Issue #16979: Fix error handling bugs in the unicode-escape-decode decoder.

Lib/test/test_codeccallbacks.py

@@ -262,12 +262,12 @@ class CodecCallbackTest(unittest.TestCase):
 
         self.assertEqual(
             "\\u3042\u3xxx".decode("unicode-escape", "test.handler1"),
-            u"\u3042[<92><117><51><120>]xx"
+            u"\u3042[<92><117><51>]xxx"
         )
 
         self.assertEqual(
             "\\u3042\u3xx".decode("unicode-escape", "test.handler1"),
-            u"\u3042[<92><117><51><120><120>]"
+            u"\u3042[<92><117><51>]xx"
         )
 
         self.assertEqual(

Lib/test/test_codecs.py

@@ -4,6 +4,11 @@ import codecs
 import locale
 import sys, StringIO, _testcapi
 
+def coding_checker(self, coder):
+    def check(input, expect):
+        self.assertEqual(coder(input), (expect, len(input)))
+    return check
+
 class Queue(object):
     """
     queue: write bytes at one end, read bytes from the other end
@@ -1786,6 +1791,84 @@ class WithStmtTest(unittest.TestCase):
             self.assertEqual(srw.read(), u"\xfc")
 
 
+class UnicodeEscapeTest(unittest.TestCase):
+    def test_empty(self):
+        self.assertEqual(codecs.unicode_escape_encode(u""), ("", 0))
+        self.assertEqual(codecs.unicode_escape_decode(""), (u"", 0))
+
+    def test_raw_encode(self):
+        encode = codecs.unicode_escape_encode
+        for b in range(32, 127):
+            if b != ord('\\'):
+                self.assertEqual(encode(unichr(b)), (chr(b), 1))
+
+    def test_raw_decode(self):
+        decode = codecs.unicode_escape_decode
+        for b in range(256):
+            if b != ord('\\'):
+                self.assertEqual(decode(chr(b) + '0'), (unichr(b) + u'0', 2))
+
+    def test_escape_encode(self):
+        encode = codecs.unicode_escape_encode
+        check = coding_checker(self, encode)
+        check(u'\t', r'\t')
+        check(u'\n', r'\n')
+        check(u'\r', r'\r')
+        check(u'\\', r'\\')
+        for b in range(32):
+            if chr(b) not in '\t\n\r':
+                check(unichr(b), '\\x%02x' % b)
+        for b in range(127, 256):
+            check(unichr(b), '\\x%02x' % b)
+        check(u'\u20ac', r'\u20ac')
+        check(u'\U0001d120', r'\U0001d120')
+
+    def test_escape_decode(self):
+        decode = codecs.unicode_escape_decode
+        check = coding_checker(self, decode)
+        check("[\\\n]", u"[]")
+        check(r'[\"]', u'["]')
+        check(r"[\']", u"[']")
+        check(r"[\\]", ur"[\]")
+        check(r"[\a]", u"[\x07]")
+        check(r"[\b]", u"[\x08]")
+        check(r"[\t]", u"[\x09]")
+        check(r"[\n]", u"[\x0a]")
+        check(r"[\v]", u"[\x0b]")
+        check(r"[\f]", u"[\x0c]")
+        check(r"[\r]", u"[\x0d]")
+        check(r"[\7]", u"[\x07]")
+        check(r"[\8]", ur"[\8]")
+        check(r"[\78]", u"[\x078]")
+        check(r"[\41]", u"[!]")
+        check(r"[\418]", u"[!8]")
+        check(r"[\101]", u"[A]")
+        check(r"[\1010]", u"[A0]")
+        check(r"[\x41]", u"[A]")
+        check(r"[\x410]", u"[A0]")
+        check(r"\u20ac", u"\u20ac")
+        check(r"\U0001d120", u"\U0001d120")
+        for b in range(256):
+            if chr(b) not in '\n"\'\\abtnvfr01234567xuUN':
+                check('\\' + chr(b), u'\\' + unichr(b))
+
+    def test_decode_errors(self):
+        decode = codecs.unicode_escape_decode
+        for c, d in ('x', 2), ('u', 4), ('U', 4):
+            for i in range(d):
+                self.assertRaises(UnicodeDecodeError, decode,
+                                  "\\" + c + "0"*i)
+                self.assertRaises(UnicodeDecodeError, decode,
+                                  "[\\" + c + "0"*i + "]")
+                data = "[\\" + c + "0"*i + "]\\" + c + "0"*i
+                self.assertEqual(decode(data, "ignore"), (u"[]", len(data)))
+                self.assertEqual(decode(data, "replace"),
+                                 (u"[\ufffd]\ufffd", len(data)))
+        self.assertRaises(UnicodeDecodeError, decode, r"\U00110000")
+        self.assertEqual(decode(r"\U00110000", "ignore"), (u"", 10))
+        self.assertEqual(decode(r"\U00110000", "replace"), (u"\ufffd", 10))
+
+
 class BomTest(unittest.TestCase):
     def test_seek0(self):
         data = u"1234567890"
@@ -1871,6 +1954,7 @@ def test_main():
         BasicStrTest,
         CharmapTest,
         WithStmtTest,
+        UnicodeEscapeTest,
         BomTest,
     )
 

Misc/NEWS

@@ -200,6 +200,8 @@ Core and Builtins
 Library
 -------
 
+- Issue #16979: Fix error handling bugs in the unicode-escape-decode decoder.
+
 - Issue #17051: Fix a memory leak in os.path.isdir() on Windows. Patch by
   Robert Xiao.
 

