bpo-20028: Empty escapechar/quotechar is not allowed for csv.Dialect (GH-28833)

corona10 · web-flow · commit ab62051152cb · 2021-10-11T20:08:15.000+09:00
diff --git a/Doc/library/csv.rst b/Doc/library/csv.rst
@@ -383,6 +383,8 @@ Dialects support the following attributes:
    :const:`False`. On reading, the *escapechar* removes any special meaning from
    the following character. It defaults to :const:`None`, which disables escaping.
 
+   .. versionchanged:: 3.11
+      An empty *escapechar* is not allowed.
 
 .. attribute:: Dialect.lineterminator
 
@@ -402,6 +404,8 @@ Dialects support the following attributes:
    as the *delimiter* or *quotechar*, or which contain new-line characters.  It
    defaults to ``'"'``.
 
+   .. versionchanged:: 3.11
+      An empty *quotechar* is not allowed.
 
 .. attribute:: Dialect.quoting
 
diff --git a/Lib/test/test_csv.py b/Lib/test/test_csv.py
@@ -44,6 +44,8 @@ def _test_arg_valid(self, ctor, arg):
                           quoting=csv.QUOTE_ALL, quotechar='')
         self.assertRaises(TypeError, ctor, arg,
                           quoting=csv.QUOTE_ALL, quotechar=None)
+        self.assertRaises(TypeError, ctor, arg,
+                          quoting=csv.QUOTE_NONE, quotechar='')
 
     def test_reader_arg_valid(self):
         self._test_arg_valid(csv.reader, [])
@@ -342,7 +344,6 @@ def test_read_escape(self):
         self._read_test(['a,^b,c'], [['a', 'b', 'c']], escapechar='^')
         self._read_test(['a,\0b,c'], [['a', 'b', 'c']], escapechar='\0')
         self._read_test(['a,\\b,c'], [['a', '\\b', 'c']], escapechar=None)
-        self._read_test(['a,\\b,c'], [['a', '\\b', 'c']], escapechar='')
         self._read_test(['a,\\b,c'], [['a', '\\b', 'c']])
 
     def test_read_quoting(self):
@@ -913,6 +914,12 @@ class mydialect(csv.Dialect):
         self.assertEqual(d.quotechar, '"')
         self.assertTrue(d.doublequote)
 
+        mydialect.quotechar = ""
+        with self.assertRaises(csv.Error) as cm:
+            mydialect()
+        self.assertEqual(str(cm.exception),
+                         '"quotechar" must be a 1-character string')
+
         mydialect.quotechar = "''"
         with self.assertRaises(csv.Error) as cm:
             mydialect()
@@ -977,6 +984,10 @@ class mydialect(csv.Dialect):
         d = mydialect()
         self.assertEqual(d.escapechar, "\\")
 
+        mydialect.escapechar = ""
+        with self.assertRaisesRegex(csv.Error, '"escapechar" must be a 1-character string'):
+            mydialect()
+
         mydialect.escapechar = "**"
         with self.assertRaisesRegex(csv.Error, '"escapechar" must be a 1-character string'):
             mydialect()
diff --git a/Misc/NEWS.d/next/Library/2021-10-10-00-25-36.bpo-20028.bPx4Z8.rst b/Misc/NEWS.d/next/Library/2021-10-10-00-25-36.bpo-20028.bPx4Z8.rst
@@ -0,0 +1,2 @@
+Empty escapechar/quotechar is not allowed when initializing
+:class:`csv.Dialect`. Patch by Vajrasky Kok and Dong-hee Na.
diff --git a/Modules/_csv.c b/Modules/_csv.c
@@ -250,16 +250,14 @@ _set_char_or_none(const char *name, Py_UCS4 *target, PyObject *src, Py_UCS4 dflt
             if (len < 0) {
                 return -1;
             }
-            if (len > 1) {
+            if (len != 1) {
                 PyErr_Format(PyExc_TypeError,
                     "\"%s\" must be a 1-character string",
                     name);
                 return -1;
             }
             /* PyUnicode_READY() is called in PyUnicode_GetLength() */
-            else if (len > 0) {
-                *target = PyUnicode_READ_CHAR(src, 0);
-            }
+            *target = PyUnicode_READ_CHAR(src, 0);
         }
     }
     return 0;
@@ -272,7 +270,6 @@ _set_char(const char *name, Py_UCS4 *target, PyObject *src, Py_UCS4 dflt)
         *target = dflt;
     }
     else {
-        *target = NOT_SET;
         if (!PyUnicode_Check(src)) {
             PyErr_Format(PyExc_TypeError,
                          "\"%s\" must be string, not %.200s", name,
@@ -283,16 +280,14 @@ _set_char(const char *name, Py_UCS4 *target, PyObject *src, Py_UCS4 dflt)
         if (len < 0) {
             return -1;
         }
-        if (len > 1) {
+        if (len != 1) {
             PyErr_Format(PyExc_TypeError,
                          "\"%s\" must be a 1-character string",
                          name);
             return -1;
         }
         /* PyUnicode_READY() is called in PyUnicode_GetLength() */
-        else if (len > 0) {
-            *target = PyUnicode_READ_CHAR(src, 0);
-        }
+        *target = PyUnicode_READ_CHAR(src, 0);
     }
     return 0;
 }

