GH-92123: Pass _elementtree state as parameter (#101189)

This commit is contained in:
Erlend E. Aasland 2023-01-24 12:45:55 +01:00 committed by GitHub
parent 38cc24f119
commit b2ac39626a
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23

View file

@ -297,11 +297,10 @@ clear_extra(ElementObject* self)
* tag and attributes. * tag and attributes.
*/ */
LOCAL(PyObject*) LOCAL(PyObject*)
create_new_element(PyObject* tag, PyObject* attrib) create_new_element(elementtreestate *st, PyObject *tag, PyObject *attrib)
{ {
ElementObject* self; ElementObject* self;
elementtreestate *st = ET_STATE_GLOBAL;
self = PyObject_GC_New(ElementObject, st->Element_Type); self = PyObject_GC_New(ElementObject, st->Element_Type);
if (self == NULL) if (self == NULL)
return NULL; return NULL;
@ -504,10 +503,10 @@ raise_type_error(PyObject *element)
} }
LOCAL(int) LOCAL(int)
element_add_subelement(ElementObject* self, PyObject* element) element_add_subelement(elementtreestate *st, ElementObject *self,
PyObject *element)
{ {
/* add a child element to a parent */ /* add a child element to a parent */
elementtreestate *st = ET_STATE_GLOBAL;
if (!Element_Check(st, element)) { if (!Element_Check(st, element)) {
raise_type_error(element); raise_type_error(element);
return -1; return -1;
@ -614,12 +613,12 @@ subelement(PyObject *self, PyObject *args, PyObject *kwds)
/* no attrib arg, no kwds, so no attribute */ /* no attrib arg, no kwds, so no attribute */
} }
elem = create_new_element(tag, attrib); elem = create_new_element(st, tag, attrib);
Py_XDECREF(attrib); Py_XDECREF(attrib);
if (elem == NULL) if (elem == NULL)
return NULL; return NULL;
if (element_add_subelement(parent, elem) < 0) { if (element_add_subelement(st, parent, elem) < 0) {
Py_DECREF(elem); Py_DECREF(elem);
return NULL; return NULL;
} }
@ -694,7 +693,8 @@ static PyObject *
_elementtree_Element_append_impl(ElementObject *self, PyObject *subelement) _elementtree_Element_append_impl(ElementObject *self, PyObject *subelement)
/*[clinic end generated code: output=54a884b7cf2295f4 input=439f2bd777288fb6]*/ /*[clinic end generated code: output=54a884b7cf2295f4 input=439f2bd777288fb6]*/
{ {
if (element_add_subelement(self, subelement) < 0) elementtreestate *st = ET_STATE_GLOBAL;
if (element_add_subelement(st, self, subelement) < 0)
return NULL; return NULL;
Py_RETURN_NONE; Py_RETURN_NONE;
@ -728,9 +728,10 @@ _elementtree_Element___copy___impl(ElementObject *self)
{ {
Py_ssize_t i; Py_ssize_t i;
ElementObject* element; ElementObject* element;
elementtreestate *st = ET_STATE_GLOBAL;
element = (ElementObject*) create_new_element( element = (ElementObject*) create_new_element(
self->tag, self->extra ? self->extra->attrib : NULL); st, self->tag, self->extra ? self->extra->attrib : NULL);
if (!element) if (!element)
return NULL; return NULL;
@ -759,7 +760,7 @@ _elementtree_Element___copy___impl(ElementObject *self)
} }
/* Helper for a deep copy. */ /* Helper for a deep copy. */
LOCAL(PyObject *) deepcopy(PyObject *, PyObject *); LOCAL(PyObject *) deepcopy(elementtreestate *, PyObject *, PyObject *);
/*[clinic input] /*[clinic input]
_elementtree.Element.__deepcopy__ _elementtree.Element.__deepcopy__
@ -781,12 +782,13 @@ _elementtree_Element___deepcopy___impl(ElementObject *self, PyObject *memo)
PyObject* tail; PyObject* tail;
PyObject* id; PyObject* id;
tag = deepcopy(self->tag, memo); elementtreestate *st = ET_STATE_GLOBAL;
tag = deepcopy(st, self->tag, memo);
if (!tag) if (!tag)
return NULL; return NULL;
if (self->extra && self->extra->attrib) { if (self->extra && self->extra->attrib) {
