2 # -*- coding: UTF-8 -*-
9 from xml.dom.minidom import parse
10 from twisted.web.client import getPage #@UnresolvedImport
11 from twisted.internet import reactor #@UnresolvedImport
27 def html2unicode(in_html):
28 #===============================================================================
31 # in_html = in_html.decode('iso-8859-1')
32 # debug("[Callhtml2utf8] Converted from latin1")
34 # debug("[Callhtml2utf8] lost in translation from latin1")
37 # in_html = in_html.decode('utf-8')
38 # debug("[Callhtml2utf8] Converted from utf-8")
40 # debug("[Callhtml2utf8] lost in translation from utf-8")
42 #===============================================================================
44 # first convert some WML codes from hex: e.g. ä -> ä
45 htmlentityhexnumbermask = re.compile('(&#x(..);)')
46 entities = htmlentityhexnumbermask.finditer(in_html)
48 in_html = in_html.replace(x.group(1), '&#' + str(int(x.group(2),16)) + ';')
50 htmlentitynamemask = re.compile('(&(\D{1,5}?);)')
52 entities = htmlentitynamemask.finditer(in_html)
54 # debug("[Callhtml2utf8] mask: found %s" %repr(x.group(2)))
55 entitydict[x.group(1)] = x.group(2)
56 for key, name in entitydict.items():
58 entitydict[key] = htmlentitydefs.name2codepoint[str(name)]
60 debug("[Callhtml2utf8] KeyError " + key + "/" + name)
62 htmlentitynumbermask = re.compile('(&#(\d{1,5}?);)')
63 entities = htmlentitynumbermask.finditer(in_html)
65 # debug("[Callhtml2utf8] number: found %s" %x.group(1))
66 entitydict[x.group(1)] = x.group(2)
67 for key, codepoint in entitydict.items():
69 debug("[nrzuname] html2utf8: replace %s with %s" %(repr(key), str(codepoint)))
70 in_html = in_html.replace(unicode(key), (unichr(int(codepoint))))
72 debug("[nrzuname] html2utf8: ValueError " + key + "/" + str(codepoint))
75 def normalizePhoneNumber(intNo):
76 found = re.match('^\+(.*)', intNo)
78 intNo = '00' + found.group(1)
79 intNo = intNo.replace('(', '').replace(')', '').replace(' ', '').replace('/', '').replace('-', '')
80 found = re.match('.*?([0-9]+)', intNo)
86 def out(number, caller):
87 debug("[nrzuname] out: %s: %s" %(number, caller))
90 name = vorname = strasse = hnr = plz = ort = ""
91 lines = caller.split(', ')
92 found = re.match("(.+?)\s+(.+)", lines[0])
95 vorname = found.group(2)
99 found = re.match("^(.+) ([-\d]+)$", lines[1], re.S)
101 strasse = found.group(1)
105 found = re.match("^(\d+) (.+)$", lines[1], re.S)
107 strasse = found.group(2)
112 for i in range(aktuell, len(lines)):
113 found = re.match("(\S+)\s+(.+)", lines[i], re.S)
119 ort = lines[aktuell].strip()
120 print "NA: %s;VN: %s;STR: %s;HNR: %s;PLZ: %s;ORT: %s" %( name,vorname,strasse,hnr,plz,ort )
122 def simpleout(number, caller):
126 from Tools.Directories import resolveFilename, SCOPE_PLUGINS
127 reverseLookupFileName = resolveFilename(SCOPE_PLUGINS, "Extensions/FritzCall/reverselookup.xml")
129 reverseLookupFileName = "reverselookup.xml"
132 reverselookupMtime = 0
134 class ReverseLookupAndNotifier:
135 def __init__(self, number, outputFunction=out, charset="cp1252", countrycode = "0049"):
136 debug("[ReverseLookupAndNotifier] reverse Lookup for %s!" %number)
138 self.outputFunction = outputFunction
140 self.currentWebsite = None
141 self.nextWebsiteNo = 0
142 #===============================================================================
143 # sorry does not work at all
145 # charset = sys.getdefaultencoding()
146 # debug("[ReverseLookupAndNotifier] set charset from system: %s!" %charset)
147 #===============================================================================
148 self.charset = charset
150 global reverselookupMtime
151 reverselookupMtimeAct = os.stat(reverseLookupFileName)[8]
152 if not countries or reverselookupMtimeAct > reverselookupMtime:
153 debug("[ReverseLookupAndNotifier] (Re-)Reading %s\n" %reverseLookupFileName)
154 reverselookupMtime = reverselookupMtimeAct
155 dom = parse(reverseLookupFileName)
156 for top in dom.getElementsByTagName("reverselookup"):
