Skip to content

Speedup struct decoding #119

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

Merged
jcrist merged 1 commit into from
Jun 10, 2022
Merged

Speedup struct decoding #119

Show file tree
Hide file tree
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
78 changes: 41 additions & 37 deletions msgspec/_core.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -81,7 +81,7 @@ ms_popcount(uint32_t i) {
/* XXX: Optimized `PyUnicode_AsUTF8AndSize`, fastpath for ascii strings. */
static inline const char *
unicode_str_and_size(PyObject *str, Py_ssize_t *size) {
if (PyUnicode_IS_COMPACT_ASCII(str)) {
if (MS_LIKELY(PyUnicode_IS_COMPACT_ASCII(str))) {
*size = ((PyASCIIObject *)str)->length;
return (char *)(((PyASCIIObject *)str) + 1);
}
Expand Down Expand Up @@ -2736,26 +2736,33 @@ Struct_dealloc(PyObject *self) {
Py_DECREF(type);
}

static Py_ssize_t
static MS_INLINE Py_ssize_t
StructMeta_get_field_index(
StructMetaObject *self, const char * key, Py_ssize_t key_size, Py_ssize_t *pos
) {
const char *field;
Py_ssize_t nfields, field_size, i, ind, offset = *pos;
Py_ssize_t nfields, field_size, i, offset = *pos;
nfields = PyTuple_GET_SIZE(self->struct_encode_fields);
for (i = 0; i < nfields; i++) {
ind = (i + offset) % nfields;
for (i = offset; i < nfields; i++) {
field = unicode_str_and_size(
PyTuple_GET_ITEM(self->struct_encode_fields, ind), &field_size
PyTuple_GET_ITEM(self->struct_encode_fields, i), &field_size
);
if (field == NULL) return -1;
if (key_size == field_size && memcmp(key, field, key_size) == 0) {
*pos = (ind + 1) % nfields;
return ind;
*pos = i < (nfields - 1) ? (i + 1) : 0;
return i;
}
}
for (i = 0; i < offset; i++) {
field = unicode_str_and_size(
PyTuple_GET_ITEM(self->struct_encode_fields, i), &field_size
);
if (key_size == field_size && memcmp(key, field, key_size) == 0) {
*pos = i + 1;
return i;
}
}
/* Not a field, check if it matches the tag field (if present) */
if (self->struct_tag_field != NULL) {
if (MS_UNLIKELY(self->struct_tag_field != NULL)) {
Py_ssize_t tag_field_size;
const char *tag_field;
tag_field = unicode_str_and_size(self->struct_tag_field, &tag_field_size);
Expand Down Expand Up @@ -9537,38 +9544,35 @@ json_decode_struct_map_inner(

/* Parse value */
field_index = StructMeta_get_field_index(st_type, key, key_size, &pos);
if (field_index < 0) {
if (MS_UNLIKELY(field_index == -2)) {
/* Matches the tag field */
Py_ssize_t tag_size, expected_size;
char *tag = NULL;
PathNode tag_path = {path, PATH_TAG, st_type->struct_tag_field};
if (MS_LIKELY(field_index >= 0)) {
field_path.index = field_index;
TypeNode *type = st_type->struct_types[field_index];
val = json_decode(self, type, &field_path);
if (val == NULL) goto error;
Struct_set_index(out, field_index, val);
}
else if (MS_UNLIKELY(field_index == -2)) {
/* Matches the tag field */
Py_ssize_t tag_size, expected_size;
char *tag = NULL;
PathNode tag_path = {path, PATH_TAG, st_type->struct_tag_field};

/* Decode the tag value */
tag_size = json_decode_cstr(self, &tag, &tag_path);
if (tag_size < 0) goto error;
/* Decode the tag value */
tag_size = json_decode_cstr(self, &tag, &tag_path);
if (tag_size < 0) goto error;

/* Check that the tag value matches the expected value */
const char *expected = unicode_str_and_size(
st_type->struct_tag_value, &expected_size
);
if (tag_size != expected_size || memcmp(tag, expected, expected_size) != 0) {
/* Tag doesn't match the expected value, error nicely */
return ms_invalid_cstr_value(tag, tag_size, &tag_path);
}
}
else {
/* Skip unknown fields */
if (json_skip(self) < 0) goto error;
/* Check that the tag value matches the expected value */
const char *expected = unicode_str_and_size(
st_type->struct_tag_value, &expected_size
);
if (tag_size != expected_size || memcmp(tag, expected, expected_size) != 0) {
/* Tag doesn't match the expected value, error nicely */
return ms_invalid_cstr_value(tag, tag_size, &tag_path);
}
}
else {
field_path.index = field_index;
val = json_decode(
self, st_type->struct_types[field_index], &field_path
);
if (val == NULL) goto error;
Struct_set_index(out, field_index, val);
/* Skip unknown fields */
if (json_skip(self) < 0) goto error;
}
}
if (Struct_fill_in_defaults(st_type, out, path) < 0) goto error;
Expand Down
18 changes: 18 additions & 0 deletions tests/test_json.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1815,6 +1815,24 @@ def test_decode_struct_missing_fields(self):
):
msgspec.json.decode(bad, type=List[Person])

def test_decode_struct_fields_mixed_order(self):
class Test(msgspec.Struct):
a: int
b: int
c: int
d: int

sol = Test(0, 1, 2, 3)

dec = msgspec.json.Decoder(Test)

pairs = list(zip("abcdef", range(6)))

for data in itertools.permutations(pairs):
msg = msgspec.json.encode(dict(data))
res = dec.decode(msg)
assert res == sol

@pytest.mark.parametrize(
"extra",
[
Expand Down