code.vuplus.com Git - vuplus_dvbapp-plugin/blob - fritzcall/src/nrzuname.py

   1 #!/usr/bin/python
   2 # -*- coding: UTF-8 -*-
   3 # $Id$
   4 # $Author$
   5 # $Revision$
   6 # $Date$
   7
   8 import re, sys, os
   9 from xml.dom.minidom import parse
  10 from twisted.web.client import getPage #@UnresolvedImport
  11 from twisted.internet import reactor #@UnresolvedImport
  12
  13 debugSetting = True
  14 def setDebug(what):
  15         global debugSetting
  16         debugSetting = what
  17
  18 def debug(str):
  19         if debugSetting:
  20                 print str
  21
  22 import htmlentitydefs
  23 def html2unicode(in_html):
  24         # sanity checks
  25         try:
  26                 in_html = in_html.decode('iso-8859-1')
  27                 debug("[Callhtml2utf8] Converted from latin1")
  28         except:
  29                 debug("[Callhtml2utf8] lost in translation from latin1")
  30                 pass
  31         try:
  32                 in_html = in_html.decode('utf-8')
  33                 debug("[Callhtml2utf8] Converted from utf-8")
  34         except:
  35                 debug("[Callhtml2utf8] lost in translation from utf-8")
  36                 pass
  37
  38         # first convert some WML codes from hex: e.g. &#xE4 -> &#228
  39         htmlentityhexnumbermask = re.compile('(&#x(..);)')
  40         entities = htmlentityhexnumbermask.finditer(in_html)
  41         for x in entities:
  42                 in_html = in_html.replace(x.group(1), '&#' + str(int(x.group(2),16)) + ';')
  43
  44         entitydict = {}
  45         # catch &uuml; and colleagues here
  46         htmlentitynamemask = re.compile('(&(\D{1,5}?);)')
  47         entities = htmlentitynamemask.finditer(in_html)
  48         for x in entities:
  49                 # debug("[Callhtml2utf8] mask: found %s" %repr(x.group(2)))
  50                 entitydict[x.group(1)] = htmlentitydefs.name2codepoint[str(x.group(2))]
  51
  52         # this is for &#288 and other numbers
  53         htmlentitynumbermask = re.compile('(&#(\d{1,5}?);)')
  54         entities = htmlentitynumbermask.finditer(in_html)
  55         for x in entities:
  56                 # debug("[Callhtml2utf8] number: found %s" %x.group(1))
  57                 entitydict[x.group(1)] = x.group(2)
  58
  59         # no go and replace all occurrences
  60         for key, codepoint in entitydict.items():
  61                 try:
  62                         debug("[Callhtml2utf8] replace %s with %s" %(repr(key), unichr(int(codepoint))))
  63                         in_html = in_html.replace(unicode(key), (unichr(int(codepoint))))
  64                         # in_html = in_html.replace(unicode(key), (unichr(int(codepoint))).decode('cp1252').encode('utf-8'))
  65                 except ValueError:
  66                         debug("[Callhtml2utf8] ValueError " + key + "/" + str(codepoint))
  67                         pass
  68         return in_html
  69
  70 def out(number, caller):
  71         debug("[out] %s: %s" %(number, caller))
  72         if not caller:
  73                 return
  74         name = vorname = strasse = hnr = plz = ort = ""
  75         lines = caller.split(', ')
  76         found = re.match("(.+?)\s+(.+)", lines[0])
  77         if found:
  78                 name = found.group(1)
  79                 vorname = found.group(2)
  80         else:
  81                 name = lines[0]
  82         aktuell = 1
  83         found = re.match("^(.+) ([-\d]+)$", lines[1], re.S)
  84         if found:
  85                 strasse = found.group(1)
  86                 hnr = found.group(2)
  87                 aktuell = 2
  88         else:
  89                 found = re.match("^(\d+) (.+)$", lines[1], re.S)
  90                 if found:
  91                         strasse = found.group(2)
  92                         hnr = found.group(1)
  93                 else:
  94                         strasse = lines[1]
  95                 aktuell = 2
  96         for i in range(aktuell, len(lines)):
  97                 found = re.match("(\S+)\s+(.+)", lines[i], re.S)
  98                 if found:
  99                         plz = found.group(1)
 100                         ort = found.group(2)
 101                         break
 102         else:
 103                 ort = lines[aktuell].strip()
 104         print "NA: %s;VN: %s;STR: %s;HNR: %s;PLZ: %s;ORT: %s" %( name,vorname,strasse,hnr,plz,ort )
 105
 106 def simpleout(number, caller):
 107         print caller
 108
 109 try:
 110         from Tools.Directories import resolveFilename, SCOPE_PLUGINS
