Remove legacy TypeScript financial surface mapping, make Rust JSON single source of truth
- Delete standard-template.ts, surface.ts, materialize.ts (dead code) - Delete financial-taxonomy.test.ts (relied on removed code) - Add missing income statement surfaces to core.surface.json - Add cost_of_revenue mapping to core.income-bridge.json - Refactor standardize.ts to remove template dependency - Simplify financial-taxonomy.ts to use only DB snapshots - Add architecture documentation
This commit is contained in:
File diff suppressed because it is too large
Load Diff
@@ -36,8 +36,7 @@ import {
|
||||
} from '@/lib/server/financials/bundles';
|
||||
import {
|
||||
buildDimensionBreakdown,
|
||||
buildLtmStandardizedRows,
|
||||
buildStandardizedRows
|
||||
buildLtmStandardizedRows
|
||||
} from '@/lib/server/financials/standardize';
|
||||
import { buildRatioRows } from '@/lib/server/financials/ratios';
|
||||
import { buildFinancialCategories, buildTrendSeries } from '@/lib/server/financials/trend-series';
|
||||
@@ -620,16 +619,7 @@ function buildQuarterlyStatementSurfaceRows(input: {
|
||||
selectedPeriodIds: input.selectedPeriodIds
|
||||
});
|
||||
|
||||
if (aggregatedRows.length > 0) {
|
||||
return aggregatedRows;
|
||||
}
|
||||
|
||||
return buildStandardizedRows({
|
||||
rows: input.faithfulRows,
|
||||
statement: input.statement,
|
||||
periods: input.sourcePeriods,
|
||||
facts: input.facts
|
||||
}) as SurfaceFinancialRow[];
|
||||
return aggregatedRows;
|
||||
}
|
||||
|
||||
function aggregatePersistedKpiRows(input: {
|
||||
@@ -1303,7 +1293,6 @@ export async function getCompanyFinancialTaxonomy(input: GetCompanyFinancialsInp
|
||||
|
||||
export const __financialTaxonomyInternals = {
|
||||
buildRows,
|
||||
buildStandardizedRows,
|
||||
buildDimensionBreakdown,
|
||||
buildNormalizationMetadata,
|
||||
aggregateSurfaceRows,
|
||||
|
||||
File diff suppressed because it is too large
Load Diff
@@ -1,31 +1,11 @@
|
||||
import type {
|
||||
DerivedFinancialRow,
|
||||
DimensionBreakdownRow,
|
||||
FinancialStatementKind,
|
||||
FinancialStatementPeriod,
|
||||
FinancialUnit,
|
||||
StandardizedFinancialRow,
|
||||
TaxonomyFactRow,
|
||||
TaxonomyStatementRow
|
||||
} from '@/lib/types';
|
||||
import {
|
||||
STANDARD_FINANCIAL_TEMPLATES,
|
||||
type StandardTemplateRowDefinition,
|
||||
type TemplateFormula
|
||||
} from '@/lib/server/financials/standard-template';
|
||||
|
||||
function normalizeToken(value: string) {
|
||||
return value.trim().toLowerCase();
|
||||
}
|
||||
|
||||
function tokenizeLabel(value: string) {
|
||||
return value
|
||||
.toLowerCase()
|
||||
.replace(/[^a-z0-9]+/g, ' ')
|
||||
.trim()
|
||||
.split(/\s+/)
|
||||
.filter((token) => token.length > 0);
|
||||
}
|
||||
|
||||
function valueOrNull(values: Record<string, number | null>, periodId: string) {
|
||||
return periodId in values ? values[periodId] : null;
|
||||
@@ -39,326 +19,6 @@ function sumValues(values: Array<number | null>, treatNullAsZero = false) {
|
||||
return values.reduce<number>((sum, value) => sum + (value ?? 0), 0);
|
||||
}
|
||||
|
||||
function subtractValues(left: number | null, right: number | null) {
|
||||
if (left === null || right === null) {
|
||||
return null;
|
||||
}
|
||||
|
||||
return left - right;
|
||||
}
|
||||
|
||||
function divideValues(left: number | null, right: number | null) {
|
||||
if (left === null || right === null || right === 0) {
|
||||
return null;
|
||||
}
|
||||
|
||||
return left / right;
|
||||
}
|
||||
|
||||
type CandidateMatchKind = 'exact_local_name' | 'secondary_local_name' | 'label_phrase';
|
||||
|
||||
type StatementRowCandidate = {
|
||||
row: TaxonomyStatementRow;
|
||||
matchKind: CandidateMatchKind;
