Update main_dl.py

main_dl.py

@@ -17,7 +17,6 @@ import csv
 import tiktoken
 from sklearn.preprocessing import LabelEncoder
 from tensorflow import keras
-# import keras
 from keras_nlp.layers import TransformerEncoder
 from tensorflow.keras import layers
 from tensorflow.keras.preprocessing.sequence import pad_sequences
@@ -46,6 +45,8 @@ def load_vocab(file_path):
 def decode_sequence_rnn(input_sentence, src, tgt):
     global translation_model
 
+    if translation_model not in globals():
+        load_rnn()
     vocab_size = 15000
     sequence_length = 50
 
@@ -180,9 +181,11 @@ class PositionalEmbedding(layers.Layer):
         })
         return config
     
-def decode_sequence_tranf(input_sentence, src, tgt):
+def decode_sequence_transf(input_sentence, src, tgt):
     global translation_model
 
+    if translation_model not in globals():
+        load_transformer()
     vocab_size = 15000
     sequence_length = 30
 
@@ -221,7 +224,7 @@ def decode_sequence_tranf(input_sentence, src, tgt):
 
 # ==== End Transforformer section ====
 
-def load_all_data():
+def load_rnn():
    
     merge = Merge( dataPath+"/rnn_en-fr_split",  dataPath, "seq2seq_rnn-model-en-fr.h5").merge(cleanup=False)
     merge = Merge( dataPath+"/rnn_fr-en_split",  dataPath, "seq2seq_rnn-model-fr-en.h5").merge(cleanup=False)
@@ -229,7 +232,9 @@ def load_all_data():
     rnn_fr_en = keras.models.load_model(dataPath+"/seq2seq_rnn-model-fr-en.h5") # , compile=False)
     rnn_en_fr.compile(optimizer="rmsprop", loss="sparse_categorical_crossentropy", metrics=["accuracy"])
     rnn_fr_en.compile(optimizer="rmsprop", loss="sparse_categorical_crossentropy", metrics=["accuracy"])
-    
+    return rnn_en_fr, rnn_fr_en
+
+def load_transformer():
     custom_objects = {"TransformerDecoder": TransformerDecoder, "PositionalEmbedding": PositionalEmbedding}
     with keras.saving.custom_object_scope(custom_objects):
         transformer_en_fr = keras.models.load_model( "data/transformer-model-en-fr.h5")
@@ -239,9 +244,10 @@ def load_all_data():
     transformer_en_fr.compile(optimizer="rmsprop", loss="sparse_categorical_crossentropy", metrics=["accuracy"])
     transformer_fr_en.compile(optimizer="rmsprop", loss="sparse_categorical_crossentropy", metrics=["accuracy"])
 
-    return rnn_en_fr, rnn_fr_en, transformer_en_fr, transformer_fr_en
+    return transformer_en_fr, transformer_fr_en
 
-rnn_en_fr, rnn_fr_en, transformer_en_fr, transformer_fr_en = load_all_data() 
+rnn_en_fr, rnn_fr_en = load_rnn() 
+transformer_en_fr, transformer_fr_en = load_transformer()
       
 #  ==== Language identifier ====
 
@@ -277,10 +283,13 @@ def init_dl_identifier():
     else: print("dl_model vide")
     return
 
+init_dl_identifier()
+
 def lang_id_dl(sentences):
     global dl_model, label_encoder, lan_to_language
     
-    print("sentences:",sentences)
+    if dl_model not in globals():
+        init_dl_identifier()
     if "str" in str(type(sentences)): predictions = dl_model.predict(encode_text([sentences]))
     else:  predictions = dl_model.predict(encode_text(sentences))
     # Décodage des prédictions en langues
@@ -293,7 +302,8 @@ def lang_id_dl(sentences):
 
 @api.get('/', name="Vérification que l'API fonctionne")
 def check_api():
-    load_all_data()
+    load_rnn()
+    load_transformer()
     init_dl_identifier()
     return {'message': "L'API fonctionne"}
 
@@ -316,10 +326,10 @@ async def trad_transformer(lang_tgt:str,
     
     if (lang_tgt=='en'):
         translation_model = transformer_fr_en
-        return decode_sequence_tranf(texte, "fr", "en")
+        return decode_sequence_transf(texte, "fr", "en")
     else:
         translation_model = transformer_en_fr
-        return decode_sequence_tranf(texte, "en", "fr")
+        return decode_sequence_transf(texte, "en", "fr")
 
 @api.get('/small_vocab/plot_model', name="Affiche le modèle")
 def affiche_modele(lang_tgt:str,
@@ -345,5 +355,5 @@ def affiche_modele(lang_tgt:str,
             return Response(content=image_data, media_type="image/png")
 
 @api.get('/lang_id_dl', name="Id de langue par DL")
-def language_id_dl(sentence:List[str] = Query(..., min_length=1)):
+async def language_id_dl(sentence:List[str] = Query(..., min_length=1)):
     return lang_id_dl(sentence)