Update app.py

app.py

@@ -3,17 +3,24 @@ import numpy as np
 import pandas as pd
 import csv
 import os
+import threading
 from itertools import combinations
 from datetime import datetime
 import fasttext
 import fasttext.util
 import enchant
+from huggingface_hub import HfApi, hf_hub_download
+from huggingface_hub.utils import EntryNotFoundError
 
 # ── Config ────────────────────────────────────────────────────────────────────
 SCORES_FILE  = "scores.csv"
 MODEL_PATH   = "cc.fr.300.bin"
 N_WORDS      = 10
 SCORE_WORDS  = 7
+DATASET_REPO = "MendoToto/dat-score"
+HF_TOKEN     = os.environ.get("HF_TOKEN", "")
+hf_api       = HfApi(token=HF_TOKEN)
+csv_lock     = threading.Lock()
 
 print("Téléchargement du modèle FastText français...")
 fasttext.util.download_model('fr', if_exists='ignore')
@@ -26,15 +33,47 @@ dic = enchant.Dict("fr_FR")
 def is_valid_word(word: str) -> bool:
     return dic.check(word)
 
-# ── Fichier de scores ─────────────────────────────────────────────────────────
+# ── Persistance via HuggingFace Dataset ──────────────────────────────────────
 def init_scores_file():
-    if not os.path.exists(SCORES_FILE):
+    """Télécharge le CSV depuis le dataset HF, ou crée un fichier vide."""
+    try:
+        path = hf_hub_download(
+            repo_id=DATASET_REPO,
+            filename="scores.csv",
+            repo_type="dataset",
+            token=HF_TOKEN
+        )
+        import shutil
+        shutil.copy(path, SCORES_FILE)
+        print("CSV chargé depuis le dataset HF.")
+    except EntryNotFoundError:
         with open(SCORES_FILE, "w", newline="") as f:
-            csv.writer(f).writerow(["score", "timestamp"])
+            csv.writer(f).writerow(["score", "timestamp", "mot1", "mot2", "mot3", "mot4", "mot5", "mot6", "mot7", "mot8", "mot9", "mot10"])
+        print("Nouveau CSV créé.")
+    except Exception as e:
+        print(f"Impossible de charger le CSV depuis HF : {e}")
+        if not os.path.exists(SCORES_FILE):
+            with open(SCORES_FILE, "w", newline="") as f:
+                csv.writer(f).writerow(["score", "timestamp"])
+
+def push_scores_to_hf():
+    """Upload le CSV local vers le dataset HF."""
+    try:
+        hf_api.upload_file(
+            path_or_fileobj=SCORES_FILE,
+            path_in_repo="scores.csv",
+            repo_id=DATASET_REPO,
+            repo_type="dataset",
+            token=HF_TOKEN
+        )
+    except Exception as e:
+        print(f"Erreur upload HF : {e}")
 
-def save_score(score: float):
-    with open(SCORES_FILE, "a", newline="") as f:
-        csv.writer(f).writerow([round(score, 1), datetime.now().strftime("%H:%M")])
+def save_score(score: float, words: list):
+    with csv_lock:
+        with open(SCORES_FILE, "a", newline="") as f:
+            csv.writer(f).writerow([round(score, 1), datetime.now().strftime("%Y-%m-%d %H:%M")] + words)
+    push_scores_to_hf()
 
 def load_stats() -> str:
     if not os.path.exists(SCORES_FILE):
@@ -195,7 +234,7 @@ def run_dat(w1, w2, w3, w4, w5, w6, w7, w8, w9, w10):
     far, close = top_pairs(emb, words)
     html_matrix = build_html_matrix(words, dist_mat)
 
-    save_score(score)
+    save_score(score, words)
     stats = load_stats()
 
     result = f"""## Votre score DAT : **{score:.1f} / 100**
@@ -271,7 +310,7 @@ with gr.Blocks(title="DAT — Créativité Divergente", css=css,
         "---\n"
         "*Basé sur [Olson et al. (2021), PNAS](https://doi.org/10.1073/pnas.2022340118) · "
         "Modèle : FastText fr (Meta/Wikipedia) · "
-        "Un outil [Fouloscopie](https://www.youtube.com/@Fouloscopie)*"
+        "Un outil [Fouloscopie](https://www.mehdimoussaid.com/)*"
     )
 
 demo.launch()