Update app.py

app.py

@@ -29,18 +29,40 @@ def predict(payload):
     
     recorte_general = ""
 
-    for chunk in cortar_en_bloques(payload, 90):
-      if model.predict([chunk]).item() == 1:
-        recorte_general += chunk + " "
+    # Crear chunks
+    chunks = cortar_en_bloques(ocr_text, 150)
+    first = -1
+    margin = int(len(chunks) * 0.25)
+    chunks_removable = chunks[:margin] + chunks[-margin:]
+    
+    for i in range(len(chunks)):
+        print('Recortando -', round((i/len(chunks))*100), '%')
+        if chunks[i] not in chunks_removable or trim_model.predict([chunks[i]]).item() == 1:
+          if first == -1:
+            first = i
+          recorte_general += chunks[i] + " "
+    
+    if first > 0:
+    recorte_general = chunks[first-1] + recorte_general
+    print(100, '%')
 
     recorte_final = ""
 
-    for chunk in cortar_en_bloques(recorte_general, 84):
-      if model.predict([chunk]).item() == 1:
-        recorte_final += chunk + " "
-          
-    # results = model.predict_proba([payload])[0]
-    # return "\n".join([f"Label: {id2label[i]},      Score: {round(result.item()*100, 2)} %" for i, result in enumerate(results)])
+    # Definir tamñano de fragmentos de texto
+    # text_splitter2 = RecursiveCharacterTextSplitter(chunk_size=100, chunk_overlap=0, length_function=len)
+    # Crear chunks
+    chunks2 = cortar_en_bloques(recorte_general, 80)
+    margin_s = int(len(chunks2) * 0.1)
+    margin_e = int(len(chunks2) * 0.1)
+    # if margin_s > 1:
+    chunks_removable2 = chunks2[:margin_s] + chunks2[-margin_e:]
+    # chunks_removable2 = chunks2[-margin_e:]
+    
+    for i in range(len(chunks2)):
+        print('Recortando -', round((i/len(chunks2))*100), '%')
+        if chunks2[i] not in chunks_removable2 or trim_model.predict([chunks2[i]]).item() == 1:
+          recorte_final += chunks2[i] + " "
+    print(100, '%')
     
     return recorte_final