[Python-3000-checkins] r58934 - in python/branches/py3k: Lib/test/test_strlit.py Python/ast.c
guido.van.rossum
python-3000-checkins at python.org
Mon Nov 12 02:13:56 CET 2007
Author: guido.van.rossum
Date: Mon Nov 12 02:13:56 2007
New Revision: 58934
Added:
python/branches/py3k/Lib/test/test_strlit.py (contents, props changed)
Modified:
python/branches/py3k/Python/ast.c
Log:
Minimal change that disables (AFAICT) the interpolation of \u and \U inside
raw string literals. I added a whole bunch of tests but am still not sure
I am testing all paths through the code. I really think the code could be
simplified quite a bit.
Added: python/branches/py3k/Lib/test/test_strlit.py
==============================================================================
--- (empty file)
+++ python/branches/py3k/Lib/test/test_strlit.py Mon Nov 12 02:13:56 2007
@@ -0,0 +1,148 @@
+r"""Test correct treatment of various string literals by the parser.
+
+There are four types of string literals:
+
+ 'abc' -- normal str
+ r'abc' -- raw str
+ b'xyz' -- normal bytes
+ br'xyz' -- raw bytes
+
+The difference between normal and raw strings is of course that in a
+raw string, \ escapes (while still used to determine the end of the
+literal) are not interpreted, so that r'\x00' contains four
+characters: a backslash, an x, and two zeros; while '\x00' contains a
+single character (code point zero).
+
+The tricky thing is what should happen when non-ASCII bytes are used
+inside literals. For bytes literals, this is considered illegal. But
+for str literals, those bytes are supposed to be decoded using the
+encoding declared for the file (UTF-8 by default).
+
+We have to test this with various file encodings. We also test it with
+exec()/eval(), which uses a different code path.
+
+This file is really about correct treatment of encodings and
+backslashes. It doens't concern itself with issues like single
+vs. double quotes or singly- vs. triply-quoted strings: that's dealt
+with elsewhere (I assume).
+"""
+
+import os
+import sys
+import shutil
+import tempfile
+import unittest
+
+
+TEMPLATE = r"""# coding: %s
+a = 'x'
+assert ord(a) == 120
+b = '\x01'
+assert ord(b) == 1
+c = r'\x01'
+assert list(map(ord, c)) == [92, 120, 48, 49]
+d = '\x81'
+assert ord(d) == 0x81
+e = r'\x81'
+assert list(map(ord, e)) == [92, 120, 56, 49]
+f = '\u1881'
+assert ord(f) == 0x1881
+g = r'\u1881'
+assert list(map(ord, g)) == [92, 117, 49, 56, 56, 49]
+"""
+
+
+def byte(i):
+ return bytes([i])
+
+
+class TestLiterals(unittest.TestCase):
+
+ def setUp(self):
+ self.save_path = sys.path[:]
+ self.tmpdir = tempfile.mkdtemp()
+ sys.path.insert(0, self.tmpdir)
+
+ def tearDown(self):
+ sys.path = self.save_path
+ shutil.rmtree(self.tmpdir, ignore_errors=True)
+
+ def test_template(self):
+ # Check that the template doesn't contain any non-printables
+ # except for \n.
+ for c in TEMPLATE:
+ assert c == '\n' or ' ' <= c <= '~', repr(c)
+
+ def test_eval_str_normal(self):
+ self.assertEqual(eval(""" 'x' """), 'x')
+ self.assertEqual(eval(r""" '\x01' """), chr(1))
+ self.assertEqual(eval(""" '\x01' """), chr(1))
+ self.assertEqual(eval(r""" '\x81' """), chr(0x81))
+ self.assertEqual(eval(""" '\x81' """), chr(0x81))
+ self.assertEqual(eval(r""" '\u1881' """), chr(0x1881))
+ self.assertEqual(eval(""" '\u1881' """), chr(0x1881))
+
+ def test_eval_str_raw(self):
+ self.assertEqual(eval(""" r'x' """), 'x')
+ self.assertEqual(eval(r""" r'\x01' """), '\\' + 'x01')
+ self.assertEqual(eval(""" r'\x01' """), chr(1))
+ self.assertEqual(eval(r""" r'\x81' """), '\\' + 'x81')
+ self.assertEqual(eval(""" r'\x81' """), chr(0x81))
+ self.assertEqual(eval(r""" r'\u1881' """), '\\' + 'u1881')
+ self.assertEqual(eval(""" r'\u1881' """), chr(0x1881))
+
+ def test_eval_bytes_normal(self):
+ self.assertEqual(eval(""" b'x' """), b'x')