attrib = deepcopy(self->extra->attrib, memo); attrib = deepcopy(st, self->extra->attrib, memo);
if (!attrib) { if (!attrib) {
Py_DECREF(tag); Py_DECREF(tag);
return NULL; return NULL;
@ -795,7 +797,7 @@ _elementtree_Element___deepcopy___impl(ElementObject *self, PyObject *memo)
attrib = NULL; attrib = NULL;
} }
element = (ElementObject*) create_new_element(tag, attrib); element = (ElementObject*) create_new_element(st, tag, attrib);
Py_DECREF(tag); Py_DECREF(tag);
Py_XDECREF(attrib); Py_XDECREF(attrib);
@ -803,12 +805,12 @@ _elementtree_Element___deepcopy___impl(ElementObject *self, PyObject *memo)
if (!element) if (!element)
return NULL; return NULL;
text = deepcopy(JOIN_OBJ(self->text), memo); text = deepcopy(st, JOIN_OBJ(self->text), memo);
if (!text) if (!text)
goto error; goto error;
_set_joined_ptr(&element->text, JOIN_SET(text, JOIN_GET(self->text))); _set_joined_ptr(&element->text, JOIN_SET(text, JOIN_GET(self->text)));
tail = deepcopy(JOIN_OBJ(self->tail), memo); tail = deepcopy(st, JOIN_OBJ(self->tail), memo);
if (!tail) if (!tail)
goto error; goto error;
_set_joined_ptr(&element->tail, JOIN_SET(tail, JOIN_GET(self->tail))); _set_joined_ptr(&element->tail, JOIN_SET(tail, JOIN_GET(self->tail)));
@ -818,9 +820,8 @@ _elementtree_Element___deepcopy___impl(ElementObject *self, PyObject *memo)
if (element_resize(element, self->extra->length) < 0) if (element_resize(element, self->extra->length) < 0)
goto error; goto error;
elementtreestate *st = ET_STATE_GLOBAL;
for (i = 0; i < self->extra->length; i++) { for (i = 0; i < self->extra->length; i++) {
PyObject* child = deepcopy(self->extra->children[i], memo); PyObject* child = deepcopy(st, self->extra->children[i], memo);
if (!child || !Element_Check(st, child)) { if (!child || !Element_Check(st, child)) {
if (child) { if (child) {
raise_type_error(child); raise_type_error(child);
@ -856,10 +857,9 @@ _elementtree_Element___deepcopy___impl(ElementObject *self, PyObject *memo)
} }
LOCAL(PyObject *) LOCAL(PyObject *)
deepcopy(PyObject *object, PyObject *memo) deepcopy(elementtreestate *st, PyObject *object, PyObject *memo)
{ {
/* do a deep copy of the given object */ /* do a deep copy of the given object */
elementtreestate *st = ET_STATE_GLOBAL;
PyObject *stack[2]; PyObject *stack[2];
/* Fast paths */ /* Fast paths */
@ -974,7 +974,8 @@ _elementtree_Element___getstate___impl(ElementObject *self)
} }
static PyObject * static PyObject *
element_setstate_from_attributes(ElementObject *self, element_setstate_from_attributes(elementtreestate *st,
ElementObject *self,
PyObject *tag, PyObject *tag,
PyObject *attrib, PyObject *attrib,
PyObject *text, PyObject *text,
@ -1032,7 +1033,6 @@ element_setstate_from_attributes(ElementObject *self,
} }
/* Copy children */ /* Copy children */
elementtreestate *st = ET_STATE_GLOBAL;
for (i = 0; i < nchildren; i++) { for (i = 0; i < nchildren; i++) {
PyObject *child = PyList_GET_ITEM(children, i); PyObject *child = PyList_GET_ITEM(children, i);
if (!Element_Check(st, child)) { if (!Element_Check(st, child)) {
@ -1065,7 +1065,8 @@ element_setstate_from_attributes(ElementObject *self,
*/ */
static PyObject * static PyObject *
element_setstate_from_Python(ElementObject *self, PyObject *state) element_setstate_from_Python(elementtreestate *st, ElementObject *self,
PyObject *state)
{ {
static char *kwlist[] = {PICKLED_TAG, PICKLED_ATTRIB, PICKLED_TEXT, static char *kwlist[] = {PICKLED_TAG, PICKLED_ATTRIB, PICKLED_TEXT,
PICKLED_TAIL, PICKLED_CHILDREN, 0}; PICKLED_TAIL, PICKLED_CHILDREN, 0};
@ -1080,7 +1081,7 @@ element_setstate_from_Python(ElementObject *self, PyObject *state)
