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 f713f52253e27e7f792e2ed4f15c0c21eae948c9..1ed2d0f70c9fc18f7afc2011a2e55381aa047ca4 100644 (file)
--- a/lib/bup/compat.py
+++ b/lib/bup/compat.py
@@ -1,8 +1,9 @@
 
 from __future__ import absolute_import, print_function
 from array import array
+from binascii import hexlify
 from traceback import print_exception
-import sys
+import os, sys
 
 # Please see CODINGSTYLE for important exception handling guidelines
 # and the rationale behind add_ex_tb(), add_ex_ctx(), etc.
@@ -12,9 +13,20 @@ py3 = py_maj >= 3
 
 if py3:
 
+    from os import environb as environ
+    from os import fsdecode, fsencode
     from shlex import quote
+    input = input
     range = range
     str_type = str
+    int_types = (int,)
+
+    def hexstr(b):
+        """Return hex string (not bytes as with hexlify) representation of b."""
+        return b.hex()
+
+    def reraise(ex):
+        raise ex.with_traceback(sys.exc_info()[2])
 
     def add_ex_tb(ex):
         """Do nothing (already handled by Python 3 infrastructure)."""
@@ -27,9 +39,16 @@ if py3:
     def items(x):
         return x.items()
 
+    def argv_bytes(x):
+        """Return the original bytes passed to main() for an argv argument."""
+        return fsencode(x)
+
     def bytes_from_uint(i):
         return bytes((i,))
 
+    def bytes_from_byte(b):  # python > 2: b[3] returns ord('x'), not b'x'
+        return bytes((b,))
+
     byte_int = lambda x: x
 
     def buffer(object, offset=None, size=None):
@@ -40,15 +59,28 @@ 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, getcwd
+
+    from bup.py2raise import reraise
+
+    input = raw_input
     range = xrange
     str_type = basestring
+    int_types = (int, long)
+
+    hexstr = hexlify
 
     def add_ex_tb(ex):
         """Add a traceback to ex if it doesn't already have one.  Return ex.
@@ -87,23 +119,54 @@ else:  # Python 2
     def items(x):
         return x.iteritems()
 
-    def bytes_from_uint(i):
-        return chr(i)
+    def argv_bytes(x):
+        """Return the original bytes passed to main() for an argv argument."""
+        return x
+
+    bytes_from_uint = chr
+
+    def bytes_from_byte(b):
+        return b
 
     byte_int = ord
 
     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
+
+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
+    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