changeset 26879:a24b98f4e03c

encoding: re-escape U+DCxx characters in toutf8b input (issue4927) This is the final missing piece in fully round-tripping random byte strings through UTF-8b. While this issue means that UTF-8 <-> UTF-8b isn't fully bijective, we don't expect to ever see U+DCxx codepoints in "real" UTF-8 data, so it should remain bijective in practice.
author Matt Mackall <mpm@selenic.com>
date Thu, 05 Nov 2015 17:30:10 -0600
parents d7e83f106459
children 8b2fbe3f59b1
files mercurial/encoding.py
diffstat 1 files changed, 14 insertions(+), 9 deletions(-) [+]
line wrap: on
line diff
--- a/mercurial/encoding.py	Thu Nov 05 17:21:43 2015 -0600
+++ b/mercurial/encoding.py	Thu Nov 05 17:30:10 2015 -0600
@@ -463,14 +463,14 @@
     internal surrogate encoding as a UTF-8 string.)
     '''
 
-    if isinstance(s, localstr):
-        return s._utf8
-
-    try:
-        s.decode('utf-8')
-        return s
-    except UnicodeDecodeError:
-        pass
+    if "\xed" not in s:
+        if isinstance(s, localstr):
+            return s._utf8
+        try:
+            s.decode('utf-8')
+            return s
+        except UnicodeDecodeError:
+            pass
 
     r = ""
     pos = 0
@@ -478,7 +478,12 @@
     while pos < l:
         try:
             c = getutf8char(s, pos)
-            pos += len(c)
+            if "\xed\xb0\x80" <= c <= "\xed\xb3\xbf":
+                # have to re-escape existing U+DCxx characters
+                c = unichr(0xdc00 + ord(s[pos])).encode('utf-8')
+                pos += 1
+            else:
+                pos += len(c)
         except UnicodeDecodeError:
             c = unichr(0xdc00 + ord(s[pos])).encode('utf-8')
             pos += 1