bh: thuban/Thuban/UI about.py,1.21,1.22 __init__.py,1.4,1.5

[cvs@intevation.de]

Author: bh

Update of /thubanrepository/thuban/Thuban/UI
In directory doto:/tmp/cvs-serv11857/Thuban/UI

Modified Files:
	about.py __init__.py 
Log Message:
First step towards unicode.  With this roughly we're at step 1
string_representation.txt

* Doc/technotes/string_representation.txt: New.  Document how
strings are represented in Thuban and how to get to a Unicode
Thuban.

* Thuban/__init__.py (set_internal_encoding)
(unicode_from_internal, internal_from_unicode): New. The first few
functions for the internal string representation

* Thuban/UI/about.py (unicodeToLocale): Removed.  Use
internal_from_unicode instead.

* Thuban/UI/__init__.py (install_wx_translation): Determine the
encoding to use for the internal string representation.  Also,
change the translation function to return strings in internal
representation even on unicode builds of wxPython

* Thuban/Model/load.py (SessionLoader.check_attrs): Decode
filenames too.
(SessionLoader.start_clrange): Use check_attrs to decode and check
the attributes.

* Thuban/Model/xmlreader.py (XMLReader.encode): Use
internal_from_unicode to convert unicode strings.

* Thuban/Model/xmlwriter.py (XMLWriter.encode): Use
unicode_from_internal when applicable

* test/runtests.py (main): New command line option:
internal-encoding to specify the internal string encoding to use
in the tests.

* test/support.py (initthuban): Set the internal encoding to
latin-1

* test/test_load.py (TestSingleLayer.test, TestClassification.test)
(TestLabelLayer.test): Use the internal string representation when
dealing with non-ascii characters

* test/test_load_1_0.py (TestSingleLayer.test)
(TestClassification.test, TestLabelLayer.test): Use the internal
string representation when dealing with non-ascii characters

* test/test_load_0_9.py (TestSingleLayer.test)
(TestClassification.test): Use the internal string representation
when dealing with non-ascii characters

* test/test_load_0_8.py (TestUnicodeStrings.test): Use the
internal string representation when dealing with non-ascii
characters

* test/test_save.py (XMLWriterTest.testEncode)
(SaveSessionTest.testClassifiedLayer): Use the internal string
representation when dealing with non-ascii characters where
applicable


Index: about.py
===================================================================
RCS file: /thubanrepository/thuban/Thuban/UI/about.py,v
retrieving revision 1.21
retrieving revision 1.22
diff -u -d -r1.21 -r1.22
--- about.py	10 Mar 2005 22:47:14 -0000	1.21
+++ about.py	1 Jul 2005 20:49:04 -0000	1.22
@@ -17,7 +17,7 @@
 
 from locale import getdefaultlocale
 
-from Thuban import _
+from Thuban import _, internal_from_unicode
 from Thuban.version import versions
 from Thuban.Model.resource import gdal_support_status
 
@@ -39,11 +39,12 @@
         lead_developer = 'Bernhard Herzog'
         developers = [ 'Jonathan Coles',
                        'Frank Koormann',
-                       unicodeToLocale(u'Martin M\xfcller'),
+                       internal_from_unicode(u'Martin M\xfcller'),
                        'Bernhard Reiter',
                        'Jan-Oliver Wagner' ]
         translators = [ ( _('French'), 'Daniel Calvelo Aros' ),
-                        ( _('German'), unicodeToLocale(u'Bj\xf6rn Broscheit')),
+                        ( _('German'),
+                          internal_from_unicode(u'Bj\xf6rn Broscheit')),
                         ( _('Hungarian'), 'Norbert Solymosi'),
                         ( _('Italian'), 'Maurizio Napolitano'),
                         ( _('Portuguese (Brazilian)'), 'Eduardo Patto Kanegae'),
@@ -163,15 +164,3 @@
 
     def OnCancel(self, event):
         self.EndModal(wxID_CANCEL)
-
-
-def unicodeToLocale(unicodeStr):
-    "Function to convert unicode to the user's locale encoding"
-    # Under a german windows 2000 getlocale returns an encoding name
-    # that's not direcly usable (it's missing a "cp" at the beginning).
-    # getdefaultlocale does return a usable encoding name so we use that
-    # instead.
-    locale=getdefaultlocale()[1]
-    if locale is None:
-        locale = 'ascii'
-    return unicodeStr.encode(locale,'replace')

Index: __init__.py
===================================================================
RCS file: /thubanrepository/thuban/Thuban/UI/__init__.py,v
retrieving revision 1.4
retrieving revision 1.5
diff -u -d -r1.4 -r1.5
--- __init__.py	27 Oct 2003 17:11:06 -0000	1.4
+++ __init__.py	1 Jul 2005 20:49:04 -0000	1.5
@@ -1,4 +1,4 @@
-# Copyright (c) 2001, 2002, 2003 by Intevation GmbH
+# Copyright (c) 2001, 2002, 2003, 2005 by Intevation GmbH
 # Authors:
 # Bernhard Herzog <bh at intevation.de>
 #
@@ -22,7 +22,23 @@
         wx.wxLocale_AddCatalogLookupPathPrefix(Thuban._message_dir)
         _locale = wx.wxLocale()
         _locale.AddCatalog("thuban")
-        Thuban.install_translation_function(wx.wxGetTranslation)
+
+        # With a unicode build of wxPython, wxGetTranslation returns a
+        # unicode object, so we have a wrapper that handles the case of
+        # not using unicode as the internal string representation of
+        # Thuban: If wxGetTranslation returns unicode and the internal
+        # string representation of Thuban is not unicode, we convert the
+        # translated string to the internal representation.  wxPython
+        # will convert such strings back to unicode if necessary,
+        # provided the internal encoding used is the one expected by
+        # wxPython, which is taken care of below.
+        def thuban_wx_translation(s):
+            t = wx.wxGetTranslation(s)
+            if isinstance(t, unicode) and Thuban._internal_encoding!="unicode":
+                t = t.encode(Thuban._internal_encoding, "replace")
+            return t
+
+        Thuban.install_translation_function(thuban_wx_translation)
 
         # Reset the python locale. This makes sure that the LC_NUMERIC
         # seen by the python interpreter and by C modules is "C" instead
@@ -31,5 +47,31 @@
         # work for some reason
         import locale
         locale.setlocale(locale.LC_NUMERIC, "")
+
+        # determine the internal encoding to use.
+        # This is very tricky.  It's probably not optimal yet.
+        encoding = None
+
+        # If we have a wxPython >= 2.5.4.1, we use the
+        # GetDefaultPyEncoding function to use exactly what wxPython
+        # also uses when converting between unicode and byte-strings ...
+        if hasattr(wx, "wxGetDefaultPyEncoding"):
+            # AFAICT from the code this will always be a usable string,
+            # although it could be "ascii".
+            internal_encoding = wx.wxGetDefaultPyEncoding()
+
+        # ... otherwise we use Python's getdefaultlocale.  This is what
+        # GetDefaultPyEncoding also uses ...
+        if encoding is None:
+            encoding = locale.getdefaultlocale()[1]
+
+        # ... finally, if we haven't figured it out yet, use latin 1 for
+        # historical reasons.  Maybe ascii would be better.
+        if encoding is None:
+            encoding = "latin1"
+
+        # set the encoding
+        Thuban.set_internal_encoding(encoding)
+
 
 install_wx_translation()