app.py

# app.py - Complete Dual-Mode Healthcare Analysis System
import os, re, json, traceback, pathlib
from functools import lru_cache
from typing import List, Dict, Any, Tuple, Optional
import pandas as pd
import numpy as np

import gradio as gr
import torch
import regex as re2

# Import necessary modules
from settings import (
    SNAPSHOT_PATH, PERSIST_CONTENT, HEALTHCARE_SETTINGS, MODEL_SETTINGS,
    HEALTHCARE_SYSTEM_PROMPT, GENERAL_CONVERSATION_PROMPT
)
from audit_log import log_event, hash_summary
from privacy import redact_text, safety_filter, refusal_reply
from data_registry import DataRegistry
from upload_ingest import extract_text_from_files
from healthcare_analysis import HealthcareAnalyzer
from response_formatter import ResponseFormatter

# ---------- Writable caches (HF Spaces-safe) ----------
HOME = pathlib.Path.home()
HF_HOME = str(HOME / ".cache" / "huggingface")
HF_HUB_CACHE = str(HOME / ".cache" / "huggingface" / "hub")
HF_TRANSFORMERS = str(HOME / ".cache" / "huggingface" / "transformers")
ST_HOME = str(HOME / ".cache" / "sentence-transformers")
GRADIO_TMP = str(HOME / "app" / "gradio")
GRADIO_CACHE = GRADIO_TMP

os.environ.setdefault("HF_HOME", HF_HOME)
os.environ.setdefault("HF_HUB_CACHE", HF_HUB_CACHE)
os.environ.setdefault("TRANSFORMERS_CACHE", HF_TRANSFORMERS)
os.environ.setdefault("SENTENCE_TRANSFORMERS_HOME", ST_HOME)
os.environ.setdefault("GRADIO_TEMP_DIR", GRADIO_TMP)
os.environ.setdefault("GRADIO_CACHE_DIR", GRADIO_CACHE)
os.environ.setdefault("HF_HUB_ENABLE_XET", "0")
os.environ.setdefault("HF_HUB_ENABLE_HF_TRANSFER", "1")

for p in [HF_HOME, HF_HUB_CACHE, HF_TRANSFORMERS, ST_HOME, GRADIO_TMP, GRADIO_CACHE]:
    try:
        os.makedirs(p, exist_ok=True)
    except Exception:
        pass

# Optional Cohere
try:
    import cohere
    _HAS_COHERE = True
except Exception:
    _HAS_COHERE = False

from transformers import AutoTokenizer, AutoModelForCausalLM
from huggingface_hub import login

# ---------- Config ----------
MODEL_ID = os.getenv("MODEL_ID", "microsoft/Phi-3-mini-4k-instruct")
HF_TOKEN = os.getenv("HUGGINGFACE_HUB_TOKEN") or os.getenv("HF_TOKEN")
COHERE_API_KEY = os.getenv("COHERE_API_KEY")
USE_HOSTED_COHERE = bool(COHERE_API_KEY and _HAS_COHERE)
MAX_NEW_TOKENS = int(os.getenv("MAX_NEW_TOKENS", MODEL_SETTINGS.get("max_new_tokens", 2048)))

# ---------- Helper Functions ----------
def find_column(df, patterns):
    """Find the first column in df that matches any of the patterns."""
    if df is None or df.empty:
        return None
    for col in df.columns:
        if any(pattern.lower() in col.lower() for pattern in patterns):
            return col
    return None

def extract_scenario_tasks(scenario_text):
    """Extract specific tasks from scenario text."""
    tasks = []
    lines = scenario_text.split('\n')
    in_tasks = False
    for line in lines:
        line = line.strip()
        if line.lower().startswith('tasks'):
            in_tasks = True
            continue
        if in_tasks:
            if line.lower().startswith('operational recommendations') or line.lower().startswith('future integration'):
                in_tasks = False
                continue
            if line and (line.startswith(('1.', '2.', '3.', '4.', '5.')) or line.startswith(('•', '-', '*'))):
                tasks.append(line)
    return tasks

# ---------- Session RAG Class ----------
class SessionRAG:
    def __init__(self):
        self.docs = []
        self.artifacts = []
        self.csv_columns = []
    
    def add_docs(self, chunks):
        self.docs.extend(chunks)
    
    def register_artifacts(self, artifacts):
        self.artifacts.extend(artifacts)
    
    def get_latest_csv_columns(self):
        return self.csv_columns
    
    def retrieve(self, query, k=5):
        return self.docs[:k] if self.docs else []
    
    def clear(self):
        self.docs.clear()
        self.artifacts.clear()
        self.csv_columns.clear()