Objects/unicodeobject.c

@@ -2738,7 +2738,6 @@ PyObject *PyUnicode_DecodeUnicodeEscape(const char *s,
     Py_ssize_t startinpos;
     Py_ssize_t endinpos;
     Py_ssize_t outpos;
-    int i;
     PyUnicodeObject *v;
     Py_UNICODE *p;
     const char *end;
@@ -2824,29 +2823,19 @@ PyObject *PyUnicode_DecodeUnicodeEscape(const char *s,
             message = "truncated \\UXXXXXXXX escape";
         hexescape:
             chr = 0;
-            outpos = p-PyUnicode_AS_UNICODE(v);
-            if (s+digits>end) {
-                endinpos = size;
-                if (unicode_decode_call_errorhandler(
-                        errors, &errorHandler,
-                        "unicodeescape", "end of string in escape sequence",
-                        starts, size, &startinpos, &endinpos, &exc, &s,
-                        &v, &outpos, &p))
-                    goto onError;
-                goto nextByte;
-            }
-            for (i = 0; i < digits; ++i) {
-                c = (unsigned char) s[i];
-                if (!isxdigit(c)) {
-                    endinpos = (s+i+1)-starts;
-                    if (unicode_decode_call_errorhandler(
-                            errors, &errorHandler,
-                            "unicodeescape", message,
-                            starts, size, &startinpos, &endinpos, &exc, &s,
-                            &v, &outpos, &p))
-                        goto onError;
-                    goto nextByte;
+            if (end - s < digits) {
+                /* count only hex digits */
+                for (; s < end; ++s) {
+                    c = (unsigned char)*s;
+                    if (!Py_ISXDIGIT(c))
+                        goto error;
                 }
+                goto error;
+            }
+            for (; digits--; ++s) {
+                c = (unsigned char)*s;
+                if (!Py_ISXDIGIT(c))
+                    goto error;
                 chr = (chr<<4) & ~0xF;
                 if (c >= '0' && c <= '9')
                     chr += c - '0';
@@ -2855,7 +2844,6 @@ PyObject *PyUnicode_DecodeUnicodeEscape(const char *s,
                 else
                     chr += 10 + c - 'A';
             }
-            s += i;
             if (chr == 0xffffffff && PyErr_Occurred())
                 /* _decoding_error will have already written into the
                    target buffer. */
@@ -2876,14 +2864,8 @@ PyObject *PyUnicode_DecodeUnicodeEscape(const char *s,
                 *p++ = 0xDC00 + (Py_UNICODE) (chr & 0x03FF);
 #endif
             } else {
-                endinpos = s-starts;
-                outpos = p-PyUnicode_AS_UNICODE(v);
-                if (unicode_decode_call_errorhandler(
-                        errors, &errorHandler,
-                        "unicodeescape", "illegal Unicode character",
-                        starts, size, &startinpos, &endinpos, &exc, &s,
-                        &v, &outpos, &p))
-                    goto onError;
+                message = "illegal Unicode character";
+                goto error;
             }
             break;
 
@@ -2910,28 +2892,13 @@ PyObject *PyUnicode_DecodeUnicodeEscape(const char *s,
                         goto store;
                 }
             }
-            endinpos = s-starts;
-            outpos = p-PyUnicode_AS_UNICODE(v);
-            if (unicode_decode_call_errorhandler(
-                    errors, &errorHandler,
-                    "unicodeescape", message,
-                    starts, size, &startinpos, &endinpos, &exc, &s,
-                    &v, &outpos, &p))
-                goto onError;
-            break;
+            goto error;
 
         default:
             if (s > end) {
                 message = "\\ at end of string";
                 s--;
-                endinpos = s-starts;
-                outpos = p-PyUnicode_AS_UNICODE(v);
-                if (unicode_decode_call_errorhandler(
-                        errors, &errorHandler,
-                        "unicodeescape", message,
-                        starts, size, &startinpos, &endinpos, &exc, &s,
-                        &v, &outpos, &p))
-                    goto onError;
+                goto error;
             }
             else {
                 *p++ = '\\';
@@ -2939,8 +2906,18 @@ PyObject *PyUnicode_DecodeUnicodeEscape(const char *s,
             }
             break;
         }
-      nextByte:
-        ;
+        continue;
+
+      error:
+        endinpos = s-starts;
+        outpos = p-PyUnicode_AS_UNICODE(v);
+        if (unicode_decode_call_errorhandler(
+                errors, &errorHandler,
+                "unicodeescape", message,
+                starts, size, &startinpos, &endinpos, &exc, &s,
+                &v, &outpos, &p))
+            goto onError;
+        continue;
     }
     if (_PyUnicode_Resize(&v, p - PyUnicode_AS_UNICODE(v)) < 0)
         goto onError;