core/data_manager.py

"""
Data Manager — CRUD operations backed by Google Sheets.
Refactored for FastAPI: module-level caching instead of streamlit.session_state.
"""
import json
import base64
import os
import time
import threading
from datetime import datetime

import gspread
from google.oauth2 import service_account


def _get_secret(key: str) -> str:
    return os.environ.get(key)


# --- Module-level caching ---
# RLock (reentrant): some helpers acquire the lock while calling another helper
# that also acquires it (e.g. _get_worksheet -> _get_spreadsheet). A plain Lock
# would deadlock.
_lock = threading.RLock()
_state: dict = {
    "spreadsheet": None,
    "worksheets": {},
    "headers": {},
    "read_cache": {},  # name -> (timestamp, list[dict])
}
_READ_TTL_SECONDS = 60


def _get_spreadsheet() -> gspread.Spreadsheet:
    with _lock:
        if _state["spreadsheet"] is None:
            creds_b64 = _get_secret("GOOGLE_CALENDAR_CREDENTIALS")
            creds_json = json.loads(base64.b64decode(creds_b64).decode("utf-8"))
            credentials = service_account.Credentials.from_service_account_info(
                creds_json,
                scopes=[
                    "https://www.googleapis.com/auth/spreadsheets",
                    "https://www.googleapis.com/auth/drive",
                ],
            )
            client = gspread.authorize(credentials)
            sheet_id = _get_secret("GOOGLE_SHEET_ID")
            _state["spreadsheet"] = client.open_by_key(sheet_id)
        return _state["spreadsheet"]


def _get_worksheet(name: str) -> gspread.Worksheet:
    with _lock:
        if name not in _state["worksheets"]:
            _state["worksheets"][name] = _get_spreadsheet().worksheet(name)
        return _state["worksheets"][name]


_JSON_FIELDS = {
    "work_hours", "lunch_break", "days_off", "recurring_days_off", "service_levels",
    "on_call",
}


def _serialize_row(data: dict) -> dict:
    row = {}
    for k, v in data.items():
        if k in _JSON_FIELDS:
            row[k] = json.dumps(v)
        elif v is None:
            row[k] = ""
        elif isinstance(v, bool):
            row[k] = str(v)
        else:
            row[k] = str(v)
    return row


def _deserialize_row(row: dict) -> dict:
    data = {}
    for k, v in row.items():
        if k in _JSON_FIELDS:
            try:
                data[k] = json.loads(v) if v else []
            except (json.JSONDecodeError, TypeError):
                data[k] = []
        elif k in ("lat", "lng"):
            if v == "" or v is None:
                data[k] = None
            else:
                try:
                    data[k] = float(v)
                except (ValueError, TypeError):
                    data[k] = None
        elif k == "active":
            data[k] = v.lower() == "true" if isinstance(v, str) else bool(v)
        elif k == "visit_level":
            try:
                data[k] = int(v)
            except (ValueError, TypeError):
                data[k] = v
        else:
            data[k] = v
    return data


def _read_all(worksheet_name: str) -> list[dict]:
    with _lock:
        cached = _state["read_cache"].get(worksheet_name)
        if cached and (time.time() - cached[0]) < _READ_TTL_SECONDS:
            return list(cached[1])
    ws = _get_worksheet(worksheet_name)
    records = ws.get_all_records()
    data = [_deserialize_row(r) for r in records]
    with _lock:
        _state["read_cache"][worksheet_name] = (time.time(), data)
    return list(data)


def _invalidate_cache(worksheet_name: str) -> None:
    with _lock:
        _state["read_cache"].pop(worksheet_name, None)


def _get_headers(worksheet_name: str) -> list[str]:
    with _lock:
        if worksheet_name in _state["headers"]:
            return list(_state["headers"][worksheet_name])
    ws = _get_worksheet(worksheet_name)
    headers = ws.row_values(1)
    with _lock:
        _state["headers"][worksheet_name] = headers
    return list(headers)


