After some testing, fixing the csv support
[theonering] / src / gvoice / backend.py
index 9996e74..69e46e4 100755 (executable)
@@ -37,6 +37,7 @@ import itertools
 import logging
 import inspect
 
+from xml.sax import saxutils
 from xml.etree import ElementTree
 
 try:
@@ -86,7 +87,7 @@ class Message(object):
                return "%s (%s): %s" % (
                        self.whoFrom,
                        self.when,
-                       "".join(str(part) for part in self.body)
+                       "".join(unicode(part) for part in self.body)
                )
 
        def to_dict(self):
@@ -187,6 +188,7 @@ class GVoiceBackend(object):
                self._XML_ACCOUNT_URL = SECURE_URL_BASE + "contacts/"
                # HACK really this redirects to the main pge and we are grabbing some javascript
                self._XML_CONTACTS_URL = "http://www.google.com/voice/inbox/search/contact"
+               self._CSV_CONTACTS_URL = "http://mail.google.com/mail/contacts/data/export"
                self._XML_RECENT_URL = SECURE_URL_BASE + "inbox/recent/"
 
                self.XML_FEEDS = (
@@ -304,6 +306,14 @@ class GVoiceBackend(object):
                self._lastAuthed = time.time()
                return True
 
+       def persist(self):
+               self._browser.save_cookies()
+
+       def shutdown(self):
+               self._browser.save_cookies()
+               self._token = None
+               self._lastAuthed = 0.0
+
        def logout(self):
                self._browser.clear_cookies()
                self._browser.save_cookies()
@@ -387,7 +397,7 @@ class GVoiceBackend(object):
                        self._sendSmsURL,
                        {
                                'phoneNumber': flattenedPhoneNumbers,
-                               'text': message
+                               'text': unicode(message).encode("utf-8"),
                        },
                )
                self._parse_with_validation(page)
@@ -473,18 +483,15 @@ class GVoiceBackend(object):
                @returns Iterable of (personsName, phoneNumber, exact date, relative date, action)
                @blocks
                """
-               for action, url in (
-                       ("Received", self._XML_RECEIVED_URL),
-                       ("Missed", self._XML_MISSED_URL),
-                       ("Placed", self._XML_PLACED_URL),
-               ):
-                       flatXml = self._get_page(url)
-
-                       allRecentHtml = self._grab_html(flatXml)
-                       allRecentData = self._parse_history(allRecentHtml)
-                       for recentCallData in allRecentData:
-                               recentCallData["action"] = action
-                               yield recentCallData
+               recentPages = [
+                       (action, self._get_page(url))
+                       for action, url in (
+                               ("Received", self._XML_RECEIVED_URL),
+                               ("Missed", self._XML_MISSED_URL),
+                               ("Placed", self._XML_PLACED_URL),
+                       )
+               ]
+               return self._parse_recent(recentPages)
 
        def get_contacts(self):
                """
@@ -492,14 +499,17 @@ class GVoiceBackend(object):
                @blocks
                """
                page = self._get_page(self._XML_CONTACTS_URL)
-               contactsBody = self._contactsBodyRe.search(page)
-               if contactsBody is None:
-                       raise RuntimeError("Could not extract contact information")
-               accountData = _fake_parse_json(contactsBody.group(1))
-               for contactId, contactDetails in accountData["contacts"].iteritems():
-                       # A zero contact id is the catch all for unknown contacts
-                       if contactId != "0":
-                               yield contactId, contactDetails
+               return self._process_contacts(page)
+
+       def get_csv_contacts(self):
+               data = {
+                       "groupToExport": "mine",
+                       "exportType": "ALL",
+                       "out": "OUTLOOK_CSV",
+               }
+               encodedData = urllib.urlencode(data)
+               contacts = self._get_page(self._CSV_CONTACTS_URL+"?"+encodedData)
+               return contacts
 