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,2 @@`
	`1`	`+Empty escapechar/quotechar is not allowed when initializing`
	`2`	+:class:`csv.Dialect`. Patch by Vajrasky Kok and Dong-hee Na.
Original file line number	Diff line number	Diff line change
`@@ -250,16 +250,14 @@ _set_char_or_none(const char name, Py_UCS4 target, PyObject *src, Py_UCS4 dflt`
`250`	`250`	`if (len < 0) {`
`251`	`251`	`return -1;`
`252`	`252`	`}`
`253`		`- if (len > 1) {`
	`253`	`+ if (len != 1) {`
`254`	`254`	`PyErr_Format(PyExc_TypeError,`
`255`	`255`	`"\"%s\" must be a 1-character string",`
`256`	`256`	`name);`
`257`	`257`	`return -1;`
`258`	`258`	`}`
`259`	`259`	`/* PyUnicode_READY() is called in PyUnicode_GetLength() */`
`260`		`- else if (len > 0) {`
`261`		`- *target = PyUnicode_READ_CHAR(src, 0);`
`262`		`- }`
	`260`	`+ *target = PyUnicode_READ_CHAR(src, 0);`
`263`	`261`	`}`
`264`	`262`	`}`
`265`	`263`	`return 0;`
`@@ -272,7 +270,6 @@ _set_char(const char name, Py_UCS4 target, PyObject *src, Py_UCS4 dflt)`
`272`	`270`	`*target = dflt;`
`273`	`271`	`}`
`274`	`272`	`else {`
`275`		`- *target = NOT_SET;`
`276`	`273`	`if (!PyUnicode_Check(src)) {`
`277`	`274`	`PyErr_Format(PyExc_TypeError,`
`278`	`275`	`"\"%s\" must be string, not %.200s", name,`
`@@ -283,16 +280,14 @@ _set_char(const char name, Py_UCS4 target, PyObject *src, Py_UCS4 dflt)`
`283`	`280`	`if (len < 0) {`
`284`	`281`	`return -1;`
`285`	`282`	`}`
`286`		`- if (len > 1) {`
	`283`	`+ if (len != 1) {`
`287`	`284`	`PyErr_Format(PyExc_TypeError,`
`288`	`285`	`"\"%s\" must be a 1-character string",`
`289`	`286`	`name);`
`290`	`287`	`return -1;`
`291`	`288`	`}`
`292`	`289`	`/* PyUnicode_READY() is called in PyUnicode_GetLength() */`
`293`		`- else if (len > 0) {`
`294`		`- *target = PyUnicode_READ_CHAR(src, 0);`
`295`		`- }`
	`290`	`+ *target = PyUnicode_READ_CHAR(src, 0);`
`296`	`291`	`}`
`297`	`292`	`return 0;`
`298`	`293`	`}`