Stop forcing LC_CTYPE=ISO-8859-1

[bup.git] / lib / bup / compat.py

diff --git a/lib/bup/compat.py b/lib/bup/compat.py
index 692bd9c1d48518b66b7750f1955a0ea05b5dbe9a..1ed2d0f70c9fc18f7afc2011a2e55381aa047ca4 100644 (file)
--- a/lib/bup/compat.py
+++ b/lib/bup/compat.py
@@ -14,19 +14,9 @@ py3 = py_maj >= 3
 if py3:
 
     from os import environb as environ
-
-    lc_ctype = environ.get(b'LC_CTYPE')
-    if lc_ctype and lc_ctype.lower() != b'iso-8859-1':
-        # Because of argv, options.py, pwd, grp, and any number of other issues
-        print('error: bup currently only works with ISO-8859-1, not LC_CTYPE=%s'
-              % lc_ctype.decode(),
-              file=sys.stderr)
-        print('error: this should already have been arranged, so indicates a bug',
-              file=sys.stderr)
-        sys.exit(2)
-
-    from os import fsencode
+    from os import fsdecode, fsencode
     from shlex import quote
+    input = input
     range = range
     str_type = str
     int_types = (int,)
@@ -69,20 +59,23 @@ if py3:
             return memoryview(object)[offset:]
         return memoryview(object)
 
-    def join_bytes(*items):
-        """Return the concatenated bytes or memoryview arguments as bytes."""
-        return b''.join(items)
+    def getcwd():
+        return fsencode(os.getcwd())
 
 else:  # Python 2
 
+    def fsdecode(x):
+        return x
+
     def fsencode(x):
         return x
 
     from pipes import quote
-    from os import environ
+    from os import environ, getcwd
 
     from bup.py2raise import reraise
 
+    input = raw_input
     range = xrange
     str_type = basestring
     int_types = (int, long)
@@ -130,8 +123,7 @@ else:  # Python 2
         """Return the original bytes passed to main() for an argv argument."""
         return x
 
-    def bytes_from_uint(i):
-        return chr(i)
+    bytes_from_uint = chr
 
     def bytes_from_byte(b):
         return b
@@ -140,35 +132,41 @@ else:  # Python 2
 
     buffer = buffer
 
-    def join_bytes(x, y):
-        """Return the concatenated bytes or buffer arguments as bytes."""
-        if type(x) == buffer:
-            assert type(y) in (bytes, buffer)
-            return x + y
-        assert type(x) == bytes
-        if type(y) == bytes:
-            return b''.join((x, y))
-        assert type(y) in (bytes, buffer)
-        return buffer(x) + y
-
-
-def restore_lc_env():
-    # Once we're up and running with iso-8859-1, undo the bup-python
-    # LC_CTYPE hackery, so we don't affect unrelated subprocesses.
-    bup_lc_all = environ.get(b'BUP_LC_ALL')
-    if bup_lc_all:
-        del environ[b'LC_COLLATE']
-        del environ[b'LC_CTYPE']
-        del environ[b'LC_MONETARY']
-        del environ[b'LC_NUMERIC']
-        del environ[b'LC_TIME']
-        del environ[b'LC_MESSAGES']
-        del environ[b'LC_MESSAGES']
-        environ[b'LC_ALL'] = bup_lc_all
+
+argv = None
+argvb = None
+
+def _configure_argv():
+    global argv, argvb
+    assert not argv
+    assert not argvb
+    if len(sys.argv) > 1:
+        if environ.get(b'BUP_ARGV_0'):
+            print('error: BUP_ARGV* set and sys.argv not empty', file=sys.stderr)
+            sys.exit(2)
+        argv = sys.argv
+        argvb = [argv_bytes(x) for x in argv]
         return
-    bup_lc_ctype = environ.get(b'BUP_LC_CTYPE')
-    if bup_lc_ctype:
-        environ[b'LC_CTYPE'] = bup_lc_ctype
+    args = []
+    i = 0
+    arg = environ.get(b'BUP_ARGV_%d' % i)
+    while arg is not None:
+        args.append(arg)
+        i += 1
+        arg = environ.get(b'BUP_ARGV_%d' % i)
+    i -= 1
+    while i >= 0:
+        del environ[b'BUP_ARGV_%d' % i]
+        i -= 1
+    argvb = args
+    # System encoding?
+    if py3:
+        argv = [x.decode(errors='surrogateescape') for x in args]
+    else:
+        argv = argvb
+
+_configure_argv()
+
 
 def wrap_main(main):
     """Run main() and raise a SystemExit with the return value if it