feat(taxonomy): add rust sidecar compact surface pipeline
This commit is contained in:
320
lib/server/financials/surface.ts
Normal file
320
lib/server/financials/surface.ts
Normal file
@@ -0,0 +1,320 @@
|
||||
import type {
|
||||
DetailFinancialRow,
|
||||
FinancialStatementKind,
|
||||
FinancialStatementPeriod,
|
||||
NormalizationSummary,
|
||||
StructuredKpiRow,
|
||||
SurfaceDetailMap,
|
||||
SurfaceFinancialRow,
|
||||
TaxonomyFactRow,
|
||||
TaxonomyStatementRow
|
||||
} from '@/lib/types';
|
||||
import { buildStandardizedRows } from '@/lib/server/financials/standardize';
|
||||
|
||||
type CompactStatement = Extract<FinancialStatementKind, 'income' | 'balance' | 'cash_flow'>;
|
||||
|
||||
type SurfaceDefinition = {
|
||||
key: string;
|
||||
label: string;
|
||||
category: string;
|
||||
order: number;
|
||||
unit: SurfaceFinancialRow['unit'];
|
||||
rowKey?: string;
|
||||
componentKeys?: string[];
|
||||
formula?: {
|
||||
kind: 'subtract';
|
||||
left: string;
|
||||
right: string;
|
||||
};
|
||||
};
|
||||
|
||||
const EMPTY_SURFACE_ROWS: Record<FinancialStatementKind, SurfaceFinancialRow[]> = {
|
||||
income: [],
|
||||
balance: [],
|
||||
cash_flow: [],
|
||||
equity: [],
|
||||
comprehensive_income: []
|
||||
};
|
||||
|
||||
const EMPTY_DETAIL_ROWS: Record<FinancialStatementKind, SurfaceDetailMap> = {
|
||||
income: {},
|
||||
balance: {},
|
||||
cash_flow: {},
|
||||
equity: {},
|
||||
comprehensive_income: {}
|
||||
};
|
||||
|
||||
const SURFACE_DEFINITIONS: Record<CompactStatement, SurfaceDefinition[]> = {
|
||||
income: [
|
||||
{ key: 'revenue', label: 'Revenue', category: 'surface', order: 10, unit: 'currency', rowKey: 'revenue' },
|
||||
{ key: 'cost_of_revenue', label: 'Cost of Revenue', category: 'surface', order: 20, unit: 'currency', rowKey: 'cost_of_revenue' },
|
||||
{ key: 'gross_profit', label: 'Gross Profit', category: 'surface', order: 30, unit: 'currency', rowKey: 'gross_profit' },
|
||||
{
|
||||
key: 'operating_expenses',
|
||||
label: 'Operating Expenses',
|
||||
category: 'surface',
|
||||
order: 40,
|
||||
unit: 'currency',
|
||||
componentKeys: ['selling_general_and_administrative', 'research_and_development', 'depreciation_and_amortization']
|
||||
},
|
||||
{ key: 'operating_income', label: 'Operating Income', category: 'surface', order: 50, unit: 'currency', rowKey: 'operating_income' },
|
||||
{
|
||||
key: 'interest_and_other',
|
||||
label: 'Interest and Other',
|
||||
category: 'surface',
|
||||
order: 60,
|
||||
unit: 'currency',
|
||||
formula: {
|
||||
kind: 'subtract',
|
||||
left: 'pretax_income',
|
||||
right: 'operating_income'
|
||||
}
|
||||
},
|
||||
{ key: 'pretax_income', label: 'Pretax Income', category: 'surface', order: 70, unit: 'currency', rowKey: 'pretax_income' },
|
||||
{ key: 'income_taxes', label: 'Income Taxes', category: 'surface', order: 80, unit: 'currency', rowKey: 'income_tax_expense' },
|
||||
{ key: 'net_income', label: 'Net Income', category: 'surface', order: 90, unit: 'currency', rowKey: 'net_income' }
|
||||
],
|
||||
balance: [
|
||||
{ key: 'cash_and_equivalents', label: 'Cash and Equivalents', category: 'surface', order: 10, unit: 'currency', rowKey: 'cash_and_equivalents' },
|
||||
{ key: 'receivables', label: 'Receivables', category: 'surface', order: 20, unit: 'currency', rowKey: 'accounts_receivable' },
|
||||
{ key: 'inventory', label: 'Inventory', category: 'surface', order: 30, unit: 'currency', rowKey: 'inventory' },
