[Assorted-commits] SF.net SVN: assorted: [804] python-commons/trunk/src/commons/strs.py

[<yan...@us...>]

Revision: 804
          http://assorted.svn.sourceforge.net/assorted/?rev=804&view=rev
Author:   yangzhang
Date:     2008-05-13 15:42:19 -0700 (Tue, 13 May 2008)

Log Message:
-----------
added cp1252_to_unicode()

Modified Paths:
--------------
    python-commons/trunk/src/commons/strs.py

Modified: python-commons/trunk/src/commons/strs.py
===================================================================
--- python-commons/trunk/src/commons/strs.py	2008-05-13 07:44:44 UTC (rev 803)
+++ python-commons/trunk/src/commons/strs.py	2008-05-13 22:42:19 UTC (rev 804)
@@ -18,3 +18,41 @@
         return str( s )
     else:
         return s
+
+cp1252_to_unicode_translations = [ (u'\x80',u'\u20AC'),
+                                   (u'\x82',u'\u201A'),
+                                   (u'\x83',u'\u0192'),
+                                   (u'\x84',u'\u201E'),
+                                   (u'\x85',u'\u2026'),
+                                   (u'\x86',u'\u2020'),
+                                   (u'\x87',u'\u2021'),
+                                   (u'\x88',u'\u02C6'),
+                                   (u'\x89',u'\u2030'),
+                                   (u'\x8A',u'\u0160'),
+                                   (u'\x8B',u'\u2039'),
+                                   (u'\x8C',u'\u0152'),
+                                   (u'\x8E',u'\u017D'),
+                                   (u'\x91',u'\u2018'),
+                                   (u'\x92',u'\u2019'),
+                                   (u'\x93',u'\u201C'),
+                                   (u'\x94',u'\u201D'),
+                                   (u'\x95',u'\u2022'),
+                                   (u'\x96',u'\u2013'),
+                                   (u'\x97',u'\u2014'),
+                                   (u'\x98',u'\u02DC'),
+                                   (u'\x99',u'\u2122'),
+                                   (u'\x9A',u'\u0161'),
+                                   (u'\x9B',u'\u203A'),
+                                   (u'\x9C',u'\u0153'),
+                                   (u'\x9E',u'\u017E'),
+                                   (u'\x9F',u'\u0178') ]
+
+def cp1252_to_unicode(x):
+    """Converts characters 0x80 through 0x9f to their proper Unicode
+    equivalents.  See
+    U{http://www.intertwingly.net/stories/2004/04/14/i18n.html} for the nice
+    translation table on which this is based."""
+    for a,b in cp1252_to_unicode_translations:
+        x = x.replace(a,b)
+    return x
+


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.