Debugging

app.py

@@ -8,14 +8,9 @@ import tempfile
 from gtts import gTTS
 import os
 
-# Device configuration
 DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
 
-# Supabase configuration
-SUPABASE_URL = "https://gptmdbhzblfybdnohqnh.supabase.co"
-SUPABASE_API_KEY = "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9..."
-
-# Load translation models
+# Load models
 model_en_to_indic = AutoModelForSeq2SeqLM.from_pretrained("ai4bharat/indictrans2-en-indic-1B", trust_remote_code=True).to(DEVICE)
 tokenizer_en_to_indic = AutoTokenizer.from_pretrained("ai4bharat/indictrans2-en-indic-1B", trust_remote_code=True)
 model_indic_to_en = AutoModelForSeq2SeqLM.from_pretrained("ai4bharat/indictrans2-indic-en-1B", trust_remote_code=True).to(DEVICE)
@@ -24,18 +19,21 @@ tokenizer_indic_to_en = AutoTokenizer.from_pretrained("ai4bharat/indictrans2-ind
 ip = IndicProcessor(inference=True)
 asr = pipeline("automatic-speech-recognition", model="openai/whisper-small")
 
+# --- Supabase settings ---
+SUPABASE_URL = "https://gptmdbhzblfybdnohqnh.supabase.co"
+SUPABASE_API_KEY = "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9..."
 
+# --- Supabase utilities ---
 def save_to_supabase(input_text, output_text, direction):
     if not input_text.strip() or not output_text.strip():
         return "Nothing to save."
 
-    table_name = "translations" if direction == "en_to_ks" else "ks_to_en_translations"
+    table = "translations" if direction == "en_to_ks" else "ks_to_en_translations"
     payload = {
         "timestamp": datetime.utcnow().isoformat(),
         "input_text": input_text,
         "output_text": output_text
     }
-
     headers = {
         "apikey": SUPABASE_API_KEY,
         "Authorization": f"Bearer {SUPABASE_API_KEY}",
@@ -43,185 +41,119 @@ def save_to_supabase(input_text, output_text, direction):
     }
 
     try:
-        response = requests.post(
-            f"{SUPABASE_URL}/rest/v1/{table_name}",
-            headers=headers,
-            json=payload,
-            timeout=10
-        )
+        response = requests.post(f"{SUPABASE_URL}/rest/v1/{table}", json=payload, headers=headers)
         return "✅ Saved successfully!" if response.status_code == 201 else "❌ Failed to save."
     except Exception as e:
-        print("SAVE EXCEPTION:", e)
-        return "❌ Save request error."
-
-
-def get_translation_history(direction="en_to_ks"):
-    table_name = "translations" if direction == "en_to_ks" else "ks_to_en_translations"
+        print("Save error:", e)
+        return "❌ Save error."
 
+def get_translation_history(direction):
+    table = "translations" if direction == "en_to_ks" else "ks_to_en_translations"
     headers = {
         "apikey": SUPABASE_API_KEY,
         "Authorization": f"Bearer {SUPABASE_API_KEY}"
     }
 
     try:
-        response = requests.get(
-            f"{SUPABASE_URL}/rest/v1/{table_name}?order=timestamp.desc&limit=10",
-            headers=headers,
-            timeout=10
-        )
-        if response.status_code == 200:
-            records = response.json()
-            return "\n\n".join(
-                [f"Input: {r['input_text']} → Output: {r['output_text']}" for r in records]
-            )
+        res = requests.get(f"{SUPABASE_URL}/rest/v1/{table}?order=timestamp.desc&limit=10", headers=headers)
+        if res.status_code == 200:
+            data = res.json()
+            return "\n\n".join([f"Input: {r['input_text']} → Output: {r['output_text']}" for r in data])
         return "Failed to load history."
     except Exception as e:
-        print("HISTORY FETCH ERROR:", e)
+        print("History error:", e)
         return "Error loading history."
 