|
||||
aliasRank: number;
|
||||
unit: FinancialUnit;
|
||||
labelTokenCount: number;
|
||||
matchedPhraseTokenCount: number;
|
||||
};
|
||||
|
||||
type FactCandidate = {
|
||||
fact: TaxonomyFactRow;
|
||||
matchKind: Exclude<CandidateMatchKind, 'label_phrase'>;
|
||||
aliasRank: number;
|
||||
unit: FinancialUnit;
|
||||
};
|
||||
|
||||
type ResolvedCandidate =
|
||||
| {
|
||||
sourceType: 'row';
|
||||
matchKind: CandidateMatchKind;
|
||||
aliasRank: number;
|
||||
unit: FinancialUnit;
|
||||
labelTokenCount: number;
|
||||
matchedPhraseTokenCount: number;
|
||||
row: TaxonomyStatementRow;
|
||||
}
|
||||
| {
|
||||
sourceType: 'fact';
|
||||
matchKind: Exclude<CandidateMatchKind, 'label_phrase'>;
|
||||
aliasRank: number;
|
||||
unit: FinancialUnit;
|
||||
fact: TaxonomyFactRow;
|
||||
};
|
||||
|
||||
type DerivedRole = 'expense' | 'addback';
|
||||
|
||||
type InternalRowMetadata = {
|
||||
derivedRoleByPeriod: Record<string, DerivedRole | null>;
|
||||
};
|
||||
|
||||
function resolvedCandidatesForPeriod(input: {
|
||||
definition: StandardTemplateRowDefinition;
|
||||
candidates: StatementRowCandidate[];
|
||||
factCandidates: FactCandidate[];
|
||||
period: FinancialStatementPeriod;
|
||||
}) {
|
||||
const rowCandidates = input.candidates
|
||||
.filter((candidate) => input.period.id in candidate.row.values && candidate.row.values[input.period.id] !== null)
|
||||
.map((candidate) => ({
|
||||
sourceType: 'row' as const,
|
||||
...candidate
|
||||
}));
|
||||
const factCandidates = input.factCandidates
|
||||
.filter((candidate) => factMatchesPeriod(candidate.fact, input.period))
|
||||
.map((candidate) => ({
|
||||
sourceType: 'fact' as const,
|
||||
...candidate
|
||||
}));
|
||||
|
||||
if (input.definition.selectionPolicy === 'aggregate_multiple_components') {
|
||||
const aggregateCandidates = [...rowCandidates, ...factCandidates]
|
||||
.sort((left, right) => compareResolvedCandidates(left, right, input.definition));
|
||||
const dedupedCandidates: ResolvedCandidate[] = [];
|
||||
const seenConcepts = new Set<string>();
|
||||
|
||||
for (const candidate of aggregateCandidates) {
|
||||
const conceptKey = candidate.sourceType === 'row'
|
||||
? candidate.row.key
|
||||
: candidate.fact.conceptKey;
|
||||
if (seenConcepts.has(conceptKey)) {
|
||||
continue;
|
||||
}
|
||||
|
||||
seenConcepts.add(conceptKey);
|
||||
dedupedCandidates.push(candidate);
|
||||
}
|
||||
|
||||
return dedupedCandidates;
|
||||
}
|
||||
|
||||
const resolvedCandidate = [...rowCandidates, ...factCandidates]
|
||||
.sort((left, right) => compareResolvedCandidates(left, right, input.definition))[0];
|
||||
|
||||
return resolvedCandidate ? [resolvedCandidate] : [];
|
||||
}
|
||||
|
||||
const GLOBAL_EXCLUDE_LABEL_PHRASES = [
|
||||
'pro forma',
|
||||
'reconciliation',
|
||||
'acquiree',
|
||||
'business combination',
|
||||
'assets acquired',
|
||||
'liabilities assumed'
|
||||
] as const;
|
||||
|
||||
function inferUnit(rawUnit: string | null, fallback: FinancialUnit) {
|
||||
const normalized = (rawUnit ?? '').toLowerCase();
|
||||
if (!normalized) {
|
||||
return fallback;
|
||||
}
|
||||
|
||||
if (normalized.includes('usd') || normalized.includes('iso4217')) {
|
||||
return 'currency';
|
||||
}
|
||||
|
||||
if (normalized.includes('shares')) {
|
||||
return 'shares';
|
||||
}
|
||||
|
||||
if (normalized.includes('pure') || normalized.includes('percent')) {
|
||||
return fallback === 'percent' ? 'percent' : 'ratio';
|
||||
}
|
||||
|
||||
return fallback;
|
||||
}
|
||||
|
||||
function rowUnit(row: TaxonomyStatementRow, fallback: FinancialUnit) {
|
||||
return inferUnit(Object.values(row.units)[0] ?? null, fallback);
|
||||
}
|
||||
|
||||
