Module 10: Search the Library, Fast (Python code)

a.py

@@ -7,67 +7,49 @@ class Words:
     def __init__(self):
         self.word = []
 
-num_buckets = 1001
-
-def bucket(s):
-    assert(len(s) != 0)
-    i = 0
-    for c in s:
-        i = (i * 217) + ord(c)
-    i = abs(i)
-    b = i % num_buckets
-    assert(b >= 0 and b < num_buckets)
-    return b
-
-
 class Library:
 
     def __init__(self):
         # master vector of word
-        self.words = Words()
-        self.counts = {}
-        self.shelves = {}
-        for i in range(num_buckets):
-            self.shelves[i] = []
+        self.words_ = Words()
+        self.counts_ = {}
+        # hash table
+        self.word_map_ = {}
 
     def IsWord(self, s):
-        for Words.word in self.shelves[bucket(s)]:
-            if s == Words.word:
-                return True
-        return False
-        
+        return s in self.word_map_
 
     def ComputeStats(self):
         # assert self.counts == {}
         for i in range(18):
-            self.counts[i] = []
-        for s in self.words:
-            _len = len(s)
+            self.counts_[i] = []
+        for s in self.words_:
+            _len = len(s.word)
             if _len <= 18:
-                self.counts[_len-1].append(_len)
+                self.counts_[_len-1].append(_len)
 
     def PrintStats(self):
         print("Here are the counts of each word length")
-        for k,v in self.counts.items():
+        for k,v in self.counts_.items():
             # print(v)
             if k != 0:
                 print(f"[{k}] {len(v)}")
 
     def GetWord(self, i):
-        assert (i >= 0 and i < len(self.words))
-        return self.words[i]
+        assert (i >= 0 and i < len(self.words_))
+        return self.words_[i]
 
     def ReadFromFile(self, filename):
         with open(filename, 'r') as f:
             for line in f:
                 line = ToUpper(line.rstrip())
-                self.words.word.append(line)
-                self.shelves[bucket(line)].append(line)
-        print(f"Read {len(self.words.word)} words from file '{filename}'")
+                self.words_.word.append(line)
+                self.word_map_[line] = line
+        print(f"Read {len(self.words_.word)} words from file '{filename}'")
 
     def DebugBuckets(self):
-        for i in range(len(self.shelves)):
-            print(f"[{i}] {len(self.shelves[i])}")
+        for i in range(len(self.word_map_.keys())):
+            print(f"[{i}]")# {self.word_map_[i]}")
 
 
 class Grid:
@@ -112,6 +94,7 @@ if __name__ == "__main__":
     print(lib.IsWord("DOG"))
     print(lib.IsWord("CAT"))
     print(lib.IsWord("THANKS"))
+    print(lib.IsWord("TFAFAS"))
     lib.DebugBuckets()
 
     # grid = Grid("MY GRID")