[3.12] gh-115712: Support CSV dialects with delimiter=' ' and skipinitialspace=True...

csv.writer() now quotes empty fields if delimiter is a space and
skipinitialspace is true and raises exception if quoting is not possible.
(cherry picked from commit 937d2821501de7adaa5ed8491eef4b7f3dc0940a)

diff --git a/Lib/test/test_csv.py b/Lib/test/test_csv.py
index 30383698d04184ea9bcd6ae2c1e0ad5c4048184e..66c13751b19d09785503bfa25e18169b1723849c 100644 (file)
--- a/Lib/test/test_csv.py
+++ b/Lib/test/test_csv.py
@@ -46,6 +46,20 @@ class Test_Csv(unittest.TestCase):
                           quoting=csv.QUOTE_ALL, quotechar=None)
         self.assertRaises(TypeError, ctor, arg,
                           quoting=csv.QUOTE_NONE, quotechar='')
+        ctor(arg, delimiter=' ')
+        ctor(arg, escapechar=' ')
+        ctor(arg, quotechar=' ')
+        ctor(arg, delimiter='\t', skipinitialspace=True)
+        ctor(arg, escapechar='\t', skipinitialspace=True)
+        ctor(arg, quotechar='\t', skipinitialspace=True)
+        ctor(arg, delimiter=' ', skipinitialspace=True)
+        ctor(arg, delimiter='^')
+        ctor(arg, escapechar='^')
+        ctor(arg, quotechar='^')
+        ctor(arg, delimiter='\x85')
+        ctor(arg, escapechar='\x85')
+        ctor(arg, quotechar='\x85')
+        ctor(arg, lineterminator='\x85')
 
     def test_reader_arg_valid(self):
         self._test_arg_valid(csv.reader, [])
@@ -152,9 +166,6 @@ class Test_Csv(unittest.TestCase):
 
     def test_write_arg_valid(self):
         self._write_error_test(csv.Error, None)
-        self._write_test((), '')
-        self._write_test([None], '""')
-        self._write_error_test(csv.Error, [None], quoting = csv.QUOTE_NONE)
         # Check that exceptions are passed up the chain
         self._write_error_test(OSError, BadIterable())
         class BadList:
@@ -168,7 +179,6 @@ class Test_Csv(unittest.TestCase):
             def __str__(self):
                 raise OSError
         self._write_error_test(OSError, [BadItem()])
-
     def test_write_bigfield(self):
         # This exercises the buffer realloc functionality
         bigstring = 'X' * 50000
@@ -275,6 +285,49 @@ class Test_Csv(unittest.TestCase):
             fileobj.seek(0)
             self.assertEqual(fileobj.read(), 'a\r\n""\r\n')
 
+
+    def test_write_empty_fields(self):
+        self._write_test((), '')
+        self._write_test([''], '""')
+        self._write_error_test(csv.Error, [''], quoting=csv.QUOTE_NONE)
+        self._write_test([''], '""', quoting=csv.QUOTE_STRINGS)
+        self._write_test([''], '""', quoting=csv.QUOTE_NOTNULL)
+        self._write_test([None], '""')
+        self._write_error_test(csv.Error, [None], quoting=csv.QUOTE_NONE)
+        self._write_error_test(csv.Error, [None], quoting=csv.QUOTE_STRINGS)
+        self._write_error_test(csv.Error, [None], quoting=csv.QUOTE_NOTNULL)
+        self._write_test(['', ''], ',')
+        self._write_test([None, None], ',')
+
+    def test_write_empty_fields_space_delimiter(self):
+        self._write_test([''], '""', delimiter=' ', skipinitialspace=False)
+        self._write_test([''], '""', delimiter=' ', skipinitialspace=True)
+        self._write_test([None], '""', delimiter=' ', skipinitialspace=False)
+        self._write_test([None], '""', delimiter=' ', skipinitialspace=True)
+
+        self._write_test(['', ''], ' ', delimiter=' ', skipinitialspace=False)
+        self._write_test(['', ''], '"" ""', delimiter=' ', skipinitialspace=True)
+        self._write_test([None, None], ' ', delimiter=' ', skipinitialspace=False)
+        self._write_test([None, None], '"" ""', delimiter=' ', skipinitialspace=True)
+
+        self._write_test(['', ''], ' ', delimiter=' ', skipinitialspace=False,
+                         quoting=csv.QUOTE_NONE)
+        self._write_error_test(csv.Error, ['', ''],
+                               delimiter=' ', skipinitialspace=True,
+                               quoting=csv.QUOTE_NONE)
+        for quoting in csv.QUOTE_STRINGS, csv.QUOTE_NOTNULL:
+            self._write_test(['', ''], '"" ""', delimiter=' ', skipinitialspace=False,
+                             quoting=quoting)
+            self._write_test(['', ''], '"" ""', delimiter=' ', skipinitialspace=True,
+                             quoting=quoting)
+
+        for quoting in csv.QUOTE_NONE, csv.QUOTE_STRINGS, csv.QUOTE_NOTNULL:
+            self._write_test([None, None], ' ', delimiter=' ', skipinitialspace=False,
+                             quoting=quoting)
+            self._write_error_test(csv.Error, [None, None],
+                                   delimiter=' ', skipinitialspace=True,
+                                   quoting=quoting)
+
     def test_writerows_errors(self):
         with TemporaryFile("w+", encoding="utf-8", newline='') as fileobj:
             writer = csv.writer(fileobj)
@@ -376,6 +429,14 @@ class Test_Csv(unittest.TestCase):
                         [['no space', 'space', 'spaces', '\ttab']],
                         skipinitialspace=True)
 
