recipe/schema.py

"""Canonical Polars schemas for every published config."""

from __future__ import annotations

from collections.abc import Mapping

import polars as pl

UTC_DATETIME = pl.Datetime(time_unit="us", time_zone="UTC")

BUILD_VERSION = "1.0.0"
PARSER_VERSION = "1.0.0"

# The Treasury's own labels, normalised. `cmb` is a cash management bill, which
# is a bill with an irregular term announced at short notice; keeping it apart
# from the regular bill calendar is the point of having the value.
SECURITY_TYPES = ("bill", "note", "bond", "tips", "frn", "cmb", "other")
CURVE_TYPES = ("nominal", "real", "bill", "long_term_composite", "long_term_real")
# How precisely the moment of publication is known. `minute` means the artifact
# stated it; `modeled_conservative` means a documented rule that is later than
# the truth and never earlier.
KNOWLEDGE_PRECISIONS = ("exact", "minute", "hour", "date", "modeled_conservative", "unknown")
STATUSES = (
    "scheduled", "discovered", "fetched", "parsed", "validated", "published",
    "quarantined", "revised", "cancelled",
)

PROVENANCE: dict[str, pl.DataType] = {
    "source": pl.String,
    "source_url": pl.String,
    "raw_sha256": pl.String,
    "parser_version": pl.String,
    "first_seen_at": UTC_DATETIME,
    "ingested_at": UTC_DATETIME,
}

AUCTION_CALENDAR_SCHEMA: dict[str, pl.DataType] = {
    "auction_id": pl.String,
    "cusip": pl.String,
    "security_type": pl.String,
    "security_term": pl.String,
    "original_security_term": pl.String,
    "reopening": pl.Boolean,
    "announcement_date": pl.Date,
    "announcement_at": UTC_DATETIME,
    "auction_date": pl.Date,
    "auction_at": UTC_DATETIME,
    "issue_date": pl.Date,
    "maturity_date": pl.Date,
    "offering_amount": pl.Float64,
    "status": pl.String,
    "knowledge_at": UTC_DATETIME,
    "knowledge_date": UTC_DATETIME,
    "knowledge_time_precision": pl.String,
    **PROVENANCE,
}

ANNOUNCEMENT_SCHEMA: dict[str, pl.DataType] = {
    "announcement_id": pl.String,
    "auction_id": pl.String,
    "release_at": UTC_DATETIME,
    "knowledge_at": UTC_DATETIME,
    "knowledge_date": UTC_DATETIME,
    "knowledge_time_precision": pl.String,
    "offering_amount": pl.Float64,
    "security_type": pl.String,
    "security_term": pl.String,
    "cusip": pl.String,
    "announcement_date": pl.Date,
    "auction_date": pl.Date,
    "issue_date": pl.Date,
    "maturity_date": pl.Date,
    "competitive_close_at": UTC_DATETIME,
    "noncompetitive_close_at": UTC_DATETIME,
    "reopening": pl.Boolean,
    "soma_holdings": pl.Float64,
    "maturing_amount": pl.Float64,
    **PROVENANCE,
}

RESULT_SCHEMA: dict[str, pl.DataType] = {
    "result_id": pl.String,
    "auction_id": pl.String,
    "cusip": pl.String,
    "security_type": pl.String,
    "security_term": pl.String,
    "auction_date": pl.Date,
    "issue_date": pl.Date,
    "maturity_date": pl.Date,
    "result_release_at": UTC_DATETIME,
    "knowledge_at": UTC_DATETIME,
    "knowledge_date": UTC_DATETIME,
    "knowledge_time_precision": pl.String,
    "auction_format": pl.String,
    "offering_amount": pl.Float64,
    "total_tendered": pl.Float64,
    "total_accepted": pl.Float64,
    "competitive_tendered": pl.Float64,
    "competitive_accepted": pl.Float64,
    "noncompetitive_tendered": pl.Float64,
    "noncompetitive_accepted": pl.Float64,
    "high_yield": pl.Float64,
    "median_yield": pl.Float64,
    "low_yield": pl.Float64,
    "high_discount_rate": pl.Float64,
    "median_discount_rate": pl.Float64,
    "low_discount_rate": pl.Float64,
    "investment_rate": pl.Float64,
    "interest_rate": pl.Float64,
    "price_per_100": pl.Float64,
    "bid_to_cover": pl.Float64,
    "allocation_percentage": pl.Float64,
    "primary_dealer_tendered": pl.Float64,
    "primary_dealer_accepted": pl.Float64,
    "direct_bidder_tendered": pl.Float64,
    "direct_bidder_accepted": pl.Float64,
    "indirect_bidder_tendered": pl.Float64,
    "indirect_bidder_accepted": pl.Float64,
    "soma_tendered": pl.Float64,
    "soma_accepted": pl.Float64,
    "treasury_retail_accepted": pl.Float64,
    # Deterministic shares of what was accepted, computed only from numbers in
    # the same result. Null where the denominator is absent, never zero.
    "dealer_accepted_pct": pl.Float64,
    "direct_accepted_pct": pl.Float64,
    "indirect_accepted_pct": pl.Float64,
    "noncompetitive_accepted_pct": pl.Float64,
    "tendered_to_offering": pl.Float64,
    # Always null. The tail is the high yield minus the when-issued market
    # yield, and no official point-in-time when-issued series exists. The daily
    # par curve is a different measurement and would not be a tail.
    "wi_yield": pl.Float64,
    "auction_tail": pl.Float64,
    "status": pl.String,
    **PROVENANCE,
}