if (PyArg_ParseTupleAndKeywords(args, state, "|$OOOOO", kwlist, &tag, if (PyArg_ParseTupleAndKeywords(args, state, "|$OOOOO", kwlist, &tag,
&attrib, &text, &tail, &children)) &attrib, &text, &tail, &children))
retval = element_setstate_from_attributes(self, tag, attrib, text, retval = element_setstate_from_attributes(st, self, tag, attrib, text,
tail, children); tail, children);
else else
retval = NULL; retval = NULL;
@ -1107,8 +1108,10 @@ _elementtree_Element___setstate__(ElementObject *self, PyObject *state)
state); state);
return NULL; return NULL;
} }
else else {
return element_setstate_from_Python(self, state); elementtreestate *st = ET_STATE_GLOBAL;
return element_setstate_from_Python(st, self, state);
}
} }
LOCAL(int) LOCAL(int)
@ -1190,9 +1193,10 @@ _elementtree_Element_extend(ElementObject *self, PyObject *elements)
return NULL; return NULL;
} }
elementtreestate *st = ET_STATE_GLOBAL;
for (i = 0; i < PySequence_Fast_GET_SIZE(seq); i++) { for (i = 0; i < PySequence_Fast_GET_SIZE(seq); i++) {
PyObject* element = Py_NewRef(PySequence_Fast_GET_ITEM(seq, i)); PyObject* element = Py_NewRef(PySequence_Fast_GET_ITEM(seq, i));
if (element_add_subelement(self, element) < 0) { if (element_add_subelement(st, self, element) < 0) {
Py_DECREF(seq); Py_DECREF(seq);
Py_DECREF(element); Py_DECREF(element);
return NULL; return NULL;
@ -1391,7 +1395,8 @@ _elementtree_Element_get_impl(ElementObject *self, PyObject *key,
} }
static PyObject * static PyObject *
create_elementiter(ElementObject *self, PyObject *tag, int gettext); create_elementiter(elementtreestate *st, ElementObject *self, PyObject *tag,
int gettext);
/*[clinic input] /*[clinic input]
@ -1416,7 +1421,8 @@ _elementtree_Element_iter_impl(ElementObject *self, PyObject *tag)
tag = Py_None; tag = Py_None;
} }
return create_elementiter(self, tag, 0); elementtreestate *st = ET_STATE_GLOBAL;
return create_elementiter(st, self, tag, 0);
} }
@ -1429,7 +1435,8 @@ static PyObject *
_elementtree_Element_itertext_impl(ElementObject *self) _elementtree_Element_itertext_impl(ElementObject *self)
/*[clinic end generated code: output=5fa34b2fbcb65df6 input=af8f0e42cb239c89]*/ /*[clinic end generated code: output=5fa34b2fbcb65df6 input=af8f0e42cb239c89]*/
{ {
return create_elementiter(self, Py_None, 1); elementtreestate *st = ET_STATE_GLOBAL;
return create_elementiter(st, self, Py_None, 1);
} }
@ -1567,7 +1574,8 @@ _elementtree_Element_makeelement_impl(ElementObject *self, PyObject *tag,
if (!attrib) if (!attrib)
return NULL; return NULL;
elem = create_new_element(tag, attrib); elementtreestate *st = ET_STATE_GLOBAL;
elem = create_new_element(st, tag, attrib);
Py_DECREF(attrib); Py_DECREF(attrib);
@ -2246,11 +2254,11 @@ static PyType_Spec elementiter_spec = {
#define INIT_PARENT_STACK_SIZE 8 #define INIT_PARENT_STACK_SIZE 8
static PyObject * static PyObject *
create_elementiter(ElementObject *self, PyObject *tag, int gettext) create_elementiter(elementtreestate *st, ElementObject *self, PyObject *tag,
int gettext)
{ {
ElementIterObject *it; ElementIterObject *it;
elementtreestate *st = ET_STATE_GLOBAL;
it = PyObject_GC_New(ElementIterObject, st->ElementIter_Type); it = PyObject_GC_New(ElementIterObject, st->ElementIter_Type);
if (!it) if (!it)
return NULL; return NULL;
@ -2506,11 +2514,11 @@ _elementtree__set_factories_impl(PyObject *module, PyObject *comment_factory,
} }
static int static int
treebuilder_extend_element_text_or_tail(PyObject *element, PyObject **data, treebuilder_extend_element_text_or_tail(elementtreestate *st, PyObject *element,
PyObject **dest, PyObject *name) PyObject **data, PyObject **dest,
PyObject *name)
{ {
/* Fast paths for the "almost always" cases. */ /* Fast paths for the "almost always" cases. */
elementtreestate *st = ET_STATE_GLOBAL;
if (Element_CheckExact(st, element)) { if (Element_CheckExact(st, element)) {
PyObject *dest_obj = JOIN_OBJ(*dest); PyObject *dest_obj = JOIN_OBJ(*dest);
if (dest_obj == Py_None) { if (dest_obj == Py_None) {
@ -2570,24 +2578,24 @@ treebuilder_flush_data(TreeBuilderObject* self)
if (!self->last_for_tail) { if (!self->last_for_tail) {
PyObject *element = self->last; PyObject *element = self->last;
return treebuilder_extend_element_text_or_tail( return treebuilder_extend_element_text_or_tail(
element, &self->data, st, element, &self->data,
&((ElementObject *) element)->text, st->str_text); &((ElementObject *) element)->text, st->str_text);
} }
else { else {
PyObject *element = self->last_for_tail; PyObject *element = self->last_for_tail;
return treebuilder_extend_element_text_or_tail( return treebuilder_extend_element_text_or_tail(
element, &self->data, st, element, &self->data,
&((ElementObject *) element)->tail, st->str_tail); &((ElementObject *) element)->tail, st->str_tail);
} }
} }
static int static int
treebuilder_add_subelement(PyObject *element, PyObject *child) treebuilder_add_subelement(elementtreestate *st, PyObject *element,
PyObject *child)
{ {
elementtreestate *st = ET_STATE_GLOBAL;
if (Element_CheckExact(st, element)) { if (Element_CheckExact(st, element)) {
ElementObject *elem = (ElementObject *) element; ElementObject *elem = (ElementObject *) element;
return element_add_subelement(elem, child); return element_add_subelement(st, elem, child);
} }
else { else {
PyObject *res; PyObject *res;
@ -2633,8 +2641,9 @@ treebuilder_handle_start(TreeBuilderObject* self, PyObject* tag,
} }
if (!self->element_factory) { if (!self->element_factory) {
node = create_new_element(tag, attrib); node = create_new_element(st, tag, attrib);
} else if (attrib == NULL) { }
else if (attrib == NULL) {
attrib = PyDict_New(); attrib = PyDict_New();
if (!attrib) if (!attrib)
return NULL; return NULL;
@ -2654,8 +2663,9 @@ treebuilder_handle_start(TreeBuilderObject* self, PyObject* tag,
Py_CLEAR(self->last_for_tail); Py_CLEAR(self->last_for_tail);
if (this != Py_None) { if (this != Py_None) {
if (treebuilder_add_subelement(this, node) < 0) if (treebuilder_add_subelement(st, this, node) < 0) {
goto error; goto error;
}
} else { } else {
if (self->root) { if (self->root) {
PyErr_SetString( PyErr_SetString(
@ -2774,8 +2784,9 @@ treebuilder_handle_comment(TreeBuilderObject* self, PyObject* text)
this = self->this; this = self->this;
if (self->insert_comments && this != Py_None) { if (self->insert_comments && this != Py_None) {
if (treebuilder_add_subelement(this, comment) < 0) if (treebuilder_add_subelement(self->state, this, comment) < 0) {
goto error; goto error;
}
Py_XSETREF(self->last_for_tail, Py_NewRef(comment)); Py_XSETREF(self->last_for_tail, Py_NewRef(comment));
} }
} else { } else {
@ -2813,8 +2824,9 @@ treebuilder_handle_pi(TreeBuilderObject* self, PyObject* target, PyObject* text)
this = self->this; this = self->this;
if (self->insert_pis && this != Py_None) { if (self->insert_pis && this != Py_None) {
if (treebuilder_add_subelement(this, pi) < 0) if (treebuilder_add_subelement(self->state, this, pi) < 0) {
goto error; goto error;
}
Py_XSETREF(self->last_for_tail, Py_NewRef(pi)); Py_XSETREF(self->last_for_tail, Py_NewRef(pi));
} }
} else { } else {
@ -3091,11 +3103,10 @@ makeuniversal(XMLParserObject* self, const char* string)
* message string is the default for the given error_code. * message string is the default for the given error_code.
