Delete app.py

app.py

@@ -1,382 +0,0 @@
-import gradio as gr
-import os
-import difflib
-from gradio_client import Client, file as gradio_file
-import time
-import google.generativeai as genai # Import Gemini library
-
-# --- Configuration & Clients ---
-
-# Function to initialize Gemini client (handles local env var and HF Secrets)
-def configure_gemini_api():
-    api_key = None
-    try:
-        # This will only work if running on Hugging Face Spaces with the secret set
-        api_key = gr.Secrets.get("GOOGLE_API_KEY")
-    except AttributeError: # Running locally, gr.Secrets not available
-        api_key = os.environ.get("GOOGLE_API_KEY")
-    except FileNotFoundError: # gr.Secrets.get can raise this if no secrets file found
-        api_key = os.environ.get("GOOGLE_API_KEY")
-
-    if api_key:
-        try:
-            genai.configure(api_key=api_key)
-            return True
-        except Exception as e:
-            print(f"Error configuring Gemini API: {e}")
-            return False
-    else:
-        print("WARN: GOOGLE_API_KEY not found in Gradio Secrets or environment. Story generation with Gemini will be disabled.")
-        return False
-
-# Configure Gemini API at startup
-GEMINI_API_CONFIGURED = configure_gemini_api()
-
-# Initialize TTS Client for Bark (suno/bark)
-try:
-    bark_tts_client = Client("suno/bark")
-except Exception as e:
-    print(f"Fatal: Could not initialize Bark TTS client (suno/bark): {e}. TTS will not work.")
-    bark_tts_client = None
-
-# Initialize STT Client for Whisper (abidlabs/whisper-large-v2)
-try:
-    whisper_stt_client = Client("abidlabs/whisper-large-v2")
-except Exception as e:
-    print(f"Fatal: Could not initialize Whisper STT client (abidlabs/whisper-large-v2): {e}. STT will not work.")
-    whisper_stt_client = None
-
-# --- Helper Functions ---
-
-def generate_story_with_gemini(name, grade, topic):
-    if not GEMINI_API_CONFIGURED:
-        return "Google Gemini API key not configured. Story generation is disabled. 🔑"
-
-    try:
-        # Choose a Gemini model. 'gemini-1.5-flash-latest' is good for speed and general tasks.
-        # Other options: 'gemini-1.0-pro', 'gemini-1.5-pro-latest' (if available and needed)
-        model = genai.GenerativeModel(model_name="gemini-1.5-flash-latest") # Or 'gemini-pro'
-
-        prompt = (
-            f"You are a super friendly and imaginative storyteller for kids. "
-            f"Please write an exciting and fun short story (around 100-120 words) for a student named {name} who is in Grade {grade}. "
-            f"The story must be about '{topic}'. "
-            f"Use simple words and sentences that a Grade {grade} student can easily read aloud and understand. "
-            f"Make the story engaging and positive. Jump right into the story without any introduction like 'Here is a story for you'."
