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

4 files changed, 117 insertions, 53 deletions

diff --git a/Lib/test/test_codeccallbacks.py b/Lib/test/test_codeccallbacks.py
index e656d2f..3b78b4a 100644
--- a/Lib/test/test_codeccallbacks.py
+++ b/Lib/test/test_codeccallbacks.py
@@ -262,12 +262,12 @@ class CodecCallbackTest(unittest.TestCase):
 
         self.assertEqual(
             b"\\u3042\u3xxx".decode("unicode-escape", "test.handler1"),
-            "\u3042[<92><117><51><120>]xx"
+            "\u3042[<92><117><51>]xxx"
         )
 
         self.assertEqual(
             b"\\u3042\u3xx".decode("unicode-escape", "test.handler1"),
-            "\u3042[<92><117><51><120><120>]"
+            "\u3042[<92><117><51>]xx"
         )
 
         self.assertEqual(
diff --git a/Lib/test/test_codecs.py b/Lib/test/test_codecs.py
index 2d1b249..e74038b 100644
--- a/Lib/test/test_codecs.py
+++ b/Lib/test/test_codecs.py
@@ -4,6 +4,11 @@ import codecs
 import locale
 import sys, _testcapi, io
 
+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
@@ -1846,6 +1851,85 @@ class TypesTest(unittest.TestCase):
         self.assertEqual(codecs.raw_unicode_escape_decode(r"\u1234"), ("\u1234", 6))
         self.assertEqual(codecs.raw_unicode_escape_decode(br"\u1234"), ("\u1234", 6))
 
+
+class UnicodeEscapeTest(unittest.TestCase):
+    def test_empty(self):
+        self.assertEqual(codecs.unicode_escape_encode(""), (b"", 0))
+        self.assertEqual(codecs.unicode_escape_decode(b""), ("", 0))
+
+    def test_raw_encode(self):
+        encode = codecs.unicode_escape_encode
+        for b in range(32, 127):
+            if b != b'\\'[0]:
+                self.assertEqual(encode(chr(b)), (bytes([b]), 1))
+
+    def test_raw_decode(self):
+        decode = codecs.unicode_escape_decode
+        for b in range(256):
+            if b != b'\\'[0]:
+                self.assertEqual(decode(bytes([b]) + b'0'), (chr(b) + '0', 2))
+
+    def test_escape_encode(self):
+        encode = codecs.unicode_escape_encode
+        check = coding_checker(self, encode)
+        check('\t', br'\t')
+        check('\n', br'\n')
+        check('\r', br'\r')
+        check('\\', br'\\')
+        for b in range(32):
+            if chr(b) not in '\t\n\r':
+                check(chr(b), ('\\x%02x' % b).encode())
+        for b in range(127, 256):
+            check(chr(b), ('\\x%02x' % b).encode())
+        check('\u20ac', br'\u20ac')
+        check('\U0001d120', br'\U0001d120')
+
+    def test_escape_decode(self):
+        decode = codecs.unicode_escape_decode
+        check = coding_checker(self, decode)
+        check(b"[\\\n]", "[]")
+        check(br'[\"]', '["]')
+        check(br"[\']", "[']")
+        check(br"[\\]", r"[\]")
+        check(br"[\a]", "[\x07]")
+        check(br"[\b]", "[\x08]")
+        check(br"[\t]", "[\x09]")
+        check(br"[\n]", "[\x0a]")
+        check(br"[\v]", "[\x0b]")
+        check(br"[\f]", "[\x0c]")
+        check(br"[\r]", "[\x0d]")
+        check(br"[\7]", "[\x07]")
+        check(br"[\8]", r"[\8]")
+        check(br"[\78]", "[\x078]")
+        check(br"[\41]", "[!]")
+        check(br"[\418]", "[!8]")
+        check(br"[\101]", "[A]")
+        check(br"[\1010]", "[A0]")
+        check(br"[\x41]", "[A]")
+        check(br"[\x410]", "[A0]")
+        check(br"\u20ac", "\u20ac")
+        check(br"\U0001d120", "\U0001d120")
+        for b in range(256):
+            if b not in b'\n"\'\\abtnvfr01234567xuUN':
+                check(b'\\' + bytes([b]), '\\' + chr(b))
+
+    def test_decode_errors(self):
+        decode = codecs.unicode_escape_decode
+        for c, d in (b'x', 2), (b'u', 4), (b'U', 4):
+            for i in range(d):
+                self.assertRaises(UnicodeDecodeError, decode,
+                                  b"\\" + c + b"0"*i)
+                self.assertRaises(UnicodeDecodeError, decode,
+                                  b"[\\" + c + b"0"*i + b"]")
+                data = b"[\\" + c + b"0"*i + b"]\\" + c + b"0"*i
+                self.assertEqual(decode(data, "ignore"), ("[]", len(data)))
+                self.assertEqual(decode(data, "replace"),
+                                 ("[\ufffd]\ufffd", len(data)))
+        self.assertRaises(UnicodeDecodeError, decode, br"\U00110000")
+        self.assertEqual(decode(br"\U00110000", "ignore"), ("", 10))
+        self.assertEqual(decode(br"\U00110000", "replace"), ("\ufffd", 10))
+
+
 class SurrogateEscapeTest(unittest.TestCase):
 
     def test_utf8(self):
@@ -2011,6 +2095,7 @@ def test_main():
         CharmapTest,
         WithStmtTest,
         TypesTest,
+        UnicodeEscapeTest,
         SurrogateEscapeTest,
         BomTest,
         TransformCodecTest,
diff --git a/Misc/NEWS b/Misc/NEWS
index daca4d3..36edcba 100644
--- a/Misc/NEWS
+++ b/Misc/NEWS
@@ -214,6 +214,8 @@ Core and Builtins
 Library
 -------
 
+- Issue #16979: Fix error handling bugs in the unicode-escape-decode decoder.
+
 - Issue #9290: In IDLE the sys.std* streams now implement io.TextIOBase
   interface and support all mandatory methods and properties.
 
diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c
index 92d1777..3a288d8 100644
--- a/Objects/unicodeobject.c
+++ b/Objects/unicodeobject.c
@@ -3760,7 +3760,6 @@ 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;
@@ -3846,29 +3845,19 @@ 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, &end, &startinpos, &endinpos, &exc, &s,
-                        &v, &outpos, &p))
-                    goto onError;
-                goto nextByte;
-            }
-            for (i = 0; i < digits; ++i) {
-                c = (unsigned char) s[i];
-                if (!Py_ISXDIGIT(c)) {
-                    endinpos = (s+i+1)-starts;
-                    if (unicode_decode_call_errorhandler(
-                            errors, &errorHandler,
-                            "unicodeescape", message,
-                            &starts, &end, &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';
@@ -3877,7 +3866,6 @@ 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. */
@@ -3898,14 +3886,8 @@ 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, &end, &startinpos, &endinpos, &exc, &s,
-                        &v, &outpos, &p))
-                    goto onError;
+                message = "illegal Unicode character";
+                goto error;
             }
             break;
 
@@ -3932,28 +3914,13 @@ 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, &end, &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, &end, &startinpos, &endinpos, &exc, &s,
-                        &v, &outpos, &p))
-                    goto onError;
+                goto error;
             }
             else {
                 *p++ = '\\';
@@ -3961,8 +3928,18 @@ 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, &end, &startinpos, &endinpos, &exc, &s,
+                &v, &outpos, &p))
+            goto onError;
+        continue;
     }
     if (_PyUnicode_Resize(&v, p - PyUnicode_AS_UNICODE(v)) < 0)
         goto onError;