# HG changeset patch
# User Paul Boddie <paul@boddie.org.uk>
# Date 1481587132 -3600
# Node ID 6d66508a5758afc010bc0963cc4338d842bca39d
# Parent  fd5cc95c936a833ebbe750318bd266582b909a66
Added encoding attributes and text encoding/decoding support to file streams.
Added a test of Unicode object creation and output.

diff -r fd5cc95c936a -r 6d66508a5758 lib/__builtins__/file.py
--- a/lib/__builtins__/file.py	Mon Dec 12 23:33:28 2016 +0100
+++ b/lib/__builtins__/file.py	Tue Dec 13 00:58:52 2016 +0100
@@ -20,16 +20,17 @@
 """
 
 from __builtins__.types import check_int, check_string
-from native import fclose, fopen, fread, fwrite
+from native import isinstance as _isinstance, fclose, fopen, fread, fwrite
 
 class filestream:
 
     "Generic file-oriented stream functionality."
 
-    def __init__(self, bufsize=1024):
+    def __init__(self, encoding=None, bufsize=1024):
 
-        "Initialise the stream with the given 'bufsize'."
+        "Initialise the stream with the given 'encoding' and 'bufsize'."
 
+        self.encoding = encoding
         self.bufsize = bufsize
         self.__data__ = None
 
@@ -42,7 +43,7 @@
         # Read any indicated number of bytes.
 
         if n > 0:
-            return fread(self.__data__, n)
+            s = fread(self.__data__, n)
 
         # Read all remaining bytes.
 
@@ -60,13 +61,26 @@
             except EOFError:
                 pass
 
-            return "".join(l)
+            s = "".join(l)
+
+        # Convert bytes to text if necessary.
+
+        if self.encoding:
+            return unicode(s, self.encoding)
+        else:
+            return s
 
     def write(self, s):
 
         "Write string 's' to the stream."
 
         check_string(s)
+
+        # Encode text as bytes if necessary.
+
+        if self.encoding and _isinstance(s, utf8string):
+            s = s.encode(self.encoding)
+
         fwrite(self.__data__, s)
 
     def close(self):
@@ -79,11 +93,14 @@
 
     "A file abstraction."
 
-    def __init__(self, filename, mode="r", bufsize=1024):
+    def __init__(self, filename, mode="r", encoding=None, bufsize=1024):
 
-        "Open the file with the given 'filename' using the given access 'mode'."
+        """
+        Open the file with the given 'filename' using the given access 'mode',
+        any specified 'encoding', and the given 'bufsize'.
+        """
 
-        get_using(filestream.__init__, self)(bufsize)
+        get_using(filestream.__init__, self)(encoding, bufsize)
         self.__data__ = fopen(filename, mode)
 
     def readline(self, size=None): pass
diff -r fd5cc95c936a -r 6d66508a5758 lib/__builtins__/unicode.py
--- a/lib/__builtins__/unicode.py	Mon Dec 12 23:33:28 2016 +0100
+++ b/lib/__builtins__/unicode.py	Tue Dec 13 00:58:52 2016 +0100
@@ -31,9 +31,11 @@
         "Encode the string to the given 'encoding'."
 
         from_utf8 = Converter("UTF-8", encoding)
+
         try:
             from_utf8.feed(self)
             return str(from_utf8)
+
         finally:
             from_utf8.close()
 
@@ -51,9 +53,11 @@
     # Convert the string to UTF-8.
 
     to_utf8 = Converter(encoding, "UTF-8")
+
     try:
         to_utf8.feed(s)
         return utf8string(str(to_utf8))
+
     finally:
         to_utf8.close()
 
diff -r fd5cc95c936a -r 6d66508a5758 lib/posix/io.py
--- a/lib/posix/io.py	Mon Dec 12 23:33:28 2016 +0100
+++ b/lib/posix/io.py	Tue Dec 13 00:58:52 2016 +0100
@@ -62,11 +62,14 @@
 
     "A system-level stream object."
 
-    def __init__(self, fd, mode="r", bufsize=1024):
+    def __init__(self, fd, mode="r", encoding=None, bufsize=1024):
 
-        "Initialise the stream with the given 'fd' and 'mode'."
+        """
+        Initialise the stream with the given 'fd', 'mode', 'encoding' and
+        'bufsize'.
+        """
 
-        get_using(filestream.__init__, self)(bufsize)
+        get_using(filestream.__init__, self)(encoding, bufsize)
         self.__data__ = fdopen(fd, mode)
 
 # Standard streams.
diff -r fd5cc95c936a -r 6d66508a5758 tests/unicode.py
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/tests/unicode.py	Tue Dec 13 00:58:52 2016 +0100
@@ -0,0 +1,20 @@
+# -*- coding: ISO-8859-1 -*-
+
+import sys
+
+# Print bytes.
+
+s = "æøå"
+print s                             # æøå
+
+# Obtain text and print it.
+
+u = unicode(s, "ISO-8859-1")
+print u                             # æøå
+print u.encode("ISO-8859-1")        # æøå
+
+# Inspect and update the encoding of stdout.
+
+print sys.stdout.encoding           # None
+sys.stdout.encoding = "ISO-8859-1"
+print u                             # æøå