Skip to content

Improved speedups #519

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
thewhaleking wants to merge 5 commits into
base: main
Choose a base branch
from
+153 −108
Open

Improved speedups #519

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
d491a1b
Improved speedups
thewhaleking Mar 10, 2026
4af4822
Add changes
thewhaleking Mar 10, 2026
3d04ec9
[pre-commit.ci lite] apply automatic fixes
pre-commit-ci-lite[bot] Mar 10, 2026
6d8337f
Missing newline
thewhaleking Mar 10, 2026
03155ed
Merge remote-tracking branch 'origin/main'
thewhaleking Mar 10, 2026
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
1 change: 1 addition & 0 deletions CHANGES.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,7 @@ Unreleased

- Drop support for Python 3.9.
- Remove previously deprecated code.
- Improve C speedups (building upon idea in https://github.com/pallets/markupsafe/pull/438)


Version 3.0.3
Expand Down
260 changes: 152 additions & 108 deletions src/markupsafe/_speedups.c
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,154 +1,198 @@
#include <Python.h>
#include <stdint.h>

/*
* Lookup tables for HTML escaping.
*
* The five special characters and their replacements:
* '"' (34) -> "&#34;" (len 5, delta +4)
* '&' (38) -> "&amp;" (len 5, delta +4)
* '\'' (39) -> "&#39;" (len 5, delta +4)
* '<' (60) -> "&lt;" (len 4, delta +3)
* '>' (62) -> "&gt;" (len 4, delta +3)
*
* REPLACE_INDEX: 0 = no escaping needed, 1-5 = index into REPLACEMENT_STR.
* All escape chars fit in a byte, so for UCS2/UCS4 we guard with c < 256.
*/
static const uint8_t REPLACE_INDEX[256] = {
['"'] = 1,
['&'] = 2,
['\''] = 3,
['<'] = 4,
['>'] = 5,
};

#define GET_DELTA(inp, inp_end, delta) \
while (inp < inp_end) { \
switch (*inp++) { \
case '"': \
case '\'': \
case '&': \
delta += 4; \
break; \
case '<': \
case '>': \
delta += 3; \
break; \
} \
}
static const char * const REPLACEMENT_STR[] = {
NULL, "&#34;", "&amp;", "&#39;", "&lt;", "&gt;"
};

static const uint8_t REPLACEMENT_LEN[] = {0, 5, 5, 5, 4, 4};

/* Extra output characters needed per input character (0 if no escaping). */
static const uint8_t DELTA_TABLE[256] = {
['"'] = 4,
['&'] = 4,
['\''] = 4,
['<'] = 3,
['>'] = 3,
};

#define DO_ESCAPE(inp, inp_end, outp) \
{ \
Py_ssize_t ncopy = 0; \
while (inp < inp_end) { \
switch (*inp) { \
case '"': \
memcpy(outp, inp-ncopy, sizeof(*outp)*ncopy); \
outp += ncopy; ncopy = 0; \
*outp++ = '&'; \
*outp++ = '#'; \
*outp++ = '3'; \
*outp++ = '4'; \
*outp++ = ';'; \
break; \
case '\'': \
memcpy(outp, inp-ncopy, sizeof(*outp)*ncopy); \
outp += ncopy; ncopy = 0; \
*outp++ = '&'; \
*outp++ = '#'; \
*outp++ = '3'; \
*outp++ = '9'; \
*outp++ = ';'; \
break; \
case '&': \
memcpy(outp, inp-ncopy, sizeof(*outp)*ncopy); \
outp += ncopy; ncopy = 0; \
*outp++ = '&'; \
*outp++ = 'a'; \
*outp++ = 'm'; \
*outp++ = 'p'; \
*outp++ = ';'; \
break; \
case '<': \
memcpy(outp, inp-ncopy, sizeof(*outp)*ncopy); \
outp += ncopy; ncopy = 0; \
*outp++ = '&'; \
*outp++ = 'l'; \
*outp++ = 't'; \
*outp++ = ';'; \
break; \
case '>': \
memcpy(outp, inp-ncopy, sizeof(*outp)*ncopy); \
outp += ncopy; ncopy = 0; \
*outp++ = '&'; \
*outp++ = 'g'; \
*outp++ = 't'; \
*outp++ = ';'; \
break; \
default: \
ncopy++; \
} \
inp++; \
} \
memcpy(outp, inp-ncopy, sizeof(*outp)*ncopy); \
/* Boolean: nonzero if this byte value requires HTML escaping. */
static const uint8_t NEEDS_ESCAPE[256] = {
['"'] = 1,
['&'] = 1,
['\''] = 1,
['<'] = 1,
['>'] = 1,
};

