UPD: nrzuname results made more generic
[vuplus_dvbapp-plugin] / fritzcall / src / nrzuname.py
index 98a6f28..add59b9 100644 (file)
@@ -5,19 +5,23 @@
 # $Revision$
 # $Date$
 
-import re, sys, os
+import re, sys, os, traceback
 from xml.dom.minidom import parse
 from twisted.web.client import getPage #@UnresolvedImport
 from twisted.internet import reactor #@UnresolvedImport
 
-debugVal = True
-def setDebug(what):
-       global debugVal
-       debugVal = what
-
-def debug(str):
-       if debugVal:
-               print str
+try:
+       from . import debug #@UnresolvedImport
+       def setDebug(what):
+               pass
+except ValueError:
+       debugVal = True
+       def setDebug(what):
+               global debugVal
+               debugVal = what
+       def debug(str):
+               if debugVal:
+                       print str
 
 import htmlentitydefs
 def html2unicode(in_html):
@@ -54,7 +58,6 @@ def html2unicode(in_html):
                        entitydict[key] = htmlentitydefs.name2codepoint[str(name)]
                except KeyError:
                        debug("[Callhtml2utf8] KeyError " + key + "/" + name)
-                       pass
 
        htmlentitynumbermask = re.compile('(&#(\d{1,5}?);)')
        entities = htmlentitynumbermask.finditer(in_html)
@@ -63,11 +66,10 @@ def html2unicode(in_html):
                entitydict[x.group(1)] = x.group(2)
        for key, codepoint in entitydict.items():
                try:
-                       debug("[Callhtml2utf8] replace %s with %s" %(repr(key), unichr(int(codepoint))))
+                       debug("[nrzuname] html2utf8: replace %s with %s" %(repr(key), str(codepoint)))
                        in_html = in_html.replace(unicode(key), (unichr(int(codepoint))))
                except ValueError:
-                       debug("[Callhtml2utf8] ValueError " + key + "/" + str(codepoint))
-                       pass
+                       debug("[nrzuname] html2utf8: ValueError " + key + "/" + str(codepoint))
        return in_html
 
 def normalizePhoneNumber(intNo):
@@ -82,42 +84,29 @@ def normalizePhoneNumber(intNo):
                return '0'
 
 def out(number, caller):
-       debug("[out] %s: %s" %(number, caller))
-       if not caller:
+       debug("[nrzuname] out: %s: %s" %(number, caller))
+       found = re.match("NA: ([^;]*);VN: ([^;]*);STR: ([^;]*);HNR: ([^;]*);PLZ: ([^;]*);ORT: ([^;]*)", caller)
+       if not found:
                return
-       name = vorname = strasse = hnr = plz = ort = ""
-       lines = caller.split(', ')
-       found = re.match("(.+?)\s+(.+)", lines[0])
-       if found:
-               name = found.group(1)
-               vorname = found.group(2)
-       else:
-               name = lines[0]
-       aktuell = 1
-       found = re.match("^(.+) ([-\d]+)$", lines[1], re.S)
-       if found:
-               strasse = found.group(1)
-               hnr = found.group(2)
-               aktuell = 2
-       else:
-               found = re.match("^(\d+) (.+)$", lines[1], re.S)
-               if found:
-                       strasse = found.group(2)
-                       hnr = found.group(1)
-               else:
-                       strasse = lines[1]
-               aktuell = 2
-       for i in range(aktuell, len(lines)):
-               found = re.match("(\S+)\s+(.+)", lines[i], re.S)
-               if found:
-                       plz = found.group(1)
-                       ort = found.group(2)
-                       break
-       else:
-               ort = lines[aktuell].strip()
-       print "NA: %s;VN: %s;STR: %s;HNR: %s;PLZ: %s;ORT: %s" %( name,vorname,strasse,hnr,plz,ort )
+       ( name,vorname,strasse,hnr,plz,ort ) = (found.group(1),
+                                                                                       found.group(2),
+                                                                                       found.group(3),
+                                                                                       found.group(4),
+                                                                                       found.group(5),
+                                                                                       found.group(6)
+                                                                                       )
+       if vorname: name += ' ' + vorname
+       if strasse or hnr or plz or ort: name += ', '
+       if strasse: name += strasse
+       if hnr: name += ' ' + hnr
+       if (strasse or hnr) and (plz or ort): name += ', '
+       if plz and ort: name += plz + ' ' + ort
+       elif plz: name += plz
+       elif ort: name += ort
+
+       print(name)
 