+ self.assertEqual(eval(r""" b'\x01' """), byte(1))
+ self.assertEqual(eval(""" b'\x01' """), byte(1))
+ self.assertEqual(eval(r""" b'\x81' """), byte(0x81))
+ self.assertRaises(SyntaxError, eval, """ b'\x81' """)
+ self.assertEqual(eval(r""" b'\u1881' """), b'\\' + b'u1881')
+ self.assertRaises(SyntaxError, eval, """ b'\u1881' """)
+
+ def test_eval_bytes_raw(self):
+ self.assertEqual(eval(""" br'x' """), b'x')
+ self.assertEqual(eval(r""" br'\x01' """), b'\\' + b'x01')
+ self.assertEqual(eval(""" br'\x01' """), byte(1))
+ self.assertEqual(eval(r""" br'\x81' """), b"\\" + b"x81")
+ self.assertRaises(SyntaxError, eval, """ br'\x81' """)
+ self.assertEqual(eval(r""" br'\u1881' """), b"\\" + b"u1881")
+ self.assertRaises(SyntaxError, eval, """ br'\u1881' """)
+
+ def check_encoding(self, encoding, extra=""):
+ modname = "xx_" + encoding.replace("-", "_")
+ fn = os.path.join(self.tmpdir, modname + ".py")
+ f = open(fn, "w", encoding=encoding)
+ try:
+ f.write(TEMPLATE % encoding)
+ f.write(extra)
+ finally:
+ f.close()
+ __import__(modname)
+ del sys.modules[modname]
+
+ def test_file_utf_8(self):
+ extra = "z = '\u1234'; assert ord(z) == 0x1234\n"
+ self.check_encoding("utf-8", extra)
+
+ def test_file_utf_8_error(self):
+ extra = "b'\x80'\n"
+ self.assertRaises(SyntaxError, self.check_encoding, "utf-8", extra)
+
+ def test_file_utf8(self):
+ self.check_encoding("utf8")
+
+ def test_file_iso_8859_1(self):
+ self.check_encoding("iso-8859-1")
+
+ def test_file_latin_1(self):
+ self.check_encoding("latin-1")
+
+ def test_file_latin9(self):
+ self.check_encoding("latin9")
+
+
+if __name__ == "__main__":
+ # Hack so that error messages containing non-ASCII can be printed
+ sys.stdout._encoding = sys.stderr._encoding = "utf-8"
+ unittest.main()
Modified: python/branches/py3k/Python/ast.c
==============================================================================
--- python/branches/py3k/Python/ast.c (original)
+++ python/branches/py3k/Python/ast.c Mon Nov 12 02:13:56 2007
@@ -1292,7 +1292,7 @@
case STRING: {
PyObject *str = parsestrplus(c, n, &bytesmode);
if (!str) {
- if (PyErr_ExceptionMatches(PyExc_UnicodeError)){
+ if (PyErr_ExceptionMatches(PyExc_UnicodeError)) {
PyObject *type, *value, *tback, *errstr;
PyErr_Fetch(&type, &value, &tback);
errstr = ((PyUnicodeErrorObject *)value)->reason;
@@ -3117,6 +3117,7 @@
char *buf;
char *p;
const char *end;
+
if (encoding == NULL) {
buf = (char *)s;
u = NULL;
@@ -3218,7 +3219,7 @@
return NULL;
}
}
- if (!*bytesmode) {
+ if (!*bytesmode && !rawmode) {
return decode_unicode(s, len, rawmode, encoding);
}
if (*bytesmode) {
@@ -3238,13 +3239,17 @@
if (rawmode || strchr(s, '\\') == NULL) {
if (need_encoding) {
PyObject *v, *u = PyUnicode_DecodeUTF8(s, len, NULL);
- if (u == NULL)
- return NULL;
+ if (u == NULL || !*bytesmode)
+ return u;
v = PyUnicode_AsEncodedString(u, encoding, NULL);
Py_DECREF(u);
return v;
- } else {
+ } else if (*bytesmode) {
return PyString_FromStringAndSize(s, len);
+ } else if (strcmp(encoding, "utf-8") == 0) {
+ return PyUnicode_FromStringAndSize(s, len);
+ } else {
+ return PyUnicode_DecodeLatin1(s, len, NULL);
}
}
@@ -3252,7 +3257,7 @@
need_encoding ? encoding : NULL);
}
-/* Build a Python string object out of a STRING atom. This takes care of
+/* Build a Python string object out of a STRING+ atom. This takes care of
* compile-time literal catenation, calling parsestr() on each piece, and
* pasting the intermediate results together.
*/
@@ -3272,8 +3277,7 @@
if (s == NULL)
goto onError;
if (*bytesmode != subbm) {
- ast_error(n, "cannot mix bytes and nonbytes"
- "literals");
+ ast_error(n, "cannot mix bytes and nonbytes literals");
goto onError;
}
if (PyString_Check(v) && PyString_Check(s)) {
More information about the Python-3000-checkins
mailing list