from transformers import AutoModelForCausalLM, AutoTokenizer ,T5ForConditionalGeneration ,T5Tokenizer
import re
import torch
torch.set_default_tensor_type(torch.cuda.FloatTensor)
import  os
import io
import warnings
from PIL import Image
from stability_sdk import client
import stability_sdk.interfaces.gooseai.generation.generation_pb2 as generation
import gradio as gr


def generate_post(model,tokenizer,company_name , description , example1 ,example2 ,example3): 
    prompt =  f""" {company_name} {description}, {example1}. 
                  {company_name} {description}, {example2}.
                  {company_name} {description}, {example3}.
                  {company_name} {description}, """

    input_ids = tokenizer(prompt, return_tensors="pt").to(0)
    sample = model.generate(**input_ids, top_k=0, temperature=0.7, do_sample = True , max_new_tokens = 70, repetition_penalty= 5.4)
    outputs = tokenizer.decode(sample[0])
    res = outputs.split(f""" {company_name} {description}, {example1}. 
                  {company_name} {description}, {example2}.
                  {company_name} {description}, {example3}.
                  {company_name} {description}, """)[1]

    res = re.sub('[#]\w+' , " ", res)
    res = re.sub('@[^\s]\w+',' ', res)
    res = re.sub(r'http\S+', ' ', res)
    res = res.replace("\n" ," ")
    res = re.sub(' +', ' ',res)
    return  res

def generate_caption(model , text_body ,tokenizer  ,max_length):
  test_sent = 'generate: ' + text_body
  input = tokenizer.encode(test_sent ,  return_tensors="pt")#.to('cuda')
  outs = model.generate(input ,
                        max_length = max_length,
                        do_sample = True ,
                        temperature = 0.7,
                        min_length = 8,
                        repetition_penalty = 5.4,
                        max_time = 12,
                        top_p = 1.0,
                        top_k = 50)
  
  
  sent = tokenizer.decode(outs[0], skip_special_tokens=True,clean_up_tokenization_spaces=True)
  return sent 



def demo_smg(company_name ,description , example1 , example2 , example3):
    
  access_token = "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"
  model_cp= T5ForConditionalGeneration.from_pretrained("Abdelmageed95/caption_model" , use_auth_token = access_token )
  tokenizer = T5Tokenizer.from_pretrained('t5-base')

  model_bm = AutoModelForCausalLM.from_pretrained("bigscience/bloom-560m" , use_auth_token = access_token)
  tokenizer_bm = AutoTokenizer.from_pretrained("bigscience/bloom-560m")
  
  res = generate_post( model_bm , tokenizer_bm, company_name , description , example1 , example2 , example3)
  generated_caption = generate_caption( model_cp , 
                  res,
                  tokenizer , 
                  30)

  os.environ['STABILITY_HOST'] = "grpc.stability.ai:443"
  os.environ['STABILITY_KEY'] = "sk-t4x1wv6WFgTANF7O1TkWDJZzxXxQZeU6X7oybl6rdCOOiHIk"
  stability_api = client.StabilityInference(
     key=os.environ['STABILITY_KEY'], 
     verbose=True)
  generated_caption = generated_caption + ", intricate, highly detailed, smooth , sharp focus, 8k"

  answers = stability_api.generate( prompt=  generated_caption ,
    #seed=34567, 
    steps= 70 )
  

  for resp in answers:
      for artifact in resp.artifacts:
          if artifact.finish_reason == generation.FILTER:
              warnings.warn(
                  "Your request activated the API's safety filters and could not be processed."
                  "Please modify the prompt and try again.")
          if artifact.type == generation.ARTIFACT_IMAGE:
              img = Image.open(io.BytesIO(artifact.binary))
              

  return res, generated_caption  ,img


company_name = "ADES Group"

description = "delivers full-scale petroleum services; from onshore and offshore drilling to full oil & gas projects and services, with emphasis on the HSE culture while maintaining excellence in operation targets."

example1 = """Throwback to ADM 680 Team during their Cyber-chair controls Course in August,
             Our development strategy at ADES does not only focus on enriching the technical expertise of our teams in their specialization in Jack- up rigs, 
             but also in providing access to latest operational models"""

example2 = """With complexity of oil & gas equipment and the seriousness of failure and its consequences confronting our people, 
            it has become a necessity to equip our Asset Management Team with leading methodologies and techniques that enable them to think and act proactively"""
  
example3 = """ Part of our people development strategy is providing our senior leadership with the latest industry technologies 
            and world class practices and standards"""
# txt , generated_caption , im = demo_smg( company_name, description , example1 , example2 , example3)

# print(txt)
# print(generated_caption)

demo = gr.Interface(
    fn= demo_smg,
    inputs=["text","text" , "text" ,"text" ,"text"],
    outputs=["text", "text", "image" ]
)
demo.launch()