statistic.py

   1 #!/usr/bin/python
   2 import sys, string
   3 from xml.dom import minidom, Node
   4 # should really be done with  http://effbot.org/zone/element.htm
   5 import os
   6 import datetime
   7
   8 from xml.dom.ext import PrettyPrint
   9 from StringIO import StringIO
  10
  11 def toprettyxml_fixed (node): #, encoding='utf-8'):
  12    tmpStream = StringIO()
  13    PrettyPrint(node, stream=tmpStream)# , encoding=encoding)
  14    return tmpStream.getvalue()
  15
  16 class Statistic:
  17    def IncreaseCounter(self, qid, how, answer, time):
  18       """
  19       Return True, is question is good
  20       """
  21       nq = self.FindQuestion (qid)
  22       if nq >= 0:
  23          for q in  self.root.getElementsByTagName("question"):
  24             id = q.getAttribute ("id")
  25             if id == qid:
  26                break
  27
  28          c = int(q.getAttribute ("c"))
  29          w = int(q.getAttribute ("w"))
  30          cs = int(q.getAttribute ("cs"))
  31          ws = int(q.getAttribute ("ws"))
  32
  33       else:
  34          print "NEU",qid
  35          q = self.stat.createElement(u'question')
  36          q.setAttribute("id",qid)
  37
  38          sss = self.root.getElementsByTagName("learning")[0]
  39          sss.appendChild(q)
  40
  41          c=0
  42          w=0
  43          cs=0
  44          ws=0
  45
  46          self.statistics.append ([qid, c, cs, ws, w, ""])
  47          nq = self.FindQuestion (qid)
  48
  49       if how == True:
  50          c += 1
  51          cs += 1
  52          ws = max(ws-1,0)
  53       else:
  54          w += 1
  55          ws += 1
  56          cs = min(cs-1,0)
  57
  58       if c < 0:
  59          c = 0
  60       if cs < 0:
  61          cs = 0
  62       if w < 0:
  63          w = 0
  64       if ws < 0:
  65          ws = 0
  66
  67       # in-memory statistics
  68       self.statistics[nq][1] = c
  69       self.statistics[nq][2] = cs
  70       self.statistics[nq][3] = ws
  71       self.statistics[nq][4] = w
  72
  73       # xml file stuff
  74       q.setAttribute("c",str(c))
  75       q.setAttribute("w",str(w))
  76       q.setAttribute("cs",str(cs))
  77       q.setAttribute("ws",str(ws))
  78
  79
  80       t = self.Timestamp()
  81       a = [1,2,4,8][(["a","b","c","d"]).index(answer)]
  82       nt = int(time*1000)
  83
  84       qq = self.stat.createElement("answer_clicked")
  85       qq.setAttribute ("datetime", str(t))
  86       qq.setAttribute ("answer_code", str(a))
  87       qq.setAttribute ("needed_time", str(nt))
  88       q.appendChild(qq)
  89
  90       if float(ws) > 0.:
  91          rr = 0.
  92       else:
  93          rr = float(cs)*self.norm
  94
  95       return rr >= self.ratio
  96
  97    def WriteFile(self):
  98       print "Writing statistics file",self.filename
  99       ss=toprettyxml_fixed(self.root)
 100       f=open(self.filename,"w")
 101       print >>f,"<!DOCTYPE AFUTrainerStatistics>"
 102       f.write(ss)
 103       f.close()
 104
 105    def OpenFile(self):
 106       print "Opening statistics file",self.filename
 107       self.stat = minidom.parse (self.filename)
 108       self.root = self.stat.documentElement
 109
 110       self.date = self.root.getAttribute ("date")
 111       self.version = self.root.getAttribute ("version")
 112       self.name = self.root.getAttribute ("name")
 113
 114
 115    def GetStatistics(self):
 116       print "Parsing statistics xml"
 117       self.statistics = []
 118
 119       self.ratio = .75
 120       self.mincorrect = 5.
 121       self.newquestion = []
 122       self.goodquestion = []
 123       self.badquestion = []
 124       self.norm = 1./self.mincorrect
 125
 126       for q in self.root.getElementsByTagName("question"):
 127          id = q.getAttribute ("id")
 128          c = q.getAttribute ("c")
 129          cs = q.getAttribute ("cs")
 130          ws = q.getAttribute ("ws")
 131          w = q.getAttribute ("w")
 132
 133          if float(ws) > 0.:
 134             rr = 0.
 135          else:
 136             rr = float(cs)*self.norm
 137
 138          if rr >= self.ratio:
 139             self.goodquestion.append (id)
 140          elif rr >= 0:
 141             self.badquestion.append (id)
 142          else: # never reached
 143             self.newquestion.append (id)
 144
 145          answers = []
 146          for a in q.childNodes:
 147             if a.nodeType == Node.ELEMENT_NODE:
 148                code = a.getAttribute("answer_code")
 149                time = a.getAttribute("needed_time")
 150                when = a.getAttribute("datetime")
 151                answers.append ([code,time,when])
 152
 153          self.statistics.append ([id, c, cs, ws, w, answers])
 154
 155    def GetPriority(self,qid):
 156       return
 157
 158    def MakeStatistics(self):
 159       ntotal = len(self.goodquestion) + len(self.newquestion) + len(self.badquestion) + len(self.goodquestion)
 160       rcorrect = float(len(self.goodquestion))/ntotal
 161       rnew = float(len(self.newquestion))/ntotal
 162       return [rnew, rcorrect]
 163
 164    def FindQuestion(self,qid):
 165       i = 0
 166       for s in self.statistics:
 167          if s[0]  == qid:
 168             return i
 169          i += 1
 170       return -1
 171
 172    def ThisQuestion(self,qid):
 173       q = self.FindQuestion (qid)
 174       if q >= 0:
 175          c = self.statistics[q][1]
 176          w = self.statistics[q][4]
 177          cs = self.statistics[q][2]
 178          ws = self.statistics[q][3]
 179       else:
 180          c=str(0)
 181          w=str(0)
 182          cs=str(0)
 183          ws=str(0)
 184       return [str(c),str(w),str(cs),str(ws)]
 185
 186    def Timestamp(self):
 187       return str(datetime.datetime.today().isoformat()).split(".")[0]
 188
 189    def __init__(self,filename="DL-A-2007.stat.xml"):
 190       self.filename=filename
 191
 192       self.OpenFile()
 193       self.GetStatistics()