-        )
-
-        # For safety, though Gemini has built-in safety settings
-        safety_settings = [
-            {"category": "HARM_CATEGORY_HARASSMENT", "threshold": "BLOCK_MEDIUM_AND_ABOVE"},
-            {"category": "HARM_CATEGORY_HATE_SPEECH", "threshold": "BLOCK_MEDIUM_AND_ABOVE"},
-            {"category": "HARM_CATEGORY_SEXUALLY_EXPLICIT", "threshold": "BLOCK_MEDIUM_AND_ABOVE"},
-            {"category": "HARM_CATEGORY_DANGEROUS_CONTENT", "threshold": "BLOCK_MEDIUM_AND_ABOVE"},
-        ]
-
-        generation_config = genai.types.GenerationConfig(
-            candidate_count=1,
-            # stop_sequences=["\n\n\n"], # Optional: if you notice overly long outputs
-            max_output_tokens=300, # Generous for a 120-word story
-            temperature=0.75, # For creativity
-            # top_p=0.9,      # Optional: nucleus sampling
-            # top_k=40        # Optional: top-k sampling
-        )
-
-        response = model.generate_content(
-            prompt,
-            generation_config=generation_config,
-            safety_settings=safety_settings
-        )
-
-        if response.candidates and response.candidates[0].content.parts:
-            story = response.text # .text directly gives the generated string
-            # Check for safety blocks
-            if response.prompt_feedback and response.prompt_feedback.block_reason:
-                 return f"Oh dear! My story idea for '{topic}' was a bit too wild and got blocked by the safety filter (Reason: {response.prompt_feedback.block_reason}). Let's try a different topic! 😊"
-            if not story.strip(): # Empty response despite no block
-                return f"Hmm, Gemini gave me a blank page for '{topic}'. Let's try a different topic or try again! ✨"
-            return story.strip()
-        else: # No valid candidates or blocked
-            if response.prompt_feedback and response.prompt_feedback.block_reason:
-                 return f"Oh dear! My story idea for '{topic}' was a bit too wild and got blocked by the safety filter (Reason: {response.prompt_feedback.block_reason}). Let's try a different topic! 😊"
-            print(f"Gemini API response issue: {response}")
-            return f"Hmm, Gemini's story magic seems to be on a little break for '{topic}'. Maybe try another topic? 🤔"
-
-    except Exception as e:
-        print(f"Error generating story with Gemini: {e}")
-        # Check for common API key related errors
-        if "API_KEY_INVALID" in str(e) or "API key not valid" in str(e):
-            return "Oops! The Google Gemini API key seems to be having a problem. Please tell the grown-ups to check it! 🔑"
-        return f"Oh no! 😟 I had a little trouble dreaming up a story with Gemini. Error: {e}"
-
-
-def text_to_speech_bark(text_to_speak):
-    if not bark_tts_client:
-        return "The Bark TTS sound machine isn't working right now. 🛠️ Please tell the grown-ups!"
-    try:
-        voice_preset = "v2/en_speaker_7"
-        job = bark_tts_client.submit(
-            text_to_speak,
-            voice_preset,
-            api_name="/generate_audio"
-        )
-        audio_result = job.result(timeout=180)
-
-        if isinstance(audio_result, tuple) and len(audio_result) > 0:
-            audio_filepath = audio_result[0]
-        elif isinstance(audio_result, str):
-            audio_filepath = audio_result
-        else:
-            print(f"Unexpected Bark TTS result format: {audio_result}")
-            return "Hmm, the sound came out a bit funny from Bark. 🤔"
-        return audio_filepath
-    except Exception as e:
-        print(f"Error with Bark TTS (suno/bark): {e}")
-        if "Queue full" in str(e).lower() or "too much pending traffic" in str(e).lower() or " सर्वर व्यस्त है" in str(e):
-            return "The Bark sound machine is super busy with other kids!  인기폭발!  очередь! Please try again in a little bit. 🕒"
-        if "generator" in str(e).lower() and "choices" in str(e).lower():
-             return f"Oops! Bark had a hiccup with the voice. Chosen: '{voice_preset}'. Maybe try later? Details: {e}"
-        return f"Oh dear, Bark couldn't make the sound. 🔇 Error: {e}"
-
-def speech_to_text_whisper_space(audio_filepath):
-    if not whisper_stt_client:
-        return "The Whisper listening ears aren't working right now. 🛠️ Please tell the grown-ups!"
