statistic.py

   1 #!/usr/bin/python
   2 import sys, string
   3 from xml.dom import minidom, Node
   4 import os
   5 import datetime
   6
   7 from xml.dom.ext import PrettyPrint
   8 from StringIO import StringIO
   9
  10 def toprettyxml_fixed (node): #, encoding='utf-8'):
  11    tmpStream = StringIO()
  12    PrettyPrint(node, stream=tmpStream)# , encoding=encoding)
  13    return tmpStream.getvalue()
  14
  15 class Statistic:
  16    def IncreaseCounter(self, qid, how, answer, time):
  17       """
  18       Return True, is question is good
  19       """
  20       nq = self.FindQuestion (qid)
  21       if nq >= 0:
  22          for q in  self.root.getElementsByTagName("question"):
  23             id = q.getAttribute ("id")
  24             if id == qid:
  25                break
  26
  27          c = int(q.getAttribute ("c"))
  28          w = int(q.getAttribute ("w"))
  29          cs = int(q.getAttribute ("cs"))
  30          ws = int(q.getAttribute ("ws"))
  31
  32       else:
  33          print "NEU",qid
  34          q = self.stat.createElement(u'question')
  35          q.setAttribute("id",qid)
  36
  37          sss = self.root.getElementsByTagName("learning")[0]
  38          sss.appendChild(q)
  39
  40          c=0
  41          w=0
  42          cs=0
  43          ws=0
  44
  45          self.statistics.append ([qid, c, cs, ws, w, ""])
  46          nq = self.FindQuestion (qid)
  47
  48       if how == True:
  49          c += 1
  50          cs += 1
  51          ws = max(ws-1,0)
  52       else:
  53          w += 1
  54          ws += 1
  55          cs = min(cs-1,0)
  56
  57       if c < 0:
  58          c = 0
  59       if cs < 0:
  60          cs = 0
  61       if w < 0:
  62          w = 0
  63       if ws < 0:
  64          ws = 0
  65
  66       # in-memory statistics
  67       self.statistics[nq][1] = c
  68       self.statistics[nq][2] = cs
  69       self.statistics[nq][3] = ws
  70       self.statistics[nq][4] = w
  71
  72       # xml file stuff
  73       q.setAttribute("c",str(c))
  74       q.setAttribute("w",str(w))
  75       q.setAttribute("cs",str(cs))
  76       q.setAttribute("ws",str(ws))
  77
  78
  79       t = self.Timestamp()
  80       a = [1,2,4,8][(["a","b","c","d"]).index(answer)]
  81       nt = int(time*1000)
  82
  83       qq = self.stat.createElement("answer_clicked")
  84       qq.setAttribute ("datetime", str(t))
  85       qq.setAttribute ("answer_code", str(a))
  86       qq.setAttribute ("needed_time", str(nt))
  87       q.appendChild(qq)
  88
  89       if float(ws) > 0.:
  90          rr = 0.
  91       else:
  92          rr = float(cs)*self.norm
  93
  94       return rr >= self.ratio
  95
  96    def WriteFile(self):
  97       print "Writing statistics file",self.filename
  98       ss=toprettyxml_fixed(self.root)
  99       f=open(self.filename,"w")
 100       print >>f,"<!DOCTYPE AFUTrainerStatistics>"
 101       f.write(ss)
 102       f.close()
 103
 104    def OpenFile(self):
 105       print "Opening statistics file",self.filename
 106       self.stat = minidom.parse (self.filename)
 107       self.root = self.stat.documentElement
 108
 109       self.date = self.root.getAttribute ("date")
 110       self.version = self.root.getAttribute ("version")
 111       self.name = self.root.getAttribute ("name")
 112
 113
 114    def GetStatistics(self):
 115       print "Parsing statistics xml"
 116       self.statistics = []
 117
 118       self.ratio = .75
 119       self.mincorrect = 5.
 120       self.newquestion = []
 121       self.goodquestion = []
 122       self.badquestion = []
 123       self.norm = 1./self.mincorrect
 124
 125       for q in self.root.getElementsByTagName("question"):
 126          id = q.getAttribute ("id")
 127          c = q.getAttribute ("c")
 128          cs = q.getAttribute ("cs")
 129          ws = q.getAttribute ("ws")
 130          w = q.getAttribute ("w")
 131
 132          if float(ws) > 0.:
 133             rr = 0.
 134          else:
 135             rr = float(cs)*self.norm
 136
 137          if rr >= self.ratio:
 138             self.goodquestion.append (id)
 139          elif rr >= 0:
 140             self.badquestion.append (id)
 141          else: # never reached
 142             self.newquestion.append (id)
 143
 144          answers = []
 145          for a in q.childNodes:
 146             if a.nodeType == Node.ELEMENT_NODE:
 147                code = a.getAttribute("answer_code")
 148                time = a.getAttribute("needed_time")
 149                when = a.getAttribute("datetime")
 150                answers.append ([code,time,when])
 151
 152          self.statistics.append ([id, c, cs, ws, w, answers])
 153
 154    def GetPriority(self,qid):
 155       return
 156
 157    def FindQuestion(self,qid):
 158       i = 0
 159       for s in self.statistics:
 160          if s[0]  == qid:
 161             return i
 162          i += 1
 163       return -1
 164
 165    def ThisQuestion(self,qid):
 166       q = self.FindQuestion (qid)
 167       if q >= 0:
 168          c = self.statistics[q][1]
 169          w = self.statistics[q][4]
 170          cs = self.statistics[q][2]
 171          ws = self.statistics[q][3]
 172       else:
 173          c=str(0)
 174          w=str(0)
 175          cs=str(0)
 176          ws=str(0)
 177       return [str(c),str(w),str(cs),str(ws)]
 178
 179    def Timestamp(self):
 180       return str(datetime.datetime.today().isoformat()).split(".")[0]
 181
 182    def __init__(self,filename="DL-A-2007.stat.xml"):
 183       self.filename=filename
 184
 185       self.OpenFile()
 186       self.GetStatistics()