Update chatbot_logic.py

chatbot_logic.py

@@ -1,238 +1,264 @@
-import os
-import re
-import json
-import pyttsx3
-import requests
-from datetime import datetime
-from dotenv import load_dotenv
-
-# === ENV SETUP ===
-load_dotenv()
-OPENROUTER_API_KEY = os.getenv("OPENROUTER_API_KEY")
-
-HEADERS = {
-    "Authorization": f"Bearer {OPENROUTER_API_KEY}",
-    "Content-Type": "application/json",
-}
-MODEL = "anthropic/claude-3-haiku"
-
-# === TTS SETUP ===
-engine = pyttsx3.init()
-engine.setProperty('rate', 180)
-tts_enabled = True
-
-def speak(text):
-    if tts_enabled:
-        try:
-            engine.say(text)
-            engine.runAndWait()
-        except Exception as e:
-            print(f"[TTS Error] {str(e)}")
-
-def set_tts(enabled: bool):
-    global tts_enabled
-    tts_enabled = enabled
-
-# === SESSION SAVE ===
-SAVE_DIR = "chat_logs"
-os.makedirs(SAVE_DIR, exist_ok=True)
-
-def save_chat_session(responses: dict, messages: list, recommendation: str):
-    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
-    filename = f"chat_{timestamp}.json"
-    path = os.path.join(SAVE_DIR, filename)
-
-    session_data = {
-        "timestamp": timestamp,
-        "user_responses": responses,
-        "chat_history": [{"user": u, "bot": b} for u, b in messages],
-        "recommendation": recommendation,
-    }
-
-    with open(path, "w", encoding="utf-8") as f:
-        json.dump(session_data, f, indent=4)
-
-    print(f"✅ Chat session saved to {path}")
-
-# === USER STATE ===
-user_state = {
-    "stage": "intro",
-    "responses": {},
-}
-
-# === QUESTION FLOW ===
-questions = [
-    {
-        "key": "age",
-        "text": "What is your age?",
-        "type": "int",
-        "validation": lambda val: val.isdigit() and 0 < int(val) < 120,
-        "error": "Please enter a valid age between 1 and 120."
-    },
-    {
-        "key": "fever",
-        "text": "Do you have a fever? (yes/no)",
-        "type": "yesno",
-        "validation": lambda val: val.lower() in ["yes", "no"],
-        "error": "Please answer with yes or no."
-    },
-    {
-        "key": "cough",
-        "text": "Are you experiencing cough or chest pain? (yes/no)",
-        "type": "yesno",
-        "validation": lambda val: val.lower() in ["yes", "no"],
-        "error": "Please answer with yes or no."
-    },
-    {
-        "key": "chronic",
-        "text": "Do you have any chronic conditions? (yes/no)",
-        "type": "yesno",
-        "validation": lambda val: val.lower() in ["yes", "no"],
-        "error": "Please answer with yes or no."
-    },
-    {
-        "key": "chills",
-        "text": "Since you have a fever, do you also have chills or body aches? (yes/no)",
-        "depends_on": {"fever": "yes"},
-        "type": "yesno",
-        "validation": lambda val: val.lower() in ["yes", "no"],
-        "error": "Please answer with yes or no."
-    }
-]
-
-def reset_user_state():
-    global user_state
-    user_state = {
-        "stage": "intro",
-        "responses": {},
-    }
-
-def infer_yes_no(user_input):
-    cleaned = user_input.lower().strip()
-    if re.search(r"\b(yes|yeah|yep|i do|i am|sure|of course)\b", cleaned):
-        return "yes"
-    elif re.search(r"\b(no|nah|nope|i don’t|i do not|not really|i am not)\b", cleaned):
-        return "no"
-    return None
-
-def query_openrouter(prompt):
-    messages = [
-        {
-            "role": "system",
-            "content": (
-                "You are a friendly virtual doctor. Respond in second person only. "
-                "Give a short, kind explanation of what the user might be facing. "
-                "Do not repeat already known symptoms. Be medically helpful but clear."
