Initial commit

markov.py

@@ -0,0 +1,112 @@
+#!/usr/bin/python
+# -*- coding: utf-8 -*-
+# \file markov.py
+# \brief TODO
+# \author Florent Guiotte <florent.guiotte@gmail.com>
+# \version 0.2
+# \date 04 sept. 2017
+#
+# TODO details
+
+import sys
+import random
+import pickle
+import getopt
+
+def main():
+    # parse command line options
+    opts, args = getopt.getopt(sys.argv[1:], "hf:n:o:l:", ["help","create-dic"])
+    create = False
+    loadpk = "out.pickle"
+    output = "out.pickle"
+    infile = "markov.log"
+    nb = 1
+    for o, a in opts:
+        if '--create-dic' == o:
+            create = True
+        elif o in ('--help', '-h'):
+            print(__doc__)
+            sys.exit(0)
+        elif o == "-o":
+            output = a
+        elif o == "-f":
+            infile = a
+        elif o == "-n":
+            nb = int(a)
+        elif o == "-l":
+            loadpk = a
+
+    if create:
+        megadic = {}
+        buildDic(megadic, infile)
+        #cslist = computeProbCSList(dic)
+        #print(randomPick(cslist))
+        megacslist = megaComputeProbCSList(megadic)
+        #print(megacslist)
+        pickle.dump(megacslist, open(output, 'wb'))
+    else:
+        try:
+            megacslist = pickle.load(open(loadpk, 'rb'))
+        except:
+            print("No pickle file provided")
+            sys.exit(2)
+
+        #print(megacslist)
+        for i in range(nb):
+            print(megaRandomPic(megacslist))
+
+def buildDic(dic, logfile):
+    log = open(logfile, "r")
+    for line in log:
+        #print("L: " + line)
+        pre = ("__firstword__", "__firstword__")
+        for word in line.split():
+            #print("W: " + word)
+            addInDic(dic, word, pre)
+            pre = (pre[1], word)
+        addInDic(dic, "__eol__", pre)
+
+def addInDic(dic, w, pre):
+    if not pre in dic:
+        dic[pre] = {}
+    if not w in dic[pre]:
+        dic[pre][w] = 1
+    else:
+        dic[pre][w] = dic[pre][w] + 1
+
+def megaComputeProbCSList(megadic):
+    megacsdic = {}
+    for dic in megadic.items():
+        cslist = computeProbCSList(dic[1])
+        megacsdic[dic[0]] = cslist
+    return megacsdic
+
+def computeProbCSList(dic):
+    tsum = sum(dic.values())
+    sortedDic = sorted(dic.items(), key=lambda x:x[1])
+    sortedDicProb = []
+    dsum = 0
+    for t in sortedDic:
+        dsum = t[1]/tsum + dsum
+        sortedDicProb.append((t[0], dsum))
+    return sortedDicProb
+
+def megaRandomPic(megacsdic):
+    index = ("__firstword__", "__firstword__")
+    currentw = randomPick(megacsdic[index])
+    string = ""
+    while currentw != "__eol__":
+        string = string + currentw + " "
+        index = (index[1], currentw)
+        currentw = randomPick(megacsdic[index])
+    return string
+
+def randomPick(CSlist):
+    r = random.random()
+    for t in CSlist:
+        if r < t[1]:
+            return t[0]
+    return None
+
+if __name__ == "__main__":
+    main()