refactor(taxonomy): remove legacy parser and add rollout checks

This commit is contained in:
2026-03-12 15:25:06 -04:00
parent 58061af006
commit c274f4d55b
11 changed files with 1344 additions and 697 deletions

View File

@@ -0,0 +1,284 @@
import type { FinancialStatementKind } from '@/lib/types';
import { hydrateFilingTaxonomySnapshot } from '@/lib/server/taxonomy/engine';
import type { TaxonomyHydrationInput } from '@/lib/server/taxonomy/types';
type ValidationCase = {
name: string;
expectedPack: string;
input: TaxonomyHydrationInput;
requiredSurfaceKeys: Partial<Record<FinancialStatementKind, string[]>>;
requiredKpiKeys?: string[];
};
type ValidationFailure = {
name: string;
issues: string[];
};
const UNIVERSAL_INCOME_KEYS = [
'revenue',
'gross_profit',
'operating_expenses',
'operating_income',
'income_tax_expense',
'net_income'
] as const;
const EXPENSE_BREAKDOWN_KEYS = [
'selling_general_and_administrative',
'research_and_development',
'other_operating_expense'
] as const;
const CORPUS: ValidationCase[] = [
{
name: 'core-msft-2026-01-28',
expectedPack: 'core',
input: {
filingId: 1,
ticker: 'MSFT',
cik: '0000789019',
accessionNumber: '0001193125-26-027207',
filingDate: '2026-01-28',
filingType: '10-Q',
filingUrl: 'https://www.sec.gov/Archives/edgar/data/789019/000119312526027207/',
primaryDocument: 'msft-20251231.htm'
},
requiredSurfaceKeys: {
income: [...UNIVERSAL_INCOME_KEYS, ...EXPENSE_BREAKDOWN_KEYS],
balance: ['total_assets']
}
},
{
name: 'bank-jpm-2026-02-13',
expectedPack: 'bank_lender',
input: {
filingId: 2,
ticker: 'JPM',
cik: '0000019617',
accessionNumber: '0001628280-26-008131',
filingDate: '2026-02-13',
filingType: '10-K',
filingUrl: 'https://www.sec.gov/Archives/edgar/data/19617/000162828026008131/',
primaryDocument: 'jpm-20251231.htm'
},
requiredSurfaceKeys: {
income: [...UNIVERSAL_INCOME_KEYS, ...EXPENSE_BREAKDOWN_KEYS, 'net_interest_income', 'noninterest_income'],
balance: ['loans', 'deposits']
},
requiredKpiKeys: ['net_interest_margin']
},
{
name: 'insurance-aig-2026-02-12',
expectedPack: 'insurance',
input: {
filingId: 3,
ticker: 'AIG',
cik: '0000005272',
accessionNumber: '0000005272-26-000023',
filingDate: '2026-02-12',
filingType: '10-K',
filingUrl: 'https://www.sec.gov/Archives/edgar/data/5272/000000527226000023/',
primaryDocument: 'aig-20251231.htm'
},
requiredSurfaceKeys: {
income: [...UNIVERSAL_INCOME_KEYS, ...EXPENSE_BREAKDOWN_KEYS, 'premiums', 'claims_and_benefits'],
balance: ['policy_liabilities']
},
requiredKpiKeys: ['combined_ratio']
},
{
name: 'reit-o-2026-02-25',
expectedPack: 'reit_real_estate',
input: {
filingId: 4,
ticker: 'O',
cik: '0000726728',
accessionNumber: '0000726728-26-000011',
filingDate: '2026-02-25',
filingType: '10-K',
filingUrl: 'https://www.sec.gov/Archives/edgar/data/726728/000072672826000011/',
primaryDocument: 'o-20251231.htm'
},
requiredSurfaceKeys: {
income: [...UNIVERSAL_INCOME_KEYS, ...EXPENSE_BREAKDOWN_KEYS, 'rental_revenue'],
balance: ['investment_property', 'total_assets']
},
requiredKpiKeys: ['property_count']
},
{
name: 'broker-blk-2026-02-25',
expectedPack: 'broker_asset_manager',
input: {
filingId: 5,
ticker: 'BLK',
cik: '0002012383',
accessionNumber: '0001193125-26-071966',
filingDate: '2026-02-25',
filingType: '10-K',
filingUrl: 'https://www.sec.gov/Archives/edgar/data/2012383/000119312526071966/',
primaryDocument: 'blk-20251231.htm'
},
requiredSurfaceKeys: {
income: [...UNIVERSAL_INCOME_KEYS, ...EXPENSE_BREAKDOWN_KEYS, 'fee_revenue'],
balance: ['total_assets', 'total_liabilities']
},
requiredKpiKeys: ['aum', 'fee_paying_aum']
}
];
const FALLBACK_WARNINGS = new Set([
'surface_rows_deferred_to_typescript',
'ts_compact_surface_fallback_used'
]);
function parseCaseFilter(argv: string[]) {
for (const arg of argv) {
if (arg === '--help' || arg === '-h') {
console.log('Validate live SEC representative filings for each active taxonomy pack.');