-def simpleout(number, caller):
+def simpleout(number, caller): #@UnusedVariable
        print caller
 
 try:
@@ -158,6 +147,12 @@ class ReverseLookupAndNotifier:
 
                self.countrycode = countrycode
 
+               if re.match('^\+', self.number):
+                       self.number = '00' + self.number[1:]
+
+               if self.number[:len(countrycode)] == countrycode:
+                       self.number = '0' + self.number[len(countrycode):]
+
                if number[0] != "0":
                        # self.caller = _("UNKNOWN")
                        self.notifyAndReset()
@@ -229,20 +224,34 @@ class ReverseLookupAndNotifier:
 
        def _gotPage(self, page):
                def cleanName(text):
-                       item = text.replace("&nbsp;"," ").replace("</b>","").replace(","," ")
-                       item = html2unicode(item).decode('iso-8859-1')
+                       item = text.replace("%20"," ").replace("&nbsp;"," ").replace("</b>","").replace(","," ").replace('\n',' ').replace('\t',' ')
+
+                       item = html2unicode(item)
+                       try: # this works under Windows
+                               item = item.decode('iso-8859-1')
+                       except:
+                               try: # this works under Enigma2
+                                       item = item.decode('utf-8')
+                               except:
+                                       try: # fall back
+                                               item = item.decode(self.charset)
+                                       except:
+                                               # debug("[ReverseLookupAndNotifier] cleanName: " + traceback.format_exc())
+                                               debug("[ReverseLookupAndNotifier] cleanName: encoding problem")
+
                        newitem = item.replace("  ", " ")
                        while newitem != item:
                                item = newitem
                                newitem = item.replace("  ", " ")
                        return newitem.strip()
-                       
+       
                debug("[ReverseLookupAndNotifier] _gotPage")
-               found = re.match('.*content=".*?charset=([^"]+)"',page,re.S)
+               found = re.match('.*<meta http-equiv="Content-Type" content="(?:application/xhtml\+xml|text/html); charset=([^"]+)" />',page, re.S)
                if found:
                        debug("[ReverseLookupAndNotifier] Charset: " + found.group(1))
                        page = page.replace("\xa0"," ").decode(found.group(1), "replace")
                else:
+                       debug("[ReverseLookupAndNotifier] Default Charset: iso-8859-1")
                        page = page.replace("\xa0"," ").decode("ISO-8859-1", "replace")
 
                for entry in self.currentWebsite.getElementsByTagName("entry"):
@@ -262,19 +271,22 @@ class ReverseLookupAndNotifier:
                                        if number != normalizePhoneNumber(found.group(1)):
                                                debug("[ReverseLookupAndNotifier] _gotPage: got unequal number '''%s''' for '''%s'''" %(found.group(1),self.number))
                                                continue
-                       details = []
                        
                        # look for <firstname> and <lastname> match, if not there look for <name>, if not there break
-                       lastname = ''
+                       name = ''
                        firstname = ''
+                       street = ''
+                       streetno = ''
+                       city = ''
+                       zipcode = ''
                        pat = self.getPattern(entry, "lastname")
                        if pat:
-                               pat = ".*?" + pat 
+                               pat = ".*?" + pat
                                debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( "lastname", pat ))
                                found = re.match(pat, page, re.S|re.M)
                                if found:
-                                       debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "lastname", repr(found.group(1)) ))
-                                       lastname = cleanName(found.group(1))
+                                       debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "lastname", found.group(1)))
+                                       name = cleanName(found.group(1))
 
                                        pat = self.getPattern(entry, "firstname")
                                        if pat:
@@ -282,52 +294,82 @@ class ReverseLookupAndNotifier:
                                                debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( "firstname", pat ))
                                                found = re.match(pat, page, re.S|re.M)
                                                if found:
-                                                       debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "firstname", repr(found.group(1)) ))
-                                               firstname = cleanName(found.group(1))
+                                                       debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "firstname", found.group(1)))
+                                               firstname = cleanName(found.group(1)).strip()
 
-                                       if firstname:
-                                               details.append(lastname + ' ' + firstname)
-                                       else:
-                                               details.append(lastname)
                        else:
                                pat = ".*?" + self.getPattern(entry, "name")
                                debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( "name", pat ))
                                found = re.match(pat, page, re.S|re.M)
                                if found:
-                                       debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "name", repr(found.group(1)) ))
+                                       debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "name", found.group(1)))
                                        item = cleanName(found.group(1))
-                                       debug("[ReverseLookupAndNotifier] _gotPage: add to details: " + repr(item))
-                                       details.append(item)
+                                       # debug("[ReverseLookupAndNotifier] _gotPage: name: " + item)
+                                       name = item.strip()
+                                       firstNameFirst = entry.getElementsByTagName('name')[0].getAttribute('swapFirstAndLastName')
+                                       # debug("[ReverseLookupAndNotifier] _gotPage: swapFirstAndLastName: " + firstNameFirst)
+                                       if firstNameFirst == 'true': # that means, the name is of the form "firstname lastname"
+                                               found = re.match('(.*?)\s+(.*)', name)
+                                               if found:
+                                                       firstname = found.group(1)
+                                                       name = found.group(2)
                                else:
                                        debug("[ReverseLookupAndNotifier] _gotPage: no name found, skipping")
                                        continue
 
