212 lines
13 KiB
Python
212 lines
13 KiB
Python
"""Join only verified series definitions; official values win independently for each field."""
|
|
from copy import deepcopy
|
|
from datetime import datetime
|
|
from decimal import Decimal, InvalidOperation
|
|
|
|
from alphapulse.calendar.catalog import FAMILIES, INDICATORS
|
|
from alphapulse.calendar.jin10 import LOCAL, period
|
|
from alphapulse.calendar.service import impact
|
|
from alphapulse.models import EconomicRelease
|
|
|
|
# (Jin10 indicator, exact name) -> (official family, indicator, release stage).
|
|
# CPI annual rates are NSA, monthly rates SA; other adjustments must not be substituted.
|
|
US = {
|
|
(234, "未季调CPI年率"): ("cpi", "cpi_yoy", None),
|
|
(230, "季调后CPI月率"): ("cpi", "cpi_mom", None),
|
|
(232, "未季调核心CPI年率"): ("cpi", "core_cpi_yoy", None),
|
|
(233, "季调后核心CPI月率"): ("cpi", "core_cpi_mom", None),
|
|
(178, "PPI年率"): ("ppi", "ppi_yoy", None),
|
|
(177, "PPI月率"): ("ppi", "ppi_mom", None),
|
|
(75, "季调后非农就业人口"): ("jobs", "payrolls", None),
|
|
(76, "失业率"): ("jobs", "unemployment", None),
|
|
(212, "PCE物价指数年率"): ("pce", "pce_yoy", None),
|
|
(213, "PCE物价指数月率"): ("pce", "pce_mom", None),
|
|
(211, "核心PCE物价指数年率"): ("pce", "core_pce_yoy", None),
|
|
(214, "核心PCE物价指数月率"): ("pce", "core_pce_mom", None),
|
|
(144, "实际GDP年化季率修正值"): ("gdp", "gdp", "second"),
|
|
(872, "实际GDP年化季率终值"): ("gdp", "gdp", "third"),
|
|
(198, "密歇根大学消费者信心指数初值"): ("umich", "umich_sentiment", "preliminary"),
|
|
(208, "密歇根大学消费者信心指数终值"): ("umich", "umich_sentiment", "final"),
|
|
(216, "实际GDP年化季率初值"): ("gdp", "gdp", "advance"),
|
|
(77, "美联储利率决定(上限)"): ("fomc", "fed_rate", None),
|
|
(78, "ADP就业人数"): ("adp_jobs", "adp_jobs", None),
|
|
(79, "当周初请失业金人数"): ("initial_claims", "initial_claims", None),
|
|
(96, "当周续请失业金人数"): ("continued_claims", "continued_claims", None),
|
|
(111, "当周EIA原油库存"): ("eia_stocks", "eia_crude", None),
|
|
(114, "当周EIA汽油库存"): ("eia_stocks", "eia_gasoline", None),
|
|
(113, "当周EIA精炼油库存"): ("eia_stocks", "eia_distillate", None),
|
|
}
|
|
EA = {
|
|
(763, "GDP季率初值"): ("ea_gdp", "gdp_qoq", "preliminary"),
|
|
(762, "GDP年率初值"): ("ea_gdp", "gdp_yoy", "preliminary"),
|
|
(930, "CPI年率初值"): ("ea_hicp", "hicp_yoy", "flash"),
|
|
(413, "CPI月率初值"): ("ea_hicp", "hicp_mom", "flash"),
|
|
(759, "核心CPI年率初值"): ("ea_hicp", "core_hicp_yoy", "flash"),
|
|
(764, "CPI年率终值"): ("ea_hicp", "hicp_yoy", "final"),
|
|
(8125, "CPI月率终值"): ("ea_hicp", "hicp_mom", "final"),
|
|
(622, "核心CPI年率终值"): ("ea_hicp", "core_hicp_yoy", "final"),
|
|
(396, "失业率"): ("ea_jobs", "ilo_unemployment", None),