# ---------- Healthcare-specific functions ----------
def is_healthcare_scenario(text: str, uploaded_files_paths) -> bool:
    """Detect if this is a healthcare scenario with specific indicators."""
    t = (text or "").lower()
    
    # Check for healthcare keywords
    has_healthcare_keywords = any(keyword in t for keyword in HEALTHCARE_SETTINGS["healthcare_keywords"])
    
    # Check for healthcare facility types
    has_facility_types = (
        any(ftype in t for ftype in ["hospital", "medical center", "health centre"]) or
        any(ftype in t for ftype in ["nursing", "residential", "care facility", "long-term care"]) or
        any(ftype in t for ftype in ["ambulatory", "clinic", "surgery center", "outpatient"])
    )
    
    # Check for healthcare-specific tasks
    has_healthcare_tasks = any(
        phrase in t for phrase in [
            "bed capacity", "occupancy rates", "facility distribution",
            "long-term care", "health operations", "resource allocation"
        ]
    )
    
    # Check for healthcare data files
    has_healthcare_files = any(
        "health" in path.lower() or "facility" in path.lower() or "bed" in path.lower()
        for path in uploaded_files_paths
    )
    
    # Check for structured scenario format
    has_scenario_structure = any(
        section in t for section in ["background", "situation", "tasks"]
    )
    
    return (has_healthcare_keywords or has_facility_types or has_healthcare_tasks) and \
           (has_healthcare_files or has_scenario_structure)

def is_general_conversation(text: str, uploaded_files_paths) -> bool:
    """Determine if this is a general conversation rather than a scenario analysis."""
    # If there are uploaded files, it's likely a scenario
    if uploaded_files_paths:
        return False
    
    # Check for scenario indicators
    scenario_indicators = [
        "scenario", "analyze", "analysis", "assess", "evaluate", "recommend",
        "tasks", "background", "situation", "dataset", "data"
    ]
    
    # If no scenario indicators, it's likely general conversation
    text_lower = text.lower()
    return not any(indicator in text_lower for indicator in scenario_indicators)

def process_healthcare_data(uploaded_files_paths, data_registry):
    """Process healthcare data files with robust error handling."""
    for file_path in uploaded_files_paths:
        try:
            if data_registry.add_path(file_path):
                print(f"Successfully processed: {file_path}")
            else:
                print(f"Failed to process: {file_path}")
        except Exception as e:
            print(f"Error processing {file_path}: {e}")
            log_event("data_processing_error", None, {
                "file": file_path,
                "error": str(e)
            })

def handle_healthcare_scenario(scenario_text, data_registry, history):
    """Handle healthcare scenarios with enhanced analysis"""
    try:
        # Initialize analyzer
        analyzer = HealthcareAnalyzer(data_registry)
        
        # Perform comprehensive analysis
        results = analyzer.comprehensive_analysis(scenario_text)
        
        # Format response
        formatter = ResponseFormatter()
        response = formatter.format_healthcare_response(scenario_text, results)
        
        return response
    except Exception as e:
        log_event("healthcare_scenario_error", None, {"error": str(e)})
        # Log the full traceback for better debugging
        import traceback
        tb_str = traceback.format_exc()
        log_event("healthcare_scenario_traceback", None, {"traceback": tb_str})
        return f"Error analyzing healthcare scenario: {str(e)}\n\nTechnical details:\n{tb_str}"

# ---------- Model loading helpers ----------
def pick_dtype_and_map():
    if torch.cuda.is_available():
        return torch.float16, "auto"
    if torch.backends.mps.is_available():
        return torch.float16, {"": "mps"}
    return torch.float32, "cpu"

@lru_cache(maxsize=1)
def load_local_model():
    if not HF_TOKEN:
        raise RuntimeError("HUGGINGFACE_HUB_TOKEN is not set.")
    login(token=HF_TOKEN, add_to_git_credential=False)
    dtype, device_map = pick_dtype_and_map()
    tok = AutoTokenizer.from_pretrained(
        MODEL_ID, token=HF_TOKEN, use_fast=True, model_max_length=8192,
        padding_side="left", trust_remote_code=True,
        cache_dir=os.environ.get("TRANSFORMERS_CACHE")
    )
    try:
        mdl = AutoModelForCausalLM.from_pretrained(
            MODEL_ID, token=HF_TOKEN, device_map=device_map,
            low_cpu_mem_usage=True, torch_dtype=dtype, trust_remote_code=True,
            cache_dir=os.environ.get("TRANSFORMERS_CACHE")
        )
    except Exception:
        mdl = AutoModelForCausalLM.from_pretrained(
            MODEL_ID, token=HF_TOKEN,
            low_cpu_mem_usage=True, torch_dtype=dtype, trust_remote_code=True,
            cache_dir=os.environ.get("TRANSFORMERS_CACHE")
        )
        mdl.to("cuda" if torch.cuda.is_available() else "cpu")
    if mdl.config.eos_token_id is None and tok.eos_token_id is not None:
        mdl.config.eos_token_id = tok.eos_token_id
    return mdl, tok

