gh-113732: Fix support of QUOTE_NOTNULL and QUOTE_STRINGS in csv.reader (GH-113738)

This commit is contained in:
Serhiy Storchaka 2024-01-30 14:21:12 +02:00 committed by GitHub
parent 58f883b91b
commit ea30a28c3e
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
4 changed files with 57 additions and 18 deletions

View file

@ -690,7 +690,7 @@ csv
* Add :const:`csv.QUOTE_NOTNULL` and :const:`csv.QUOTE_STRINGS` flags to * Add :const:`csv.QUOTE_NOTNULL` and :const:`csv.QUOTE_STRINGS` flags to
provide finer grained control of ``None`` and empty strings by provide finer grained control of ``None`` and empty strings by
:class:`csv.writer` objects. :class:`~csv.reader` and :class:`~csv.writer` objects.
dis dis
--- ---

View file

@ -392,10 +392,26 @@ class Test_Csv(unittest.TestCase):
# will this fail where locale uses comma for decimals? # will this fail where locale uses comma for decimals?
self._read_test([',3,"5",7.3, 9'], [['', 3, '5', 7.3, 9]], self._read_test([',3,"5",7.3, 9'], [['', 3, '5', 7.3, 9]],
quoting=csv.QUOTE_NONNUMERIC) quoting=csv.QUOTE_NONNUMERIC)
self._read_test([',3,"5",7.3, 9'], [[None, '3', '5', '7.3', ' 9']],
quoting=csv.QUOTE_NOTNULL)
self._read_test([',3,"5",7.3, 9'], [[None, 3, '5', 7.3, 9]],
quoting=csv.QUOTE_STRINGS)
self._read_test([',,"",'], [['', '', '', '']])
self._read_test([',,"",'], [['', '', '', '']],
quoting=csv.QUOTE_NONNUMERIC)
self._read_test([',,"",'], [[None, None, '', None]],
quoting=csv.QUOTE_NOTNULL)
self._read_test([',,"",'], [[None, None, '', None]],
quoting=csv.QUOTE_STRINGS)
self._read_test(['"a\nb", 7'], [['a\nb', ' 7']]) self._read_test(['"a\nb", 7'], [['a\nb', ' 7']])
self.assertRaises(ValueError, self._read_test, self.assertRaises(ValueError, self._read_test,
['abc,3'], [[]], ['abc,3'], [[]],
quoting=csv.QUOTE_NONNUMERIC) quoting=csv.QUOTE_NONNUMERIC)
self.assertRaises(ValueError, self._read_test,
['abc,3'], [[]],
quoting=csv.QUOTE_STRINGS)
self._read_test(['1,@,3,@,5'], [['1', ',3,', '5']], quotechar='@') self._read_test(['1,@,3,@,5'], [['1', ',3,', '5']], quotechar='@')
self._read_test(['1,\0,3,\0,5'], [['1', ',3,', '5']], quotechar='\0') self._read_test(['1,\0,3,\0,5'], [['1', ',3,', '5']], quotechar='\0')
@ -403,6 +419,15 @@ class Test_Csv(unittest.TestCase):
self._read_test(['no space, space, spaces,\ttab'], self._read_test(['no space, space, spaces,\ttab'],
[['no space', 'space', 'spaces', '\ttab']], [['no space', 'space', 'spaces', '\ttab']],
skipinitialspace=True) skipinitialspace=True)
self._read_test([' , , '],
[['', '', '']],
skipinitialspace=True)
self._read_test([' , , '],
[[None, None, None]],
skipinitialspace=True, quoting=csv.QUOTE_NOTNULL)
self._read_test([' , , '],
[[None, None, None]],
skipinitialspace=True, quoting=csv.QUOTE_STRINGS)
def test_read_bigfield(self): def test_read_bigfield(self):
# This exercises the buffer realloc functionality and field size # This exercises the buffer realloc functionality and field size

View file

@ -0,0 +1,2 @@
Fix support of :data:`~csv.QUOTE_NOTNULL` and :data:`~csv.QUOTE_STRINGS` in
:func:`csv.reader`.