/*
* Count the total extra characters needed for escaping a UCS1 string.
* Processes 4 bytes at a time: if none of the four need escaping, the
* entire chunk is skipped with a single OR across four table lookups.
* Falls back to per-byte delta lookup only for chunks containing a
* special character. Returns 0 if no escaping is needed at all.
*/
static Py_ssize_t
count_delta_1(const Py_UCS1 *inp, Py_ssize_t len)
{
Py_ssize_t i = 0;
Py_ssize_t delta = 0;

for (; i + 4 <= len; i += 4) {
if (NEEDS_ESCAPE[inp[i]] | NEEDS_ESCAPE[inp[i+1]] |
NEEDS_ESCAPE[inp[i+2]] | NEEDS_ESCAPE[inp[i+3]]) {
delta += DELTA_TABLE[inp[i]] + DELTA_TABLE[inp[i+1]] +
DELTA_TABLE[inp[i+2]] + DELTA_TABLE[inp[i+3]];
}
}
for (; i < len; i++)
delta += DELTA_TABLE[inp[i]];

static PyObject*
return delta;
}

static PyObject *
escape_unicode_kind1(PyUnicodeObject *in)
{
Py_UCS1 *inp = PyUnicode_1BYTE_DATA(in);
Py_UCS1 *inp_end = inp + PyUnicode_GET_LENGTH(in);
Py_UCS1 *outp;
PyObject *out;
Py_ssize_t delta = 0;
const Py_UCS1 *inp = PyUnicode_1BYTE_DATA(in);
Py_ssize_t len = PyUnicode_GET_LENGTH(in);
Py_ssize_t delta = count_delta_1(inp, len);

GET_DELTA(inp, inp_end, delta);
if (!delta) {
Py_INCREF(in);
return (PyObject*)in;
return (PyObject *)in;
}

out = PyUnicode_New(PyUnicode_GET_LENGTH(in) + delta,
PyUnicode_IS_ASCII(in) ? 127 : 255);
PyObject *out = PyUnicode_New(len + delta,
PyUnicode_IS_ASCII(in) ? 127 : 255);
if (!out)
return NULL;

inp = PyUnicode_1BYTE_DATA(in);
outp = PyUnicode_1BYTE_DATA(out);
DO_ESCAPE(inp, inp_end, outp);
Py_UCS1 *outp = PyUnicode_1BYTE_DATA(out);
Py_ssize_t prev = 0;
for (Py_ssize_t i = 0; i < len; i++) {
uint8_t ri = REPLACE_INDEX[inp[i]];
if (ri) {
if (i > prev) {
memcpy(outp, inp + prev, i - prev);
outp += i - prev;
}
uint8_t rlen = REPLACEMENT_LEN[ri];
memcpy(outp, REPLACEMENT_STR[ri], rlen);
outp += rlen;
prev = i + 1;
}
}
if (len > prev)
memcpy(outp, inp + prev, len - prev);

return out;
}

static PyObject*
static PyObject *
escape_unicode_kind2(PyUnicodeObject *in)
{
Py_UCS2 *inp = PyUnicode_2BYTE_DATA(in);
Py_UCS2 *inp_end = inp + PyUnicode_GET_LENGTH(in);
Py_UCS2 *outp;
PyObject *out;
const Py_UCS2 *inp = PyUnicode_2BYTE_DATA(in);
Py_ssize_t len = PyUnicode_GET_LENGTH(in);
Py_ssize_t delta = 0;

