Unescaping HTML escaping for names and message content

diff --git a/src/backends/gvoice.py b/src/backends/gvoice.py
index 19275f8..187c394 100755 (executable)
--- a/src/backends/gvoice.py
+++ b/src/backends/gvoice.py
@@ -37,6 +37,7 @@ import itertools
 import logging
 import inspect
 
 import logging
 import inspect
 

+from xml.sax import saxutils

 from xml.etree import ElementTree
 
 try:
 from xml.etree import ElementTree
 
 try:

@@ -479,6 +480,8 @@ class GVoiceBackend(object):
                for contactId, contactDetails in accountData["contacts"].iteritems():
                        # A zero contact id is the catch all for unknown contacts
                        if contactId != "0":
                for contactId, contactDetails in accountData["contacts"].iteritems():
                        # A zero contact id is the catch all for unknown contacts
                        if contactId != "0":

+                               if "name" in contactDetails:
+                                       contactDetails["name"] = unescape(contactDetails["name"])

                                yield contactId, contactDetails
 
        def get_voicemails(self):
                                yield contactId, contactDetails
 
        def get_voicemails(self):

@@ -575,12 +578,12 @@ class GVoiceBackend(object):
                        yield {
                                "id": messageId.strip(),
                                "contactId": contactId,
                        yield {
                                "id": messageId.strip(),
                                "contactId": contactId,

-                               "name": name,
+                               "name": unescape(name),

                                "time": exactTime,
                                "relTime": relativeTime,
                                "prettyNumber": prettyNumber,
                                "number": number,
                                "time": exactTime,
                                "relTime": relativeTime,
                                "prettyNumber": prettyNumber,
                                "number": number,

-                               "location": location,
+                               "location": unescape(location),

                        }
 
        @staticmethod
                        }
 
        @staticmethod

@@ -589,7 +592,7 @@ class GVoiceBackend(object):
                text = MessageText()
                if quality is not None and content is not None:
                        text.accuracy = quality
                text = MessageText()
                if quality is not None and content is not None:
                        text.accuracy = quality

-                       text.text = content
+                       text.text = unescape(content)

                        return text
                elif number is not None:
                        text.accuracy = MessageText.ACCURACY_HIGH
                        return text
                elif number is not None:
                        text.accuracy = MessageText.ACCURACY_HIGH

@@ -609,10 +612,10 @@ class GVoiceBackend(object):
                        relativeTimeGroup = self._relativeVoicemailTimeRegex.search(messageHtml)
                        conv.relTime = relativeTimeGroup.group(1).strip() if relativeTimeGroup else ""
                        locationGroup = self._voicemailLocationRegex.search(messageHtml)
                        relativeTimeGroup = self._relativeVoicemailTimeRegex.search(messageHtml)
                        conv.relTime = relativeTimeGroup.group(1).strip() if relativeTimeGroup else ""
                        locationGroup = self._voicemailLocationRegex.search(messageHtml)

-                       conv.location = locationGroup.group(1).strip() if locationGroup else ""
+                       conv.location = unescape(locationGroup.group(1).strip() if locationGroup else "")

 
                        nameGroup = self._voicemailNameRegex.search(messageHtml)
 
                        nameGroup = self._voicemailNameRegex.search(messageHtml)

-                       conv.name = nameGroup.group(1).strip() if nameGroup else ""
+                       conv.name = unescape(nameGroup.group(1).strip() if nameGroup else "")

                        numberGroup = self._voicemailNumberRegex.search(messageHtml)
                        conv.number = numberGroup.group(1).strip() if numberGroup else ""
                        prettyNumberGroup = self._prettyVoicemailNumberRegex.search(messageHtml)
                        numberGroup = self._voicemailNumberRegex.search(messageHtml)
                        conv.number = numberGroup.group(1).strip() if numberGroup else ""
                        prettyNumberGroup = self._prettyVoicemailNumberRegex.search(messageHtml)

@@ -637,7 +640,7 @@ class GVoiceBackend(object):
        def _interpret_sms_message_parts(fromPart, textPart, timePart):
                text = MessageText()
                text.accuracy = MessageText.ACCURACY_MEDIUM
        def _interpret_sms_message_parts(fromPart, textPart, timePart):
                text = MessageText()
                text.accuracy = MessageText.ACCURACY_MEDIUM

-               text.text = textPart
+               text.text = unescape(textPart)

 
                message = Message()
                message.body = (text, )
 
                message = Message()
                message.body = (text, )

@@ -661,7 +664,7 @@ class GVoiceBackend(object):
                        conv.location = ""
 
                        nameGroup = self._voicemailNameRegex.search(messageHtml)
                        conv.location = ""
 
                        nameGroup = self._voicemailNameRegex.search(messageHtml)

-                       conv.name = nameGroup.group(1).strip() if nameGroup else ""
+                       conv.name = unescape(nameGroup.group(1).strip() if nameGroup else "")

                        numberGroup = self._voicemailNumberRegex.search(messageHtml)
                        conv.number = numberGroup.group(1).strip() if numberGroup else ""
                        prettyNumberGroup = self._prettyVoicemailNumberRegex.search(messageHtml)
                        numberGroup = self._voicemailNumberRegex.search(messageHtml)
                        conv.number = numberGroup.group(1).strip() if numberGroup else ""
                        prettyNumberGroup = self._prettyVoicemailNumberRegex.search(messageHtml)

@@ -722,6 +725,18 @@ class GVoiceBackend(object):
                return json
 
 
                return json
 
 

+_UNESCAPE_ENTITIES = {
+ """: '"',
+ " ": " ",
+ "'": "'",
+}
+
+
+def unescape(text):
+       plain = saxutils.unescape(text, _UNESCAPE_ENTITIES)
+       return plain
+
+

 def google_strptime(time):
        """
        Hack: Google always returns the time in the same locale.  Sadly if the
 def google_strptime(time):
        """
        Hack: Google always returns the time in the same locale.  Sadly if the