console.log('');
console.log('Usage:');
console.log(' bun run scripts/validate-taxonomy-packs.ts');
console.log(' bun run scripts/validate-taxonomy-packs.ts --case=bank-jpm-2026-02-13');
process.exit(0);
}
if (arg.startsWith('--case=')) {
const value = arg.slice('--case='.length).trim();
return value.length > 0 ? value : null;
}
}
return null;
}
function keysForStatement(
result: Awaited<ReturnType<typeof hydrateFilingTaxonomySnapshot>>,
statement: FinancialStatementKind
) {
return (result.surface_rows[statement] ?? []).map((row) => row.key);
}
async function validateCase(testCase: ValidationCase): Promise<ValidationFailure | null> {
const startedAt = Date.now();
const result = await hydrateFilingTaxonomySnapshot(testCase.input);
const issues: string[] = [];
const warnings = result.normalization_summary.warnings ?? [];
const kpiKeys = result.kpi_rows.map((row) => row.key);
if (result.parse_status !== 'ready') {
issues.push(`parse_status=${result.parse_status}${result.parse_error ? ` parse_error=${result.parse_error}` : ''}`);
}
if (result.fiscal_pack !== testCase.expectedPack) {
issues.push(`fiscal_pack=${result.fiscal_pack ?? 'null'} expected=${testCase.expectedPack}`);
}
if ((Object.values(result.surface_rows) as Array<Array<{ key: string }>>).every((rows) => rows.length === 0)) {
issues.push('surface_rows are empty');
}
const fallbackWarning = warnings.find((warning) => FALLBACK_WARNINGS.has(warning));
if (fallbackWarning) {
issues.push(`unexpected fallback warning=${fallbackWarning}`);
}
for (const [statement, requiredKeys] of Object.entries(testCase.requiredSurfaceKeys) as Array<
[FinancialStatementKind, string[]]
>) {
const actualKeys = new Set(keysForStatement(result, statement));
for (const requiredKey of requiredKeys) {
if (!actualKeys.has(requiredKey)) {
issues.push(`${statement} missing surface key=${requiredKey}`);
}
}
}
for (const requiredKpiKey of testCase.requiredKpiKeys ?? []) {
if (!kpiKeys.includes(requiredKpiKey)) {
issues.push(`missing kpi key=${requiredKpiKey}`);
}
}
const durationMs = Date.now() - startedAt;
const incomeKeys = keysForStatement(result, 'income');
const balanceKeys = keysForStatement(result, 'balance');
console.log(
[
`[validate-taxonomy-packs] ${testCase.name}`,
`status=${issues.length === 0 ? 'pass' : 'fail'}`,
`parse=${result.parse_status}`,
`pack=${result.fiscal_pack ?? 'null'}`,
`income=${incomeKeys.join(',') || '-'}`,
`balance=${balanceKeys.join(',') || '-'}`,
`kpis=${kpiKeys.join(',') || '-'}`,
`warnings=${warnings.join(',') || '-'}`,
`durationMs=${durationMs}`
].join(' ')
);
if (issues.length === 0) {
return null;
}
return {
name: testCase.name,
issues
};
}
async function main() {
process.env.XBRL_ENGINE_TIMEOUT_MS = process.env.XBRL_ENGINE_TIMEOUT_MS ?? '180000';
const requestedCase = parseCaseFilter(process.argv.slice(2));
const selectedCases = requestedCase
? CORPUS.filter((testCase) => testCase.name === requestedCase)
: CORPUS;
if (selectedCases.length === 0) {
console.error(`[validate-taxonomy-packs] unknown case: ${requestedCase}`);
process.exitCode = 1;
return;
}
const failures: ValidationFailure[] = [];
const startedAt = Date.now();
for (const testCase of selectedCases) {
try {
const failure = await validateCase(testCase);
if (failure) {
failures.push(failure);
}
} catch (error) {
failures.push({
name: testCase.name,
issues: [error instanceof Error ? error.message : String(error)]
});
}
await Bun.sleep(150);
}
console.log(
`[validate-taxonomy-packs] completed cases=${selectedCases.length} failures=${failures.length} durationSec=${(
(Date.now() - startedAt) /
1000
).toFixed(1)}`
);
if (failures.length === 0) {
return;
}
for (const failure of failures) {
console.error(`[validate-taxonomy-packs] ${failure.name}`);
for (const issue of failure.issues) {
console.error(` - ${issue}`);
}
}
process.exitCode = 1;
}
void main();