Module 9: Search the Library (Python code only)

a.py

@@ -1,9 +1,41 @@
+def ToUpper(s):
+    return s.upper()
+
+
+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 = []
+        self.words = Words()
         self.counts = {}
+        self.shelves = {}
+        for i in range(num_buckets):
+            self.shelves[i] = []
+
+    def IsWord(self, s):
+        for Words.word in self.shelves[bucket(s)]:
+            if s == Words.word:
+                return True
+        return False
+        
 
     def ComputeStats(self):
         # assert self.counts == {}
@@ -28,8 +60,14 @@ class Library:
     def ReadFromFile(self, filename):
         with open(filename, 'r') as f:
             for line in f:
-                self.words.append(line)
-        print(f"Read {len(self.words)} words from file '{filename}'")
+                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}'")
+
+    def DebugBuckets(self):
+        for i in range(len(self.shelves)):
+            print(f"[{i}] {len(self.shelves[i])}")
 
 
 class Grid:
@@ -68,8 +106,13 @@ class Grid:
 if __name__ == "__main__":
     lib = Library()
     lib.ReadFromFile("top_12000.txt")
-    lib.ComputeStats()
-    lib.PrintStats()
+    # lib.ComputeStats()
+    # lib.PrintStats()
+
+    print(lib.IsWord("DOG"))
+    print(lib.IsWord("CAT"))
+    print(lib.IsWord("THANKS"))
+    lib.DebugBuckets()
 
     # grid = Grid("MY GRID")
     # grid.LoadFromFile("test")