        def get_voicemails(self):
                """
@@ -508,6 +518,8 @@ class GVoiceBackend(object):
                voicemailPage = self._get_page(self._XML_VOICEMAIL_URL)
                voicemailHtml = self._grab_html(voicemailPage)
                voicemailJson = self._grab_json(voicemailPage)
+               if voicemailJson is None:
+                       return ()
                parsedVoicemail = self._parse_voicemail(voicemailHtml)
                voicemails = self._merge_conversation_sources(parsedVoicemail, voicemailJson)
                return voicemails
@@ -519,6 +531,8 @@ class GVoiceBackend(object):
                smsPage = self._get_page(self._XML_SMS_URL)
                smsHtml = self._grab_html(smsPage)
                smsJson = self._grab_json(smsPage)
+               if smsJson is None:
+                       return ()
                parsedSms = self._parse_sms(smsHtml)
                smss = self._merge_conversation_sources(parsedSms, smsJson)
                return smss
@@ -584,6 +598,24 @@ class GVoiceBackend(object):
                        raise RuntimeError("Not Authenticated")
                return number
 
+       def _parse_recent(self, recentPages):
+               for action, flatXml in recentPages:
+                       allRecentHtml = self._grab_html(flatXml)
+                       allRecentData = self._parse_history(allRecentHtml)
+                       for recentCallData in allRecentData:
+                               recentCallData["action"] = action
+                               yield recentCallData
+
+       def _process_contacts(self, page):
+               contactsBody = self._contactsBodyRe.search(page)
+               if contactsBody is None:
+                       raise RuntimeError("Could not extract contact information")
+               accountData = _fake_parse_json(contactsBody.group(1))
+               for contactId, contactDetails in accountData["contacts"].iteritems():
+                       # A zero contact id is the catch all for unknown contacts
+                       if contactId != "0":
+                               yield contactId, contactDetails
+
        def _parse_history(self, historyHtml):
                splitVoicemail = self._seperateVoicemailsRegex.split(historyHtml)
                for messageId, messageHtml in itergroup(splitVoicemail[1:], 2):
@@ -607,12 +639,12 @@ class GVoiceBackend(object):
                        yield {
                                "id": messageId.strip(),
                                "contactId": contactId,
-                               "name": name,
+                               "name": unescape(name),
                                "time": exactTime,
                                "relTime": relativeTime,
                                "prettyNumber": prettyNumber,
                                "number": number,
-                               "location": location,
+                               "location": unescape(location),
                        }
 
        @staticmethod
@@ -641,10 +673,10 @@ class GVoiceBackend(object):
                        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)
-                       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)
@@ -693,7 +725,7 @@ class GVoiceBackend(object):
                        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)
@@ -754,6 +786,18 @@ class GVoiceBackend(object):
                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
@@ -762,7 +806,7 @@ def google_strptime(time):
        """
        abbrevTime = time[:-3]
        parsedTime = datetime.datetime.strptime(abbrevTime, "%m/%d/%y %I:%M")
-       if time[-2] == "PN":
+       if time.endswith("PM"):
                parsedTime += datetime.timedelta(hours=12)
        return parsedTime
 
@@ -803,9 +847,16 @@ def itergroup(iterator, count, padValue = None):
 def safe_eval(s):
        _TRUE_REGEX = re.compile("true")
        _FALSE_REGEX = re.compile("false")
+       _COMMENT_REGEX = re.compile("^\s+//.*$", re.M)
        s = _TRUE_REGEX.sub("True", s)
        s = _FALSE_REGEX.sub("False", s)
-       return eval(s, {}, {})
+       s = _COMMENT_REGEX.sub("#", s)
+       try:
+               results = eval(s, {}, {})
+       except SyntaxError:
+               _moduleLogger.exception("Oops")
+               results = None
+       return results
 
 
 def _fake_parse_json(flattened):
@@ -840,7 +891,9 @@ def validate_response(response):
        Validates that the JSON response is A-OK
        """
        try:
-               assert 'ok' in response and response['ok']
+               assert response is not None
+               assert 'ok' in response
+               assert response['ok']
        except AssertionError:
                raise RuntimeError('There was a problem with GV: %s' % response)
 
@@ -926,6 +979,7 @@ def grab_debug_info(username, password):
                ("isdnd", backend._isDndURL),
                ("account", backend._XML_ACCOUNT_URL),
                ("contacts", backend._XML_CONTACTS_URL),
+               ("csv", backend._CSV_CONTACTS_URL),
 
                ("voicemail", backend._XML_VOICEMAIL_URL),
                ("sms", backend._XML_SMS_URL),