-            )
-        },
-        {"role": "user", "content": prompt}
-    ]
-
-    data = {
-        "model": MODEL,
-        "messages": messages,
-    }
-
-    try:
-        response = requests.post(
-            "https://openrouter.ai/api/v1/chat/completions",
-            headers=HEADERS,
-            json=data,
-            timeout=20
-        )
-        result = response.json()
-        return result["choices"][0]["message"]["content"]
-    except Exception as e:
-        return f"LLM Error: {str(e)}"
-
-def get_next_question():
-    for q in questions:
-        key = q["key"]
-        if key not in user_state["responses"]:
-            if "depends_on" in q:
-                if not all(user_state["responses"].get(k) == v for k, v in q["depends_on"].items()):
-                    continue
-            return q
-    return None
-
-def format_summary_prompt():
-    summary = "\n".join([f"{k}: {v}" for k, v in user_state["responses"].items()])
-    return f"Based on these user symptoms, provide a helpful diagnosis and suggestions:\n{summary}"
-
-def handle_user_message(user_input, chat_history):
-    global user_state
-
-    user_input = user_input.strip()
-
-    if user_input.lower() == "hi" and user_state["stage"] == "done":
-        reset_user_state()
-
-    if user_state["stage"] == "intro":
-        bot_msg = (
-            "👋 Hello! I'm your virtual health assistant. I'll ask a few questions and suggest what to do.\n\n"
-            + questions[0]["text"]
-        )
-        user_state["stage"] = "questions"
-        chat_history.append((user_input, bot_msg))
-        speak(bot_msg)
-        return bot_msg, chat_history
-
-    elif user_state["stage"] == "questions":
-        question = get_next_question()
-        if question:
-            key = question["key"]
-            input_value = user_input.lower()
-
-            # Yes/No intent inference
-            if question["type"] == "yesno":
-                inferred = infer_yes_no(input_value)
-                if inferred:
-                    input_value = inferred
-                else:
-                    bot_msg = question["error"]
-                    chat_history.append((user_input, bot_msg))
-                    speak(bot_msg)
-                    return bot_msg, chat_history
-
-            # Age validation
-            if question["type"] == "int" and not question["validation"](input_value):
-                bot_msg = question["error"]
-                chat_history.append((user_input, bot_msg))
-                speak(bot_msg)
-                return bot_msg, chat_history
-
-            # Save valid input
-            user_state["responses"][key] = int(input_value) if question["type"] == "int" else input_value
-
-            # Red flag → end early
-            if (
-                user_state["responses"].get("fever") == "yes"
-                and user_state["responses"].get("cough") == "yes"
-                and user_state["responses"].get("chills") == "yes"
-            ):
-                prompt = format_summary_prompt()
-                advice = query_openrouter(prompt)
-                user_state["stage"] = "done"
-                chat_history.append((user_input, advice))
-                speak(advice)
-                save_chat_session(user_state["responses"], chat_history, advice)
-                return advice, chat_history
-
-            next_q = get_next_question()
-            if next_q:
-                bot_msg = next_q["text"]
-            else:
-                user_state["stage"] = "done"
-                prompt = format_summary_prompt()
-                bot_msg = query_openrouter(prompt)
-                save_chat_session(user_state["responses"], chat_history, bot_msg)
-
-            chat_history.append((user_input, bot_msg))
-            speak(bot_msg)
-            return bot_msg, chat_history
-
-    elif user_state["stage"] == "done":
-        bot_msg = "✅ You've completed the health check. Refresh or type 'hi' to start over."