def _col_letter(n: int) -> str:
    result = ""
    while n > 0:
        n, remainder = divmod(n - 1, 26)
        result = chr(65 + remainder) + result
    return result


def _ensure_columns(worksheet_name: str, required: list[str]) -> list[str]:
    headers = _get_headers(worksheet_name)
    missing = [c for c in required if c not in headers]
    if not missing:
        return headers
    ws = _get_worksheet(worksheet_name)
    new_headers = headers + missing
    end_col = _col_letter(len(new_headers))
    ws.update(values=[new_headers], range_name=f"A1:{end_col}1", value_input_option="RAW")
    with _lock:
        _state["headers"][worksheet_name] = new_headers
    _invalidate_cache(worksheet_name)
    return new_headers


def _find_row_index(worksheet_name: str, id_column: str, id_value: str) -> int | None:
    records = _read_all(worksheet_name)
    for i, record in enumerate(records):
        if record.get(id_column) == id_value:
            return i + 2
    return None


def _append_row(worksheet_name: str, data: dict) -> None:
    ws = _get_worksheet(worksheet_name)
    headers = _get_headers(worksheet_name)
    row_values = [_serialize_row(data).get(h, "") for h in headers]
    ws.append_row(row_values, value_input_option="RAW")
    _invalidate_cache(worksheet_name)


def _update_row(worksheet_name: str, row_index: int, data: dict) -> None:
    ws = _get_worksheet(worksheet_name)
    headers = _get_headers(worksheet_name)
    serialized = _serialize_row(data)
    row_values = [serialized.get(h, "") for h in headers]
    end_col = _col_letter(len(headers))
    ws.update(values=[row_values], range_name=f"A{row_index}:{end_col}{row_index}", value_input_option="RAW")
    _invalidate_cache(worksheet_name)


def _generate_id(prefix: str, existing_ids: list[str]) -> str:
    if not existing_ids:
        return f"{prefix}_001"
    nums = []
    for eid in existing_ids:
        parts = eid.split("_")
        if len(parts) >= 2:
            try:
                nums.append(int(parts[-1]))
            except ValueError:
                pass
    next_num = max(nums, default=0) + 1
    return f"{prefix}_{next_num:03d}"


# --- Providers ---

def get_all_providers(active_only: bool = True) -> list[dict]:
    providers = _read_all("Providers")
    if active_only:
        return [p for p in providers if p.get("active", True)]
    return providers


def get_provider_by_id(provider_id: str) -> dict | None:
    for p in _read_all("Providers"):
        if p.get("id") == provider_id:
            return p
    return None


def get_provider_by_name(name: str) -> dict | None:
    for p in _read_all("Providers"):
        if p.get("name", "").lower() == name.lower():
            return p
    return None


def add_provider(provider_data: dict) -> dict:
    _ensure_columns("Providers", ["on_call"])
    providers = _read_all("Providers")
    existing_ids = [p["id"] for p in providers if "id" in p]
    provider_data["id"] = _generate_id("prov", existing_ids)
    provider_data.setdefault("active", True)
    provider_data.setdefault("days_off", [])
    provider_data.setdefault("recurring_days_off", [])
    provider_data.setdefault("on_call", [])
    provider_data.setdefault("lat", None)
    provider_data.setdefault("lng", None)
    _append_row("Providers", provider_data)
    return provider_data


def update_provider(provider_id: str, updates: dict) -> dict | None:
    if "on_call" in updates:
        _ensure_columns("Providers", ["on_call"])
    row_idx = _find_row_index("Providers", "id", provider_id)
    if row_idx is None:
        return None
    providers = _read_all("Providers")
    for p in providers:
        if p.get("id") == provider_id:
            p.update(updates)
            _update_row("Providers", row_idx, p)
            return p
    return None


def delete_provider(provider_id: str) -> bool:
    row_idx = _find_row_index("Providers", "id", provider_id)
    if row_idx is None:
        return False
    ws = _get_worksheet("Providers")
    ws.delete_rows(row_idx)
    _invalidate_cache("Providers")
    return True


# --- Patients ---

def get_all_patients() -> list[dict]:
    return _read_all("Patients")


def get_patient_by_id(patient_id: str) -> dict | None:
    for p in _read_all("Patients"):
        if p.get("id") == patient_id:
            return p
    return None


def add_patient(patient_data: dict) -> dict:
    patients = _read_all("Patients")
    existing_ids = [p["id"] for p in patients if "id" in p]
    patient_data["id"] = _generate_id("pat", existing_ids)
    patient_data.setdefault("lat", None)
    patient_data.setdefault("lng", None)
    patient_data.setdefault("notes", "")
    _append_row("Patients", patient_data)
    return patient_data


def update_patient(patient_id: str, updates: dict) -> dict | None:
    row_idx = _find_row_index("Patients", "id", patient_id)
    if row_idx is None:
        return None
    patients = _read_all("Patients")
    for p in patients:
        if p.get("id") == patient_id:
            p.update(updates)
            _update_row("Patients", row_idx, p)
            return p
    return None


def delete_patient(patient_id: str) -> bool:
    row_idx = _find_row_index("Patients", "id", patient_id)
    if row_idx is None:
        return False
    ws = _get_worksheet("Patients")
    ws.delete_rows(row_idx)
    _invalidate_cache("Patients")
    return True


# --- Appointments ---

def get_all_appointments(status: str | None = None) -> list[dict]:
    appointments = _read_all("Appointments")
    if status:
        return [a for a in appointments if a.get("status") == status]
    return appointments


def get_appointments_for_provider(provider_id: str, date: str | None = None) -> list[dict]:
    appointments = _read_all("Appointments")
    result = [a for a in appointments if a.get("provider_id") == provider_id and a.get("status") == "scheduled"]
    if date:
        result = [a for a in result if a.get("date") == date]
    return result


def get_appointments_for_patient(patient_id: str) -> list[dict]:
    appointments = _read_all("Appointments")
    return [a for a in appointments if a.get("patient_id") == patient_id and a.get("status") == "scheduled"]


def add_appointment(appointment_data: dict) -> dict:
    appointments = _read_all("Appointments")
    existing_ids = [a["id"] for a in appointments if "id" in a]
    appointment_data["id"] = _generate_id("appt", existing_ids)
    appointment_data["created_at"] = datetime.now().isoformat()
    appointment_data.setdefault("status", "scheduled")
    _append_row("Appointments", appointment_data)
    return appointment_data


def update_appointment(appointment_id: str, updates: dict) -> dict | None:
    row_idx = _find_row_index("Appointments", "id", appointment_id)
    if row_idx is None:
        return None
    appointments = _read_all("Appointments")
    for a in appointments:
        if a.get("id") == appointment_id:
            a.update(updates)
            _update_row("Appointments", row_idx, a)
            return a
    return None


def cancel_appointment(appointment_id: str) -> dict | None:
    return update_appointment(appointment_id, {"status": "cancelled", "cancelled_at": datetime.now().isoformat()})


def complete_appointment(appointment_id: str) -> dict | None:
    return update_appointment(appointment_id, {"status": "completed", "completed_at": datetime.now().isoformat()})


# --- Audit Log ---

def log_action(user: str, action: str, details: str, appointment_id: str | None = None) -> None:
    entry = {
        "timestamp": datetime.now().isoformat(),
        "user": user,
        "action": action,
        "details": details,
        "appointment_id": appointment_id or "",
    }
    _append_row("AuditLog", entry)


def get_audit_log(limit: int = 100) -> list[dict]:
    log = _read_all("AuditLog")
    log.sort(key=lambda x: x.get("timestamp", ""), reverse=True)
    return log[:limit]