# ---------- Chat helpers ----------
def is_identity_query(message, history):
    patterns = [
        r"\bwho\s+are\s+you\b", r"\bwhat\s+are\s+you\b", r"\bwhat\s+is\s+your\s+name\b",
        r"\bwho\s+is\s+this\b", r"\bidentify\s+yourself\b", r"\btell\s+me\s+about\s+yourself\b",
        r"\bdescribe\s+yourself\b", r"\band\s+you\s*\?\b", r"\byour\s+name\b",
        r"\bwho\s+am\s+i\s+chatting\s+with\b",
    ]
    def match(t): return any(re.search(p, (t or "").strip().lower()) for p in patterns)
    if match(message): return True
    if history:
        last_user = history[-1][0] if isinstance(history[-1], (list, tuple)) else None
        if match(last_user): return True
    return False

def _iter_user_assistant(history):
    for item in (history or []):
        if isinstance(item, (list, tuple)):
            u = item[0] if len(item) > 0 else ""
            a = item[1] if len(item) > 1 else ""
            yield u, a

def _sanitize_text(s: str) -> str:
    if not isinstance(s, str):
        return s
    return re2.sub(r'[\p{C}--[\n\t]]+', '', s)

def cohere_chat(message, history):
    if not USE_HOSTED_COHERE:
        return None
    try:
        client = cohere.Client(api_key=COHERE_API_KEY)
        parts = []
        for u, a in _iter_user_assistant(history):
            if u: parts.append(f"User: {u}")
            if a: parts.append(f"Assistant: {a}")
        parts.append(f"User: {message}")
        prompt = "\n".join(parts) + "\nAssistant:"
        resp = client.chat(
            model="command-r7b-12-2024",
            message=prompt,
            temperature=MODEL_SETTINGS.get("temperature", 0.3),
            max_tokens=MAX_NEW_TOKENS,
        )
        if hasattr(resp, "text") and resp.text: return resp.text.strip()
        if hasattr(resp, "reply") and resp.reply: return resp.reply.strip()
        if hasattr(resp, "generations") and resp.generations: return resp.generations[0].text.strip()
        return None
    except Exception:
        return None

def build_inputs(tokenizer, message, history, system_prompt):
    msgs = [{"role": "system", "content": system_prompt}]
    for u, a in _iter_user_assistant(history):
        if u: msgs.append({"role": "user", "content": u})
        if a: msgs.append({"role": "assistant", "content": a})
    msgs.append({"role": "user", "content": message})
    return tokenizer.apply_chat_template(
        msgs, tokenize=True, add_generation_prompt=True, return_tensors="pt"
    )

def local_generate(model, tokenizer, input_ids, max_new_tokens=MAX_NEW_TOKENS):
    input_ids = input_ids.to(model.device)
    with torch.no_grad():
        out = model.generate(
            input_ids=input_ids, max_new_tokens=max_new_tokens,
            do_sample=True, temperature=MODEL_SETTINGS.get("temperature", 0.3),
            top_p=MODEL_SETTINGS.get("top_p", 0.9),
            repetition_penalty=MODEL_SETTINGS.get("repetition_penalty", 1.15),
            pad_token_id=tokenizer.eos_token_id,
            eos_token_id=tokenizer.eos_token_id,
        )
    gen_only = out[0, input_ids.shape[-1]:]
    return tokenizer.decode(gen_only, skip_special_tokens=True).strip()

# ---------- Core chat logic ----------
def clarityops_reply(user_msg, history, tz, uploaded_files_paths, awaiting_answers=False):
    try:
        log_event("user_message", None, {"sizes": {"chars": len(user_msg or "")}})

        safe_in, blocked_in, reason_in = safety_filter(user_msg, mode="input")
        if blocked_in:
            ans = refusal_reply(reason_in)
            return history + [(user_msg, ans)], awaiting_answers

        if is_identity_query(safe_in, history):
            ans = "I am an AI analytical system designed to help with both general conversations and healthcare scenario analysis. I can answer your questions and also analyze healthcare data when you upload files and describe a scenario."
            return history + [(user_msg, ans)], awaiting_answers

