prepping text for voice/dollars

ai_voice/voice_handling.py

@@ -24,10 +24,27 @@ def SetVoiceId(newVoice):
     print("SetVoiceId: voice_id: "+voice_id + " newVoice "+newVoice)
     voice_id = GetVoiceId(newVoice)
 
+def UpdateDF():
+    global dataframe
+    dataframe.value=pd.DataFrame({"role": [""], "content": [""] })
+
 def PrepareForVoice(text):
-    prepped_text = text.replace('"', '').replace('401k', '4 oh 1 k').replace('10k', 'ten thousand').replace('slalom', "slallum").replace('Slalom', "slallum").replace('IT 101', "IT 1 oh 1")
-    prepped_text = re.sub(r'\$(\d+)', r'\1 dollars', prepped_text)    
-    return clean_text.ReplaceNumbersWithWords(prepped_text)
+    print("prepped_text before: "+text)
+    p = inflect.engine()
+    prepped_text = text.replace('"', '').replace('401k', '4 oh 1 k').replace('slalom', "slallum").replace('Slalom', "slallum").replace('IT 101', "IT 1 oh 1")
+    prepped_text = re.sub(r'(\d+)m', lambda m: p.number_to_words(int(m.group(1)) * 1000000), prepped_text, flags=re.IGNORECASE)
+    prepped_text = re.sub(r'(\d+)k', lambda m: p.number_to_words(int(m.group(1)) * 1000), prepped_text, flags=re.IGNORECASE)
+    prepped_text = re.sub(r'\$(\d+(?:,\d{3})*(?:\.\d{2})?)', lambda m: p.number_to_words(int(m.group(1).replace(',', '')) if '.' not in m.group(1) else float(m.group(1).replace(',', ''))) + ' dollars', prepped_text, flags=re.IGNORECASE)
+    print("prepped_text: after "+prepped_text) 
+    return clean_text.ReplaceNumbersWithWords(prepped_text)  # Change to use ReplaceNumbersWithWords directly
+
+    
+'''def PrepareForVoice(text):
+    prepped_text = text.replace('"', '').replace(',', '').replace('401k', '4 oh 1 k').replace('10k', 'ten thousand').replace('slalom', "slallum").replace('Slalom', "slallum").replace('IT 101', "IT 1 oh 1")
+    print("prepped_text before: "+prepped_text)
+    prepped_text = re.sub(r'\$(\d+)', r'\1 dollars', prepped_text)   
+    print("prepped_text: "+prepped_text) 
+    return clean_text.ReplaceNumbersWithWords(prepped_text)'''
 
 def GetVoiceId(voice):
     if voice==constants.VOICE_2:
@@ -89,5 +106,4 @@ def TranslateAndProcessAudio(audio, prompt, eval_sentiment, eval_emotion, messag
     processedAudio=PrepareForVoice(system_message)
     audio_html=ProcessAudio(processedAudio)
     df = pd.DataFrame(messages)
-    return messages, df, audio_html
-   
+    return messages, df, audio_html   

app.py

@@ -1,15 +1,16 @@
 import gradio as gr
 import pandas as pd
 import openai
+from image_gen import image_creation
+from summarization import summarize
 from utilities import constants,api_keys
 from ui.app_theme import SoftBlue
-from ui import prompt_builder,summarization,image_creation
+from prompts import prompt_builder
 
 openai.api_key = api_keys.APIKeys().get_key('OPENAI_API_KEY')
 
-
 ui1=prompt_builder.ui
-ui2=summarization.ui
+ui2=summarize.ui
 ui3=image_creation.ui
 ui = gr.TabbedInterface([ui1,ui2,ui3], (constants.UI_1, constants.UI_2, constants.UI_3), theme=SoftBlue())
 

{ui → image_gen}/image_creation.py

@@ -5,6 +5,10 @@ from image_gen import image_generation
 
 IMG_SIZE=256
 
+def InitDF():
+    global imageDF
+    imageDF=pd.DataFrame({"role": [""], "content": [""] })
+
 def UpdateArtSetting(setting):
     global settingOptions
     settingOptions.value=setting
@@ -49,4 +53,6 @@ with gr.Blocks() as ui:
         imageDF = gr.DataFrame(type="pandas", value=pd.DataFrame({"role": [""], "content": [""] }), wrap=True, label=constants.OPENAI_LOG)
     imageSubmit.click(Generate, inputs=[imageDescription], outputs=[image1,image2,image3,imageDF])
     artOptions.input(UpdateArtStyle,inputs=[artOptions], outputs=[])
-    settingOptions.input(UpdateArtSetting,inputs=[settingOptions], outputs=[])
+    settingOptions.input(UpdateArtSetting,inputs=[settingOptions], outputs=[])
+
+    InitDF()