-        chat_history.append((user_input, bot_msg))
-        speak(bot_msg)
-        return bot_msg, chat_history
+import os
+import re
+import json
+import pyttsx3
+import requests
+from datetime import datetime
+from dotenv import load_dotenv
+
+# === ENV SETUP ===
+load_dotenv()
+OPENROUTER_API_KEY = os.getenv("OPENROUTER_API_KEY")
+
+HEADERS = {
+    "Authorization": f"Bearer {OPENROUTER_API_KEY}",
+    "Content-Type": "application/json",
+}
+MODEL = "anthropic/claude-3-haiku"
+
+# # === TTS SETUP ===
+# engine = pyttsx3.init()
+# engine.setProperty('rate', 180)
+# tts_enabled = True
+
+# def speak(text):
+#     if tts_enabled:
+#         try:
+#             engine.say(text)
+#             engine.runAndWait()
+#         except Exception as e:
+#             print(f"[TTS Error] {str(e)}")
+
+# def set_tts(enabled: bool):
+#     global tts_enabled
+#     tts_enabled = enabled
+
+# speech_engine.py
+import os
+from gtts import gTTS
+from tempfile import NamedTemporaryFile
+
+tts_enabled = True
+tts_supported = os.getenv("SPACE_ENVIRONMENT") != "huggingface"
+
+def set_tts(enabled: bool):
+    global tts_enabled
+    tts_enabled = enabled
+
+def speak(text: str):
+    if not tts_enabled or not tts_supported:
+        return
+
+    try:
+        tts = gTTS(text)
+        with NamedTemporaryFile(delete=False, suffix=".mp3") as tmp:
+            tts.save(tmp.name)
+            os.system(f"start {tmp.name}" if os.name == 'nt' else f"mpg123 {tmp.name}")
+            os.remove(tmp.name)
+    except Exception as e:
+        print(f"[TTS Error] {str(e)}")
+
+
+# === SESSION SAVE ===
+SAVE_DIR = "chat_logs"
+os.makedirs(SAVE_DIR, exist_ok=True)
+
+def save_chat_session(responses: dict, messages: list, recommendation: str):
+    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+    filename = f"chat_{timestamp}.json"
+    path = os.path.join(SAVE_DIR, filename)
+
+    session_data = {
+        "timestamp": timestamp,
+        "user_responses": responses,
+        "chat_history": [{"user": u, "bot": b} for u, b in messages],
+        "recommendation": recommendation,
+    }
+
+    with open(path, "w", encoding="utf-8") as f:
+        json.dump(session_data, f, indent=4)
+
+    print(f"✅ Chat session saved to {path}")
+
+# === USER STATE ===
+user_state = {
+    "stage": "intro",
+    "responses": {},
+}
+
+# === QUESTION FLOW ===
+questions = [
+    {
+        "key": "age",
+        "text": "What is your age?",
+        "type": "int",
+        "validation": lambda val: val.isdigit() and 0 < int(val) < 120,
+        "error": "Please enter a valid age between 1 and 120."
+    },
+    {
+        "key": "fever",
+        "text": "Do you have a fever? (yes/no)",
+        "type": "yesno",
+        "validation": lambda val: val.lower() in ["yes", "no"],
+        "error": "Please answer with yes or no."
+    },
+    {
+        "key": "cough",
+        "text": "Are you experiencing cough or chest pain? (yes/no)",
+        "type": "yesno",
+        "validation": lambda val: val.lower() in ["yes", "no"],
+        "error": "Please answer with yes or no."
+    },
+    {
+        "key": "chronic",
+        "text": "Do you have any chronic conditions? (yes/no)",
+        "type": "yesno",
+        "validation": lambda val: val.lower() in ["yes", "no"],
+        "error": "Please answer with yes or no."
+    },
+    {
+        "key": "chills",
+        "text": "Since you have a fever, do you also have chills or body aches? (yes/no)",
+        "depends_on": {"fever": "yes"},
+        "type": "yesno",
+        "validation": lambda val: val.lower() in ["yes", "no"],
+        "error": "Please answer with yes or no."
+    }
+]
+
+def reset_user_state():
+    global user_state
+    user_state = {
+        "stage": "intro",
+        "responses": {},
+    }
+
+def infer_yes_no(user_input):
+    cleaned = user_input.lower().strip()
+    if re.search(r"\b(yes|yeah|yep|i do|i am|sure|of course)\b", cleaned):
+        return "yes"
+    elif re.search(r"\b(no|nah|nope|i don’t|i do not|not really|i am not)\b", cleaned):
+        return "no"
+    return None
+
+def query_openrouter(prompt):
+    messages = [
+        {
+            "role": "system",
+            "content": (
+                "You are a friendly virtual doctor. Respond in second person only. "
+                "Give a short, kind explanation of what the user might be facing. "
+                "Do not repeat already known symptoms. Be medically helpful but clear."
