Update app.py

counting image gens, logging, stats

app.py

@@ -49,6 +49,7 @@ def genUsageStats(do_reset=False):
     ttotal4mini_out = 0
     totalAudio = 0
     totalSpeech = 0
+    totalImages = 0
     for user in unames:
         tokens4o_in = 0
         tokens4o_out = 0
@@ -128,8 +129,29 @@ def genUsageStats(do_reset=False):
                     sleep(3)
             if not accessOk:
                 return f'File access failed reading speech stats for user: {user}'
-        result.append([user, f'{tokens4mini_in}/{tokens4mini_out}', f'{tokens4o_in}/{tokens4o_out}', f'audio:{userAudio}',f'speech:{userSpeech}'])
-    result.append(['totals', f'{ttotal4mini_in}/{ttotal4mini_out}', f'{ttotal4o_in}/{ttotal4o_out}', f'audio:{totalAudio}',f'speech:{totalSpeech}'])
+        user_images = 0
+        fp = image_count_path(user)
+        if os.path.exists(fp):
+            accessOk = False
+            for i in range(3):
+                try:
+                    with open(fp) as f:
+                        dataList = f.readlines()
+                    if do_reset:
+                        os.remove(fp)
+                    else:
+                        for line in dataList:
+                            cnt = line.strip()
+                            user_images += int(cnt)
+                        totalImages += int(user_images)
+                    accessOk = True
+                    break
+                except:
+                    sleep(3)
+            if not accessOk:
+                return f'File access failed reading image gen stats for user: {user}'
+        result.append([user, f'{tokens4mini_in}/{tokens4mini_out}', f'{tokens4o_in}/{tokens4o_out}', f'audio:{userAudio}',f'speech:{userSpeech}', f'images:{user_images}'])
+    result.append(['totals', f'{ttotal4mini_in}/{ttotal4mini_out}', f'{ttotal4o_in}/{ttotal4o_out}', f'audio:{totalAudio}',f'speech:{totalSpeech}', f'images:{totalImages}'])
     return result       
 
 def new_conversation(user):
@@ -204,6 +226,10 @@ def new_func(user):
     dataFile = dataDir + user + '_log.txt'
     return dataFile
 
+def image_count_path(user):
+    fpath = dataDir + user + '_image_count.txt'
+    return fpath
+
 def transcribe(user, pwd, fpath):
     user = user.lower().strip()
     pwd = pwd.lower().strip()
@@ -298,7 +324,7 @@ def make_image(prompt, user, pwd):
     fpath = None
     if user in unames and pwd == pwdList[unames.index(user)]:
         if len(prompt.strip()) == 0:
-            return [None, 'You must provide a prompt']
+            return [None, 'You must provide a prompt describing image you desire']
         try:
             response = client.images.generate(model='dall-e-2', prompt=prompt,size='512x512',
                quality='standard', response_format='b64_json')
@@ -306,6 +332,8 @@ def make_image(prompt, user, pwd):
             image = Image.open(BytesIO(base64.b64decode(image_data)))
             fpath = dataDir + user + '.png'
             image.save(fpath)
+            with open(image_count_path(user), 'at') as fp:
+                fp.write('1\n')
             msg = 'Image created!'
         except:
             return [None, msg]
@@ -314,6 +342,31 @@ def make_image(prompt, user, pwd):
         return [None, msg]
     return [fpath, msg]
 
+def show_help():
+    return '''
+    1.  Login with user name and password (not case-sensitive)
+    2.  Type prompts (questions, instructions) into prompt window (OR) you can speak prompts by
+        tapping the audio "Record" button, saying your prompt, then tapping the "Stop" button.
+        Your prompt will appear in the Prompt window, and you can edit it there if needed.
+    3.  Chat:
+          1.1 tap the "Submit Prompt/Question" button.  The response will appear in the Dialog window.
+          1.2 To speak the response, tap the "Speak Dialog" button.
+          1.3 Enter follow-up questions in the Prompt window either by typing or speaking. Tap the voice
+              entry "Reset Voice Entry" button to enable additional voice entry. Then tap "Submit Prompt/Question".
+          1.4 If topic changes or when done chatting, tap the "Restart Conversation" button.
+    4.  Make Image:
+          1.1 Enter description of desired image in prompt window via either typing or voice entry
+          1.2 Tap the "Make Image" button.  This can take a few seconds.
+          1.3 There is a download button on the image display if your system supports file downloads.
+          1.4 When done viewing image, tap the "Restart Conversation" button
+
+    Hints:
+        1. Better chat and image results are obtained by including detailed descriptions and instructions
+            in the prompt.
+        2. Always tap "Restart Conversation" before requesting an image or changing chat topics.
+        3. Audio input and output functions depend on the hardware capability of your device.'''
+
+
 
 with gr.Blocks() as demo:
     history = gr.State([])
@@ -419,15 +472,13 @@ with gr.Blocks() as demo:
 
 
     gr.Markdown('# GPT Chat')
-    gr.Markdown('Enter user name & password then enter prompt and click submit button.  Restart conversation if topic changes.  ' +
-      'You can enter prompts by voice.  Tap "Record", speak, then tap "Stop".  ' +
-      'Tap "Reset Voice Entry" to enter more voice. Tap "Speak Dialog" to hear dialog.  ' + 
-               'Note: first voice response may take a longer time.')
+    gr.Markdown('Enter user name & password.  Tap "Help & Hints" button for more instructions.')
     with gr.Row():
         user_window = gr.Textbox(label = "User Name")
         user_window.blur(fn=update_user, inputs=user_window, outputs=[user, user_window])
         pwd_window = gr.Textbox(label = "Password")
         pwd_window.blur(updatePassword, inputs = pwd_window, outputs = [password, pwd_window])
+        help_button = gr.Button(value='Help & Hints')
     with gr.Row():
         audio_widget = gr.Audio(type='filepath', format='wav',waveform_options=gr.WaveformOptions(
            show_recording_waveform=True), sources=['microphone'], scale = 3, label="Prompt/Question Voice Entry", max_length=120)
@@ -455,5 +506,6 @@ with gr.Blocks() as demo:
     output_window.change(fn=set_speak_button, inputs=output_window,outputs=speak_output)
     button_do_image.click(fn=make_image, inputs=[prompt_window,user_window, password],outputs=[image_window, output_window])
     image_window.change(fn=delete_image, inputs=[user])
+    help_button.click(fn=show_help, outputs=output_window)
     # demo.unload(final_clean_up(user))
 demo.launch(share=True)