-
+# --- Translation ---
 def translate(text, direction):
     if not text.strip():
-        return "Please enter some text.", gr.update(), gr.update()
+        return "Enter some text.", gr.update(), gr.update()
 
     if direction == "en_to_ks":
-        src_lang = "eng_Latn"
-        tgt_lang = "kas_Arab"
-        model = model_en_to_indic
-        tokenizer = tokenizer_en_to_indic
+        src_lang, tgt_lang = "eng_Latn", "kas_Arab"
+        model, tokenizer = model_en_to_indic, tokenizer_en_to_indic
     else:
-        src_lang = "kas_Arab"
-        tgt_lang = "eng_Latn"
-        model = model_indic_to_en
-        tokenizer = tokenizer_indic_to_en
+        src_lang, tgt_lang = "kas_Arab", "eng_Latn"
+        model, tokenizer = model_indic_to_en, tokenizer_indic_to_en
 
     try:
-        processed = ip.preprocess_batch([text], src_lang=src_lang, tgt_lang=tgt_lang)
-        batch = tokenizer(processed, return_tensors="pt", padding=True).to(DEVICE)
-
+        batch = ip.preprocess_batch([text], src_lang=src_lang, tgt_lang=tgt_lang)
+        tokens = tokenizer(batch, return_tensors="pt", padding=True).to(DEVICE)
         with torch.no_grad():
-            outputs = model.generate(
-                **batch,
-                max_length=256,
-                num_beams=5,
-                num_return_sequences=1
-            )
-
-        translated = tokenizer.batch_decode(outputs, skip_special_tokens=True)
-        result = ip.postprocess_batch(translated, lang=tgt_lang)[0]
-
-        return result, gr.update(), gr.update()
+            output = model.generate(**tokens, max_length=256, num_beams=5)
+        result = tokenizer.batch_decode(output, skip_special_tokens=True)
+        final = ip.postprocess_batch(result, lang=tgt_lang)[0]
+        return final, gr.update(), gr.update()
     except Exception as e:
-        print("Translation Error:", e)
+        print("Translation error:", e)
         return "⚠️ Translation failed.", gr.update(), gr.update()
 
-
-def transcribe_audio(audio_path):
+# --- TTS for English output only ---
+def synthesize_tts(text):
     try:
-        result = asr(audio_path)
-        return result["text"]
+        tts = gTTS(text=text, lang="en")
+        with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as f:
+            tts.save(f.name)
+            return f.name
     except Exception as e:
-        print("STT Error:", e)
-        return "⚠️ Transcription failed."
+        print("TTS error:", e)
+        return None
 
+# --- STT only for en_to_ks ---
+def handle_audio_translation(audio_path, direction):
+    if direction != "en_to_ks":
+        return "⚠️ Audio input is only supported for English to Kashmiri.", "", "", None
 
-def synthesize_tts(text, direction):
-    if direction == "ks_to_en" and text.strip():
-        try:
-            tts = gTTS(text=text, lang="en")
-            tmp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".mp3")
-            tts.save(tmp_file.name)
-            tmp_file.close()
-            return tmp_file.name
-        except Exception as e:
-            print("TTS Error:", e)
-    return None
+    try:
+        transcription = asr(audio_path)["text"]
+    except Exception as e:
+        print("STT error:", e)
+        return "⚠️ Transcription failed.", "", "", None
 
+    translated, _, _ = translate(transcription, direction)
+    return transcription, translated, transcription, None
 
+# --- Switch UI direction ---
 def switch_direction(direction, input_text_val, output_text_val):
     new_direction = "ks_to_en" if direction == "en_to_ks" else "en_to_ks"
     input_label = "Kashmiri Text" if new_direction == "ks_to_en" else "English Text"
     output_label = "English Translation" if new_direction == "ks_to_en" else "Kashmiri Translation"
-    return (
-        new_direction,
-        gr.update(value=output_text_val, label=input_label),
-        gr.update(value=input_text_val, label=output_label)
-    )
-
-
-def handle_audio_translation(audio_path, direction):
-    if direction == "en_to_ks":
-        transcription = transcribe_audio(audio_path)
-        if transcription.startswith("⚠️"):
-            return transcription, "", "", None
-        translation, _, _ = translate(transcription, direction)
-        return transcription, translation, transcription, None
-    else:
-        # Assume audio_path is not used; rely on text in the input box
-        transcription = transcribe_audio(audio_path)
-        translation, _, _ = translate(transcription, direction)
-        tts_audio = synthesize_tts(translation, direction)
-        return transcription, translation, transcription, tts_audio
-
+    return new_direction, gr.update(value=output_text_val, label=input_label), gr.update(value=input_text_val, label=output_label)
 
