#!/usr/bin/python
# -*- coding: UTF-8 -*-
-# $Id$
-# $Author$
-# $Revision$
-# $Date$
+'''
+$Id$
+$Author$
+$Revision$
+$Date$
+'''
import re, sys, os
from xml.dom.minidom import parse
from twisted.web.client import getPage #@UnresolvedImport
from twisted.internet import reactor #@UnresolvedImport
-debugSetting = True
-def setDebug(what):
- global debugSetting
- debugSetting = what
-
-def debug(str):
- if debugSetting:
- print str
+try:
+ from . import debug #@UnresolvedImport # pylint: disable-msg=W0613,F0401
+ def setDebug(what): # pylint: disable-msg=W0613
+ pass
+except ValueError:
+ debugVal = True
+ def setDebug(what):
+ global debugVal
+ debugVal = what
+ def debug(message):
+ if debugVal:
+ print message
import htmlentitydefs
-def html2unicode(in_html):
- # sanity checks
- try:
- in_html = in_html.decode('iso-8859-1')
- debug("[Callhtml2utf8] Converted from latin1")
- except:
- debug("[Callhtml2utf8] lost in translation from latin1")
- pass
- try:
- in_html = in_html.decode('utf-8')
- debug("[Callhtml2utf8] Converted from utf-8")
- except:
- debug("[Callhtml2utf8] lost in translation from utf-8")
- pass
+def html2unicode(in_html, charset):
+#===============================================================================
+# # sanity checks
+# try:
+# in_html = in_html.decode('iso-8859-1')
+# debug("[Callhtml2utf8] Converted from latin1")
+# except:
+# debug("[Callhtml2utf8] lost in translation from latin1")
+# pass
+# try:
+# in_html = in_html.decode('utf-8')
+# debug("[Callhtml2utf8] Converted from utf-8")
+# except:
+# debug("[Callhtml2utf8] lost in translation from utf-8")
+# pass
+#===============================================================================
# first convert some WML codes from hex: e.g. ä -> ä
htmlentityhexnumbermask = re.compile('(&#x(..);)')
entities = htmlentityhexnumbermask.finditer(in_html)
for x in entities:
- in_html = in_html.replace(x.group(1), '&#' + str(int(x.group(2),16)) + ';')
+ in_html = in_html.replace(x.group(1), '&#' + str(int(x.group(2), 16)) + ';')
- entitydict = {}
- # catch ü and colleagues here
htmlentitynamemask = re.compile('(&(\D{1,5}?);)')
+ entitydict = {}
entities = htmlentitynamemask.finditer(in_html)
for x in entities:
# debug("[Callhtml2utf8] mask: found %s" %repr(x.group(2)))
- entitydict[x.group(1)] = htmlentitydefs.name2codepoint[str(x.group(2))]
+ entitydict[x.group(1)] = x.group(2)
+ for key, name in entitydict.items():
+ try:
+ entitydict[key] = htmlentitydefs.name2codepoint[str(name)]
+ except KeyError:
+ debug("[Callhtml2utf8] KeyError " + key + "/" + name)
- # this is for Ġ and other numbers
htmlentitynumbermask = re.compile('(&#(\d{1,5}?);)')
entities = htmlentitynumbermask.finditer(in_html)
for x in entities:
# debug("[Callhtml2utf8] number: found %s" %x.group(1))
entitydict[x.group(1)] = x.group(2)
-
- # no go and replace all occurrences
for key, codepoint in entitydict.items():
try:
- debug("[Callhtml2utf8] replace %s with %s" %(repr(key), unichr(int(codepoint))))
- in_html = in_html.replace(unicode(key), (unichr(int(codepoint))))
- # in_html = in_html.replace(unicode(key), (unichr(int(codepoint))).decode('cp1252').encode('utf-8'))
- except ValueError:
- debug("[Callhtml2utf8] ValueError " + key + "/" + str(codepoint))
- pass
+ uml = unichr(int(codepoint))
+ debug("[nrzuname] html2utf8: replace %s with %s in %s" %(repr(key), repr(uml), repr(in_html[0:20]+'...')))
