make transcoding more robust

default to ASCII
abort if unknown encoding
wrap abort strings with _
add test

mercurial/util.py

@@ -17,7 +17,8 b' from demandload import *'
 demandload(globals(), "cStringIO errno getpass popen2 re shutil sys tempfile")
 demandload(globals(), "os threading time calendar ConfigParser locale")
 
-_encoding = os.environ.get("HGENCODING") or locale.getpreferredencoding()
+_encoding = os.environ.get("HGENCODING") or locale.getpreferredencoding() \
+            or "ascii"
 _encodingmode = os.environ.get("HGENCODINGMODE", "strict")
 _fallbackencoding = 'ISO-8859-1'
 
@@ -35,6 +36,8 b' def tolocal(s):'
         try:
             u = s.decode(e) # attempt strict decoding
             return u.encode(_encoding, "replace")
+        except LookupError, k:
+            raise Abort(_("%s, please check your locale settings") % k)
         except UnicodeDecodeError:
             pass
     u = s.decode("utf-8", "replace") # last ditch
@@ -54,7 +57,9 b' def fromlocal(s):'
         return s.decode(_encoding, _encodingmode).encode("utf-8")
     except UnicodeDecodeError, inst:
         sub = s[max(0, inst.start-10):inst.start+10]
-        raise Abort("decoding near '%s': %s!\n" % (sub, inst))
+        raise Abort("decoding near '%s': %s!" % (sub, inst))
+    except LookupError, k:
+        raise Abort(_("%s, please check your locale settings") % k)
 
 def locallen(s):
     """Find the length in characters of a local string"""
@@ -70,7 +75,7 b' def localsub(s, a, b=None):'
         return u.encode(_encoding, _encodingmode)
     except UnicodeDecodeError, inst:
         sub = s[max(0, inst.start-10), inst.start+10]
-        raise Abort("decoding near '%s': %s!\n" % (sub, inst))
+        raise Abort(_("decoding near '%s': %s!\n") % (sub, inst))
 
 # used by parsedate
 defaultdateformats = (

tests/test-encoding

@@ -52,3 +52,5 b" echo '[ui]' >> .hg/hgrc"
 echo 'fallbackencoding = koi8-r' >> .hg/hgrc
 echo % utf-8
 HGENCODING=utf-8 hg log
+
+HGENCODING=dolphin hg log
  No newline at end of file

tests/test-encoding.out

@@ -10,7 +10,6 b' M a'
 ? latin-1-tag
 ? utf-8
 abort: decoding near ' encoded: �': 'ascii' codec can't decode byte 0xe9 in position 20: ordinal not in range(128)!
-
 transaction abort!
 rollback completed
 % these should work
@@ -165,3 +164,4 b' user: test'
 date:        Mon Jan 12 13:46:40 1970 +0000
 summary:     latin-1 e': И = u'\xe9'
 
+abort: unknown encoding: dolphin, please check your locale settings