CURVE_SCHEMA: dict[str, pl.DataType] = {
    "curve_id": pl.String,
    "curve_date": pl.Date,
    "curve_type": pl.String,
    "maturity": pl.String,
    "source_label": pl.String,
    "maturity_years": pl.Float64,
    "yield_pct": pl.Float64,
    "market_observation_at": UTC_DATETIME,
    "publication_at": UTC_DATETIME,
    "knowledge_at": UTC_DATETIME,
    "knowledge_date": UTC_DATETIME,
    "knowledge_time_precision": pl.String,
    "availability_time_basis": pl.String,
    "methodology_id": pl.String,
    **PROVENANCE,
}

BILL_RATE_SCHEMA: dict[str, pl.DataType] = {
    "bill_rate_id": pl.String,
    "rate_date": pl.Date,
    "maturity": pl.String,
    "source_label": pl.String,
    "maturity_weeks": pl.Float64,
    "bank_discount_rate": pl.Float64,
    "coupon_equivalent_rate": pl.Float64,
    "market_observation_at": UTC_DATETIME,
    "knowledge_at": UTC_DATETIME,
    "knowledge_date": UTC_DATETIME,
    "knowledge_time_precision": pl.String,
    "availability_time_basis": pl.String,
    **PROVENANCE,
}

CURVE_FEATURE_SCHEMA: dict[str, pl.DataType] = {
    "feature_id": pl.String,
    "curve_date": pl.Date,
    "knowledge_at": UTC_DATETIME,
    "knowledge_date": UTC_DATETIME,
    "knowledge_time_precision": pl.String,
    "slope_3m10y_bp": pl.Float64,
    "slope_2y10y_bp": pl.Float64,
    "slope_5y30y_bp": pl.Float64,
    "real_5y10y_slope_bp": pl.Float64,
    # A spread between two official curves published for the same day. It is
    # not a market breakeven: both legs are par yields struck from indicative
    # quotations, and neither is a traded inflation swap.
    "nominal_10y_minus_real_10y_bp": pl.Float64,
    "derived_from": pl.String,
    "parser_version": pl.String,
    "ingested_at": UTC_DATETIME,
}

# One row per public number, in the shape ziplime mounts a bundle from.
PIT_SCHEMA: dict[str, pl.DataType] = {
    "entity_id": pl.String,
    "event_date": pl.Date,
    "knowledge_date": UTC_DATETIME,
    "subset": pl.String,
    "record_id": pl.String,
    "actual": pl.Float64,
    "previous": pl.Float64,
    "change": pl.Float64,
    # Always null: consensus estimates are licensed data and none are included.
    "forecast": pl.Float64,
    "unit": pl.String,
    "value_method": pl.String,
    "knowledge_time_precision": pl.String,
    "source_url": pl.String,
    "parser_version": pl.String,
    "ingested_at": UTC_DATETIME,
}

CONFIG_SCHEMAS: dict[str, dict[str, pl.DataType]] = {
    "auction_calendar": AUCTION_CALENDAR_SCHEMA,
    "auction_announcements": ANNOUNCEMENT_SCHEMA,
    "auction_results": RESULT_SCHEMA,
    "nominal_yield_curve": CURVE_SCHEMA,
    "real_yield_curve": CURVE_SCHEMA,
    "bill_rates": BILL_RATE_SCHEMA,
    "curve_features": CURVE_FEATURE_SCHEMA,
    "pit": PIT_SCHEMA,
}


def empty_frame(schema: Mapping[str, pl.DataType]) -> pl.DataFrame:
    return pl.DataFrame(schema=dict(schema))


def align(frame: pl.DataFrame, schema: Mapping[str, pl.DataType]) -> pl.DataFrame:
    missing = [
        pl.lit(None, dtype=dtype).alias(name)
        for name, dtype in schema.items()
        if name not in frame.columns
    ]
    if missing:
        frame = frame.with_columns(missing)
    return frame.select([pl.col(n).cast(d, strict=False) for n, d in schema.items()])


__all__ = [
    "ANNOUNCEMENT_SCHEMA",
    "AUCTION_CALENDAR_SCHEMA",
    "BILL_RATE_SCHEMA",
    "BUILD_VERSION",
    "CONFIG_SCHEMAS",
    "CURVE_FEATURE_SCHEMA",
    "CURVE_SCHEMA",
    "CURVE_TYPES",
    "KNOWLEDGE_PRECISIONS",
    "PARSER_VERSION",
    "PIT_SCHEMA",
    "PROVENANCE",
    "RESULT_SCHEMA",
    "SECURITY_TYPES",
    "STATUSES",
    "UTC_DATETIME",
    "align",
    "empty_frame",
]