+ in_html = in_html.replace(key, uml)
+ except ValueError, e:
+ debug("[nrzuname] html2utf8: ValueError " + repr(key) + ":" + repr(codepoint) + " (" + str(e) + ")")
return in_html
-def out(number, caller):
- debug("[out] %s: %s" %(number, caller))
- if not caller:
- return
- name = vorname = strasse = hnr = plz = ort = ""
- lines = caller.split(', ')
- found = re.match("(.+?)\s+(.+)", lines[0])
+def normalizePhoneNumber(intNo):
+ found = re.match('^\+(.*)', intNo)
if found:
- name = found.group(1)
- vorname = found.group(2)
- else:
- name = lines[0]
- aktuell = 1
- found = re.match("^(.+) ([-\d]+)$", lines[1], re.S)
+ intNo = '00' + found.group(1)
+ intNo = intNo.replace('(', '').replace(')', '').replace(' ', '').replace('/', '').replace('-', '')
+ found = re.match('.*?([0-9]+)', intNo)
if found:
- strasse = found.group(1)
- hnr = found.group(2)
- aktuell = 2
- else:
- found = re.match("^(\d+) (.+)$", 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
+ return found.group(1)
else:
- ort = lines[aktuell].strip()
- print "NA: %s;VN: %s;STR: %s;HNR: %s;PLZ: %s;ORT: %s" %( name,vorname,strasse,hnr,plz,ort )
+ return '0'
-def simpleout(number, caller):
+def out(number, 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 ) = (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): #@UnusedVariable # pylint: disable-msg=W0613
print caller
try:
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()
number = self.number
url = website.getAttribute("url")
- if re.search('$AREACODE',url) or re.search('$PFXAREACODE',url):
+ if re.search('$AREACODE', url) or re.search('$PFXAREACODE', url):
debug("[ReverseLookupAndNotifier] handleWebsite: (PFX)ARECODE cannot be handled")
# self.caller = _("UNKNOWN")
self.notifyAndReset()
# Apparently, there is no attribute called (pfx)areacode anymore
# So, this below will not work.
#
- if re.search('\\$AREACODE',url) and website.hasAttribute("areacode"):
+ if re.search('\\$AREACODE', url) and website.hasAttribute("areacode"):
areaCodeLen = int(website.getAttribute("areacode"))
- url = url.replace("$AREACODE","%(areacode)s").replace("$NUMBER","%(number)s")
- url = url %{ 'areacode':number[:areaCodeLen], 'number':number[areaCodeLen:] }
- elif re.search('\\$PFXAREACODE',url) and website.hasAttribute("pfxareacode"):
+ url = url.replace("$AREACODE", number[:areaCodeLen]).replace("$NUMBER", number[areaCodeLen:])
+ elif re.search('\\$PFXAREACODE', url) and website.hasAttribute("pfxareacode"):
areaCodeLen = int(website.getAttribute("pfxareacode"))
- url = url.replace("$PFXAREACODE","%(pfxareacode)s").replace("$NUMBER","%(number)s")
- url = url %{ 'pfxareacode':number[:areaCodeLen], 'number':number[areaCodeLen:] }
- elif re.search('\\$NUMBER',url):
+ url = url.replace("$PFXAREACODE","%(pfxareacode)s").replace("$NUMBER", "%(number)s")
+ url = url % { 'pfxareacode': number[:areaCodeLen], 'number': number[areaCodeLen:] }
+ elif re.search('\\$NUMBER', url):
url = url.replace("$NUMBER","%s") %number
else:
debug("[ReverseLookupAndNotifier] handleWebsite: cannot handle websites with no $NUMBER in url")
def _gotPage(self, page):
+ def cleanName(text):
+ item = text.replace("%20"," ").replace(" "," ").replace("</b>","").replace(","," ").replace('\n',' ').replace('\t',' ')
+
+ item = html2unicode(item, self.charset)
+ #===================================================================
+ # try: # this works under Windows
+ # item = item.encode('iso-8859-1')
+ # except UnicodeEncodeError:
+ # debug("[ReverseLookupAndNotifier] cleanName: encoding problem with iso8859")
+ # try: # this works under Enigma2
+ # item = item.encode('utf-8')
+ # except UnicodeEncodeError:
+ # debug("[ReverseLookupAndNotifier] cleanName: encoding problem with utf-8")
+ # try: # fall back
+ # item = item.encode(self.charset)
+ # except UnicodeEncodeError:
+ # # 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"):
- # debug("[ReverseLookupAndNotifier] _gotPage: try entry")
- details = []
- for what in ["name", "street", "city", "zipcode"]:
- pat = ".*?" + self.getPattern(entry, what)
