Skip to content

gh-158451: Optimize PyUnicode_Join() #158693

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
vstinner wants to merge 1 commit into
base: main
Choose a base branch
from
+57 −33
Open

gh-158451: Optimize PyUnicode_Join() #158693

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
90 changes: 57 additions & 33 deletions Objects/unicodeobject.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1835,8 +1835,7 @@ PyUnicode_Resize(PyObject **p_unicode, Py_ssize_t length)
static PyObject*
get_latin1_char(Py_UCS1 ch)
{
PyObject *o = LATIN1(ch);
return o;
return LATIN1(ch);
}

static PyObject*
Expand Down Expand Up @@ -10489,9 +10488,7 @@ _PyUnicode_JoinArray(PyObject *separator, PyObject *const *items, Py_ssize_t seq
/* Set up sep and seplen */
if (separator == NULL) {
/* fall back to a blank space separator */
sep = PyUnicode_FromOrdinal(' ');
if (!sep)
goto onError;
sep = get_latin1_char(' ');

@eendebakpt eendebakpt Oct 4, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Note: this is a borrowed ref, but gets decref'ed below. Ok since the result is an immortal singleton.

A few lines below there is a comment /* inc refcount ... that is a bit misleading now.

seplen = 1;
maxchar = 32;
}
Expand Down Expand Up @@ -10562,51 +10559,78 @@ _PyUnicode_JoinArray(PyObject *separator, PyObject *const *items, Py_ssize_t seq
use_memcpy = 0;
#else
if (use_memcpy) {
res_data = PyUnicode_1BYTE_DATA(res);
res_data = PyUnicode_DATA(res);
kind = PyUnicode_KIND(res);
if (seplen != 0)
sep_data = PyUnicode_1BYTE_DATA(sep);
sep_data = PyUnicode_DATA(sep);
}
#endif
if (use_memcpy) {
for (i = 0; i < seqlen; ++i) {
Py_ssize_t itemlen;
item = items[i];

/* Copy item, and maybe the separator. */
if (i && seplen != 0) {
memcpy(res_data,
sep_data,
kind * seplen);
res_data += kind * seplen;
}

itemlen = PyUnicode_GET_LENGTH(item);
if (seplen != 0) {
item = items[0];
Py_ssize_t itemlen = PyUnicode_GET_LENGTH(item);
if (itemlen != 0) {
memcpy(res_data,
PyUnicode_DATA(item),
kind * itemlen);
memcpy(res_data, PyUnicode_DATA(item), kind * itemlen);
res_data += kind * itemlen;
}

for (i = 1; i < seqlen; ++i) {
/* Copy item, and maybe the separator. */
memcpy(res_data, sep_data, kind * seplen);
res_data += kind * seplen;

item = items[i];
itemlen = PyUnicode_GET_LENGTH(item);
if (itemlen != 0) {
memcpy(res_data, PyUnicode_DATA(item), kind * itemlen);
res_data += kind * itemlen;
}
}
}
else {
for (i = 0; i < seqlen; ++i) {
item = items[i];
Py_ssize_t itemlen = PyUnicode_GET_LENGTH(item);
if (itemlen != 0) {
memcpy(res_data, PyUnicode_DATA(item), kind * itemlen);
res_data += kind * itemlen;
}
}
}
assert(res_data == PyUnicode_1BYTE_DATA(res)
+ kind * PyUnicode_GET_LENGTH(res));
}
else {
for (i = 0, res_offset = 0; i < seqlen; ++i) {
Py_ssize_t itemlen;
item = items[i];
if (seplen != 0) {
res_offset = 0;
item = items[0];
Py_ssize_t itemlen = PyUnicode_GET_LENGTH(item);
if (itemlen != 0) {
_PyUnicode_FastCopyCharacters(res, res_offset, item, 0, itemlen);
res_offset += itemlen;
}

/* Copy item, and maybe the separator. */
if (i && seplen != 0) {
for (i = 1; i < seqlen; ++i) {
/* Copy item, and maybe the separator. */
_PyUnicode_FastCopyCharacters(res, res_offset, sep, 0, seplen);
res_offset += seplen;
}

itemlen = PyUnicode_GET_LENGTH(item);
if (itemlen != 0) {
_PyUnicode_FastCopyCharacters(res, res_offset, item, 0, itemlen);
res_offset += itemlen;
item = items[i];
itemlen = PyUnicode_GET_LENGTH(item);
if (itemlen != 0) {
_PyUnicode_FastCopyCharacters(res, res_offset, item, 0, itemlen);
res_offset += itemlen;
}
}
}
else {
for (i = 0, res_offset = 0; i < seqlen; ++i) {
item = items[i];
Py_ssize_t itemlen = PyUnicode_GET_LENGTH(item);
if (itemlen != 0) {
_PyUnicode_FastCopyCharacters(res, res_offset, item, 0, itemlen);
res_offset += itemlen;
}
}
}
assert(res_offset == PyUnicode_GET_LENGTH(res));
Expand Down
Loading