Fix crash if invalid character is inserted into the text.

manuskript/models/abstractItem.py

@@ -9,6 +9,7 @@
 from PyQt5.QtGui import QIcon, QFont
 from PyQt5.QtWidgets import QTextEdit, qApp
 from lxml import etree as ET
+import re
 
 from manuskript import enums
 
@@ -21,6 +22,9 @@ class abstractItem():
     # Used for XML export
     name = "abstractItem"
 
+    # Regexp from https://stackoverflow.com/questions/8733233/filtering-out-certain-bytes-in-python
+    valid_xml_re = re.compile(u'[^\u0020-\uD7FF\u0009\u000A\u000D\uE000-\uFFFD\U00010000-\U0010FFFF]+')
+
     def __init__(self, model=None, title="", _type="abstract", xml=None, parent=None, ID=None):
 
         self._data = {}
@@ -255,6 +259,9 @@ def setData(self, column, data, role=Qt.DisplayRole):
     # We want to force some data even if they're empty
     XMLForce = []
 
+    def cleanTextForXML(self, text):
+        return self.valid_xml_re.sub('', text)
+
     def toXML(self):
         """
         Returns a string containing the item (and children) in XML.
@@ -269,7 +276,7 @@ def toXML(self):
                 continue
             val = self.data(attrib)
             if val or attrib in self.XMLForce:
-                item.set(attrib.name, str(val))
+                item.set(attrib.name, self.cleanTextForXML(str(val)))
 
         # Saving lastPath
         item.set("lastPath", self._lastPath)

manuskript/models/outlineItem.py

@@ -483,7 +483,7 @@ def toXMLProcessItem(self, item):
         for r in rev:
             revItem = ET.Element("revision")
             revItem.set("timestamp", str(r[0]))
-            revItem.set("text", r[1])
+            revItem.set("text", self.cleanTextForXML(r[1]))
             item.append(revItem)
 
         return item