fix

Findingword.py

@@ -1,9 +1,8 @@
 import openai
 from flask import Flask, jsonify, request, send_from_directory, send_file, Blueprint, current_app
 import os
-from google.cloud import texttospeech
 from flask_cors import CORS
-import io  # NEW: for streaming S3 bytes in HF/AWS mode
+import io  # for streaming S3 bytes in HF/AWS mode
 
 # Optional (only used in AWS mode)
 try:
@@ -17,17 +16,15 @@ except Exception:
 app = Flask(__name__)
 CORS(app)
 
-# --- NEW: Blueprint ---
+# --- Blueprint ---
 finding_bp = Blueprint("findingword", __name__)
 
 # Directories for video, audio, and transcripts
 VIDEO_FOLDER = 'static/videos'
 AUDIO_FOLDER = 'static/audio'  # Ensure this folder exists
 TRANSCRIPT_FOLDER = 'static/transcripts'
+os.makedirs(AUDIO_FOLDER, exist_ok=True)
 
-# Set your OpenAI API key (left as-is per your request)
-# openai.api_key = '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'
-# openai.api_key = os.getenv("OPENAI_API_KEY", "")
 # --- OpenAI key handling (same as vocab builder) ---
 _OPENAI_API_KEY_FALLBACK = os.getenv("OPENAI_API_KEY", "")
 
@@ -36,13 +33,8 @@ def _ensure_openai_key():
     api_key = (current_app.config.get("OPENAI_API_KEY") if current_app else None) or _OPENAI_API_KEY_FALLBACK
     if api_key:
         openai.api_key = api_key
-# Initialize Google Cloud TTS client (local mode)
-client = texttospeech.TextToSpeechClient()
-
-# Ensure required folders exist (used in local mode)
-os.makedirs(AUDIO_FOLDER, exist_ok=True)
 
-# ---------------------- audio-mode helpers (UNCHANGED) ----------------------
+# ---------------------- audio-mode helpers ----------------------
 def _is_aws_mode() -> bool:
     """
     Switch to AWS Polly + S3 on Hugging Face / prod.
@@ -64,8 +56,11 @@ def _sanitize_filename(word: str) -> str:
 
 @finding_bp.route('/generate-vocabulary', methods=['GET'])
 def get_vocabulary_word_from_openai():
-    #prompt = "Pick a random vocabulary word suitable for children and provide its meaning. Do not repeat words from previous responses. Format: 'Word: [word]. Meaning: [meaning].'"
-    prompt = "Pick a simple vocabulary word suitable for children (ages 6–8) and provide its meaning in very easy English. Do not repeat words from previous responses. Format: 'Word: [word]. Meaning: [meaning].'"
+    prompt = (
+        "Pick a simple vocabulary word suitable for children (ages 6–8) "
+        "and provide its meaning in very easy English. Do not repeat words from previous responses. "
+        "Format: 'Word: [word]. Meaning: [meaning].'"
+    )
 
     try:
         _ensure_openai_key()
@@ -83,31 +78,30 @@ def get_vocabulary_word_from_openai():
         if "Word:" in result and "Meaning:" in result:
             parts = result.split("Meaning:")
             word = parts[0].replace("Word:", "").strip()
-            # ✅ Ensure the extracted word does not contain trailing dots or special characters
-            word = word.rstrip('.')
+            word = word.rstrip('.')  # avoid trailing dot
             meaning = parts[1].strip()
 
             # Generate the sentence
             sentence = generate_sentence(word, meaning)
 
             # Generate audio file for the vocabulary word
-            audio_file_path_or_name = generate_audio(word)  # may be local path or just filename in AWS mode
+            audio_file_path_or_name = generate_audio(word)  # local path or just filename in AWS mode
 