|
|
(395, "欧洲央行存款机制利率"): ("ecb_rate", "ecb_deposit", None),
|
|
(753, "欧洲央行主要再融资利率"): ("ecb_rate", "ecb_main", None),
|
|
(419, "欧洲央行边际贷款利率"): ("ecb_rate", "ecb_marginal", None),
|
|
}
|
|
|
|
FR = {
|
|
(831, "GDP年率初值"): ("fr_gdp", "gdp_yoy", "initial"),
|
|
(829, "GDP年率终值"): ("fr_gdp", "gdp_yoy", "final"),
|
|
(830, "GDP季率终值"): ("fr_gdp", "gdp_qoq", "final"),
|
|
(8111, "CPI年率初值"): ("fr_cpi", "eu_cpi_yoy", "preliminary"),
|
|
(8112, "CPI月率初值"): ("fr_cpi", "eu_cpi_mom", "preliminary"),
|
|
(8113, "调和CPI年率初值"): ("fr_cpi", "hicp_yoy", "preliminary"),
|
|
(8114, "调和CPI月率初值"): ("fr_cpi", "hicp_mom", "preliminary"),
|
|
(567, "CPI年率终值"): ("fr_cpi", "eu_cpi_yoy", "final"),
|
|
(568, "CPI月率终值"): ("fr_cpi", "eu_cpi_mom", "final"),
|
|
(569, "调和CPI年率终值"): ("fr_cpi", "hicp_yoy", "final"),
|
|
(570, "调和CPI月率终值"): ("fr_cpi", "hicp_mom", "final"),
|
|
(832, "GDP季率初值"): ("fr_gdp", "gdp_qoq", "initial"),
|
|
}
|
|
DE = {
|
|
(270, "季调后GDP季率终值"): ("de_gdp", "gdp_qoq", "final"),
|
|
(842, "工作日调整后GDP年率终值"): ("de_gdp", "gdp_yoy", "final"),
|
|
(846, "CPI年率初值"): ("de_cpi", "eu_cpi_yoy", "preliminary"),
|
|
(847, "CPI月率初值"): ("de_cpi", "eu_cpi_mom", "preliminary"),
|
|
(839, "调和CPI年率初值"): ("de_cpi", "hicp_yoy", "preliminary"),
|
|
(840, "调和CPI月率初值"): ("de_cpi", "hicp_mom", "preliminary"),
|
|
(844, "CPI年率终值"): ("de_cpi", "eu_cpi_yoy", "final"),
|
|
(845, "CPI月率终值"): ("de_cpi", "eu_cpi_mom", "final"),
|
|
(837, "调和CPI年率终值"): ("de_cpi", "hicp_yoy", "final"),
|
|
(838, "调和CPI月率终值"): ("de_cpi", "hicp_mom", "final"),
|
|
(245, "季调后GDP季率初值"): ("de_gdp", "gdp_qoq", "initial"),
|
|
(843, "工作日调整后GDP年率初值"): ("de_gdp", "gdp_yoy", "initial"),
|
|
}
|
|
GB = {
|
|
(649, "三个月ILO失业率"): ("uk_jobs", "uk_unemployment", None),
|
|
(390, "GDP年率初值"): ("uk_gdp", "gdp_yoy", "initial"),
|
|
(391, "GDP季率初值"): ("uk_gdp", "gdp_qoq", "initial"),
|
|
(385, "GDP年率终值"): ("uk_gdp", "gdp_yoy", "final"),
|
|
(658, "GDP季率终值"): ("uk_gdp", "gdp_qoq", "final"),
|
|
(344, "CPI年率"): ("uk_cpi", "eu_cpi_yoy", None),
|
|
(348, "CPI月率"): ("uk_cpi", "eu_cpi_mom", None),
|
|
(349, "核心CPI年率"): ("uk_cpi", "eu_core_cpi_yoy", None),
|
|
(638, "央行利率决定"): ("boe_rate", "bank_rate", None),
|
|
}
|
|
MAPPINGS = {"US": US, "EA": EA, "FR": FR, "DE": DE, "GB": GB}
|
|
|
|
|
|
def definition(meta: dict) -> tuple[str, str, str | None] | None:
|
|
if meta["kind"] != "data" or meta["precision"] != "minute":
|
|
return None
|
|
mapping = MAPPINGS.get(meta["country_code"], {})
|
|
match = mapping.get((int(meta["indicator"].removeprefix("jin10_")), meta["name"]))
