Using trigram reference table to early stop searches, much faster computes

src/wordhunt.py

@@ -56,6 +56,47 @@ def wordSearch2(start_index, G, dictionary, words, frames, prefix="", visited=No
 
     return frames 
 
+# early stopping using trigram matching
+def wordSearch3(start_index, G, dictionary, trigram_dict ,words, frames, prefix="", visited=None):
+    if visited==None:
+        visited = []
+
+    start = G.get_node_val(start_index)
+    prefix += start
+    visited.append(start_index)
+
+    if dictionary.search(prefix) and not prefix in words:
+        words.append(prefix)
+
+    for n in G.get_neighbours(start_index):
+        letter = G.get_node_val(n)
+        if dictionary.is_prefix(prefix + letter) and not n in visited:
+
+            if len(prefix + letter) >= 3:
+                trigram = (prefix + letter)[-3:]
+                if not trigram_dict.search(trigram):
+                    continue
+
+            wordSearch3(start_index=n, 
+                       G=G, 
+                       dictionary=dictionary, 
+                       trigram_dict=trigram_dict,
+                       frames=frames, 
+                       prefix=prefix, 
+                       visited=visited.copy(),
+                       words=words)
+        else:
+            # only add meaningful frames
+            if not n in visited:
+                visited2 = visited.copy()
+                visited2.append(n)
+                frames.append((visited2, prefix + letter, words.copy())) 
+
+    return frames 
+
+
+
+
 def solveBoard(n, G, dictionary, words):
     for i in range(n * n):
         wordSearch(i, G, dictionary, words)
@@ -64,6 +105,10 @@ def solveBoard2(n, G, dictionary, words, frames):
     for i in range(n * n):
         wordSearch2(i, G, dictionary, words, frames)
 
+def solveBoard3(n, G, dictionary, trigram_dict ,words, frames):
+    for i in range(n * n):
+        wordSearch3(i, G, dictionary, trigram_dict, words, frames)
+
 def createDictionary(filepath, dictionary):
     f = open(filepath, 'r')
     lines = f.readlines()
@@ -75,6 +120,7 @@ def createDictionary(filepath, dictionary):
 board_letters = "mdacofhraueumnne"
 G = create_board(board_letters, 4)
 dictionary = createDictionary('vocabulary/scrabble_wordbank_2019.txt', Trie())
+trigram_dict = createDictionary('vocabulary/trigram_table.txt', Trie())
 
 # prints board
 print_board(4, G)
@@ -99,6 +145,6 @@ def createDictionary(filepath, dictionary):
 font_scale = 2
 # frames_data = [[0, 1, 4, 9], [1, 2, 5, 9]] # JUST FOR TESTING
 
-solveBoard2(4, G, dictionary, words, frames)
+solveBoard3(4, G, dictionary, trigram_dict, words, frames)
 rendered_frames = draw_frames(frames, img_dims, lett_per_row, thickness, font_face, font_scale, board_letters, font_face)
 create_mp4(rendered_frames, img_dims)