initial commit

rankwords.py

@@ -0,0 +1,168 @@
+#! /usr/bin/python
+
+import re
+
+letterscore = {
+   "s": 26,
+   "e": 25,
+   "a": 24,
+   "o": 23,
+   "r": 22,
+   "i": 21,
+   "l": 20,
+   "t": 19,
+   "n": 18,
+   "u": 17,
+   "d": 16,
+   "c": 15,
+   "y": 14,
+   "p": 13,
+   "m": 12,
+   "h": 11,
+   "g": 10,
+   "b": 9,
+   "k": 8,
+   "f": 7,
+   "w": 6,
+   "v": 5,
+   "z": 4,
+   "x": 3,
+   "j": 2,
+   "q": 1,
+   "\n": 0
+}
+
+def trimWordList(toRemove, words):
+   for word in toRemove:
+      try:
+         words.pop(word)
+      except(KeyError):
+         pass
+   
+def rankWord(word, words):
+   word=word.lower()
+   score=0
+   for letter in word:
+      score=score+letterscore[letter]
+
+   multiplier=countDistinct(word)
+   
+   words[word] = score*multiplier
+   return score*multiplier
+
+def rankLetters(words):
+   
+   lettercount = {
+      "\n": 0
+   }
+   for word in words:
+      for letter in word:
+         if letter in lettercount.keys():
+            lettercount[letter]=lettercount[letter]+1
+         else:
+            lettercount[letter]=1
+   sort_letters = sorted(lettercount.items(), key=lambda x: x[1])   
+   
+   for key in letterscore:
+      letterscore[key] = 0
+   
+   rank=0
+
+   for i in sort_letters:
+      letterscore[i[0]] = rank
+      rank+=1
+
+def matchKnown(mask, words):
+   toRemove = []
+   for word in words:
+      if not re.search(mask, word):
+         toRemove.append(word)
+   trimWordList(toRemove, words)
+
+def removeExcluded(mask, words):
+   toRemove = []
+   for word in words:
+      for letter in mask:
+         if re.search(letter, word):
+            toRemove.append(word)
+   trimWordList(toRemove, words)
+
+
+def matchIncluded(mask, words):
+   toRemove = []
+   for word in words:
+      for letter in mask:
+         if not re.search(letter, word):
+            toRemove.append(word)
+   trimWordList(toRemove, words)
+         
+
+   
+def removeMask(masks, words):
+   toRemove = []
+   for word in words:
+      index=0
+      for letter in word:
+         for excluded in masks[index]:
+            if letter == excluded:
+               toRemove.append(word)
+         index+=1
+         
+
+   trimWordList(toRemove, words)
+   
+
+
+def countDistinct(str):
+    s = set(str)
+    return len(s)
+
+def loadWords(file, words):
+   with open(file) as f:
+      for line in f:
+         words[line.rstrip()]=0;
+
+def parseMask(mask, masks):
+   index=0
+   for letter in mask:
+      if letter != ".":
+         masks[index].append(letter)
+         index+=1
+      else:
+         index+=1
+
+
+if __name__ == "__main__":
+   import sys
+   
+   words = {}
+
+   
+   loadWords(sys.argv[1], words)
+
+   masks=[]
+   maskList=[[]]
+   for index in range(0,len(next(iter(words)))-1):
+      new=[]
+      maskList.append(new)
+
+   matchKnown(sys.argv[2], words)
+   removeExcluded(sys.argv[4], words)
+   matchIncluded(sys.argv[3], words)
+
+   for index in range(5, len(sys.argv)):
+      masks.append(sys.argv[index])
+   for mask in masks:
+      parseMask(mask, maskList)
+
+   removeMask(maskList, words)
+
+   rankLetters(words)
+
+   for word in words:
+      words[word] = rankWord(word, words)
+
+    
+   for word in words:
+      print str(words[word]) + " " + word
+