 111         reverseLookupFileName = resolveFilename(SCOPE_PLUGINS, "Extensions/FritzCall/reverselookup.xml")
 112 except ImportError:
 113         reverseLookupFileName = "reverselookup.xml"
 114
 115 countries = { }
 116 reverselookupMtime = 0
 117
 118 class ReverseLookupAndNotifier:
 119         def __init__(self, number, outputFunction=out, charset="cp1252", countrycode = "0049"):
 120                 debug("[ReverseLookupAndNotifier] reverse Lookup for %s!" %number)
 121                 self.number = number
 122                 self.outputFunction = outputFunction
 123                 self.caller = ""
 124                 self.currentWebsite = None
 125                 self.nextWebsiteNo = 0
 126 #===============================================================================
 127 # sorry does not work at all
 128 #               if not charset:
 129 #                       charset = sys.getdefaultencoding()
 130 #                       debug("[ReverseLookupAndNotifier] set charset from system: %s!" %charset)
 131 #===============================================================================
 132                 self.charset = charset
 133
 134                 global reverselookupMtime
 135                 reverselookupMtimeAct = os.stat(reverseLookupFileName)[8]
 136                 if not countries or reverselookupMtimeAct > reverselookupMtime:
 137                         debug("[ReverseLookupAndNotifier] (Re-)Reading %s\n" %reverseLookupFileName)
 138                         reverselookupMtime = reverselookupMtimeAct
 139                         dom = parse(reverseLookupFileName)
 140                         for top in dom.getElementsByTagName("reverselookup"):
 141                                 for country in top.getElementsByTagName("country"):
 142                                         code = country.getAttribute("code").replace("+","00")
 143                                         countries[code] = country.getElementsByTagName("website")
 144
 145                 self.countrycode = countrycode
 146
 147                 if number[0] != "0":
 148                         # self.caller = _("UNKNOWN")
 149                         self.notifyAndReset()
 150                         return
 151
 152                 if self.number[:2] == "00":
 153                         if countries.has_key(self.number[:3]):   #      e.g. USA
 154                                 self.countrycode = self.number[:3]
 155                         elif countries.has_key(self.number[:4]):
 156                                 self.countrycode = self.number[:4]
 157                         elif countries.has_key(self.number[:5]):
 158                                 self.countrycode = self.number[:5]
 159                         else:
 160                                 debug("[ReverseLookupAndNotifier] Country cannot be reverse handled")
 161                                 # self.caller = _("UNKNOWN")
 162                                 self.notifyAndReset()
 163                                 return
 164
 165                 if countries.has_key(self.countrycode):
 166                         debug("[ReverseLookupAndNotifier] Found website for reverse lookup")
 167                         self.websites = countries[self.countrycode]
 168                         self.nextWebsiteNo = 1
 169                         self.handleWebsite(self.websites[0])
 170                 else:
 171                         debug("[ReverseLookupAndNotifier] Country cannot be reverse handled")
 172                         # self.caller = _("UNKNOWN")
 173                         self.notifyAndReset()
 174                         return
 175
 176         def handleWebsite(self, website):
 177                 debug("[ReverseLookupAndNotifier] handleWebsite: " + website.getAttribute("name"))
 178                 if self.number[:2] == "00":
 179                         number = website.getAttribute("prefix") + self.number.replace(self.countrycode,"")
 180                 else:
 181                         number = self.number
 182
 183                 url = website.getAttribute("url")
 184                 if re.search('$AREACODE',url) or re.search('$PFXAREACODE',url):
 185                         debug("[ReverseLookupAndNotifier] handleWebsite: (PFX)ARECODE cannot be handled")
 186                         # self.caller = _("UNKNOWN")
 187                         self.notifyAndReset()
 188                         return
 189                 #
 190                 # Apparently, there is no attribute called (pfx)areacode anymore
 191                 # So, this below will not work.