157 for country in top.getElementsByTagName("country"):
158 code = country.getAttribute("code").replace("+","00")
159 countries[code] = country.getElementsByTagName("website")
161 self.countrycode = countrycode
164 # self.caller = _("UNKNOWN")
165 self.notifyAndReset()
168 if self.number[:2] == "00":
169 if countries.has_key(self.number[:3]): # e.g. USA
170 self.countrycode = self.number[:3]
171 elif countries.has_key(self.number[:4]):
172 self.countrycode = self.number[:4]
173 elif countries.has_key(self.number[:5]):
174 self.countrycode = self.number[:5]
176 debug("[ReverseLookupAndNotifier] Country cannot be reverse handled")
177 # self.caller = _("UNKNOWN")
178 self.notifyAndReset()
181 if countries.has_key(self.countrycode):
182 debug("[ReverseLookupAndNotifier] Found website for reverse lookup")
183 self.websites = countries[self.countrycode]
184 self.nextWebsiteNo = 1
185 self.handleWebsite(self.websites[0])
187 debug("[ReverseLookupAndNotifier] Country cannot be reverse handled")
188 # self.caller = _("UNKNOWN")
189 self.notifyAndReset()
192 def handleWebsite(self, website):
193 debug("[ReverseLookupAndNotifier] handleWebsite: " + website.getAttribute("name"))
194 if self.number[:2] == "00":
195 number = website.getAttribute("prefix") + self.number.replace(self.countrycode,"")
199 url = website.getAttribute("url")
200 if re.search('$AREACODE',url) or re.search('$PFXAREACODE',url):
201 debug("[ReverseLookupAndNotifier] handleWebsite: (PFX)ARECODE cannot be handled")
202 # self.caller = _("UNKNOWN")
203 self.notifyAndReset()
206 # Apparently, there is no attribute called (pfx)areacode anymore
207 # So, this below will not work.
209 if re.search('\\$AREACODE',url) and website.hasAttribute("areacode"):
210 areaCodeLen = int(website.getAttribute("areacode"))
211 url = url.replace("$AREACODE","%(areacode)s").replace("$NUMBER","%(number)s")
212 url = url %{ 'areacode':number[:areaCodeLen], 'number':number[areaCodeLen:] }
213 elif re.search('\\$PFXAREACODE',url) and website.hasAttribute("pfxareacode"):
214 areaCodeLen = int(website.getAttribute("pfxareacode"))
215 url = url.replace("$PFXAREACODE","%(pfxareacode)s").replace("$NUMBER","%(number)s")
216 url = url %{ 'pfxareacode':number[:areaCodeLen], 'number':number[areaCodeLen:] }
217 elif re.search('\\$NUMBER',url):
218 url = url.replace("$NUMBER","%s") %number
220 debug("[ReverseLookupAndNotifier] handleWebsite: cannot handle websites with no $NUMBER in url")
221 # self.caller = _("UNKNOWN")
222 self.notifyAndReset()
224 debug("[ReverseLookupAndNotifier] Url to query: " + url)
225 url = url.encode("UTF-8", "replace")
226 self.currentWebsite = website
228 agent="Mozilla/5.0 (Windows; U; Windows NT 6.0; de; rv:1.9.0.5) Gecko/2008120122 Firefox/3.0.5"
229 ).addCallback(self._gotPage).addErrback(self._gotError)
232 def _gotPage(self, page):
235 item = text.replace(" "," ").replace("</b>","").replace(","," ")
236 item = html2unicode(item).decode('iso-8859-1')
237 # item = html2unicode(item)
238 newitem = item.replace(" ", " ")
239 while newitem != item:
241 newitem = item.replace(" ", " ")
242 return newitem.strip()
246 debug("[ReverseLookupAndNotifier] _gotPage")
247 found = re.match('.*content=".*?charset=([^"]+)"',page,re.S)
249 debug("[ReverseLookupAndNotifier] Charset: " + found.group(1))
250 page = page.replace("\xa0"," ").decode(found.group(1), "replace")
252 page = page.replace("\xa0"," ").decode("ISO-8859-1", "replace")
254 for entry in self.currentWebsite.getElementsByTagName("entry"):
256 # for the sites delivering fuzzy matches, we check against the returned number
258 pat = self.getPattern(entry, "number")
261 debug("[ReverseLookupAndNotifier] _gotPage: look for number with '''%s'''" %( pat ))
262 found = re.match(pat, page, re.S|re.M)
264 if self.number[:2] == '00':
265 number = '0' + self.number[4:]
268 if number != normalizePhoneNumber(found.group(1)):