        # Initialize data registry and session RAG
        data_registry = DataRegistry()
        session_rag = SessionRAG()

        # Process uploaded files if any
        if uploaded_files_paths:
            process_healthcare_data(uploaded_files_paths, data_registry)
            
            # Also extract text for RAG
            ing = extract_text_from_files(uploaded_files_paths)
            if ing.get("chunks"):
                session_rag.add_docs(ing["chunks"])
            if ing.get("artifacts"):
                session_rag.register_artifacts(ing["artifacts"])
            
            # Update session RAG with CSV columns
            for file_name in data_registry.names():
                if file_name.endswith('.csv'):
                    df = data_registry.get(file_name)
                    session_rag.csv_columns = list(df.columns)

        # Determine the mode: healthcare scenario or general conversation
        if is_healthcare_scenario(safe_in, uploaded_files_paths):
            # Healthcare scenario mode
            response = handle_healthcare_scenario(safe_in, data_registry, history)
            return history + [(user_msg, response)], False
        else:
            # General conversation mode with enhanced handling
            if USE_HOSTED_COHERE:
                out = cohere_chat(safe_in, history)
                if out:
                    out = _sanitize_text(out)
                    safe_out, blocked_out, reason_out = safety_filter(out, mode="output")
                    if blocked_out:
                        safe_out = refusal_reply(reason_out)
                    log_event("assistant_reply", None, {
                        **hash_summary("prompt", safe_in if not PERSIST_CONTENT else ""),
                        **hash_summary("reply", safe_out if not PERSIST_CONTENT else ""),
                        "mode": "general_cohere",
                    })
                    return history + [(user_msg, safe_out)], False
            
            # Enhanced local model generation
            try:
                model, tokenizer = load_local_model()
                
                # Use general conversation prompt
                inputs = build_inputs(tokenizer, safe_in, history, GENERAL_CONVERSATION_PROMPT)
                out = local_generate(model, tokenizer, inputs, max_new_tokens=MAX_NEW_TOKENS)
                
                if isinstance(out, str):
                    for tag in ("Assistant:", "System:", "User:"):
                        if out.startswith(tag):
                            out = out[len(tag):].strip()
                
                out = _sanitize_text(out or "")
                safe_out, blocked_out, reason_out = safety_filter(out, mode="output")
                if blocked_out:
                    safe_out = refusal_reply(reason_out)
                
                log_event("assistant_reply", None, {
                    **hash_summary("prompt", safe_in if not PERSIST_CONTENT else ""),
                    **hash_summary("reply", safe_out if not PERSIST_CONTENT else ""),
                    "mode": "general_local",
                })
                
                return history + [(user_msg, safe_out)], False
            except Exception as e:
                err = f"Error generating response: {str(e)}"
                log_event("model_error", None, {"error": str(e)})
                return history + [(user_msg, err)], False

    except Exception as e:
        err = f"Error: {e}"
        try:
            traceback.print_exc()
        except Exception:
            pass
        return history + [(user_msg, err)], awaiting_answers

# ---------- UI Setup ----------
theme = gr.themes.Soft(primary_hue="teal", neutral_hue="slate", radius_size=gr.themes.sizes.radius_lg)
custom_css = """
:root { --brand-bg: #0f172a; --brand-accent: #0d9488; --brand-text: #0f172a; --brand-text-light: #ffffff; }
html, body, .gradio-container { height: 100vh; }
.gradio-container { background: var(--brand-bg); display: flex; flex-direction: column; }
/* HERO (landing) */
#hero-wrap { height: 70vh; display: grid; place-items: center; }
#hero { text-align: center; }
#hero h2 { color: #0f172a; font-weight: 800; font-size: 32px; margin-bottom: 22px; }
#hero .search-row { width: min(860px, 92vw); margin: 0 auto; display: flex; gap: 8px; align-items: stretch; }
#hero .search-row .hero-box { flex: 1 1 auto; }
#hero .search-row .hero-box textarea { height: 52px !important; }
#hero-send > button { height: 52px !important; padding: 0 18px !important; border-radius: 12px !important; }
#hero .hint { color: #334155; margin-top: 10px; font-size: 13px; opacity: 0.9; }
/* CHAT */
#chat-container { position: relative; }
.chatbot header, .chatbot .label, .chatbot .label-wrap { display: none !important; }
.message.user, .message.bot { background: var(--brand-accent) !important; color: var(--brand-text-light) !important; border-radius: 12px !important; padding: 8px 12px !important; }
textarea, input, .gr-input { border-radius: 12px !important; }
/* Chat input row equal heights */
#chat-input-row { align-items: stretch; }
#chat-msg textarea { height: 52px !important; }
#chat-send > button, #chat-clear > button { height: 52px !important; padding: 0 18px !important; border-radius: 12px !important; }
"""