-            # Convert for frontend exactly as before
+            # URL for frontend remains identical
             audio_url = f"/static/audio/{os.path.basename(audio_file_path_or_name)}"
 
             return jsonify({
                 "word": word,
                 "meaning": meaning,
                 "sentence": sentence,
-                "audio_file_path": audio_url  # Returning the same field as before
+                "audio_file_path": audio_url
             })
 
         else:
             return jsonify({"response": result, "message": "Meaning not provided in the expected format"})
 
     except Exception as e:
-        return jsonify({"error": str(e)})
+        return jsonify({"error": str(e)}), 500
 
 
 def generate_sentence(word, meaning):
@@ -120,14 +114,13 @@ def generate_sentence(word, meaning):
             {"role": "user", "content": prompt},
         ]
     )
-
     sentence = response.choices[0].message.content.strip()
     return sentence
 
 
 def generate_audio(word):
     """
-    Local (default): Google TTS → write MP3 to ./static/audio/<word>.mp3 → return full path (unchanged).
+    Local (default): Google TTS → write MP3 to ./static/audio/<word>.mp3 → return full path.
     Hugging Face / AWS mode: Polly → upload to S3 (findingword/<word>.mp3) → return just the filename,
     and let /static/audio/<filename> stream from S3 (see route below).
     """
@@ -171,18 +164,31 @@ def generate_audio(word):
         # Return only the filename; /static/audio/<filename> will proxy from S3
         return filename
 
-    # ---- Local Google TTS path (unchanged) ----
+    # ---- Local Google TTS path (lazy import; no GCP on HF) ----
     os.makedirs(AUDIO_FOLDER, exist_ok=True)
     audio_file_path = os.path.join(AUDIO_FOLDER, filename)
 
     if not os.path.exists(audio_file_path):
+        try:
+            # Import only in local mode to avoid HF credential errors
+            from google.cloud import texttospeech
+            gcp_client = texttospeech.TextToSpeechClient()
+        except Exception as e:
+            raise RuntimeError(
+                "Google TTS is required in local mode but missing. "
+                "Install google-cloud-texttospeech and set GOOGLE_APPLICATION_CREDENTIALS. "
+                f"Details: {e}"
+            )
+
         synthesis_input = texttospeech.SynthesisInput(text=word)
         voice = texttospeech.VoiceSelectionParams(
             language_code="en-US", ssml_gender=texttospeech.SsmlVoiceGender.NEUTRAL
         )
         audio_config = texttospeech.AudioConfig(audio_encoding=texttospeech.AudioEncoding.MP3)
 
-        response = client.synthesize_speech(input=synthesis_input, voice=voice, audio_config=audio_config)
+        response = gcp_client.synthesize_speech(
+            input=synthesis_input, voice=voice, audio_config=audio_config
+        )
 
         with open(audio_file_path, "wb") as out:
             out.write(response.audio_content)
@@ -196,33 +202,27 @@ def generate_audio(word):
 def validate_word():
     try:
         data = request.get_json()
-        print("📥 Received data for validation:", data)  # Log input data
+        print("📥 Received data for validation:", data)
 
         if not data or 'user_input' not in data or 'correct_word' not in data:
-            print("❌ Missing user_input or correct_word in request data.")
             return jsonify({"error": "Invalid request, missing fields"}), 400
 
         user_input = data.get('user_input', '').strip()
         correct_word = data.get('correct_word', '').strip()
 
-        print(f"🔍 Validating: User Input - '{user_input}' | Correct Word - '{correct_word}'")
-
         if user_input.lower() == correct_word.lower():
-            print("✅ Validation Success")
             return jsonify({"status": "success", "message": "Correct! You typed the word correctly."})
         else:
-            print("❌ Validation Failed")
             return jsonify({"status": "failure", "message": f"Incorrect. The correct word was '{correct_word}'."})
 
     except Exception as e:
-        print("❌ Exception in validate-word API:", str(e))
         return jsonify({"error": str(e)}), 500
 
 
 @finding_bp.route('/static/audio/<filename>')
 def serve_audio(filename):
     """
-    Local: serve from disk (unchanged).
+    Local: serve from disk.
     AWS mode (HF): fetch the object from S3 and stream it (no local storage).
     """
     if _is_aws_mode():
@@ -240,7 +240,6 @@ def serve_audio(filename):
         try:
             obj = s3.get_object(Bucket=bucket, Key=key)
             data = obj["Body"].read()
-            # Stream as an MP3 without saving locally
             return send_file(
                 io.BytesIO(data),
                 mimetype="audio/mpeg",