Add remote server

requirements.txt

@@ -1,2 +1,4 @@
 colorama
 gensim
+json
+requests

src/cemantix.py

@@ -1,36 +1,118 @@
 from gensim.models import KeyedVectors
 from colorama import Fore, Back, Style
 import argparse
+import requests
 import readline
 import random
+import json
 import time
 
+import numpy as np
+np.seterr(divide='ignore', invalid='ignore')
 
-def random_word(model, k=5, dist=100):
+class Server():
+    inverse_order = False
+    def __init__(self):
+        pass
+
+    def init_word(self):
+        pass
+
+    def get_rank(self, guess):
+        pass
+
+    def get_temp(self, guess):
+        pass
+
+    def _help(self):
+        raise NotImplementedError
+
+    def _reveal_word(self):
+        raise NotImplementedError
+
+class LocalServer(Server):
+    inverse_order = True
+    def __init__(self, word=None, file="models/selected_word2vec_model.bin"):
+        self.model = KeyedVectors.load_word2vec_format(
+            file,
+            binary=True,
+            unicode_errors="ignore"
+        )
+        self.word = word
+        self.nearest = []
+
+
+    def init_word(self, k=1, dist=100):
+        while (self.word is None or len(self.word) < 5
+               or '-' in self.word or '_' in self.word):
             base_words = [
-        model.index_to_key[random.randint(0, len(model))]
+                self.model.index_to_key[random.randint(0, len(self.model))]
                 for _ in range(k)
             ]
 
             complete_list = base_words.copy()
             for word in base_words:
-        complete_list += [i[0] for i in model.most_similar(word, topn=dist)]
+                complete_list += [i[0] for i in self.model.most_similar(word, topn=dist)]
 
-    rk_words = model.rank_by_centrality(complete_list)
+            rk_words = self.model.rank_by_centrality(complete_list)
-    return rk_words[random.randint(0,5)%len(rk_words)][1]
 
+            self.word = rk_words[random.randint(0,5)%len(rk_words)][1]
+            self.nearest = [word]+[i[0] for i in self.model.most_similar(self.word, topn=1000)]
 
-def cemantix(model, word=None):
+    def get_rank(self, guess):
-    while word is None or len(word) < 5 or '-' in word or '_' in word:
+        if guess not in self.nearest:
-        word = random_word(model, k=1, dist=0) # augment dist for a "smoother selection"
-
-    nearest = [word]+[i[0] for i in model.most_similar(word, topn=1000)]
-    guesses = [] # guess, temp, rank
-    def get_rank(guess):
-        if guess not in nearest:
             return None
-        return 1000 - nearest.index(guess)
+        return 1000 - self.nearest.index(guess)
 
+    def get_temp(self, guess):
+        return round(self.model.distance(self.word, guess)*100, 2)
+
+    def _help(self, rk):
+        return self.nearest[rk]
+
+    def _reveal_word(self):
+        return self.word
+
+class CemantixServer(Server):
+    def __init__(self):
+        self.words = {}
+        self.solvers = None
+        self.num = None
+
+    def _try(self, word):
+        url = 'https://cemantix.certitudes.org/score'
+        headers = {"Origin": "https://cemantix.certitudes.org"}
+        data = {'word': word}
+        # Need some additional cookies to be a valid client
+        r = requests.post(url, headers=headers, data=data)
+
+        assert r.ok
+
+        data = json.loads(r.text)
+        self.solvers = int(data["solvers"])
+        self.num = int(data["num"])
+        self.words[word] = {"score" : float(data["score"])}
+        if "percentile" in data:
+            self.words[word]["percentile"] = data["percentile"]
+
+
+    def get_rank(self, guess):
+        if guess not in self.words:
+            self._try(guess)
+
+        return self.words[guess].get("percentile", None)
+
+    def get_temp(self, guess):
+        if guess not in self.words:
+            self._try(guess)
+
+        return round(self.words[guess]["score"]*100, 2)
+
+
+def cemantix(server: Server):
+    server.init_word()
+
+    guesses = [] # guess, temp, rank
     def formatted_status(guesses, last=None):
         text = ""
         for w, temp, rank in guesses:
@@ -46,15 +128,19 @@ def cemantix(model, word=None):
         return text[:-1]
 
     def tried(word, guessed):
-        return word in [i[0] for i in guessed]
+        return (word in [i[0] for i in guessed])
 
     def interpret_command(cmd, guesses):
         match cmd:
             case "clear":
                 guesses = [g for g in guesses if g[1] <= 75.]
             case "help":
+                try:
                     best_rk = max([rk for _, _, rk in guesses if rk is not None]+[749])
-                print("Maybe try "+Back.YELLOW+Fore.BLACK+nearest[999-best_rk]+Style.RESET_ALL)
+                    print("Maybe try "+Back.YELLOW+Fore.BLACK+server._help(999-best_rk)
+                        +Style.RESET_ALL)
+                except NotImplementedError:
+                    print(Fore.RED+"No help available"+Style.RESET_ALL)
             case _:
                 print(Fore.RED+"Unknown command"+Style.RESET_ALL)
 
@@ -69,21 +155,24 @@ def cemantix(model, word=None):
                 guesses = interpret_command(guess[:-2], guesses)
                 continue
         except (EOFError, KeyboardInterrupt):
-            print("The word was "+Style.BRIGHT+word+Style.RESET_ALL)
+            try:
+                print("The word was "+Style.BRIGHT+server._reveal_word()+Style.RESET_ALL)
+            except NotImplementedError:
+                pass
             print("Goodbye!")
             return -1
         try:
-            dist = round(round(model.distance(word, guess), 4)*100, 2)
+            dist = server.get_temp(guess)
         except KeyError:
             print(Fore.RED+"Key not present"+Style.RESET_ALL)
             continue
         
         if not tried(guess, guesses):
-            guesses.append((guess, dist, get_rank(guess)))
+            guesses.append((guess, dist, server.get_rank(guess)))
-            guesses.sort(key=lambda x:-x[1])
+            guesses.sort(key=lambda x:-x[1] if server.inverse_order else x[1])
         print(chr(27) + "[2J")
         print(formatted_status(guesses, last=guess))
-        if guess == word:
+        if server.get_rank(guess) == 1000:
             time.sleep(1)
             print(Fore.GREEN+"Correct!"+Style.RESET_ALL+f" {len(guesses)} tries.")
             return len(guesses)
@@ -98,14 +187,15 @@ def main():
     parser.add_argument("-m", "--model", dest="model", action="store",
                         default="models/selected_word2vec_model.bin",
                         help="Specify model to use")
+    parser.add_argument("-r", "--remote", dest="remote", action="store_true",
+                        default=False,
+                        help="Use cemantix.certitudes.org instead of local model")
+
     args = parser.parse_args()
 
-    model = KeyedVectors.load_word2vec_format(
+    if args.remote:
-        args.model,
+        return cemantix(CemantixServer())
-        binary=True,
+    return cemantix(LocalServer(word=args.word, file=args.model))
-        unicode_errors="ignore"
-    )
-    cemantix(model, word=args.word)
 
 if __name__ == "__main__":
     main()