-    if not audio_filepath:
-        return "Oops! I didn't get any recording to listen to. 🎤"
-    try:
-        job = whisper_stt_client.submit(
-            gradio_file(audio_filepath),
-            "transcribe",
-            "English",
-            api_name="/predict"
-        )
-        result_dict = job.result(timeout=120)
-
-        if isinstance(result_dict, dict) and 'text' in result_dict:
-            return result_dict['text']
-        elif isinstance(result_dict, str):
-            return result_dict
-        else:
-            print(f"Unexpected Whisper STT result format: {result_dict}")
-            return "Hmm, I couldn't quite understand the words from Whisper. 🤔"
-    except Exception as e:
-        print(f"Error transcribing audio with Whisper Space: {e}")
-        if "Queue full" in str(e).lower() or "too much pending traffic" in str(e).lower():
-            return "The Whisper listening ears are super busy! 인기폭발! очередь! Please try again in a bit. 🕒"
-        return f"Oh no! Whisper had trouble hearing that. 🙉 Error: {e}"
-
-def clean_text_for_comparison(text):
-    if not isinstance(text, str): return []
-    text = text.lower()
-    punctuation_to_remove = "!\"#$%&()*+,-./:;<=>?@[\\]^_`{|}~"
-    text = text.translate(str.maketrans('', '', punctuation_to_remove))
-    return text.split()
-
-def compare_texts_for_feedback(original_text, student_text):
-    original_words = clean_text_for_comparison(original_text)
-    student_words = clean_text_for_comparison(student_text)
-
-    if not student_words:
-        return "It sounds like you didn't record anything, or maybe it was super quiet! 🤫 Try recording again nice and clear!", ""
-
-    matcher = difflib.SequenceMatcher(None, original_words, student_words, autojunk=False)
-    feedback_lines = []
-    highlighted_passage_parts = []
-
-    for tag, i1, i2, j1, j2 in matcher.get_opcodes():
-        original_segment = original_words[i1:i2]
-        student_segment = student_words[j1:j2]
-
-        if tag == 'equal':
-            highlighted_passage_parts.append(" ".join(original_segment))
-        elif tag == 'replace':
-            if len(original_segment) == len(student_segment):
-                for i in range(len(original_segment)):
-                    o_word = original_segment[i]
-                    s_word = student_segment[i]
-                    feedback_lines.append(f"- You said: \"*{s_word}*\" instead of: \"**{o_word}**\"")
-                    highlighted_passage_parts.append(f"~~{o_word}~~ **{s_word}**")
-            else:
-                feedback_lines.append(f"- Instead of: \"**{' '.join(original_segment)}**\", you said: \"*{' '.join(student_segment)}*\"")
-                highlighted_passage_parts.append(f"~~{' '.join(original_segment)}~~ **{' '.join(student_segment)}**")
-        elif tag == 'delete':
-            feedback_lines.append(f"- You missed: \"**{' '.join(original_segment)}**\"")
-            highlighted_passage_parts.append(f"~~{' '.join(original_segment)}~~ (*skipped*)")
-        elif tag == 'insert':
-            feedback_lines.append(f"- You added: \"*{' '.join(student_segment)}*\" (which wasn't in the story)")
-            highlighted_passage_parts.append(f"(*added:* **{' '.join(student_segment)}**)")
-
-    final_highlighted_text = " ".join(highlighted_passage_parts)
-
-    if not feedback_lines:
-        return "🎉🥳 WOOHOO! Amazing reading! You got all the words spot on! 🥳🎉", final_highlighted_text
-    else:
-        feedback_summary = "Great try! Here are a few words to practice to make it even better:\n" + "\n".join(feedback_lines)
-        return feedback_summary, final_highlighted_text
-
-# --- Gradio UI Functions ---
-def generate_story_and_audio_for_ui(name, grade, topic, progress=gr.Progress(track_tqdm=True)):
-    if not name or not grade or not topic:
-        return "Oops! Please tell me your name, grade, and a fun topic first! 😊", None, gr.update(visible=False), ""
-
-    progress(0.1, desc="📖 Asking Gemini to dream up a cool story for you...")
-    story_text = generate_story_with_gemini(name, grade, topic) # USE GEMINI FUNCTION
-    
-    # Check for Gemini specific error messages or general failure indicators
-    gemini_error_keywords = ["Gemini API key not configured", "Oh no!", "Oops!", "Hmm,"]
-    if any(keyword in story_text for keyword in gemini_error_keywords) or not story_text.strip() :
-         # Keep recording area hidden if story generation failed
-        return story_text, None, gr.update(visible=False), story_text
-
-    progress(0.5, desc="🎧 Warming up the Bark sound machine... (this can take a moment, like magic!)")