 192                 #
 193                 if re.search('\\$AREACODE',url) and website.hasAttribute("areacode"):
 194                         areaCodeLen = int(website.getAttribute("areacode"))
 195                         url = url.replace("$AREACODE","%(areacode)s").replace("$NUMBER","%(number)s")
 196                         url = url %{ 'areacode':number[:areaCodeLen], 'number':number[areaCodeLen:] }
 197                 elif re.search('\\$PFXAREACODE',url) and website.hasAttribute("pfxareacode"):
 198                         areaCodeLen = int(website.getAttribute("pfxareacode"))
 199                         url = url.replace("$PFXAREACODE","%(pfxareacode)s").replace("$NUMBER","%(number)s")
 200                         url = url %{ 'pfxareacode':number[:areaCodeLen], 'number':number[areaCodeLen:] }
 201                 elif re.search('\\$NUMBER',url):
 202                         url = url.replace("$NUMBER","%s") %number
 203                 else:
 204                         debug("[ReverseLookupAndNotifier] handleWebsite: cannot handle websites with no $NUMBER in url")
 205                         # self.caller = _("UNKNOWN")
 206                         self.notifyAndReset()
 207                         return
 208                 debug("[ReverseLookupAndNotifier] Url to query: " + url)
 209                 url = url.encode("UTF-8", "replace")
 210                 self.currentWebsite = website
 211                 getPage(url,
 212                         agent="Mozilla/5.0 (Windows; U; Windows NT 6.0; de; rv:1.9.0.5) Gecko/2008120122 Firefox/3.0.5"
 213                         ).addCallback(self._gotPage).addErrback(self._gotError)
 214
 215
 216         def _gotPage(self, page):
 217                 debug("[ReverseLookupAndNotifier] _gotPage")
 218                 found = re.match('.*content=".*?charset=([^"]+)"',page,re.S)
 219                 if found:
 220                         debug("[ReverseLookupAndNotifier] Charset: " + found.group(1))
 221                         page = page.replace("\xa0"," ").decode(found.group(1), "replace")
 222                 else:
 223                         page = page.replace("\xa0"," ").decode("ISO-8859-1", "replace")
 224
 225                 for entry in self.currentWebsite.getElementsByTagName("entry"):
 226                         # debug("[ReverseLookupAndNotifier] _gotPage: try entry")
 227                         details = []
 228                         for what in ["name", "street", "city", "zipcode"]:
 229                                 pat = ".*?" + self.getPattern(entry, what)
 230                                 debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( what, pat ))
 231                                 found = re.match(pat, page, re.S|re.M)
 232                                 if found:
 233                                         debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( what, repr(found.group(1)) ))
 234                                         item = found.group(1).replace("&nbsp;"," ").replace("</b>","").replace(","," ")
 235                                         item = html2unicode(item)
 236                                         newitem = item.replace("  ", " ")
 237                                         while newitem != item:
 238                                                 item = newitem
 239                                                 newitem = item.replace("  ", " ")
 240                                         debug("[ReverseLookupAndNotifier] _gotPage: add to details: " + repr(item))
 241                                         details.append(item.strip())
 242                                 else:
 243                                         break
 244
 245                         if len(details) != 4:
 246                                 continue
 247                         else:
 248                                 name = details[0]
 249                                 address =  details[1] + ", " + details[3] + " " + details[2]
 250                                 debug("[ReverseLookupAndNotifier] _gotPage: Reverse lookup succeeded:\nName: %s\nAddress: %s" %(name, address))
 251                                 self.caller = "%s, %s" %(name, address)
 252                                 # if self.number != 0 and config.plugins.Call.addcallers.value and self.event == "RING":
 253                                         # phonebook.add(self.number, self.caller)
 254
 255                                 self.notifyAndReset()
 256                                 return True
 257                 else:
 258                         self._gotError("[ReverseLookupAndNotifier] _gotPage: Nothing found at %s" %self.currentWebsite.getAttribute("name"))
 259
 260         def _gotError(self, error = ""):
 261                 debug("[ReverseLookupAndNotifier] _gotError - Error: %s" %error)
 262                 if self.nextWebsiteNo >= len(self.websites):
 263                         debug("[ReverseLookupAndNotifier] _gotError: I give up")
 264                         # self.caller = _("UNKNOWN")
 265                         self.notifyAndReset()
 266                         return
 267                 else:
 268                         debug("[ReverseLookupAndNotifier] _gotError: try next website")
 269                         self.nextWebsiteNo = self.nextWebsiteNo+1
 270                         self.handleWebsite(self.websites[self.nextWebsiteNo-1])
 271
 272         def getPattern(self, website, which):
 273                 pat1 = website.getElementsByTagName(which)
 274                 if len(pat1) > 1:
 275                         debug("[ReverseLookupAndNotifier] getPattern: Something strange: more than one %s for website %s" %(which, website.getAttribute("name")))
 276                 return pat1[0].childNodes[0].data
 277
 278         def notifyAndReset(self):
 279                 debug("[ReverseLookupAndNotifier] notifyAndReset: Number: " + self.number + "; Caller: " + self.caller)
 280                 # debug("1: " + repr(self.caller))
 281                 if self.caller:
 282                         try:
 283                                 # debug("2: " + repr(self.caller))
 284                                 self.caller = self.caller.encode(self.charset)
 285                                 # debug("3: " + repr(self.caller))
 286                         except:
 287                                 debug("[ReverseLookupAndNotifier] cannot encode?!?!")
 288                                 pass
 289                         # self.caller = unicode(self.caller)
 290                         # debug("4: " + repr(self.caller))
 291                         self.outputFunction(self.number, self.caller)
 292                 else:
 293                         self.outputFunction(self.number, "")
 294                 if __name__ == '__main__':
 295                         reactor.stop() #@UndefinedVariable
 296
 297 if __name__ == '__main__':
 298         cwd = os.path.dirname(sys.argv[0])
 299         if (len(sys.argv) == 2):
 300                 # nrzuname.py Nummer
 301                 ReverseLookupAndNotifier(sys.argv[1])
 302                 reactor.run() #@UndefinedVariable
 303         elif (len(sys.argv) == 3):
 304                 # nrzuname.py Nummer Charset
 305                 setDebug(False)
 306                 ReverseLookupAndNotifier(sys.argv[1], simpleout, sys.argv[2])
 307                 reactor.run() #@UndefinedVariable