 # === Gradio Interface ===
 with gr.Blocks() as interface:
-    gr.HTML("""
-    <div style="display: flex; justify-content: space-between; align-items: center; padding: 10px;">
-        <img src="https://raw.githubusercontent.com/BurhaanRasheedZargar/Images/211321a234613a9c3dd944fe9367cf13d1386239/assets/left_logo.png" style="height:150px; width:auto;">
-        <h2 style="margin: 0; text-align: center;">English ↔ Kashmiri Translator</h2>
-        <img src="https://raw.githubusercontent.com/BurhaanRasheedZargar/Images/77797f7f7cbee328fa0f9d31cf3e290441e04cd3/assets/right_logo.png">
-    </div>
-    """)
-
+    gr.HTML("<h2 style='text-align:center;'>English ↔ Kashmiri Translator</h2>")
     translation_direction = gr.State(value="en_to_ks")
 
     with gr.Row():
-        input_text = gr.Textbox(lines=2, label="English Text", placeholder="Enter text....")
-        output_text = gr.Textbox(lines=2, label="Kashmiri Translation", placeholder="Translated text....")
+        input_text = gr.Textbox(label="English Text", placeholder="Enter text here...")
+        output_text = gr.Textbox(label="Kashmiri Translation", placeholder="Translated text...")
 
     with gr.Row():
         translate_button = gr.Button("Translate")
         save_button = gr.Button("Save Translation")
-        switch_button = gr.Button("Switch")
+        switch_button = gr.Button("Switch Direction")
 
     save_status = gr.Textbox(label="Save Status", interactive=False)
-    history_box = gr.Textbox(lines=10, label="Translation History", interactive=False)
+    history = gr.Textbox(label="Translation History", lines=8, interactive=False)
 
     with gr.Row():
-        audio_input = gr.Audio(type="filepath", label="🎙️ Upload or record English audio")
-        audio_output = gr.Audio(label="🔊 English Output Audio")
+        audio_input = gr.Audio(type="filepath", label="🎙️ Record English audio")
+        audio_output = gr.Audio(label="🔊 English TTS")
 
-    stt_translate_button = gr.Button("🎤 Transcribe & Translate")
+    stt_button = gr.Button("🎤 Transcribe & Translate (EN → KS Only)")
 
     # Events
-    translate_button.click(
-        fn=translate,
-        inputs=[input_text, translation_direction],
-        outputs=[output_text, input_text, output_text]
-    )
-
-    save_button.click(
-        fn=save_to_supabase,
-        inputs=[input_text, output_text, translation_direction],
-        outputs=save_status
-    ).then(
-        fn=get_translation_history,
-        inputs=translation_direction,
-        outputs=history_box
-    )
-
-    switch_button.click(
-        fn=switch_direction,
-        inputs=[translation_direction, input_text, output_text],
-        outputs=[translation_direction, input_text, output_text]
-    )
-
-    stt_translate_button.click(
-        fn=handle_audio_translation,
-        inputs=[audio_input, translation_direction],
-        outputs=[input_text, output_text, input_text, audio_output]
-    )
+    translate_button.click(fn=translate, inputs=[input_text, translation_direction], outputs=[output_text, input_text, output_text])
+
+    save_button.click(fn=save_to_supabase, inputs=[input_text, output_text, translation_direction], outputs=save_status)\
+               .then(fn=get_translation_history, inputs=translation_direction, outputs=history)
+
+    switch_button.click(fn=switch_direction, inputs=[translation_direction, input_text, output_text],
+                        outputs=[translation_direction, input_text, output_text])
+
+    stt_button.click(fn=handle_audio_translation, inputs=[audio_input, translation_direction],
+                     outputs=[input_text, output_text, input_text, audio_output])
 
 if __name__ == "__main__":
-    interface.queue().launch(share=True)
+    interface.queue().launch()