|
|
# Jin10 omits units for the verified Michigan index, whose official series is an index in points.
|
|
unit = "点" if meta["unit"] is None and meta["country_code"] == "US" and meta["indicator"] in ("jin10_198", "jin10_208") else meta["unit"]
|
|
return match if match and unit == INDICATORS[match[1]].unit else None
|
|
|
|
|
|
def provenance(source: str, at: str | datetime | None, url: str) -> dict:
|
|
return {"source": source, "observed_at": at, "url": url}
|
|
|
|
|
|
def jin10_item(row: EconomicRelease, now: datetime) -> dict:
|
|
meta = row.values["_jin10"]
|
|
key = meta["indicator"]
|
|
value = row.values.get(key, {})
|
|
actual, previous, forecast = value.get("actual"), value.get("previous"), value.get("forecast")
|
|
revised = value.get("previous_revised")
|
|
match = definition(meta)
|
|
eligible = value.get("forecast_before_release", False)
|
|
if row.schedule_status == "withdrawn":
|
|
status = "withdrawn"
|
|
direction, explanation = "uncertain", "金十已移除此日程；已保存值仅供追溯，不作为当前有效日程或方向判断。"
|
|
elif meta["kind"] != "data":
|
|
status = "recorded"
|
|
direction, explanation = "uncertain", "日程记录不代表事件已经发生；需结合实际消息判断 BTC/ETH 影响。"
|
|
else:
|
|
status = "released" if actual is not None else "historical" if (now - row.scheduled_at).days > 45 else "scheduled" if now < row.scheduled_at else "delayed"
|
|
direction, explanation = direction_for(match[1] if match else None, actual, revised if revised is not None else previous, forecast if eligible else None)
|
|
observed = value.get("observed_at", {})
|
|
fields = {field: provenance("jin10", observed.get(field), row.schedule_url)
|
|
for field in ("actual", "previous", "forecast", "previous_revised") if value.get(field) is not None}
|
|
fields["scheduled_at"] = provenance("jin10", row.schedule_checked_at, row.schedule_url)
|
|
if revised is not None:
|
|
if "previous" in fields:
|
|
fields["previous_original"] = fields["previous"]
|
|
fields["previous"] = fields["previous_revised"]
|
|
return {"release_id": row.id, "indicator": key, "name": meta["name"], "unit": meta["unit"],
|
|
"country_code": meta["country_code"], "country_name": meta["country_name"],
|
|
"event_name": f"{meta['country_name']}{meta['period_label'] or ''}{meta['name']}",
|
|
"period": None, "period_label": meta["period_label"], "stage": None, "stage_name": None,
|
|
"title": meta["description"] or meta["name"], "source": "jin10", "scheduled_at": row.scheduled_at,
|
|
"calendar_date": row.scheduled_at.astimezone(LOCAL).date().isoformat(), "time_precision": meta["precision"], "kind": meta["kind"],
|
|
"importance": meta["importance"], "importance_source": "jin10", "status": status,
|
|
"actual": actual, "previous": revised if revised is not None else previous,
|
|
"previous_original": previous, "previous_revised": revised, "forecast": forecast,
|
|
"previous_basis": "金十列示的修订前值（原前值单独保留）" if revised is not None else "金十本次记录列示前值" if previous is not None else None,
|
|
"forecast_before_release": eligible, "forecast_observed_at": observed.get("forecast"),
|
|
"impact": direction, "explanation": explanation, "impact_basis": "forecast" if forecast is not None and eligible else "previous" if previous is not None or revised is not None else None,
|
|
"release_url": None, "schedule_url": row.schedule_url, "first_received_at": row.first_received_at,
|
|
"last_checked_at": row.last_checked_at, "error": row.error, "revision_no": row.revision_no,
|
|
"field_sources": fields, "related_release_id": None, "related_revision_no": None}
|
|
|
|
|
|