+    def test_read_space_delimiter(self):
+        self._read_test(['a   b', '  a  ', '  ', ''],
+                        [['a', '', '', 'b'], ['', '', 'a', '', ''], ['', '', ''], []],
+                        delimiter=' ', skipinitialspace=False)
+        self._read_test(['a   b', '  a  ', '  ', ''],
+                        [['a', 'b'], ['a', ''], [''], []],
+                        delimiter=' ', skipinitialspace=True)
+
     def test_read_bigfield(self):
         # This exercises the buffer realloc functionality and field size
         # limits.
@@ -502,10 +563,10 @@ class TestDialectRegistry(unittest.TestCase):
             escapechar = "\\"
 
         with TemporaryFile("w+", encoding="utf-8") as fileobj:
-            fileobj.write("abc def\nc1ccccc1 benzene\n")
+            fileobj.write("abc   def\nc1ccccc1 benzene\n")
             fileobj.seek(0)
             reader = csv.reader(fileobj, dialect=space())
-            self.assertEqual(next(reader), ["abc", "def"])
+            self.assertEqual(next(reader), ["abc", "", "", "def"])
             self.assertEqual(next(reader), ["c1ccccc1", "benzene"])
 
     def compare_dialect_123(self, expected, *writeargs, **kwwriteargs):

diff --git a/Misc/NEWS.d/next/Library/2024-02-20-16-42-54.gh-issue-115712.EXVMXw.rst b/Misc/NEWS.d/next/Library/2024-02-20-16-42-54.gh-issue-115712.EXVMXw.rst
new file mode 100644 (file)
index 0000000..70243dc
--- /dev/null
+++ b/Misc/NEWS.d/next/Library/2024-02-20-16-42-54.gh-issue-115712.EXVMXw.rst
@@ -0,0 +1,3 @@
+:func:`csv.writer()` now quotes empty fields if delimiter is a
+space and skipinitialspace is true and raises exception if quoting is not
+possible.

diff --git a/Modules/_csv.c b/Modules/_csv.c
index 91cb63628a1f7b49beef2aa0c26ae14893800167..23437a3bfacdff2e18cd8e31b618cb940ff2d0a2 100644 (file)
--- a/Modules/_csv.c
+++ b/Modules/_csv.c
@@ -1180,6 +1180,7 @@ join_check_rec_size(WriterObj *self, Py_ssize_t rec_len)
 static int
 join_append(WriterObj *self, PyObject *field, int quoted)
 {
+    DialectObj *dialect = self->dialect;
     int field_kind = -1;
     const void *field_data = NULL;
     Py_ssize_t field_len = 0;
@@ -1192,6 +1193,19 @@ join_append(WriterObj *self, PyObject *field, int quoted)
         field_data = PyUnicode_DATA(field);
         field_len = PyUnicode_GET_LENGTH(field);
     }
+    if (!field_len && dialect->delimiter == ' ' && dialect->skipinitialspace) {
+        if (dialect->quoting == QUOTE_NONE ||
+            (field == NULL &&
+             (dialect->quoting == QUOTE_STRINGS ||
+              dialect->quoting == QUOTE_NOTNULL)))
+        {
+            PyErr_Format(self->error_obj,
+                         "empty field must be quoted if delimiter is a space "
+                         "and skipinitialspace is true");
+            return 0;
+        }
+        quoted = 1;
+    }
     rec_len = join_append_data(self, field_kind, field_data, field_len,
                                &quoted, 0);
     if (rec_len < 0)
@@ -1243,6 +1257,7 @@ csv_writerow(WriterObj *self, PyObject *seq)
 {
     DialectObj *dialect = self->dialect;
     PyObject *iter, *field, *line, *result;
+    bool null_field = false;
 
     iter = PyObject_GetIter(seq);
     if (iter == NULL) {
@@ -1279,11 +1294,12 @@ csv_writerow(WriterObj *self, PyObject *seq)
             break;
         }
 
+        null_field = (field == Py_None);
         if (PyUnicode_Check(field)) {
             append_ok = join_append(self, field, quoted);
             Py_DECREF(field);
         }
-        else if (field == Py_None) {
+        else if (null_field) {
             append_ok = join_append(self, NULL, quoted);
             Py_DECREF(field);
         }
@@ -1309,7 +1325,11 @@ csv_writerow(WriterObj *self, PyObject *seq)
         return NULL;
 
     if (self->num_fields > 0 && self->rec_len == 0) {
-        if (dialect->quoting == QUOTE_NONE) {
+        if (dialect->quoting == QUOTE_NONE ||
+            (null_field &&
+             (dialect->quoting == QUOTE_STRINGS ||
+              dialect->quoting == QUOTE_NOTNULL)))
+        {
             PyErr_Format(self->error_obj,
                 "single empty field record must be quoted");
             return NULL;