gh-91576: Speed up iteration of strings (#91574)

This commit is contained in:
Kumar Aditya 2022-04-18 19:48:27 +05:30 committed by GitHub
parent a29f858124
commit 8c54c3dacc
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
5 changed files with 79 additions and 6 deletions

View file

@ -15697,7 +15697,7 @@ unicodeiter_traverse(unicodeiterobject *it, visitproc visit, void *arg)
static PyObject *
unicodeiter_next(unicodeiterobject *it)
{
PyObject *seq, *item;
PyObject *seq;
assert(it != NULL);
seq = it->it_seq;
@ -15709,10 +15709,8 @@ unicodeiter_next(unicodeiterobject *it)
int kind = PyUnicode_KIND(seq);
const void *data = PyUnicode_DATA(seq);
Py_UCS4 chr = PyUnicode_READ(kind, data, it->it_index);
item = PyUnicode_FromOrdinal(chr);
if (item != NULL)
++it->it_index;
return item;
it->it_index++;
return unicode_char(chr);
}
it->it_seq = NULL;
@ -15720,6 +15718,29 @@ unicodeiter_next(unicodeiterobject *it)
return NULL;
}
static PyObject *
unicode_ascii_iter_next(unicodeiterobject *it)
{
assert(it != NULL);
PyObject *seq = it->it_seq;
if (seq == NULL) {
return NULL;
}
assert(_PyUnicode_CHECK(seq));
assert(PyUnicode_IS_COMPACT_ASCII(seq));
if (it->it_index < PyUnicode_GET_LENGTH(seq)) {
const void *data = ((void*)(_PyASCIIObject_CAST(seq) + 1));
Py_UCS1 chr = (Py_UCS1)PyUnicode_READ(PyUnicode_1BYTE_KIND,
data, it->it_index);
it->it_index++;
PyObject *item = (PyObject*)&_Py_SINGLETON(strings).ascii[chr];
return Py_NewRef(item);
}
it->it_seq = NULL;
Py_DECREF(seq);
return NULL;
}
static PyObject *
unicodeiter_len(unicodeiterobject *it, PyObject *Py_UNUSED(ignored))
{
@ -15808,6 +15829,19 @@ PyTypeObject PyUnicodeIter_Type = {
0,
};
PyTypeObject _PyUnicodeASCIIIter_Type = {
PyVarObject_HEAD_INIT(&PyType_Type, 0)
.tp_name = "str_ascii_iterator",
.tp_basicsize = sizeof(unicodeiterobject),
.tp_dealloc = (destructor)unicodeiter_dealloc,
.tp_getattro = PyObject_GenericGetAttr,
.tp_flags = Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC,
.tp_traverse = (traverseproc)unicodeiter_traverse,
.tp_iter = PyObject_SelfIter,
.tp_iternext = (iternextfunc)unicode_ascii_iter_next,
.tp_methods = unicodeiter_methods,
};
static PyObject *
unicode_iter(PyObject *seq)
{
@ -15819,7 +15853,12 @@ unicode_iter(PyObject *seq)
}
if (PyUnicode_READY(seq) == -1)
return NULL;
it = PyObject_GC_New(unicodeiterobject, &PyUnicodeIter_Type);
if (PyUnicode_IS_COMPACT_ASCII(seq)) {
it = PyObject_GC_New(unicodeiterobject, &_PyUnicodeASCIIIter_Type);
}
else {
it = PyObject_GC_New(unicodeiterobject, &PyUnicodeIter_Type);
}
if (it == NULL)
return NULL;
it->it_index = 0;