-    tts_audio_path = text_to_speech_bark(story_text)
-
-    error_conditions_tts = [
-        "couldn't make the sound", "sound came out a bit funny", "sound machine isn't working",
-        "sound machine is super busy", "Bark had a hiccup"
-    ]
-    if any(err in (tts_audio_path or "") for err in error_conditions_tts):
-        # Show story, but show TTS error and hide recording parts
-        return story_text, tts_audio_path, gr.update(visible=False), story_text
-
-    progress(1.0, desc="✅ Story and sound are ready! Let's go!")
-    return (
-        story_text,
-        tts_audio_path,
-        gr.update(visible=True), # Show recording_assessment_area
-        story_text               # Pass story_text to gr.State
-    )
-
-def assess_student_reading_ui(original_passage_state, student_audio_path, progress=gr.Progress(track_tqdm=True)):
-    if not student_audio_path:
-        return "🎤 Whoops! Did you forget to record your awesome reading? Try again!", ""
-    if not original_passage_state:
-        return "Hmm, I lost the story! 😟 Please generate a new story first.", ""
-
-    progress(0.2, desc="👂 Whisper is listening carefully to your recording...")
-    transcribed_text = speech_to_text_whisper_space(student_audio_path)
-
-    error_conditions_stt = [
-        "couldn't understand the words", "had trouble hearing that", "listening ears aren't working",
-        "listening ears are super busy", "didn't get any recording"
-    ]
-    if any(err in (transcribed_text or "") for err in error_conditions_stt):
-        return transcribed_text, ""
-
-    progress(0.7, desc="🧠 Thinking about the words...")
-    feedback, highlighted_passage = compare_texts_for_feedback(original_passage_state, transcribed_text)
-    progress(1.0, desc="⭐ Feedback is ready!")
-    return feedback, highlighted_passage
-
-# --- Gradio Interface ---
-css = """
-body { font-family: 'Comic Sans MS', 'Chalkboard SE', 'Comic Neue', cursive; background-color: #F0F8FF; } /* AliceBlue background */
-.gr-button {
-    background-color: #FF69B4 !important; /* HotPink */
-    color: white !important;
-    border-radius: 20px !important;
-    font-weight: bold !important;
-    border: 2px solid #FF1493 !important; /* DeepPink border */
-    box-shadow: 0px 3px 5px rgba(0,0,0,0.2) !important;
-}
-.gr-button:hover { background-color: #FF1493 !important; } /* DeepPink on hover */
-.gr-panel {
-    border-radius: 15px !important;
-    box-shadow: 5px 5px 15px rgba(0,0,0,0.1) !important;
-    background-color: #FFFACD !important; /* LemonChiffon panel background */
-    border: 2px dashed #FFD700 !important; /* Gold dashed border */
-}
-label, .gr-checkbox-label { color: #4B0082 !important; font-weight: bold !important; } /* Indigo */
-.gr-textbox, .gr-dropdown { border-radius: 10px !important; border: 1px solid #DDA0DD !important; } /* Plum border for inputs */
-#student_audio_input audio { background-color: #E6E6FA; border-radius: 10px; } /* Lavender for audio player */
-#feedback_output, #highlighted_passage_output {
-    background-color: #FFFFE0; /* LightYellow */
-    padding: 15px;
-    border-radius: 10px;
-    border: 1px solid #FAFAD2; /* LightGoldenrodYellow */
-}
-"""
-
-with gr.Blocks(theme=gr.themes.Soft(primary_hue=gr.themes.colors.pink, secondary_hue=gr.themes.colors.purple), css=css) as app:
-    gr.Markdown(
-        """
-        <div style="text-align: center; padding: 20px 0;">
-            <h1 style="color: #FF6347; font-size: 3em; text-shadow: 2px 2px #D3D3D3;">🌈🦄✨ AI Reading Buddy ✨🦄🌈</h1>
-            <p style="font-size: 1.3em; color: #483D8B;">Let's read a super fun story from Gemini and practice our words!</p>
-        </div>
-        """
-    )
-
-    original_passage_state = gr.State("")
-
-    with gr.Row():
-        with gr.Column(scale=1):
-            gr.Markdown("### <span style='color:#DB7093;'>✏️ Tell Me About You!</span>")
-            student_name_input = gr.Textbox(label="👑 Your Awesome Name:", placeholder="E.g., Princess Lily")
-            student_grade_input = gr.Dropdown(
-                label="🧑‍🎓 Your Grade:",
-                choices=[f"{i}" for i in range(1, 11)],
-                value="3"
-            )
-            topic_input = gr.Textbox(label="🚀 Story Topic Idea:", placeholder="E.g., brave little astronaut")
-            generate_button = gr.Button(value="🎈 Get My Gemini Story!")