269 debug("[ReverseLookupAndNotifier] _gotPage: got unequal number '''%s''' for '''%s'''" %(found.group(1),self.number))
273 # look for <firstname> and <lastname> match, if not there look for <name>, if not there break
276 pat = self.getPattern(entry, "lastname")
279 debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( "lastname", pat ))
280 found = re.match(pat, page, re.S|re.M)
282 debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "lastname", found.group(1)))
283 lastname = cleanName(found.group(1))
285 pat = self.getPattern(entry, "firstname")
288 debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( "firstname", pat ))
289 found = re.match(pat, page, re.S|re.M)
291 debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "firstname", found.group(1)))
292 firstname = cleanName(found.group(1))
295 details.append(lastname + ' ' + firstname)
297 details.append(lastname)
299 pat = ".*?" + self.getPattern(entry, "name")
300 debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( "name", pat ))
301 found = re.match(pat, page, re.S|re.M)
303 debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "name", found.group(1)))
304 item = cleanName(found.group(1))
305 debug("[ReverseLookupAndNotifier] _gotPage: add to details: " + item)
308 debug("[ReverseLookupAndNotifier] _gotPage: no name found, skipping")
311 for what in ["street", "city", "zipcode"]:
312 pat = ".*?" + self.getPattern(entry, what)
313 debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( what, pat ))
314 found = re.match(pat, page, re.S|re.M)
316 debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( what, found.group(1)))
317 item = cleanName(found.group(1))
318 debug("[ReverseLookupAndNotifier] _gotPage: add to details: " + item)
319 details.append(item.strip())
323 if len(details) != 4:
327 address = details[1] + ", " + details[3] + " " + details[2]
328 debug("[ReverseLookupAndNotifier] _gotPage: Reverse lookup succeeded:\nName: %s\nAddress: %s" %(name, address))
329 self.caller = "%s, %s" %(name, address)
330 # if self.number != 0 and config.plugins.Call.addcallers.value and self.event == "RING":
331 # phonebook.add(self.number, self.caller)
333 self.notifyAndReset()
336 self._gotError("[ReverseLookupAndNotifier] _gotPage: Nothing found at %s" %self.currentWebsite.getAttribute("name"))
338 def _gotError(self, error = ""):
339 debug("[ReverseLookupAndNotifier] _gotError - Error: %s" %error)
340 if self.nextWebsiteNo >= len(self.websites):
341 debug("[ReverseLookupAndNotifier] _gotError: I give up")
342 # self.caller = _("UNKNOWN")
343 self.notifyAndReset()
346 debug("[ReverseLookupAndNotifier] _gotError: try next website")
347 self.nextWebsiteNo = self.nextWebsiteNo+1
348 self.handleWebsite(self.websites[self.nextWebsiteNo-1])
350 def getPattern(self, website, which):
351 pat1 = website.getElementsByTagName(which)
356 debug("[ReverseLookupAndNotifier] getPattern: Something strange: more than one %s for website %s" %(which, website.getAttribute("name")))
357 return pat1[0].childNodes[0].data
359 def notifyAndReset(self):
360 debug("[ReverseLookupAndNotifier] notifyAndReset: Number: " + self.number + "; Caller: " + self.caller)
361 # debug("1: " + repr(self.caller))
364 # debug("2: " + repr(self.caller))
365 self.caller = self.caller.encode(self.charset)
366 # debug("3: " + repr(self.caller))
368 debug("[ReverseLookupAndNotifier] cannot encode?!?!")
370 # self.caller = unicode(self.caller)
371 # debug("4: " + repr(self.caller))
372 self.outputFunction(self.number, self.caller)
374 self.outputFunction(self.number, "")
375 if __name__ == '__main__':
376 reactor.stop() #@UndefinedVariable
378 if __name__ == '__main__':
379 cwd = os.path.dirname(sys.argv[0])
380 if (len(sys.argv) == 2):
382 ReverseLookupAndNotifier(sys.argv[1])
383 reactor.run() #@UndefinedVariable
384 elif (len(sys.argv) == 3):
385 # nrzuname.py Nummer Charset
387 ReverseLookupAndNotifier(sys.argv[1], simpleout, sys.argv[2])
388 reactor.run() #@UndefinedVariable