From b1556c537d7c49978fa40594a9c9f40c6f88cdde Mon Sep 17 00:00:00 2001
From: Victor Stinner <victor.stinner@haypocalc.com>
Date: Thu, 20 May 2010 11:29:45 +0000
Subject: [PATCH] libpython.py: fix support of non-BMP unicode characters

Forward port some code from Python3:

 * join surrogate pairs if sizeof(Py_UNICODE)==2
 * Enable non-BMP test on narrow builds using u"\U0001D121" instead of
   unichr(0x1D121)
---
 Lib/test/test_gdb.py   | 10 ++--------
 Tools/gdb/libpython.py | 28 ++++++++++++++++++++++++++++
 2 files changed, 30 insertions(+), 8 deletions(-)

diff --git a/Lib/test/test_gdb.py b/Lib/test/test_gdb.py
index 388f24059e..2a7057c8e5 100644
--- a/Lib/test/test_gdb.py
+++ b/Lib/test/test_gdb.py
@@ -243,14 +243,8 @@ class PrettyPrintTests(DebuggerTests):
         # This is:
         # UTF-8: 0xF0 0x9D 0x84 0xA1
         # UTF-16: 0xD834 0xDD21
-        try:
-            # This will only work on wide-unicode builds:
-            self.assertGdbRepr(unichr(0x1D121))
-        except ValueError, e:
-            # We're probably on a narrow-unicode build; if we're seeing a
-            # different problem, then re-raise it:
-            if e.args != ('unichr() arg not in range(0x10000) (narrow Python build)',):
-                raise e
+        # This will only work on wide-unicode builds:
+        self.assertGdbRepr(u"\U0001D121")
 
     def test_sets(self):
         'Verify the pretty-printing of sets'
diff --git a/Tools/gdb/libpython.py b/Tools/gdb/libpython.py
index f62735fb2a..3481f71c15 100644
--- a/Tools/gdb/libpython.py
+++ b/Tools/gdb/libpython.py
@@ -1013,6 +1013,10 @@ class PyTypeObjectPtr(PyObjectPtr):
 class PyUnicodeObjectPtr(PyObjectPtr):
     _typename = 'PyUnicodeObject'
 
+    def char_width(self):
+        _type_Py_UNICODE = gdb.lookup_type('Py_UNICODE')
+        return _type_Py_UNICODE.sizeof
+
     def proxyval(self, visited):
         # From unicodeobject.h:
         #     Py_ssize_t length;  /* Length of raw Unicode data in buffer */
@@ -1029,6 +1033,30 @@ class PyUnicodeObjectPtr(PyObjectPtr):
         result = u''.join([unichr(ucs) for ucs in Py_UNICODEs])
         return result
 
+    def write_repr(self, out, visited):
+        proxy = self.proxyval(visited)
+        if self.char_width() == 2:
+            # sizeof(Py_UNICODE)==2: join surrogates
+            proxy2 = []
+            i = 0
+            while i < len(proxy):
+                ch = proxy[i]
+                i += 1
+                if (i < len(proxy)
+                and 0xD800 <= ord(ch) < 0xDC00 \
+                and 0xDC00 <= ord(proxy[i]) <= 0xDFFF):
+                    # Get code point from surrogate pair
+                    ch2 = proxy[i]
+                    code = (ord(ch) & 0x03FF) << 10
+                    code |= ord(ch2) & 0x03FF
+                    code += 0x00010000
+                    i += 1
+                    proxy2.append(unichr(code))
+                else:
+                    proxy2.append(ch)
+            proxy = u''.join(proxy2)
+        out.write(repr(proxy))
+
 
 def int_from_int(gdbval):
     return int(str(gdbval))
-- 
2.50.1