*/ */
static void static void
expat_set_error(enum XML_Error error_code, Py_ssize_t line, Py_ssize_t column, expat_set_error(elementtreestate *st, enum XML_Error error_code,
const char *message) Py_ssize_t line, Py_ssize_t column, const char *message)
{ {
PyObject *errmsg, *error, *position, *code; PyObject *errmsg, *error, *position, *code;
elementtreestate *st = ET_STATE_GLOBAL;
errmsg = PyUnicode_FromFormat("%s: line %zd, column %zd", errmsg = PyUnicode_FromFormat("%s: line %zd, column %zd",
message ? message : EXPAT(ErrorString)(error_code), message ? message : EXPAT(ErrorString)(error_code),
@ -3160,8 +3171,8 @@ expat_default_handler(XMLParserObject* self, const XML_Char* data_in,
value = PyDict_GetItemWithError(self->entity, key); value = PyDict_GetItemWithError(self->entity, key);
elementtreestate *st = self->state;
if (value) { if (value) {
elementtreestate *st = self->state;
if (TreeBuilder_CheckExact(st, self->target)) if (TreeBuilder_CheckExact(st, self->target))
res = treebuilder_handle_data( res = treebuilder_handle_data(
(TreeBuilderObject*) self->target, value (TreeBuilderObject*) self->target, value
@ -3176,6 +3187,7 @@ expat_default_handler(XMLParserObject* self, const XML_Char* data_in,
char message[128] = "undefined entity "; char message[128] = "undefined entity ";
strncat(message, data_in, data_len < 100?data_len:100); strncat(message, data_in, data_len < 100?data_len:100);
expat_set_error( expat_set_error(
st,
XML_ERROR_UNDEFINED_ENTITY, XML_ERROR_UNDEFINED_ENTITY,
EXPAT(GetErrorLineNumber)(self->parser), EXPAT(GetErrorLineNumber)(self->parser),
EXPAT(GetErrorColumnNumber)(self->parser), EXPAT(GetErrorColumnNumber)(self->parser),
@ -3774,7 +3786,8 @@ _check_xmlparser(XMLParserObject* self)
} }
LOCAL(PyObject*) LOCAL(PyObject*)
expat_parse(XMLParserObject* self, const char* data, int data_len, int final) expat_parse(elementtreestate *st, XMLParserObject *self, const char *data,
int data_len, int final)
{ {
int ok; int ok;
@ -3786,6 +3799,7 @@ expat_parse(XMLParserObject* self, const char* data, int data_len, int final)
if (!ok) { if (!ok) {
expat_set_error( expat_set_error(
st,
EXPAT(GetErrorCode)(self->parser), EXPAT(GetErrorCode)(self->parser),
EXPAT(GetErrorLineNumber)(self->parser), EXPAT(GetErrorLineNumber)(self->parser),
EXPAT(GetErrorColumnNumber)(self->parser), EXPAT(GetErrorColumnNumber)(self->parser),
@ -3813,11 +3827,11 @@ _elementtree_XMLParser_close_impl(XMLParserObject *self)
if (!_check_xmlparser(self)) { if (!_check_xmlparser(self)) {
return NULL; return NULL;
} }
res = expat_parse(self, "", 0, 1); elementtreestate *st = self->state;
res = expat_parse(st, self, "", 0, 1);
if (!res) if (!res)
return NULL; return NULL;
elementtreestate *st = self->state;
