2 # -*- coding: UTF-8 -*-
8 import re, sys, os, traceback
9 from xml.dom.minidom import parse
10 from twisted.web.client import getPage #@UnresolvedImport
11 from twisted.internet import reactor #@UnresolvedImport
14 from . import debug #@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)
100 if len(lines) > 2: # this means, we have street and city
101 found = re.match("^(.+) ([-\d]+)$", lines[1], re.S)
103 strasse = found.group(1)
106 found = re.match("^(\d+) (.+)$", lines[1], re.S)
108 strasse = found.group(2)
112 for i in range(2, len(lines)):
113 found = re.match("(\S+)\s+(.+)", lines[i], re.S)
114 if found and re.search('\d', found.group(1)):
118 else: # only two lines, the second must be the city...
119 ort = lines[1].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
163 if re.match('^\+', self.number):
164 self.number = '00' + self.number[1:]
166 if self.number[:len(countrycode)] == countrycode:
167 self.number = '0' + self.number[len(countrycode):]
170 # self.caller = _("UNKNOWN")
171 self.notifyAndReset()
174 if self.number[:2] == "00":
175 if countries.has_key(self.number[:3]): # e.g. USA
176 self.countrycode = self.number[:3]
177 elif countries.has_key(self.number[:4]):
178 self.countrycode = self.number[:4]
179 elif countries.has_key(self.number[:5]):
180 self.countrycode = self.number[:5]
182 debug("[ReverseLookupAndNotifier] Country cannot be reverse handled")
183 # self.caller = _("UNKNOWN")
184 self.notifyAndReset()
187 if countries.has_key(self.countrycode):
188 debug("[ReverseLookupAndNotifier] Found website for reverse lookup")
189 self.websites = countries[self.countrycode]
190 self.nextWebsiteNo = 1
191 self.handleWebsite(self.websites[0])
193 debug("[ReverseLookupAndNotifier] Country cannot be reverse handled")
194 # self.caller = _("UNKNOWN")
195 self.notifyAndReset()
198 def handleWebsite(self, website):
199 debug("[ReverseLookupAndNotifier] handleWebsite: " + website.getAttribute("name"))
200 if self.number[:2] == "00":
201 number = website.getAttribute("prefix") + self.number.replace(self.countrycode,"")
205 url = website.getAttribute("url")
206 if re.search('$AREACODE',url) or re.search('$PFXAREACODE',url):
207 debug("[ReverseLookupAndNotifier] handleWebsite: (PFX)ARECODE cannot be handled")
208 # self.caller = _("UNKNOWN")
209 self.notifyAndReset()
212 # Apparently, there is no attribute called (pfx)areacode anymore
213 # So, this below will not work.
215 if re.search('\\$AREACODE',url) and website.hasAttribute("areacode"):
216 areaCodeLen = int(website.getAttribute("areacode"))
217 url = url.replace("$AREACODE","%(areacode)s").replace("$NUMBER","%(number)s")
218 url = url %{ 'areacode':number[:areaCodeLen], 'number':number[areaCodeLen:] }
219 elif re.search('\\$PFXAREACODE',url) and website.hasAttribute("pfxareacode"):
220 areaCodeLen = int(website.getAttribute("pfxareacode"))
221 url = url.replace("$PFXAREACODE","%(pfxareacode)s").replace("$NUMBER","%(number)s")
222 url = url %{ 'pfxareacode':number[:areaCodeLen], 'number':number[areaCodeLen:] }
223 elif re.search('\\$NUMBER',url):
224 url = url.replace("$NUMBER","%s") %number
226 debug("[ReverseLookupAndNotifier] handleWebsite: cannot handle websites with no $NUMBER in url")
227 # self.caller = _("UNKNOWN")
228 self.notifyAndReset()
230 debug("[ReverseLookupAndNotifier] Url to query: " + url)
231 url = url.encode("UTF-8", "replace")
232 self.currentWebsite = website
234 agent="Mozilla/5.0 (Windows; U; Windows NT 6.0; de; rv:1.9.0.5) Gecko/2008120122 Firefox/3.0.5"
235 ).addCallback(self._gotPage).addErrback(self._gotError)
238 def _gotPage(self, page):
240 item = text.replace(" "," ").replace("</b>","").replace(","," ").replace('\n',' ').replace('\t',' ')
242 item = html2unicode(item).decode('iso-8859-1')
243 # item = html2unicode(item)
244 newitem = item.replace(" ", " ")
245 while newitem != item:
247 newitem = item.replace(" ", " ")
248 return newitem.strip()
250 debug("[ReverseLookupAndNotifier] cleanName: " + traceback.format_exc())
253 debug("[ReverseLookupAndNotifier] _gotPage")
254 found = re.match('.*content=".*?charset=([^"]+)"',page, re.S)
256 debug("[ReverseLookupAndNotifier] Charset: " + found.group(1))
257 page = page.replace("\xa0"," ").decode(found.group(1), "replace")
259 debug("[ReverseLookupAndNotifier] Charset: iso-8859-1")
260 page = page.replace("\xa0"," ").decode("ISO-8859-1", "replace")
262 for entry in self.currentWebsite.getElementsByTagName("entry"):
264 # for the sites delivering fuzzy matches, we check against the returned number
266 pat = self.getPattern(entry, "number")
269 debug("[ReverseLookupAndNotifier] _gotPage: look for number with '''%s'''" %( pat ))
270 found = re.match(pat, page, re.S|re.M)
272 if self.number[:2] == '00':