+            )
+        },
+        {"role": "user", "content": prompt}
+    ]
+
+    data = {
+        "model": MODEL,
+        "messages": messages,
+    }
+
+    try:
+        response = requests.post(
+            "https://openrouter.ai/api/v1/chat/completions",
+            headers=HEADERS,
+            json=data,
+            timeout=20
+        )
+        result = response.json()
+        return result["choices"][0]["message"]["content"]
+    except Exception as e:
+        return f"LLM Error: {str(e)}"
+
+def get_next_question():
+    for q in questions:
+        key = q["key"]
+        if key not in user_state["responses"]:
+            if "depends_on" in q:
+                if not all(user_state["responses"].get(k) == v for k, v in q["depends_on"].items()):
+                    continue
+            return q
+    return None
+
+def format_summary_prompt():
+    summary = "\n".join([f"{k}: {v}" for k, v in user_state["responses"].items()])
+    return f"Based on these user symptoms, provide a helpful diagnosis and suggestions:\n{summary}"
+
+def handle_user_message(user_input, chat_history):
+    global user_state
+
+    user_input = user_input.strip()
+
+    if user_input.lower() == "hi" and user_state["stage"] == "done":
+        reset_user_state()
+
+    if user_state["stage"] == "intro":
+        bot_msg = (
+            "👋 Hello! I'm your virtual health assistant. I'll ask a few questions and suggest what to do.\n\n"
+            + questions[0]["text"]
+        )
+        user_state["stage"] = "questions"
+        chat_history.append((user_input, bot_msg))
+        speak(bot_msg)
+        return bot_msg, chat_history
+
+    elif user_state["stage"] == "questions":
+        question = get_next_question()
+        if question:
+            key = question["key"]
+            input_value = user_input.lower()
+
+            # Yes/No intent inference
+            if question["type"] == "yesno":
+                inferred = infer_yes_no(input_value)
+                if inferred:
+                    input_value = inferred
+                else:
+                    bot_msg = question["error"]
+                    chat_history.append((user_input, bot_msg))
+                    speak(bot_msg)
+                    return bot_msg, chat_history
+
+            # Age validation
+            if question["type"] == "int" and not question["validation"](input_value):
+                bot_msg = question["error"]
+                chat_history.append((user_input, bot_msg))
+                speak(bot_msg)
+                return bot_msg, chat_history
+
+            # Save valid input
+            user_state["responses"][key] = int(input_value) if question["type"] == "int" else input_value
+
+            # Red flag → end early
+            if (
+                user_state["responses"].get("fever") == "yes"
+                and user_state["responses"].get("cough") == "yes"
+                and user_state["responses"].get("chills") == "yes"
+            ):
+                prompt = format_summary_prompt()
+                advice = query_openrouter(prompt)
+                user_state["stage"] = "done"
+                chat_history.append((user_input, advice))
+                speak(advice)
+                save_chat_session(user_state["responses"], chat_history, advice)
+                return advice, chat_history
+
+            next_q = get_next_question()
+            if next_q:
+                bot_msg = next_q["text"]
+            else:
+                user_state["stage"] = "done"
+                prompt = format_summary_prompt()
+                bot_msg = query_openrouter(prompt)
+                save_chat_session(user_state["responses"], chat_history, bot_msg)
+
+            chat_history.append((user_input, bot_msg))
+            speak(bot_msg)
+            return bot_msg, chat_history
+
+    elif user_state["stage"] == "done":
+        bot_msg = "✅ You've completed the health check. Refresh or type 'hi' to start over."
+        chat_history.append((user_input, bot_msg))
+        speak(bot_msg)
+        return bot_msg, chat_history