{ui → prompts}/prompt_builder.py

@@ -13,6 +13,10 @@ def UpdatePersonas(person_new):
     label, messages, persona.value = system_prompts.ChangePersona(person_new,[],persona)
     return label
 
+def InitDF():
+    global dataframe
+    dataframe=pd.DataFrame({"role": [""], "content": [""] })
+
 def Clear():
     global audio
     audio.value=None
@@ -87,4 +91,5 @@ with gr.Blocks() as ui:
     sentiment.change(ToggleSentiment, inputs=[sentiment], outputs=[])
     emotion.change(ToggleEmotion, inputs=[emotion], outputs=[])
 
-UpdatePersonas(constants.PERSONA_HR_EXPERT)
+UpdatePersonas(constants.PERSONA_HR_EXPERT)
+InitDF()

prompts/prompt_constants.py

@@ -8,8 +8,8 @@ SPANISH_PROMPT="Provide responses entirely in Spanish."
 FRENCH_PROMPT="Provide responses entirely in French."
 ENGLISH_PROMPT="Provide responses entirely in English."  
 SUMMARY_PROMPT="Identify the language (code) or genre (literature) as Type, summarize what this text does as simply as possible in the Summary. Identify key areas of interest in Key Concepts. Format response as follows:\nType: Python\nSummary: The code performs web scraping, text processing, tokenization, embedding generation, and question-answering tasks using OpenAI models and libraries.\nKey Concepts: Key concepts include the web scraping logic, text processing techniques, and the usage of OpenAI's embedding and question-answering models. Content: "
-SENTIMENT_PROMPT="At the end of the response, Provide a sentiment evaluation of the user's input as \nSentiment: [Positive, Neutral, Negative]."
-EMOTION_PROMPT = "At the end of the response, Provide an emotion evaluation of the user's input as \nEmotion: [Happy, Angry, Sad, etc]."
+SENTIMENT_PROMPT="At the end of the response, Provide a sentiment evaluation of the user's input as \nSentiment: [Positive, Neutral, Negative]. Include '.' at the end of the sentence."
+EMOTION_PROMPT = "At the end of the response, Provide an emotion evaluation of the user's input as \nEmotion: [Happy, Angry, Sad, etc]. Include '.' at the end of the sentence."
 IMPRESSIONISM_PROMPT="Impressionist Art. Use short, thick strokes of paint to capture the essence of the Subject rather than the details. Focus on the changing effects of light on the subject. Ex. Monet"
 POP_ART_PROMPT="Pop Art. Use bold, vibrant colors and iconic imagery to capture the spirit of the Subject, including consumerism, mass media, and popular culture. Ex. Warhol"
 CUBISM_PROMPT="Cubist Art. Fragment and reassemble the Subject into an abstracted form, focus on geometric shapes and multiple viewpoints to depict the subject across dimensions. Ex.Picasso."

ui/summarization.py → summarization/summarize.py

@@ -11,6 +11,10 @@ def Summary(code):
     df = pd.DataFrame(sum_messages) 
     return sum_text,df
 
+def InitDF():
+    global summaryDF
+    summaryDF=pd.DataFrame({"role": [""], "content": [""] })
+
 def UpdateWithExample(newExample):
     fileName=newExample.replace(" ","_").lower()
     if os.path.exists(f"{constants.TXT_PREFIX}{fileName}.txt"):
@@ -46,3 +50,5 @@ with gr.Blocks() as ui:
                                wrap=True, show_label=False, label=constants.OPENAI_LOG)
     summarize.click(Summary, inputs=[code], outputs=[summary,summaryDF])
     examples.input(UpdateWithExample,inputs=[examples], outputs=[code])
+
+    InitDF()

utilities/clean_text.py

@@ -7,10 +7,14 @@ def RemoveRole(text):
 def ReplaceNumbersWithWords(text):
     p = inflect.engine()
     words = text.split()
+    print("words: "+str(words))
     for i, word in enumerate(words):
         if word.isdigit():
+            print("word-isdigit: "+word)
             words[i] = p.number_to_words(word)
         elif word in constants.SYMBOL_TO_WORD:
+            print("word - else: "+word+" symbol: "+constants.SYMBOL_TO_WORD[word])
             words[i] = constants.SYMBOL_TO_WORD[word]
     reply=' '.join(words)
+    print('returning: '+reply)
     return reply