GET_DELTA(inp, inp_end, delta);
for (Py_ssize_t i = 0; i < len; i++)
delta += (inp[i] < 256) ? DELTA_TABLE[inp[i]] : 0;

if (!delta) {
Py_INCREF(in);
return (PyObject*)in;
return (PyObject *)in;
}

out = PyUnicode_New(PyUnicode_GET_LENGTH(in) + delta, 65535);
PyObject *out = PyUnicode_New(len + delta, 65535);
if (!out)
return NULL;

inp = PyUnicode_2BYTE_DATA(in);
outp = PyUnicode_2BYTE_DATA(out);
DO_ESCAPE(inp, inp_end, outp);
Py_UCS2 *outp = PyUnicode_2BYTE_DATA(out);
Py_ssize_t prev = 0;
for (Py_ssize_t i = 0; i < len; i++) {
uint8_t ri = (inp[i] < 256) ? REPLACE_INDEX[inp[i]] : 0;
if (ri) {
if (i > prev) {
memcpy(outp, inp + prev, (i - prev) * sizeof(Py_UCS2));
outp += i - prev;
}
const char *repl = REPLACEMENT_STR[ri];
uint8_t rlen = REPLACEMENT_LEN[ri];
for (uint8_t j = 0; j < rlen; j++)
*outp++ = (Py_UCS2)(unsigned char)repl[j];
prev = i + 1;
}
}
if (len > prev)
memcpy(outp, inp + prev, (len - prev) * sizeof(Py_UCS2));

return out;
}


static PyObject*
static PyObject *
escape_unicode_kind4(PyUnicodeObject *in)
{
Py_UCS4 *inp = PyUnicode_4BYTE_DATA(in);
Py_UCS4 *inp_end = inp + PyUnicode_GET_LENGTH(in);
Py_UCS4 *outp;
PyObject *out;
const Py_UCS4 *inp = PyUnicode_4BYTE_DATA(in);
Py_ssize_t len = PyUnicode_GET_LENGTH(in);
Py_ssize_t delta = 0;

GET_DELTA(inp, inp_end, delta);
for (Py_ssize_t i = 0; i < len; i++)
delta += (inp[i] < 256) ? DELTA_TABLE[inp[i]] : 0;

if (!delta) {
Py_INCREF(in);
return (PyObject*)in;
return (PyObject *)in;
}

out = PyUnicode_New(PyUnicode_GET_LENGTH(in) + delta, 1114111);
PyObject *out = PyUnicode_New(len + delta, 1114111);
if (!out)
return NULL;

inp = PyUnicode_4BYTE_DATA(in);
outp = PyUnicode_4BYTE_DATA(out);
DO_ESCAPE(inp, inp_end, outp);
Py_UCS4 *outp = PyUnicode_4BYTE_DATA(out);
Py_ssize_t prev = 0;
for (Py_ssize_t i = 0; i < len; i++) {
uint8_t ri = (inp[i] < 256) ? REPLACE_INDEX[inp[i]] : 0;
if (ri) {
if (i > prev) {
memcpy(outp, inp + prev, (i - prev) * sizeof(Py_UCS4));
outp += i - prev;
}
const char *repl = REPLACEMENT_STR[ri];
uint8_t rlen = REPLACEMENT_LEN[ri];
for (uint8_t j = 0; j < rlen; j++)
*outp++ = (Py_UCS4)(unsigned char)repl[j];
prev = i + 1;
}
}
if (len > prev)
memcpy(outp, inp + prev, (len - prev) * sizeof(Py_UCS4));

return out;
}

static PyObject*
static PyObject *
escape_unicode(PyObject *self, PyObject *s)
{
if (!PyUnicode_Check(s))
Expand All @@ -160,11 +204,11 @@ escape_unicode(PyObject *self, PyObject *s)

switch (PyUnicode_KIND(s)) {
case PyUnicode_1BYTE_KIND:
return escape_unicode_kind1((PyUnicodeObject*) s);
return escape_unicode_kind1((PyUnicodeObject *) s);
case PyUnicode_2BYTE_KIND:
return escape_unicode_kind2((PyUnicodeObject*) s);
return escape_unicode_kind2((PyUnicodeObject *) s);
case PyUnicode_4BYTE_KIND:
return escape_unicode_kind4((PyUnicodeObject*) s);
return escape_unicode_kind4((PyUnicodeObject *) s);
}
assert(0); /* shouldn't happen */
return NULL;
Expand Down