# ---------- Main App ----------
with gr.Blocks(theme=theme, css=custom_css, analytics_enabled=False) as demo:
    # --- HERO (initial screen) ---
    with gr.Column(elem_id="hero-wrap", visible=True) as hero_wrap:
        with gr.Column(elem_id="hero"):
            gr.HTML("<h2>How can I help you today?</h2>")
            with gr.Row(elem_classes="search-row"):
                hero_msg = gr.Textbox(
                    placeholder="Ask me anything or upload healthcare data files for scenario analysis…",
                    show_label=False,
                    lines=1,
                    elem_classes="hero-box"
                )
                hero_send = gr.Button("➤", scale=0, elem_id="hero-send")
            gr.Markdown('<div class="hint">I can help with general questions or analyze healthcare scenarios when you upload data files and describe your analysis needs.</div>')

    # --- MAIN APP (hidden until first message) ---
    with gr.Column(elem_id="chat-container", visible=False) as app_wrap:
        chat = gr.Chatbot(label="", show_label=False, height="80vh")
        with gr.Row():
            uploads = gr.Files(
                label="Upload healthcare data files",
                file_types=HEALTHCARE_SETTINGS["supported_file_types"],
                file_count="multiple", height=68
            )
        with gr.Row(elem_id="chat-input-row"):
            msg = gr.Textbox(
                label="",
                show_label=False,
                placeholder="Ask me anything or continue your healthcare scenario analysis…",
                scale=10,
                elem_id="chat-msg",
                lines=1,
            )
            send = gr.Button("Send", scale=1, elem_id="chat-send")
            clear = gr.Button("Clear chat", scale=1, elem_id="chat-clear")

    # ---- State
    state_history = gr.State(value=[])
    state_uploaded = gr.State(value=[])
    state_awaiting = gr.State(value=False)

    # ---- Uploads
    def _store_uploads(files, current):
        paths = []
        for f in (files or []):
            paths.append(getattr(f, "name", None) or f)
        return (current or []) + paths

    uploads.change(fn=_store_uploads, inputs=[uploads, state_uploaded], outputs=state_uploaded)

    # ---- Core send (used by both hero input and chat input)
    def _on_send(user_msg, history, up_paths, awaiting):
        try:
            if not user_msg or not user_msg.strip():
                return history, "", history, awaiting
            new_history, new_awaiting = clarityops_reply(
                user_msg.strip(), history or [], None, up_paths or [], awaiting_answers=awaiting
            )
            return new_history, "", new_history, new_awaiting
        except Exception as e:
            err = f"Error: {e}"
            try: traceback.print_exc()
            except Exception: pass
            new_hist = (history or []) + [(user_msg or "", err)]
            return new_hist, "", new_hist, awaiting

    # ---- Hero -> App transition + first send
    def _hero_start(user_msg, history, up_paths, awaiting):
        chat_o, msg_o, hist_o, await_o = _on_send(user_msg, history, up_paths, awaiting)
        return (
            chat_o, msg_o, hist_o, await_o,
            gr.update(visible=False),
            gr.update(visible=True),
            ""
        )

    hero_send.click(
        _hero_start,
        inputs=[hero_msg, state_history, state_uploaded, state_awaiting],
        outputs=[chat, msg, state_history, state_awaiting, hero_wrap, app_wrap, hero_msg],
        concurrency_limit=2, queue=True
    )
    hero_msg.submit(
        _hero_start,
        inputs=[hero_msg, state_history, state_uploaded, state_awaiting],
        outputs=[chat, msg, state_history, state_awaiting, hero_wrap, app_wrap, hero_msg],
        concurrency_limit=2, queue=True
    )

    # ---- Normal chat interactions after hero is gone
    send.click(_on_send, inputs=[msg, state_history, state_uploaded, state_awaiting],
               outputs=[chat, msg, state_history, state_awaiting],
               concurrency_limit=2, queue=True)
    msg.submit(_on_send, inputs=[msg, state_history, state_uploaded, state_awaiting],
               outputs=[chat, msg, state_history, state_awaiting],
               concurrency_limit=2, queue=True)

    def _on_clear():
        return (
            [], "", [], False,
            gr.update(visible=True),
            gr.update(visible=False),
            ""
        )

    clear.click(_on_clear, None, [chat, msg, state_history, state_awaiting, hero_wrap, app_wrap, hero_msg])

if __name__ == "__main__":
    port = int(os.environ.get("PORT", "7860"))
    demo.launch(server_name="0.0.0.0", server_port=port, show_api=False, max_threads=40)