Updating from skeleton

[gonvert] / src / util / io.py

diff --git a/src/util/io.py b/src/util/io.py
index aac896d..4198f4b 100644 (file)
--- a/src/util/io.py
+++ b/src/util/io.py
@@ -8,6 +8,7 @@ import pickle
 import contextlib
 import itertools
 import codecs
+from xml.sax import saxutils
 import csv
 try:
        import cStringIO as StringIO
@@ -207,3 +208,24 @@ def unicode_csv_reader(unicode_csv_data, dialect=csv.excel, **kwargs):
 def utf_8_encoder(unicode_csv_data):
        for line in unicode_csv_data:
                yield line.encode('utf-8')
+
+
+_UNESCAPE_ENTITIES = {
+ """: '"',
+ " ": " ",
+ "'": "'",
+}
+
+
+_ESCAPE_ENTITIES = dict((v, k) for (v, k) in zip(_UNESCAPE_ENTITIES.itervalues(), _UNESCAPE_ENTITIES.iterkeys()))
+del _ESCAPE_ENTITIES[" "]
+
+
+def unescape(text):
+       plain = saxutils.unescape(text, _UNESCAPE_ENTITIES)
+       return plain
+
+
+def escape(text):
+       fancy = saxutils.escape(text, _ESCAPE_ENTITIES)
+       return fancy