bpo-37050: Remove expr_text from FormattedValue ast node, use Constant node instead (GH-13597)

When using the "=" debug functionality of f-strings, use another Constant node (or a merged constant node) instead of adding expr_text to the FormattedValue node.
This commit is contained in:
Eric V. Smith 2019-05-27 15:31:52 -04:00 committed by GitHub
parent 695b1dd8cb
commit 6f6ff8a565
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
9 changed files with 87 additions and 100 deletions

View file

@ -5006,10 +5006,16 @@ fstring_parse(const char **str, const char *end, int raw, int recurse_lvl,
closing brace doesn't match an opening paren, for example. It
doesn't need to error on all invalid expressions, just correctly
find the end of all valid ones. Any errors inside the expression
will be caught when we parse it later. */
will be caught when we parse it later.
*expression is set to the expression. For an '=' "debug" expression,
*expr_text is set to the debug text (the original text of the expression,
*including the '=' and any whitespace around it, as a string object). If
*not a debug expression, *expr_text set to NULL. */
static int
fstring_find_expr(const char **str, const char *end, int raw, int recurse_lvl,
expr_ty *expression, struct compiling *c, const node *n)
PyObject **expr_text, expr_ty *expression,
struct compiling *c, const node *n)
{
/* Return -1 on error, else 0. */
@ -5020,9 +5026,6 @@ fstring_find_expr(const char **str, const char *end, int raw, int recurse_lvl,
int conversion = -1; /* The conversion char. Use default if not
specified, or !r if using = and no format
spec. */
int equal_flag = 0; /* Are we using the = feature? */
PyObject *expr_text = NULL; /* The text of the expression, used for =. */
const char *expr_text_end;
/* 0 if we're not in a string, else the quote char we're trying to
match (single or double quote). */
@ -5198,7 +5201,6 @@ fstring_find_expr(const char **str, const char *end, int raw, int recurse_lvl,
expr_text. */
if (**str == '=') {
*str += 1;
equal_flag = 1;
/* Skip over ASCII whitespace. No need to test for end of string
here, since we know there's at least a trailing quote somewhere
@ -5206,7 +5208,14 @@ fstring_find_expr(const char **str, const char *end, int raw, int recurse_lvl,
while (Py_ISSPACE(**str)) {
*str += 1;
}
expr_text_end = *str;
/* Set *expr_text to the text of the expression. */
*expr_text = PyUnicode_FromStringAndSize(expr_start, *str-expr_start);
if (!*expr_text) {
goto error;
}
} else {
*expr_text = NULL;
}
/* Check for a conversion char, if present. */
@ -5227,17 +5236,6 @@ fstring_find_expr(const char **str, const char *end, int raw, int recurse_lvl,
}
}
if (equal_flag) {
Py_ssize_t len = expr_text_end - expr_start;
expr_text = PyUnicode_FromStringAndSize(expr_start, len);
if (!expr_text) {
goto error;
}
if (PyArena_AddPyObject(c->c_arena, expr_text) < 0) {
Py_DECREF(expr_text);
goto error;
}
}
/* Check for the format spec, if present. */
if (*str >= end)
@ -5261,16 +5259,16 @@ fstring_find_expr(const char **str, const char *end, int raw, int recurse_lvl,
assert(**str == '}');
*str += 1;
/* If we're in = mode, and have no format spec and no explict conversion,
set the conversion to 'r'. */
if (equal_flag && format_spec == NULL && conversion == -1) {
/* If we're in = mode (detected by non-NULL expr_text), and have no format
spec and no explict conversion, set the conversion to 'r'. */
if (*expr_text && format_spec == NULL && conversion == -1) {
conversion = 'r';
}
/* And now create the FormattedValue node that represents this
entire expression with the conversion and format spec. */
*expression = FormattedValue(simple_expression, conversion,
format_spec, expr_text, LINENO(n),
format_spec, LINENO(n),
n->n_col_offset, n->n_end_lineno,
n->n_end_col_offset, c->c_arena);
if (!*expression)
@ -5313,7 +5311,7 @@ error:
static int
fstring_find_literal_and_expr(const char **str, const char *end, int raw,
int recurse_lvl, PyObject **literal,
expr_ty *expression,
PyObject **expr_text, expr_ty *expression,
struct compiling *c, const node *n)
{
int result;
@ -5341,7 +5339,8 @@ fstring_find_literal_and_expr(const char **str, const char *end, int raw,
/* We must now be the start of an expression, on a '{'. */
assert(**str == '{');
if (fstring_find_expr(str, end, raw, recurse_lvl, expression, c, n) < 0)
if (fstring_find_expr(str, end, raw, recurse_lvl, expr_text,
expression, c, n) < 0)
goto error;
return 0;
@ -5604,7 +5603,7 @@ FstringParser_ConcatFstring(FstringParser *state, const char **str,
/* Parse the f-string. */
while (1) {
PyObject *literal = NULL;
PyObject *literal[2] = {NULL, NULL};
expr_ty expression = NULL;
/* If there's a zero length literal in front of the
@ -5612,31 +5611,34 @@ FstringParser_ConcatFstring(FstringParser *state, const char **str,
the f-string, expression will be NULL (unless result == 1,
see below). */
int result = fstring_find_literal_and_expr(str, end, raw, recurse_lvl,
&literal, &expression,
c, n);
&literal[0], &literal[1],
&expression, c, n);
if (result < 0)
return -1;
/* Add the literal, if any. */
if (!literal) {
/* Do nothing. Just leave last_str alone (and possibly
NULL). */
} else if (!state->last_str) {
/* Note that the literal can be zero length, if the
input string is "\\\n" or "\\\r", among others. */
state->last_str = literal;
literal = NULL;
} else {
/* We have a literal, concatenate it. */
assert(PyUnicode_GET_LENGTH(literal) != 0);
if (FstringParser_ConcatAndDel(state, literal) < 0)
return -1;
literal = NULL;
/* Add the literals, if any. */
for (int i = 0; i < 2; i++) {
if (!literal[i]) {
/* Do nothing. Just leave last_str alone (and possibly
NULL). */
} else if (!state->last_str) {
/* Note that the literal can be zero length, if the
input string is "\\\n" or "\\\r", among others. */
state->last_str = literal[i];
literal[i] = NULL;
} else {
/* We have a literal, concatenate it. */
assert(PyUnicode_GET_LENGTH(literal[i]) != 0);
if (FstringParser_ConcatAndDel(state, literal[i]) < 0)
return -1;
literal[i] = NULL;
}
}
/* We've dealt with the literal now. It can't be leaked on further
/* We've dealt with the literals now. They can't be leaked on further
errors. */
assert(literal == NULL);
assert(literal[0] == NULL);
assert(literal[1] == NULL);
/* See if we should just loop around to get the next literal
and expression, while ignoring the expression this