def direction_for(key: str | None, actual: str | None, previous: str | None, forecast: str | None) -> tuple[str, str]:
|
|
if key is None:
|
|
return "uncertain", "尚无经核实的 BTC/ETH 传导判断；金十的黄金/美元影响标签未用于加密资产。"
|
|
try:
|
|
if any(not Decimal(v).is_finite() for v in (actual, previous, forecast) if v is not None):
|
|
raise InvalidOperation()
|
|
return impact(key, actual, previous, forecast)
|
|
except InvalidOperation:
|
|
return "uncertain", "公布值或比较基准不是单一数值，暂不能判断 BTC/ETH 方向。"
|
|
|
|
|
|
def merge_items(official: list[tuple[EconomicRelease, dict]], third_party: list[EconomicRelease], now: datetime) -> list[dict]:
|
|
output = []
|
|
used = set()
|
|
for row, original in official:
|
|
item = deepcopy(original)
|
|
item.update(kind="data", time_precision="minute", calendar_date=row.scheduled_at.astimezone(LOCAL).date().isoformat(),
|
|
importance_source="project", field_sources={field: provenance(row.source, row.first_received_at, row.release_url or row.schedule_url)
|
|
for field in ("actual", "previous", "forecast") if item[field] is not None},
|
|
related_release_id=None, related_revision_no=None, period_label=None, previous_original=None, previous_revised=None,
|
|
forecast_before_release=False, forecast_observed_at=None)
|
|
item["field_sources"]["scheduled_at"] = provenance(row.source, row.schedule_checked_at, row.schedule_url)
|
|
matches = []
|
|
for extra in third_party:
|
|
if extra.schedule_status == "withdrawn":
|
|
continue
|
|
meta = extra.values["_jin10"]
|
|
match = definition(meta)
|
|
if not match or match != (row.family, item["indicator"], row.stage):
|
|
continue
|
|
if FAMILIES[row.family].source != row.source or extra.scheduled_at != row.scheduled_at:
|
|
continue
|
|
# Rate decisions have no statistical reference period. All data series must agree on it.
|
|
if row.family not in ("fomc", "ecb_rate", "boe_rate") and (row.period is None or period(meta["period_label"], extra.scheduled_at) != row.period):
|
|
continue
|
|
matches.append(extra)
|
|
if len(matches) == 1:
|
|
extra = matches[0]
|
|
fallback = jin10_item(extra, now)
|
|
used.add(extra.id)
|
|
item.update(related_release_id=extra.id, related_revision_no=extra.revision_no,
|
|
previous_original=fallback["previous_original"], previous_revised=fallback["previous_revised"],
|
|
forecast_before_release=fallback["forecast_before_release"], forecast_observed_at=fallback["forecast_observed_at"])
|
|
for field in ("actual", "previous", "forecast"):
|
|
if item[field] is None and fallback[field] is not None:
|
|
item[field] = fallback[field]
|
|
provider_field = "previous_revised" if field == "previous" and fallback["previous_revised"] is not None else field
|
|
item["field_sources"][field] = fallback["field_sources"][provider_field]
|
|
if field == "previous":
|
|
item["previous_basis"] = fallback["previous_basis"]
|
|
if field == "actual":
|
|
item["first_received_at"] = fallback["first_received_at"]
|
|
if item["actual"] is not None and item["status"] != "withdrawn":
|
|
item["status"] = "released"
|
|
forecast = item["forecast"] if item.get("forecast_before_release") else None
|
|
item["impact"], item["explanation"] = direction_for(item["indicator"], item["actual"], item["previous"], forecast)
|
|
item["impact_basis"] = "forecast" if forecast is not None else "previous" if item["previous"] is not None else None
|
|
output.append(item)
|
|
output.extend(jin10_item(row, now) for row in third_party if row.id not in used)
|
|
return sorted(output, key=lambda r: (r["calendar_date"], r["scheduled_at"], r["release_id"], r["indicator"]))
|