View file

@ -131,7 +131,7 @@ typedef struct {
Py_UCS4 *field; /* temporary buffer */ Py_UCS4 *field; /* temporary buffer */
Py_ssize_t field_size; /* size of allocated buffer */ Py_ssize_t field_size; /* size of allocated buffer */
Py_ssize_t field_len; /* length of current field */ Py_ssize_t field_len; /* length of current field */
int numeric_field; /* treat field as numeric */ bool unquoted_field; /* true if no quotes around the current field */
unsigned long line_num; /* Source-file line number */ unsigned long line_num; /* Source-file line number */
} ReaderObj; } ReaderObj;
@ -644,22 +644,33 @@ _call_dialect(_csvstate *module_state, PyObject *dialect_inst, PyObject *kwargs)
static int static int
parse_save_field(ReaderObj *self) parse_save_field(ReaderObj *self)
{ {
int quoting = self->dialect->quoting;
PyObject *field; PyObject *field;
field = PyUnicode_FromKindAndData(PyUnicode_4BYTE_KIND, if (self->unquoted_field &&
(void *) self->field, self->field_len); self->field_len == 0 &&
if (field == NULL) (quoting == QUOTE_NOTNULL || quoting == QUOTE_STRINGS))
return -1; {
self->field_len = 0; field = Py_NewRef(Py_None);
if (self->numeric_field) { }
PyObject *tmp; else {
field = PyUnicode_FromKindAndData(PyUnicode_4BYTE_KIND,
self->numeric_field = 0; (void *) self->field, self->field_len);
tmp = PyNumber_Float(field); if (field == NULL) {
Py_DECREF(field);
if (tmp == NULL)
return -1; return -1;
field = tmp; }
if (self->unquoted_field &&
self->field_len != 0 &&
(quoting == QUOTE_NONNUMERIC || quoting == QUOTE_STRINGS))
{
PyObject *tmp = PyNumber_Float(field);
Py_DECREF(field);
if (tmp == NULL) {
return -1;
}
field = tmp;
}
self->field_len = 0;
} }
if (PyList_Append(self->fields, field) < 0) { if (PyList_Append(self->fields, field) < 0) {
Py_DECREF(field); Py_DECREF(field);
@ -721,6 +732,7 @@ parse_process_char(ReaderObj *self, _csvstate *module_state, Py_UCS4 c)
/* fallthru */ /* fallthru */
case START_FIELD: case START_FIELD:
/* expecting field */ /* expecting field */
self->unquoted_field = true;
if (c == '\n' || c == '\r' || c == EOL) { if (c == '\n' || c == '\r' || c == EOL) {
/* save empty field - return [fields] */ /* save empty field - return [fields] */
if (parse_save_field(self) < 0) if (parse_save_field(self) < 0)
@ -730,10 +742,12 @@ parse_process_char(ReaderObj *self, _csvstate *module_state, Py_UCS4 c)
else if (c == dialect->quotechar && else if (c == dialect->quotechar &&
dialect->quoting != QUOTE_NONE) { dialect->quoting != QUOTE_NONE) {
/* start quoted field */ /* start quoted field */
self->unquoted_field = false;
self->state = IN_QUOTED_FIELD; self->state = IN_QUOTED_FIELD;
} }
else if (c == dialect->escapechar) { else if (c == dialect->escapechar) {
/* possible escaped character */ /* possible escaped character */
self->unquoted_field = false;
self->state = ESCAPED_CHAR; self->state = ESCAPED_CHAR;
} }
else if (c == ' ' && dialect->skipinitialspace) else if (c == ' ' && dialect->skipinitialspace)
@ -746,8 +760,6 @@ parse_process_char(ReaderObj *self, _csvstate *module_state, Py_UCS4 c)
} }
else { else {
/* begin new unquoted field */ /* begin new unquoted field */
if (dialect->quoting == QUOTE_NONNUMERIC)
self->numeric_field = 1;
if (parse_add_char(self, module_state, c) < 0) if (parse_add_char(self, module_state, c) < 0)
return -1; return -1;
self->state = IN_FIELD; self->state = IN_FIELD;
@ -892,7 +904,7 @@ parse_reset(ReaderObj *self)
return -1; return -1;
self->field_len = 0; self->field_len = 0;
self->state = START_RECORD; self->state = START_RECORD;
self->numeric_field = 0; self->unquoted_field = false;
return 0; return 0;
} }