBytes_AS_STRING(obj), PyBytes_GET_SIZE(obj),
+                    encoding, errors);
+        }
+        return v;
+    }
+
     if (PyUnicode_Check(obj)) {
         PyErr_SetString(PyExc_TypeError,
                         "decoding str is not supported");
         return NULL;
     }
 
-    /* Coerce object */
-    if (PyBytes_Check(obj)) {
-        s = PyBytes_AS_STRING(obj);
-        len = PyBytes_GET_SIZE(obj);
-    }
-    else if (PyByteArray_Check(obj)) {
-        s = PyByteArray_AS_STRING(obj);
-        len = PyByteArray_GET_SIZE(obj);
-    }
-    else if (PyObject_AsCharBuffer(obj, &s, &len)) {
-        /* Overwrite the error message with something more useful in
-           case of a TypeError. */
-        if (PyErr_ExceptionMatches(PyExc_TypeError))
-            PyErr_Format(PyExc_TypeError,
-                         "coercing to str: need string or buffer, "
-                         "%.80s found",
-                         Py_TYPE(obj)->tp_name);
-        goto onError;
+    /* Retrieve a bytes buffer view through the PEP 3118 buffer interface */
+    if (PyObject_GetBuffer(obj, &buffer, PyBUF_SIMPLE) < 0) {
+        PyErr_Format(PyExc_TypeError,
+                     "coercing to str: need bytes, bytearray "
+                     "or buffer-like object, %.80s found",
+                     Py_TYPE(obj)->tp_name);
+        return NULL;
     }
 
-    /* Convert to Unicode */
-    if (len == 0) {
+    if (buffer.len == 0) {
         Py_INCREF(unicode_empty);
-        v = (PyObject *)unicode_empty;
+        v = (PyObject *) unicode_empty;
     }
     else
-        v = PyUnicode_Decode(s, len, encoding, errors);
+        v = PyUnicode_Decode((char*) buffer.buf, buffer.len, encoding, errors);
 
+    PyBuffer_Release(&buffer);
     return v;
-
-  onError:
-    return NULL;
 }
 
 PyObject *PyUnicode_Decode(const char *s,