create app.py

app.py

@@ -0,0 +1,130 @@
+import os
+import random
+from PIL import Image
+from bs4 import BeautifulSoup
+
+def whisper_stt(audio):
+  print("inside whisper_stt")
+  print(f'audio is - {audio}')
+  print(f'audio is - {type(audio)}')
+  #Connect to the Space
+  client = Client("abidlabs/whisper", hf_token="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") 
+  print ("now here") 
+  text = client.predict(audio, api_name = '/predict')
+  print(f"text is - {text}")
+  return text
+
+def ttimg(prompt):
+  print("inside ttimg")
+  #Connect to the Space
+  client = Client("stabilityai/stable-diffusion")  
+  img_dir = client.predict(prompt,"blur",7, fn_index = 1)
+  print(f'image dir  - {img_dir}')
+  jpg_files = [f for f in os.listdir(img_dir) if f.endswith('.jpg')]
+  #if len(jpg_files) > 0:
+  #    chosen_file = random.choice(jpg_files)
+  file_path = os.path.join(img_dir, jpg_files[0])
+  print(f'file_path - {file_path}')
+  return file_path
+  #with open(file_path, 'rb') as f:
+  #    pil_img = Image.open(f)
+  #    return pil_img.show()
+
+def controlnet_img2img(img, cnet_prompt):
+  print("inside controlnet_img2img")
+  #Connect to the Space
+  client = Client("ysharma/ControlNetwithSlider")  
+  img_dir = client.predict(img, cnet_prompt, "", "blur",1,512,20,9,123,0,100,200,api_name = '/canny')
+  png_files = [os.path.join(img_dir, f) for f in os.listdir(img_dir) if f.endswith('.png')]
+  return png_files[0], png_files[1]
+
+def img2para(cnet_img2):
+  print("inside img2para")
+  #Connect to the Space
+  client = Client("Awiny/Image2Paragraph")  
+  txt = client.predict(cnet_img2, [], api_name="/predict")
+  print(f'img2para text is - {txt}')
+  soup = BeautifulSoup(txt, 'html.parser')
+  gpt4_section = soup.find_all('div', {'style': 'display: flex; flex-wrap: wrap;'})[1].find_all('p')[0].text
+  return gpt4_section
+
+
+# demo with different Gradio Clients
+demo = gr.Blocks()
+
+with demo:
+  with gr.Row():
+    audio_mic = gr.Audio(source="upload", type="filepath")
+    with gr.Column():
+      btn_whisper = gr.Button("Transcribe👇").style(full_width=True)
+      prompt = gr.Textbox(label="prompt for SD image generation", interactive=False) #, value='a black swan')
+  with gr.Row():
+    btn_sd = gr.Button("Generate Image👉").style(full_width=True)
+    sd_img = gr.Image(type='filepath') #value="/content/e321040c-b2eb-4550-8aaa-4cd416a7602d/tmpwyiwdr_1.jpg",
+  with gr.Row():
+    with gr.Column():
+      cnet_prompt = gr.Textbox(label="prompt for SD image generation")
+      btn_cnet = gr.Button("ControlNet Output👉").style(full_width=True)
+    cnet_img1 = gr.Image(type='filepath')
+    cnet_img2 = gr.Image(type='filepath')
+  with gr.Row():
+    btn_Img2para = gr.Button("Lets understand Image with Text👉").style(full_width=True)
+    paragraph = gr.Textbox(label="Image to Paragraph generation")
+    
+  btn_whisper.click(whisper_stt, audio_mic, prompt)
+  btn_sd.click(ttimg, prompt, sd_img)
+  btn_cnet.click(controlnet_img2img, [sd_img, cnet_prompt], [cnet_img1, cnet_img2])
+  btn_Img2para.click(img2para, cnet_img1, paragraph)
+  
+#demo.launch()
+
+# demo with different Gradio Clients Chained together 
+demo_chain_events = gr.Blocks()
+
+with demo_chain_events:
+  #with gr.Row():
+  audio_mic = gr.Audio(source="upload", type="filepath")
+  btn_whisper = gr.Button("Trigger the Chained Events🚀🚀").style(full_width=True)
+  prompt = gr.Textbox(label="Getting the Prompt for Stable Diffusion using Whisper Gradio Client", interactive=False,) #value='a black swan'
+  
+  with gr.Row():
+    with gr.Column():
+      html_sd = gr.HTML("<h1><center>Stable Diffusion Generated Image👇</center></h1>") #.style(full_width=True)
+      sd_img = gr.Image(type='filepath') #value="/content/e321040c-b2eb-4550-8aaa-4cd416a7602d/tmpwyiwdr_1.jpg",
+  #with gr.Row():
+    with gr.Column():
+      with gr.Row():
+        html_cnet = gr.HTML("<h1><center>ControlNet Output👇</center></h1>") #.style(full_width=True)
+        cnet_prompt = gr.Textbox(label="Prompt for ControlNet", value='a beautiful scenery')
+      with gr.Row():
+        cnet_img1 = gr.Image(type='filepath')
+        cnet_img2 = gr.Image(type='filepath')
+  with gr.Row():
+    with gr.Column():
+      html_Img2para = gr.HTML("<h1><center>Image to Paragraph output👇</center></h1>") #.style(full_width=True)
+      paragraph = gr.Textbox(label="Image to Paragraph generation")
+    
+  """
+  btn_whisper.click(whisper_stt, audio_mic, prompt)
+  btn_sd.click(ttimg, prompt, sd_img)
+  btn_cnet.click(controlnet_img2img, [sd_img, cnet_prompt], [cnet_img1, cnet_img2])
+  btn_Img2para.click(img2para, cnet_img1, paragraph)
+  """
+  btn_whisper.click(whisper_stt, audio_mic, prompt)\
+  .success(ttimg, prompt, sd_img)\
+  .success(controlnet_img2img, [sd_img, cnet_prompt], [cnet_img1, cnet_img2])\
+  .success(img2para, cnet_img1, paragraph)
+
+#demo_chain_events.launch()
+
+
+demo_joint = gr.Blocks()
+with demo_joint:
+  with gr.Row():
+    with gr.Box():
+      demo.render()
+    with gr.Box():
+      demo_chain_events.render()
+    
+
+demo_joint.launch(debug=True)