gdb: fix representation of non-printable surrogate pairs, and workaround

a bug in ascii().

diff --git a/Lib/test/test_gdb.py b/Lib/test/test_gdb.py
index bac8d36c18ee6a89f1afcbab111681d00f3285d5..d80acfe1da47c19e5f2aa1434bb85a09cf0e9acb 100644 (file)
--- a/Lib/test/test_gdb.py
+++ b/Lib/test/test_gdb.py
@@ -234,7 +234,9 @@ class PrettyPrintTests(DebuggerTests):
                 text.encode(encoding)
                 printable = True
             except UnicodeEncodeError:
-                self.assertGdbRepr(text, ascii(text))
+                # Workaround ascii() bug on UCS-2 builds: issue #9804
+                asc = "'" + text.encode('unicode-escape').decode('ascii') + "'"
+                self.assertGdbRepr(text, asc)
             else:
                 self.assertGdbRepr(text)
 

diff --git a/Tools/gdb/libpython.py b/Tools/gdb/libpython.py
index 361fb033037fd50e63332d31ecd99035a53a07f4..8a6980388e098766ed869b9db8844e979659e2e0 100644 (file)
--- a/Tools/gdb/libpython.py
+++ b/Tools/gdb/libpython.py
@@ -1171,9 +1171,8 @@ class PyUnicodeObjectPtr(PyObjectPtr):
             # Non-ASCII characters
             else:
                 ucs = ch
-                orig_ucs = None
                 ch2 = None
-                if self.char_width() == 2:
+                if sys.maxunicode < 0x10000:
                     # If sizeof(Py_UNICODE) is 2 here (in gdb), join
                     # surrogate pairs before calling _unichr_is_printable.
                     if (i < len(proxy)
@@ -1183,22 +1182,26 @@ class PyUnicodeObjectPtr(PyObjectPtr):
                         ucs = ch + ch2
                         i += 1
 
+                # Unfortuately, Python 2's unicode type doesn't seem
+                # to expose the "isprintable" method
                 printable = _unichr_is_printable(ucs)
                 if printable:
                     try:
                         ucs.encode(ENCODING)
                     except UnicodeEncodeError:
                         printable = False
-                        if orig_ucs is not None:
-                            ucs = orig_ucs
-                            i -= 1
 
                 # Map Unicode whitespace and control characters
                 # (categories Z* and C* except ASCII space)
                 if not printable:
-                    # Unfortuately, Python 2's unicode type doesn't seem
-                    # to expose the "isprintable" method
-                    code = ord(ucs)
+                    if ch2 is not None:
+                        # Match Python 3's representation of non-printable
+                        # wide characters.
+                        code = (ord(ch) & 0x03FF) << 10
+                        code |= ord(ch2) & 0x03FF
+                        code += 0x00010000
+                    else:
+                        code = ord(ucs)
 
                     # Map 8-bit characters to '\\xhh'
                     if code <= 0xff: