"""Canonical data contracts.

Bars: `timestamp_utc` marks the **bar open** (vendor convention, enforced). `session_date` is the trading
session date in the exchange timezone (America/New_York for US listings). Volumes are float64 because vendor
adjusted volumes may be fractional; raw volumes are integral and validated as such.
"""
from __future__ import annotations

import json
from enum import StrEnum

import pyarrow as pa

BARS_SCHEMA_VERSION = 1


class AssetClass(StrEnum):
    stock = "stock"
    etf = "etf"
    futures = "futures"
    index = "index"
    fx = "fx"
    crypto = "crypto"
    options = "options"


class Interval(StrEnum):
    m1 = "1m"
    m5 = "5m"
    m30 = "30m"
    h1 = "1h"
    d1 = "1d"


INTERVAL_SECONDS = {Interval.m1: 60, Interval.m5: 300, Interval.m30: 1800, Interval.h1: 3600, Interval.d1: 86400}


class Adjustment(StrEnum):
    raw = "raw"              # vendor UNADJUSTED
    split = "split"          # vendor adj_split
    split_div = "split_div"  # vendor adj_splitdiv
    continuous = "continuous"


class QualityFlag(StrEnum):
    ohlc_inconsistent = "ohlc_inconsistent"       # high < max(open, close) or low > min(open, close)
    nonpositive_price = "nonpositive_price"
    negative_volume = "negative_volume"
    zero_volume = "zero_volume"
    fractional_volume = "fractional_volume"
    off_interval = "off_interval"                 # timestamp not aligned to the interval grid
    duplicate = "duplicate"
    outside_session = "outside_session"           # outside 04:00–20:00 ET extended hours
    non_trading_day = "non_trading_day"
    price_jump = "price_jump"                     # |close/prev_close - 1| > threshold within a session
    gap = "gap"                                   # missing regular-hours minute(s) before this bar
    out_of_order = "out_of_order"


# The vendor CSV layout for intraday/daily bars (no header).
RAW_BAR_COLUMNS = ["datetime", "open", "high", "low", "close", "volume"]

BARS_SCHEMA = pa.schema(
    [
        pa.field("instrument_id", pa.string(), nullable=False),
        pa.field("symbol_as_reported", pa.string(), nullable=False),
        pa.field("venue", pa.string(), nullable=True),
        pa.field("asset_class", pa.string(), nullable=False),
        pa.field("interval", pa.string(), nullable=False),
        pa.field("timestamp_utc", pa.timestamp("us", tz="UTC"), nullable=False),
        pa.field("session_date", pa.date32(), nullable=False),
        pa.field("open", pa.float64(), nullable=False),
        pa.field("high", pa.float64(), nullable=False),
        pa.field("low", pa.float64(), nullable=False),
        pa.field("close", pa.float64(), nullable=False),
        pa.field("volume", pa.float64(), nullable=False),
        pa.field("currency", pa.string(), nullable=False),
        pa.field("trade_count", pa.int32(), nullable=True),
        pa.field("vwap", pa.float64(), nullable=True),
        pa.field("adjustment_type", pa.string(), nullable=False),
        pa.field("source_id", pa.string(), nullable=False),
        pa.field("ingested_at", pa.timestamp("us", tz="UTC"), nullable=False),
        pa.field("quality_flags", pa.list_(pa.string()), nullable=False),
        pa.field("revision_id", pa.int32(), nullable=False),
    ],
    metadata={b"findata.schema": b"bars", b"findata.schema_version": str(BARS_SCHEMA_VERSION).encode(),
              b"findata.timestamp_convention": b"bar_open"},
)

FUNDAMENTALS_SCHEMA = pa.schema(
    [
        pa.field("entity_id", pa.string(), nullable=False),
        pa.field("security_id", pa.string(), nullable=True),
        pa.field("metric", pa.string(), nullable=False),
        pa.field("value", pa.float64(), nullable=True),
        pa.field("unit", pa.string(), nullable=False),
        pa.field("currency", pa.string(), nullable=True),
        pa.field("fiscal_period_start", pa.date32(), nullable=True),
        pa.field("fiscal_period_end", pa.date32(), nullable=False),
        pa.field("fiscal_period_type", pa.string(), nullable=False),
        pa.field("fiscal_year", pa.int32(), nullable=True),
        pa.field("fiscal_period", pa.string(), nullable=True),
        pa.field("filing_date", pa.date32(), nullable=True),
        pa.field("published_at", pa.timestamp("us", tz="UTC"), nullable=True),
        pa.field("available_at", pa.timestamp("us", tz="UTC"), nullable=False),
        pa.field("observed_at", pa.timestamp("us", tz="UTC"), nullable=False),
        pa.field("source_document", pa.string(), nullable=True),
        pa.field("source_id", pa.string(), nullable=False),
        pa.field("revision_id", pa.int32(), nullable=False),
        pa.field("restatement_flag", pa.bool_(), nullable=False),
    ]
)


def bars_schema_json() -> str:
    return json.dumps([{"name": f.name, "type": str(f.type), "nullable": f.nullable} for f in BARS_SCHEMA], indent=2)


# End-of-day option chains (vendor CBOE feed). trade_date is the session; not split-adjusted at source.
OPTIONS_SCHEMA = pa.schema(
    [
        pa.field("underlying", pa.string(), nullable=False),
        pa.field("trade_date", pa.date32(), nullable=False),
        pa.field("strike", pa.float64(), nullable=False),
        pa.field("expiry", pa.date32(), nullable=False),
        pa.field("call_put", pa.string(), nullable=False),
        pa.field("last_price", pa.float64(), nullable=True),
        pa.field("bid", pa.float64(), nullable=True),
        pa.field("ask", pa.float64(), nullable=True),
        pa.field("bid_iv", pa.float64(), nullable=True),
        pa.field("ask_iv", pa.float64(), nullable=True),
        pa.field("open_interest", pa.float64(), nullable=True),
        pa.field("volume", pa.float64(), nullable=True),
        pa.field("delta", pa.float64(), nullable=True),
        pa.field("gamma", pa.float64(), nullable=True),
        pa.field("vega", pa.float64(), nullable=True),
        pa.field("theta", pa.float64(), nullable=True),
        pa.field("rho", pa.float64(), nullable=True),
        pa.field("source_id", pa.string(), nullable=False),
        pa.field("ingested_at", pa.timestamp("us", tz="UTC"), nullable=False),
    ],
    metadata={b"findata.schema": b"options_eod", b"findata.schema_version": b"1"},
)