|
||||
{ key: 'current_assets', label: 'Current Assets', category: 'surface', order: 40, unit: 'currency', rowKey: 'current_assets' },
|
||||
{ key: 'ppe', label: 'Property, Plant & Equipment', category: 'surface', order: 50, unit: 'currency', rowKey: 'property_plant_equipment' },
|
||||
{
|
||||
key: 'goodwill_and_intangibles',
|
||||
label: 'Goodwill and Intangibles',
|
||||
category: 'surface',
|
||||
order: 60,
|
||||
unit: 'currency',
|
||||
componentKeys: ['goodwill', 'intangible_assets']
|
||||
},
|
||||
{ key: 'total_assets', label: 'Total Assets', category: 'surface', order: 70, unit: 'currency', rowKey: 'total_assets' },
|
||||
{ key: 'current_liabilities', label: 'Current Liabilities', category: 'surface', order: 80, unit: 'currency', rowKey: 'current_liabilities' },
|
||||
{ key: 'debt', label: 'Debt', category: 'surface', order: 90, unit: 'currency', rowKey: 'total_debt' },
|
||||
{ key: 'total_liabilities', label: 'Total Liabilities', category: 'surface', order: 100, unit: 'currency', rowKey: 'total_liabilities' },
|
||||
{ key: 'shareholders_equity', label: 'Shareholders Equity', category: 'surface', order: 110, unit: 'currency', rowKey: 'total_equity' }
|
||||
],
|
||||
cash_flow: [
|
||||
{ key: 'operating_cash_flow', label: 'Operating Cash Flow', category: 'surface', order: 10, unit: 'currency', rowKey: 'operating_cash_flow' },
|
||||
{ key: 'capital_expenditures', label: 'Capital Expenditures', category: 'surface', order: 20, unit: 'currency', rowKey: 'capital_expenditures' },
|
||||
{ key: 'acquisitions', label: 'Acquisitions', category: 'surface', order: 30, unit: 'currency', rowKey: 'acquisitions' },
|
||||
{ key: 'investing_cash_flow', label: 'Investing Cash Flow', category: 'surface', order: 40, unit: 'currency', rowKey: 'investing_cash_flow' },
|
||||
{ key: 'financing_cash_flow', label: 'Financing Cash Flow', category: 'surface', order: 50, unit: 'currency', rowKey: 'financing_cash_flow' },
|
||||
{ key: 'free_cash_flow', label: 'Free Cash Flow', category: 'surface', order: 60, unit: 'currency', rowKey: 'free_cash_flow' }
|
||||
]
|
||||
};
|
||||
|
||||
function rowHasAnyValue(row: { values: Record<string, number | null> }) {
|
||||
return Object.values(row.values).some((value) => value !== null);
|
||||
}
|
||||
|
||||
function sumValues(values: Array<number | null>) {
|
||||
if (values.every((value) => value === null)) {
|
||||
return null;
|
||||
}
|
||||
|
||||
return values.reduce<number>((sum, value) => sum + (value ?? 0), 0);
|
||||
}
|
||||
|
||||
function valueForPeriod(
|
||||
rowByKey: Map<string, SurfaceFinancialRow>,
|
||||
rowKey: string,
|
||||
periodId: string
|
||||
) {
|
||||
return rowByKey.get(rowKey)?.values[periodId] ?? null;
|
||||
}
|
||||
|
||||
function maxAbsValue(values: Record<string, number | null>) {
|
||||
return Object.values(values).reduce<number>((max, value) => Math.max(max, Math.abs(value ?? 0)), 0);
|
||||
}
|
||||
|
||||
function detailUnit(row: SurfaceFinancialRow, faithfulRow: TaxonomyStatementRow | undefined) {
|
||||
if (faithfulRow) {
|
||||
return Object.values(faithfulRow.units)[0] ?? null;
|
||||
}
|
||||
|
||||
switch (row.unit) {
|
||||
case 'currency':
|
||||
return 'USD';
|
||||
case 'shares':
|
||||
return 'shares';
|
||||
case 'percent':
|
||||
return 'pure';
|
||||
default:
|
||||
return null;
|
||||
}
|
||||
}
|
||||
|
||||
function buildDetailRow(input: {
|
||||
row: SurfaceFinancialRow;
|
||||
parentSurfaceKey: string;