273 number = '0' + self.number[4:]
276 if number != normalizePhoneNumber(found.group(1)):
277 debug("[ReverseLookupAndNotifier] _gotPage: got unequal number '''%s''' for '''%s'''" %(found.group(1),self.number))
280 # look for <firstname> and <lastname> match, if not there look for <name>, if not there break
283 pat = self.getPattern(entry, "lastname")
286 debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( "lastname", pat ))
287 found = re.match(pat, page, re.S|re.M)
289 debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "lastname", found.group(1)))
290 lastname = cleanName(found.group(1))
292 pat = self.getPattern(entry, "firstname")
295 debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( "firstname", pat ))
296 found = re.match(pat, page, re.S|re.M)
298 debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "firstname", found.group(1)))
299 firstname = cleanName(found.group(1))
302 name = lastname + ' ' + firstname
306 pat = ".*?" + self.getPattern(entry, "name")
307 debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( "name", pat ))
308 found = re.match(pat, page, re.S|re.M)
310 debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "name", found.group(1)))
311 item = cleanName(found.group(1))
312 debug("[ReverseLookupAndNotifier] _gotPage: name: " + item)
315 debug("[ReverseLookupAndNotifier] _gotPage: no name found, skipping")
320 pat = ".*?" + self.getPattern(entry, "city")
321 debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( "city", pat ))
322 found = re.match(pat, page, re.S|re.M)
324 debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "city", found.group(1)))
325 item = cleanName(found.group(1))
326 debug("[ReverseLookupAndNotifier] _gotPage: city: " + item)
327 address = item.strip()
329 pat = ".*?" + self.getPattern(entry, "zipcode")
330 debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( "zipcode", pat ))
331 found = re.match(pat, page, re.S|re.M)
332 if found and found.group(1):
333 debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "zipcode", found.group(1)))
334 item = cleanName(found.group(1))
335 debug("[ReverseLookupAndNotifier] _gotPage: zipcode: " + item)
336 address = item.strip() + ' ' + address
338 pat = ".*?" + self.getPattern(entry, "street")
339 debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( "street", pat ))
340 found = re.match(pat, page, re.S|re.M)
341 if found and found.group(1):
342 debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "street", found.group(1)))
343 item = cleanName(found.group(1))
344 debug("[ReverseLookupAndNotifier] _gotPage: street: " + item)
345 address = item.strip() + ', ' + address
348 debug("[ReverseLookupAndNotifier] _gotPage: Reverse lookup succeeded:\nName: %s\nAddress: %s" %(name, address))
349 self.caller = "%s, %s" %(name, address)
351 debug("[ReverseLookupAndNotifier] _gotPage: Reverse lookup succeeded:\nName: %s" %(name))
354 self.notifyAndReset()
359 self._gotError("[ReverseLookupAndNotifier] _gotPage: Nothing found at %s" %self.currentWebsite.getAttribute("name"))
362 def _gotError(self, error = ""):
363 debug("[ReverseLookupAndNotifier] _gotError - Error: %s" %error)
364 if self.nextWebsiteNo >= len(self.websites):
365 debug("[ReverseLookupAndNotifier] _gotError: I give up")
366 # self.caller = _("UNKNOWN")
367 self.notifyAndReset()
370 debug("[ReverseLookupAndNotifier] _gotError: try next website")
371 self.nextWebsiteNo = self.nextWebsiteNo+1
372 self.handleWebsite(self.websites[self.nextWebsiteNo-1])
374 def getPattern(self, website, which):
375 pat1 = website.getElementsByTagName(which)
380 debug("[ReverseLookupAndNotifier] getPattern: Something strange: more than one %s for website %s" %(which, website.getAttribute("name")))
381 return pat1[0].childNodes[0].data
383 def notifyAndReset(self):
384 debug("[ReverseLookupAndNotifier] notifyAndReset: Number: " + self.number + "; Caller: " + self.caller)
385 # debug("1: " + repr(self.caller))
388 # debug("2: " + repr(self.caller))
389 self.caller = self.caller.encode(self.charset, 'replace')
390 # debug("3: " + repr(self.caller))
392 debug("[ReverseLookupAndNotifier] cannot encode?!?!")
394 # self.caller = unicode(self.caller)
395 # debug("4: " + repr(self.caller))
396 self.outputFunction(self.number, self.caller)
398 self.outputFunction(self.number, "")
399 if __name__ == '__main__':
400 reactor.stop() #@UndefinedVariable
402 if __name__ == '__main__':
403 cwd = os.path.dirname(sys.argv[0])
404 if (len(sys.argv) == 2):
406 ReverseLookupAndNotifier(sys.argv[1])
407 reactor.run() #@UndefinedVariable
408 elif (len(sys.argv) == 3):
409 # nrzuname.py Nummer Charset
411 ReverseLookupAndNotifier(sys.argv[1], simpleout, sys.argv[2])
412 reactor.run() #@UndefinedVariable