-
-        with gr.Column(scale=2):
-            gr.Markdown("### <span style='color:#DB7093;'>📖 Your Special Story (from Gemini AI):</span>")
-            passage_output = gr.Textbox(label="Read this aloud:", lines=10, interactive=False)
-            gr.Markdown("### <span style='color:#DB7093;'>🔊 Listen to the Story:</span>")
-            audio_output = gr.Audio(label="Hear how it sounds (with Bark TTS Bark️)", type="filepath")
-
-    gr.Markdown("<hr style='border:1px dashed #FFB6C1;'>") # LightPink dashed separator
-
-    with gr.Row(visible=False) as recording_assessment_area:
-        with gr.Column(scale=1):
-            gr.Markdown("### <span style='color:#32CD32;'>🤩 Your Turn to Shine! 🤩</span>")
-            student_audio_input = gr.Audio(sources=["microphone"], type="filepath", label="🎤 Record yourself reading the story! Press the mic, then stop.", elem_id="student_audio_input")
-            assess_button = gr.Button(value="🧐 Check My Reading!", elem_id="assess_button")
-
-        with gr.Column(scale=2):
-            gr.Markdown("### <span style='color:#32CD32;'>💡 Word Detective Feedback:</span>")
-            feedback_output = gr.Markdown(value="Your amazing feedback will pop up here! ✨", elem_id="feedback_output")
-            highlighted_passage_output = gr.Markdown(value="See your reading journey here! 🗺️", elem_id="highlighted_passage_output")
-
-
-    generate_button.click(
-        fn=generate_story_and_audio_for_ui,
-        inputs=[student_name_input, student_grade_input, topic_input],
-        outputs=[
-            passage_output,
-            audio_output,
-            recording_assessment_area, # Directly control visibility of the row
-            original_passage_state
-        ]
-    )
-
-    assess_button.click(
-        fn=assess_student_reading_ui,
-        inputs=[original_passage_state, student_audio_input],
-        outputs=[feedback_output, highlighted_passage_output]
-    )
-
-    gr.Markdown(
-        """
-        ---
-        <div style="text-align: center; font-size: 0.9em; color: #555;">
-        Built with ❤️ for the Agentic Demo Track Hackathon! Tag: <code>agent-demo-track</code>
-        <br>Stories by Google Gemini, voices by Suno Bark @ HF, and listening by Whisper @ HF.
-        </div>
-        """
-    )
-
-# --- Launching the App ---
-if __name__ == "__main__":
-    if not GEMINI_API_CONFIGURED:
-        print("🚨 GOOGLE_API_KEY not configured for local testing or failed to initialize!")
-        print("Please set it: export GOOGLE_API_KEY='your_key_here'")
-
-    if not bark_tts_client:
-        print("🚨 Bark TTS client (suno/bark) could not be initialized. TTS will not work.")
-    if not whisper_stt_client:
-        print("🚨 Whisper STT client (abidlabs/whisper-large-v2) could not be initialized. STT will not work.")
-
-    app.launch(debug=True)