s4/scripting/bin: open unicode files with utf8 encoding and write unicode string

In files like `libcli/util/werror_err_table.txt` and `libcli/util/ntstatus_err_table.txt`,
there were unicode quote symbols at line 6:

    ...(“this documentation”)...

In `libcli/util/wscript_build`, it will run `gen_werror.py` and `gen_ntstatus.py`
to `open` above files, read content from them and write to other files.

When encoding not specified, `open` in both python 2/3 will guess encoding from locale.

When locale is not set, it defaults to POSIX or C, and then python will use
encoding `ANSI_X3.4-1968`.

So, on a system locale is not set, `make` will fail with encoding error
for both python 2 and 3:

    File "/home/ubuntu/samba/source4/scripting/bin/gen_werror.py", line 139, in main
        errors = parseErrorDescriptions(input_file, True, transformErrorName)
      File "/home/ubuntu/samba/source4/scripting/bin/gen_error_common.py", line 52, in parseErrorDescriptions
        for line in file_contents:
      File "/usr/lib/python3.5/encodings/ascii.py", line 26, in decode
        return codecs.ascii_decode(input, self.errors)[0]
    UnicodeDecodeError: 'ascii' codec can't decode byte 0xe2 in position 318: ordinal not in range(128)

In this case, we have to use `io.open` with `encoding='utf8'`.
However, then we got unicode strs and try to write them with other strs
into new file, which means the new file must also open with utf-8 and
all other strs have to be unicode, too.

Instead of prefix `u` to all strs, a more easier/elegant way is to enable
unicode literals for the python scripts, which we normally didn't do in samba.

Since both `gen_werror.py` and `gen_ntstatus.py` are bin scripts and no
other modules import them, it should be ok for this case.

Signed-off-by: Joe Guo <joeg@catalyst.net.nz>
Autobuild-User(master): Douglas Bagnall <dbagnall@samba.org>
Autobuild-Date(master): Fri Feb  8 06:34:47 CET 2019 on sn-devel-144

diff --git a/source4/scripting/bin/gen_ntstatus.py b/source4/scripting/bin/gen_ntstatus.py
index e0d4fe3b7a16a96d66bb8e6dfdecdf4a8e10596e..7691e7bc5d1daf238919a07ae1bc555cebe25d21 100755 (executable)
--- a/source4/scripting/bin/gen_ntstatus.py
+++ b/source4/scripting/bin/gen_ntstatus.py
@@ -20,6 +20,9 @@
 # You should have received a copy of the GNU General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
+from __future__ import unicode_literals
+# this file is a bin script and was not imported by any other modules
+# so it should be fine to enable unicode string for python2
 
 import sys, os.path, io, string
 from gen_error_common import parseErrorDescriptions, ErrorDef
@@ -134,20 +137,20 @@ def main ():
         sys.exit()
 
     # read in the data
-    file_contents = open(input_file, "r")
+    file_contents = io.open(input_file, "rt", encoding='utf8')
 
     errors = parseErrorDescriptions(file_contents, False, transformErrorName)
 
     print("writing new header file: %s" % gen_headerfile_name)
-    out_file = open(gen_headerfile_name, "w")
+    out_file = io.open(gen_headerfile_name, "wt", encoding='utf8')
     generateHeaderFile(out_file, errors)
     out_file.close()
     print("writing new source file: %s" % gen_sourcefile_name)
-    out_file = open(gen_sourcefile_name, "w")
+    out_file = io.open(gen_sourcefile_name, "wt", encoding='utf8')
     generateSourceFile(out_file, errors)
     out_file.close()
     print("writing new python file: %s" % gen_pythonfile_name)
-    out_file = open(gen_pythonfile_name, "w")
+    out_file = io.open(gen_pythonfile_name, "wt", encoding='utf8')
     generatePythonFile(out_file, errors)
     out_file.close()
 

diff --git a/source4/scripting/bin/gen_werror.py b/source4/scripting/bin/gen_werror.py
index 700ccaf140073b232bbf4cd91adb2e7c5ccfbdc4..6280d2d388379538fde047ff2e6df07418c004a4 100755 (executable)
--- a/source4/scripting/bin/gen_werror.py
+++ b/source4/scripting/bin/gen_werror.py
@@ -20,6 +20,9 @@
 # You should have received a copy of the GNU General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
+from __future__ import unicode_literals
+# this file is a bin script and was not imported by any other modules
+# so it should be fine to enable unicode string for python2
 
 import sys, os.path, io, string
 from gen_error_common import parseErrorDescriptions, ErrorDef
@@ -135,20 +138,20 @@ def main():
         print("usage: %s winerrorfile headerfile sourcefile pythonfile" % sys.argv[0])
         sys.exit()
 
-    input_file = open(input_file_name, "r")
+    input_file = io.open(input_file_name, "rt", encoding='utf8')
     errors = parseErrorDescriptions(input_file, True, transformErrorName)
     input_file.close()
 
     print("writing new header file: %s" % gen_headerfile_name)
-    out_file = open(gen_headerfile_name, "w")
+    out_file = io.open(gen_headerfile_name, "wt", encoding='utf8')
     generateHeaderFile(out_file, errors)
     out_file.close()
     print("writing new source file: %s" % gen_sourcefile_name)
-    out_file = open(gen_sourcefile_name, "w")
+    out_file = io.open(gen_sourcefile_name, "wt", encoding='utf8')
     generateSourceFile(out_file, errors)
     out_file.close()
     print("writing new python file: %s" % gen_pythonfile_name)
-    out_file = open(gen_pythonfile_name, "w")
+    out_file = io.open(gen_pythonfile_name, "wt", encoding='utf8')
     generatePythonFile(out_file, errors)
     out_file.close()