-                       for what in ["street", "city", "zipcode"]:
-                               pat = ".*?" + self.getPattern(entry, what)
-                               debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( what, pat ))
-                               found = re.match(pat, page, re.S|re.M)
-                               if found:
-                                       debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( what, repr(found.group(1)) ))
-                                       item = cleanName(found.group(1))
-                                       debug("[ReverseLookupAndNotifier] _gotPage: add to details: " + repr(item))
-                                       details.append(item.strip())
-                               else:
-                                       break
+                       if not name:
+                               continue
 
-                       if len(details) != 4:
+                       pat = ".*?" + self.getPattern(entry, "city")
+                       debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( "city", pat ))
+                       found = re.match(pat, page, re.S|re.M)
+                       if found:
+                               debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "city", found.group(1)))
+                               item = cleanName(found.group(1))
+                               debug("[ReverseLookupAndNotifier] _gotPage: city: " + item)
+                               city = item.strip()
+
+                       if not city:
                                continue
-                       else:
-                               name = details[0]
-                               address =  details[1] + ", " + details[3] + " " + details[2]
-                               debug("[ReverseLookupAndNotifier] _gotPage: Reverse lookup succeeded:\nName: %s\nAddress: %s" %(name, address))
-                               self.caller = "%s, %s" %(name, address)
-                               # if self.number != 0 and config.plugins.Call.addcallers.value and self.event == "RING":
-                                       # phonebook.add(self.number, self.caller)
 
-                               self.notifyAndReset()
-                               return True
+                       pat = ".*?" + self.getPattern(entry, "zipcode")
+                       debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( "zipcode", pat ))
+                       found = re.match(pat, page, re.S|re.M)
+                       if found and found.group(1):
+                               debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "zipcode", found.group(1)))
+                               item = cleanName(found.group(1))
+                               debug("[ReverseLookupAndNotifier] _gotPage: zipcode: " + item)
+                               zipcode = item.strip()
+
+                       pat = ".*?" + self.getPattern(entry, "street")
+                       debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( "street", pat ))
+                       found = re.match(pat, page, re.S|re.M)
+                       if found and found.group(1):
+                               debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "street", found.group(1)))
+                               item = cleanName(found.group(1))
+                               debug("[ReverseLookupAndNotifier] _gotPage: street: " + item)
+                               street = item.strip()
+                               streetno = ''
+                               found = re.match("^(.+) ([-\d]+)$", street, re.S)
+                               if found:
+                                       street = found.group(1)
+                                       streetno= found.group(2)
+                               #===============================================================
+                               # else:
+                               #       found = re.match("^(\d+) (.+)$", street, re.S)
+                               #       if found:
+                               #               street = found.group(2)
+                               #               streetno = found.group(1)
+                               #===============================================================
+
+                       self.caller = "NA: %s;VN: %s;STR: %s;HNR: %s;PLZ: %s;ORT: %s" %( name,firstname,street,streetno,zipcode,city )
+                       debug("[ReverseLookupAndNotifier] _gotPage: Reverse lookup succeeded:\nName: %s" %(self.caller))
+
+                       self.notifyAndReset()
+                       return True
                else:
                        self._gotError("[ReverseLookupAndNotifier] _gotPage: Nothing found at %s" %self.currentWebsite.getAttribute("name"))
+                       return False
                        
        def _gotError(self, error = ""):
                debug("[ReverseLookupAndNotifier] _gotError - Error: %s" %error)
@@ -356,7 +398,7 @@ class ReverseLookupAndNotifier:
                if self.caller:
                        try:
                                # debug("2: " + repr(self.caller))
-                               self.caller = self.caller.encode(self.charset)
+                               self.caller = self.caller.encode(self.charset, 'replace')
                                # debug("3: " + repr(self.caller))
                        except:
                                debug("[ReverseLookupAndNotifier] cannot encode?!?!")
@@ -373,10 +415,10 @@ if __name__ == '__main__':
        cwd = os.path.dirname(sys.argv[0])
        if (len(sys.argv) == 2):
                # nrzuname.py Nummer
-               ReverseLookupAndNotifier(sys.argv[1])
+               ReverseLookupAndNotifier(sys.argv[1], simpleout)
                reactor.run() #@UndefinedVariable
        elif (len(sys.argv) == 3):
                # nrzuname.py Nummer Charset
                setDebug(False)
-               ReverseLookupAndNotifier(sys.argv[1], simpleout, sys.argv[2])
+               ReverseLookupAndNotifier(sys.argv[1], out, sys.argv[2])
                reactor.run() #@UndefinedVariable