if (TreeBuilder_CheckExact(st, self->target)) { if (TreeBuilder_CheckExact(st, self->target)) {
Py_DECREF(res); Py_DECREF(res);
return treebuilder_done((TreeBuilderObject*) self->target); return treebuilder_done((TreeBuilderObject*) self->target);
@ -3848,6 +3862,7 @@ _elementtree_XMLParser_feed(XMLParserObject *self, PyObject *data)
if (!_check_xmlparser(self)) { if (!_check_xmlparser(self)) {
return NULL; return NULL;
} }
elementtreestate *st = self->state;
if (PyUnicode_Check(data)) { if (PyUnicode_Check(data)) {
Py_ssize_t data_len; Py_ssize_t data_len;
const char *data_ptr = PyUnicode_AsUTF8AndSize(data, &data_len); const char *data_ptr = PyUnicode_AsUTF8AndSize(data, &data_len);
@ -3859,7 +3874,8 @@ _elementtree_XMLParser_feed(XMLParserObject *self, PyObject *data)
} }
/* Explicitly set UTF-8 encoding. Return code ignored. */ /* Explicitly set UTF-8 encoding. Return code ignored. */
(void)EXPAT(SetEncoding)(self->parser, "utf-8"); (void)EXPAT(SetEncoding)(self->parser, "utf-8");
return expat_parse(self, data_ptr, (int)data_len, 0);
return expat_parse(st, self, data_ptr, (int)data_len, 0);
} }
else { else {
Py_buffer view; Py_buffer view;
@ -3871,7 +3887,7 @@ _elementtree_XMLParser_feed(XMLParserObject *self, PyObject *data)
PyErr_SetString(PyExc_OverflowError, "size does not fit in an int"); PyErr_SetString(PyExc_OverflowError, "size does not fit in an int");
return NULL; return NULL;
} }
res = expat_parse(self, view.buf, (int)view.len, 0); res = expat_parse(st, self, view.buf, (int)view.len, 0);
PyBuffer_Release(&view); PyBuffer_Release(&view);
return res; return res;
} }
@ -3903,6 +3919,7 @@ _elementtree_XMLParser__parse_whole(XMLParserObject *self, PyObject *file)
return NULL; return NULL;
/* read from open file object */ /* read from open file object */
elementtreestate *st = self->state;
for (;;) { for (;;) {
buffer = PyObject_CallFunction(reader, "i", 64*1024); buffer = PyObject_CallFunction(reader, "i", 64*1024);
@ -3940,8 +3957,8 @@ _elementtree_XMLParser__parse_whole(XMLParserObject *self, PyObject *file)
return NULL; return NULL;
} }
res = expat_parse( res = expat_parse(
self, PyBytes_AS_STRING(buffer), (int)PyBytes_GET_SIZE(buffer), 0 st, self, PyBytes_AS_STRING(buffer), (int)PyBytes_GET_SIZE(buffer),
); 0);
Py_DECREF(buffer); Py_DECREF(buffer);
@ -3955,9 +3972,8 @@ _elementtree_XMLParser__parse_whole(XMLParserObject *self, PyObject *file)
Py_DECREF(reader); Py_DECREF(reader);
res = expat_parse(self, "", 0, 1); res = expat_parse(st, self, "", 0, 1);
elementtreestate *st = self->state;
if (res && TreeBuilder_CheckExact(st, self->target)) { if (res && TreeBuilder_CheckExact(st, self->target)) {
Py_DECREF(res); Py_DECREF(res);
return treebuilder_done((TreeBuilderObject*) self->target); return treebuilder_done((TreeBuilderObject*) self->target);