app.py

import gradio as gr
from PIL import Image

import requests
import base64

import os
import openai
openai.api_type = "azure"
openai.api_base = "https://kidgpt.openai.azure.com/"
openai.api_version = "2023-03-15-preview"
#openai.api_key = os.getenv("OPENAI_API_KEY")
openai.api_key = "c10faac20a3f4008b63a573ab3afc30b"


def get_completion(text, model="Demo-gpt-35-turbo"): #  this is the latest 3.5 turbo 16k model

    prompt = f"""
    对以下```符号内是通过图片ocr识别的资料，请帮我完成以下任务：
    1. 分析博主背景信息
    2. 分析这位博主潜在受众
    3. 帮这位博主设计5个未来值得做内容标题
    4. 如果作为粉丝想与这位博主互动，给出5个博主会关心很可能会回答的问题（用平等的语气）

    输出格式如下：
    #background#
    <分析博主背景信息>

    #audience#
    <分析博主潜在受众>

    #topic#
    <设计5个未来值得做的内容标题>

    #questions#
    <设计5个能与博主关心的提问>

    ```{text}```
    \n\n\n
    """

    messages = [{"role":"system","content":"You are an AI assistant that helps people find information."},
                {"role":"user","content":prompt}]
    response = openai.ChatCompletion.create(
        engine=model,
        messages=messages,
        temperature=0, # this is the degree of randomness of the model's output
        max_tokens=1500,
        #top_p=0.95,
        #frequency_penalty=0,
        #presence_penalty=0,
        stop="\n\n\n"   #避免重复
    )
    return response.choices[0].message["content"]




def recognize_text(input_image1, input_image2):

    ocr = PaddleOCR(use_angle_cls=True, lang='ch') # need to run only once to download and load model into memory

    # 将上传的图像转换为PIL图像对象
    #image = Image.open(input_image.name)

    if input_image1 is None:
      text = "请上传图片"
      response = "请上传图片"
      return text, response

    # 使用Tesseract进行OCR识别
    result = ocr.ocr(input_image1, cls=True)
    text = ""
    text1 = ""
    for idx in range(len(result)):
        res = result[idx]
        for line in res:
            txts = line[1][0]
            text = text + txts + '\n'

    text1 = text
    print(text1)

    text2 = ""
    if input_image2 is not None:
      result = ocr.ocr(input_image1, cls=True)
      text = ""
      for idx in range(len(result)):
          res = result[idx]
          for line in res:
              txts = line[1][0]
              text = text + txts + '\n'

      text2 = text
      print(text2)

    text = text1 + text2

    response = get_completion(text)
    #response = get_completion(result)
    print(response)

    return text, response



# # 创建输入组件和输出组件
# input_images = gr.inputs.File(file_count="multiple")
input_image1 = gr.inputs.Image(type="filepath", label="上传图片1")
print("type：", type(input_image1))

input_image2 = gr.inputs.Image(type="filepath", label="上传图片2")

output_text1 = gr.outputs.Textbox(label="识别的文字")
output_text2 = gr.outputs.Textbox(label="GhatGPT文字")



# # 创建Gradio接口
iface = gr.Interface(fn=recognize_text, inputs=[input_image1, input_image2], outputs=[output_text1, output_text2], title="图片文字识别", allow_flagging=False)
iface.launch(share=True, debug = True)

#print(pytesseract.image_to_string(image, lang='chi_sim', config='--psm 6'))