|
||||
faithfulRowByKey: Map<string, TaxonomyStatementRow>;
|
||||
}): DetailFinancialRow {
|
||||
const sourceRowKey = input.row.sourceRowKeys.find((key) => input.faithfulRowByKey.has(key)) ?? input.row.sourceRowKeys[0] ?? input.row.key;
|
||||
const faithfulRow = sourceRowKey ? input.faithfulRowByKey.get(sourceRowKey) : undefined;
|
||||
const qname = faithfulRow?.qname ?? input.row.sourceConcepts[0] ?? input.row.key;
|
||||
const [prefix, ...rest] = qname.split(':');
|
||||
const localName = faithfulRow?.localName ?? (rest.length > 0 ? rest.join(':') : qname);
|
||||
|
||||
return {
|
||||
key: input.row.key,
|
||||
parentSurfaceKey: input.parentSurfaceKey,
|
||||
label: input.row.label,
|
||||
conceptKey: faithfulRow?.conceptKey ?? sourceRowKey,
|
||||
qname,
|
||||
namespaceUri: faithfulRow?.namespaceUri ?? (prefix && rest.length > 0 ? `urn:unknown:${prefix}` : 'urn:surface'),
|
||||
localName,
|
||||
unit: detailUnit(input.row, faithfulRow),
|
||||
values: { ...input.row.values },
|
||||
sourceFactIds: [...input.row.sourceFactIds],
|
||||
isExtension: faithfulRow?.isExtension ?? false,
|
||||
dimensionsSummary: faithfulRow?.hasDimensions ? ['has_dimensions'] : [],
|
||||
residualFlag: input.parentSurfaceKey === 'unmapped'
|
||||
};
|
||||
}
|
||||
|
||||
function baselineForStatement(statement: CompactStatement, rowByKey: Map<string, SurfaceFinancialRow>) {
|
||||
const anchorKey = statement === 'balance' ? 'total_assets' : 'revenue';
|
||||
return maxAbsValue(rowByKey.get(anchorKey)?.values ?? {});
|
||||
}
|
||||
|
||||
function materialityThreshold(statement: CompactStatement, baseline: number) {
|
||||
if (statement === 'balance') {
|
||||
return Math.max(5_000_000, baseline * 0.005);
|
||||
}
|
||||
|
||||
return Math.max(1_000_000, baseline * 0.01);
|
||||
}
|
||||
|
||||
export function buildCompactHydrationModel(input: {
|
||||
periods: FinancialStatementPeriod[];
|
||||
faithfulRows: Record<FinancialStatementKind, TaxonomyStatementRow[]>;
|
||||
facts: TaxonomyFactRow[];
|
||||
kpiRows?: StructuredKpiRow[];
|
||||
}) {
|
||||
const surfaceRows = structuredClone(EMPTY_SURFACE_ROWS);
|
||||
const detailRows = structuredClone(EMPTY_DETAIL_ROWS);
|
||||
let surfaceRowCount = 0;
|
||||
let detailRowCount = 0;
|
||||
let unmappedRowCount = 0;
|
||||
let materialUnmappedRowCount = 0;
|
||||
|
||||
for (const statement of Object.keys(SURFACE_DEFINITIONS) as CompactStatement[]) {
|
||||
const faithfulRows = input.faithfulRows[statement] ?? [];
|
||||
const facts = input.facts.filter((fact) => fact.statement === statement);
|
||||
const fullRows = buildStandardizedRows({
|
||||
rows: faithfulRows,
|
||||
statement,
|
||||
periods: input.periods,
|
||||
facts
|
||||
});
|
||||
const rowByKey = new Map(fullRows.map((row) => [row.key, row]));
|
||||
const faithfulRowByKey = new Map(faithfulRows.map((row) => [row.key, row]));
|
||||
const statementDetails: SurfaceDetailMap = {};
|
||||
|
||||
for (const definition of SURFACE_DEFINITIONS[statement]) {
|
||||
const contributingRows = definition.rowKey
|
||||
? [rowByKey.get(definition.rowKey)].filter((row): row is SurfaceFinancialRow => row !== undefined)
|
||||
: (definition.componentKeys ?? [])
|
||||
.map((key) => rowByKey.get(key))
|
||||
.filter((row): row is SurfaceFinancialRow => row !== undefined);
|
||||
|
||||
const values = Object.fromEntries(input.periods.map((period) => {
|
||||
const nextValue = definition.rowKey
|
||||
