app.py

#import gradio as gr
#gr.load("models/tclopess/bart_samsum").launch()
import gradio as gr
import nltk
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

# Carregue o modelo e o tokenizer
checkpoint = "tclopess/bart_samsum"
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
model = AutoModelForSeq2SeqLM.from_pretrained(checkpoint)

# Função para fragmentar o texto
def fragment_text(text, tokenizer):
  sentences = nltk.tokenize.sent_tokenize(text)
  max_len = tokenizer.max_len_single_sentence

  chunks = []
  chunk = ""
  count = -1

  for sentence in sentences:
    count += 1
    combined_length = len(tokenizer.tokenize(sentence)) + len(chunk)

    if combined_length <= max_len:
      chunk += sentence + " "
    else:
      chunks.append(chunk.strip())
      chunk = sentence + " "

  if chunk != "":
    chunks.append(chunk.strip())

  return chunks

def generate_summaries(text):
  chunks = fragment_text(text, tokenizer)
  summaries = []
  for chunk in chunks:
    input = tokenizer(chunk, return_tensors='pt')
    output = model.generate(**input)
    summary = tokenizer.decode(*output, skip_special_tokens=True)
    summaries.append(summary)
  return summaries

# Função para exibir o resumo final
def display_summary(summaries):
  summary = " ".join(summaries)
  gr.text("Resumo final:", summary)

# Função `input_fn()` para retornar o campo de input
def input_fn():
  # Crie um campo de input do tipo `text`
  input_text = gr.inputs.Textbox(label="Insira ou cole o texto aqui:")

  # Retorne o campo de input
  return input_text

# Altere a linha `gr.load()` para usar a função `input_fn()`
gr.load("models/tclopess/bart_samsum").launch(input_fn=input_fn)