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):
  17       nq = self.FindQuestion (qid)
  18       if nq >= 0:
  19          for q in  self.root.getElementsByTagName("question"):
  20             id = q.getAttribute ("id")
  21             if id == qid:
  22                break
  23
  24          c = int(q.getAttribute ("c"))
  25          w = int(q.getAttribute ("w"))
  26          cs = int(q.getAttribute ("cs"))
  27          ws = int(q.getAttribute ("ws"))
  28
  29       else:
  30          print "NEU",qid
  31          q = self.stat.createElement(u'question')
  32          q.setAttribute("id",qid)
  33
  34          sss = self.root.getElementsByTagName("learning")[0]
  35          sss.appendChild(q)
  36
  37          c=0
  38          w=0
  39          cs=0
  40          ws=0
  41
  42          self.statistics.append ([qid, c, cs, ws, w, ""])
  43          nq = self.FindQuestion (qid)
  44
  45       if how == True:
  46          c += 1
  47          cs += 1
  48       else:
  49          w += 1
  50          ws += 1
  51
  52       # in-memory statistics
  53       self.statistics[nq][1] = c
  54       self.statistics[nq][2] = cs
  55       self.statistics[nq][3] = ws
  56       self.statistics[nq][4] = w
  57
  58       # xml file stuff
  59       q.setAttribute("c",str(c))
  60       q.setAttribute("w",str(w))
  61       q.setAttribute("cs",str(cs))
  62       q.setAttribute("ws",str(ws))
  63
  64
  65       t = self.Timestamp()
  66       a = [1,2,4,8][(["a","b","c","d"]).index(answer)]
  67       nt = 15000 # FIXME
  68
  69       qq = self.stat.createElement("answer_clicked")
  70       qq.setAttribute ("datetime", str(t))
  71       qq.setAttribute ("answer_code", str(a))
  72       qq.setAttribute ("needed_time", str(nt))
  73       q.appendChild(qq)
  74
  75       # FIXME: UPDATE QUESTION PRIORITIES
  76
  77    def WriteFile(self):
  78       print "Writing statistics file",self.filename
  79       ss=toprettyxml_fixed(self.root)
  80       f=open(self.filename,"w")
  81       print >>f,"<!DOCTYPE AFUTrainerStatistics>"
  82       f.write(ss)
  83       f.close()
  84
  85    def OpenFile(self):
  86       print "Opening statistics file",self.filename
  87       self.stat = minidom.parse (self.filename)
  88       self.root = self.stat.documentElement
  89
  90       self.date = self.root.getAttribute ("date")
  91       self.version = self.root.getAttribute ("version")
  92       self.name = self.root.getAttribute ("name")
  93
  94
  95    def GetStatistics(self):
  96       print "Parsing statistics xml"
  97       self.statistics = []
  98
  99       self.ratio = .75
 100       self.new = []
 101       self.good = []
 102       self.bad = []
 103       self.nnew = 0
 104       self.ngood = 0
 105       self.nbad = 0
 106
 107       for q in self.root.getElementsByTagName("question"):
 108          id = q.getAttribute ("id")
 109          c = q.getAttribute ("c")
 110          cs = q.getAttribute ("cs")
 111          ws = q.getAttribute ("ws")
 112          w = q.getAttribute ("w")
 113
 114          if ws > 0:
 115             rr = 1.*cs/ws
 116          else:
 117             rr = .5
 118
 119          if rr >= self.ratio:
 120             self.good.append (id)
 121             self.ngood += 1
 122          elif rr >= 0:
 123             self.bad.append (id)
 124             self.nbad += 1
 125          else:
 126             self.new.append (id)
 127             self.nnew += 1
 128
 129          answers = []
 130          for a in q.childNodes:
 131             if a.nodeType == Node.ELEMENT_NODE:
 132                code = a.getAttribute("answer_code")
 133                time = a.getAttribute("needed_time")
 134                when = a.getAttribute("datetime")
 135                answers.append ([code,time,when])
 136
 137          self.statistics.append ([id, c, cs, ws, w, answers])
 138
 139    def FindQuestion(self,qid):
 140       i = 0
 141       for s in self.statistics:
 142          if s[0]  == qid:
 143             return i
 144          i += 1
 145       return -1
 146
 147    def ThisQuestion(self,qid):
 148       q = self.FindQuestion (qid)
 149       if q >= 0:
 150          c = self.statistics[q][1]
 151          w = self.statistics[q][4]
 152       else:
 153          c=str(0)
 154          w=str(0)
 155       return [c,w]
 156
 157    def Timestamp(self):
 158       return str(datetime.datetime.today().isoformat()).split(".")[0]
 159
 160    def __init__(self,filename="DL-A-2007.stat.xml"):
 161       self.filename=filename
 162
 163       self.OpenFile()
 164       self.GetStatistics()