? valueForPeriod(rowByKey, definition.rowKey, period.id)
|
||||
: definition.formula
|
||||
? (() => {
|
||||
const left = valueForPeriod(rowByKey, definition.formula!.left, period.id);
|
||||
const right = valueForPeriod(rowByKey, definition.formula!.right, period.id);
|
||||
return left === null || right === null ? null : left - right;
|
||||
})()
|
||||
: sumValues(contributingRows.map((row) => row.values[period.id] ?? null));
|
||||
|
||||
return [period.id, nextValue];
|
||||
})) satisfies Record<string, number | null>;
|
||||
|
||||
if (!rowHasAnyValue({ values })) {
|
||||
continue;
|
||||
}
|
||||
|
||||
const sourceConcepts = [...new Set(contributingRows.flatMap((row) => row.sourceConcepts))].sort((left, right) => left.localeCompare(right));
|
||||
const sourceRowKeys = [...new Set(contributingRows.flatMap((row) => row.sourceRowKeys))].sort((left, right) => left.localeCompare(right));
|
||||
const sourceFactIds = [...new Set(contributingRows.flatMap((row) => row.sourceFactIds))].sort((left, right) => left - right);
|
||||
const hasDimensions = contributingRows.some((row) => row.hasDimensions);
|
||||
const resolvedSourceRowKeys = Object.fromEntries(input.periods.map((period) => [
|
||||
period.id,
|
||||
definition.rowKey
|
||||
? rowByKey.get(definition.rowKey)?.resolvedSourceRowKeys[period.id] ?? null
|
||||
: null
|
||||
]));
|
||||
|
||||
const rowsForDetail = definition.componentKeys
|
||||
? contributingRows
|
||||
: [];
|
||||
const details = rowsForDetail
|
||||
.filter((row) => rowHasAnyValue(row))
|
||||
.map((row) => buildDetailRow({
|
||||
row,
|
||||
parentSurfaceKey: definition.key,
|
||||
faithfulRowByKey
|
||||
}));
|
||||
|
||||
statementDetails[definition.key] = details;
|
||||
detailRowCount += details.length;
|
||||
|
||||
surfaceRows[statement].push({
|
||||
key: definition.key,
|
||||
label: definition.label,
|
||||
category: definition.category,
|
||||
templateSection: definition.category,
|
||||
order: definition.order,
|
||||
unit: definition.unit,
|
||||
values,
|
||||
sourceConcepts,
|
||||
sourceRowKeys,
|
||||
sourceFactIds,
|
||||
formulaKey: definition.formula ? definition.key : null,
|
||||
hasDimensions,
|
||||
resolvedSourceRowKeys,
|
||||
statement,
|
||||
detailCount: details.length
|
||||
});
|
||||
surfaceRowCount += 1;
|
||||
}
|
||||
|
||||
const baseline = baselineForStatement(statement, rowByKey);
|
||||
const threshold = materialityThreshold(statement, baseline);
|
||||
const residualRows = fullRows
|
||||
.filter((row) => row.key.startsWith('other:'))
|
||||
.filter((row) => rowHasAnyValue(row))
|
||||
.map((row) => buildDetailRow({
|
||||
row,
|
||||
parentSurfaceKey: 'unmapped',
|
||||
faithfulRowByKey
|
||||
}));
|
||||
|
||||
if (residualRows.length > 0) {
|
||||
statementDetails.unmapped = residualRows;
|
||||
detailRowCount += residualRows.length;
|
||||
unmappedRowCount += residualRows.length;
|
||||
materialUnmappedRowCount += residualRows.filter((row) => maxAbsValue(row.values) >= threshold).length;
|
||||
}
|
||||
|
||||
detailRows[statement] = statementDetails;
|
||||
}
|
||||
|
||||
const normalizationSummary: NormalizationSummary = {
|
||||
surfaceRowCount,
|
||||
detailRowCount,
|
||||
kpiRowCount: input.kpiRows?.length ?? 0,
|
||||
unmappedRowCount,
|
||||
materialUnmappedRowCount,
|
||||
warnings: []
|
||||
};
|
||||
|
||||
return {
|
||||
surfaceRows,
|
||||
detailRows,
|
||||
normalizationSummary
|
||||
};
|
||||
}
|
||||
Reference in New Issue
Block a user