- debug("[ReverseLookupAndNotifier] _gotPage: look for '''%s''' with '''%s'''" %( what, pat ))
+ #
+ # for the sites delivering fuzzy matches, we check against the returned number
+ #
+ pat = self.getPattern(entry, "number")
+ if pat:
+ pat = ".*?" + pat
+ debug("[ReverseLookupAndNotifier] _gotPage: look for number with '''%s'''" %( pat ))
+ found = re.match(pat, page, re.S|re.M)
+ if found:
+ if self.number[:2] == '00':
+ number = '0' + self.number[4:]
+ else:
+ number = self.number
+ if number != normalizePhoneNumber(found.group(1)):
+ debug("[ReverseLookupAndNotifier] _gotPage: got unequal number '''%s''' for '''%s'''" %(found.group(1), self.number))
+ continue
+
+ # look for <firstname> and <lastname> match, if not there look for <name>, if not there break
+ name = ''
+ firstname = ''
+ street = ''
+ streetno = ''
+ city = ''
+ zipcode = ''
+ pat = self.getPattern(entry, "lastname")
+ if 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", found.group(1)))
+ name = cleanName(found.group(1))
+
+ pat = self.getPattern(entry, "firstname")
+ if pat:
+ pat = ".*?" + pat
+ 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", found.group(1)))
+ firstname = cleanName(found.group(1)).strip()
+
+ 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'''" %( what, repr(found.group(1)) ))
- item = found.group(1).replace(" "," ").replace("</b>","").replace(","," ")
- item = html2unicode(item)
- newitem = item.replace(" ", " ")
- while newitem != item:
- item = newitem
- newitem = item.replace(" ", " ")
- debug("[ReverseLookupAndNotifier] _gotPage: add to details: " + repr(item))
- details.append(item.strip())
+ debug("[ReverseLookupAndNotifier] _gotPage: found for '''%s''': '''%s'''" %( "name", found.group(1)))
+ item = cleanName(found.group(1))
+ # 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:
- break
+ debug("[ReverseLookupAndNotifier] _gotPage: no name found, skipping")
+ continue
- if len(details) != 4:
+ if not name:
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, "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
+
+ 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)
def getPattern(self, website, which):
pat1 = website.getElementsByTagName(which)
- if len(pat1) > 1:
- debug("[ReverseLookupAndNotifier] getPattern: Something strange: more than one %s for website %s" %(which, website.getAttribute("name")))
- return pat1[0].childNodes[0].data
+ if len(pat1) == 0:
+ return ''
+ else:
+ if len(pat1) > 1:
+ debug("[ReverseLookupAndNotifier] getPattern: Something strange: more than one %s for website %s" %(which, website.getAttribute("name")))
+ return pat1[0].childNodes[0].data
def notifyAndReset(self):
debug("[ReverseLookupAndNotifier] notifyAndReset: Number: " + self.number + "; Caller: " + self.caller)
# debug("1: " + repr(self.caller))
if self.caller:
try:
- # debug("2: " + repr(self.caller))
- self.caller = self.caller.encode(self.charset)
- # debug("3: " + repr(self.caller))
- except:
+ debug("2: " + repr(self.caller))
+ self.caller = self.caller.encode(self.charset, 'replace')
+ debug("3: " + repr(self.caller))
+ except UnicodeDecodeError:
debug("[ReverseLookupAndNotifier] cannot encode?!?!")
- pass
# self.caller = unicode(self.caller)
# debug("4: " + repr(self.caller))
self.outputFunction(self.number, self.caller)
else:
self.outputFunction(self.number, "")
if __name__ == '__main__':
- reactor.stop() #@UndefinedVariable
+ reactor.stop() #@UndefinedVariable # pylint: disable-msg=E1101
if __name__ == '__main__':
cwd = os.path.dirname(sys.argv[0])
if (len(sys.argv) == 2):
# nrzuname.py Nummer
- ReverseLookupAndNotifier(sys.argv[1])
- reactor.run() #@UndefinedVariable
+ ReverseLookupAndNotifier(sys.argv[1], simpleout)
+ reactor.run() #@UndefinedVariable # pylint: disable-msg=E1101
elif (len(sys.argv) == 3):
# nrzuname.py Nummer Charset
setDebug(False)
- ReverseLookupAndNotifier(sys.argv[1], simpleout, sys.argv[2])
- reactor.run() #@UndefinedVariable
+ ReverseLookupAndNotifier(sys.argv[1], out, sys.argv[2])
+ reactor.run() #@UndefinedVariable # pylint: disable-msg=E1101
projects / vuplus_dvbapp-plugin / blobdiff