function isUnitCompatible(expected: FinancialUnit, actual: FinancialUnit) {
|
||||
if (expected === actual) {
|
||||
return true;
|
||||
}
|
||||
|
||||
if ((expected === 'percent' || expected === 'ratio') && (actual === 'percent' || actual === 'ratio')) {
|
||||
return true;
|
||||
}
|
||||
|
||||
return false;
|
||||
}
|
||||
|
||||
function phraseTokens(phrase: string) {
|
||||
return tokenizeLabel(phrase);
|
||||
}
|
||||
|
||||
function labelContainsPhrase(labelTokens: string[], phrase: string) {
|
||||
const target = phraseTokens(phrase);
|
||||
if (target.length === 0 || target.length > labelTokens.length) {
|
||||
return false;
|
||||
}
|
||||
|
||||
for (let index = 0; index <= labelTokens.length - target.length; index += 1) {
|
||||
let matched = true;
|
||||
for (let offset = 0; offset < target.length; offset += 1) {
|
||||
if (labelTokens[index + offset] !== target[offset]) {
|
||||
matched = false;
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
if (matched) {
|
||||
return true;
|
||||
}
|
||||
}
|
||||
|
||||
return false;
|
||||
}
|
||||
|
||||
function matchRank(matchKind: CandidateMatchKind) {
|
||||
switch (matchKind) {
|
||||
case 'exact_local_name':
|
||||
return 0;
|
||||
case 'secondary_local_name':
|
||||
return 1;
|
||||
case 'label_phrase':
|
||||
return 2;
|
||||
}
|
||||
}
|
||||
|
||||
function aliasRank(localName: string, aliases: readonly string[] | undefined) {
|
||||
const normalizedLocalName = normalizeToken(localName);
|
||||
const matchIndex = (aliases ?? []).findIndex((alias) => normalizeToken(alias) === normalizedLocalName);
|
||||
return matchIndex === -1 ? Number.MAX_SAFE_INTEGER : matchIndex;
|
||||
}
|
||||
|
||||
function applySignTransform(value: number | null, transform: StandardTemplateRowDefinition['signTransform']) {
|
||||
if (value === null || !transform) {
|
||||
return value;
|
||||
}
|
||||
|
||||
if (transform === 'invert') {
|
||||
return value * -1;
|
||||
}
|
||||
|
||||
return Math.abs(value);
|
||||
}
|
||||
|
||||
function classifyStatementRowCandidate(
|
||||
row: TaxonomyStatementRow,
|
||||
definition: StandardTemplateRowDefinition
|
||||
) {
|
||||
if (definition.selectionPolicy === 'formula_only') {
|
||||
return null;
|
||||
}
|
||||
|
||||
const rowLocalName = normalizeToken(row.localName);
|
||||
if ((definition.matchers.excludeLocalNames ?? []).some((localName) => normalizeToken(localName) === rowLocalName)) {
|
||||
return null;
|
||||
}
|
||||
|
||||
const labelTokens = tokenizeLabel(row.label);
|
||||
const excludedLabelPhrases = [
|
||||
...GLOBAL_EXCLUDE_LABEL_PHRASES,
|
||||
...(definition.matchers.excludeLabelPhrases ?? [])
|
||||
];
|
||||
if (excludedLabelPhrases.some((phrase) => labelContainsPhrase(labelTokens, phrase))) {
|
||||
return null;
|
||||
}
|
||||
|
||||
const unit = rowUnit(row, definition.unit);
|
||||
if (!isUnitCompatible(definition.unit, unit)) {
|
||||
return null;
|
||||
}
|
||||
|
||||
if ((definition.matchers.exactLocalNames ?? []).some((localName) => normalizeToken(localName) === rowLocalName)) {
|
||||
return {
|
||||
row,
|
||||
matchKind: 'exact_local_name',
|
||||
aliasRank: aliasRank(row.localName, definition.matchers.exactLocalNames),
|
||||
unit,
|
||||
labelTokenCount: labelTokens.length,
|
||||
matchedPhraseTokenCount: 0
|
||||
} satisfies StatementRowCandidate;
|
||||
}
|
||||
|
||||
if ((definition.matchers.secondaryLocalNames ?? []).some((localName) => normalizeToken(localName) === rowLocalName)) {
|
||||
return {
|
||||
row,
|
||||
matchKind: 'secondary_local_name',
|
||||
aliasRank: aliasRank(row.localName, definition.matchers.secondaryLocalNames),
|
||||
unit,
|
||||
labelTokenCount: labelTokens.length,
|
||||
matchedPhraseTokenCount: 0
|
||||
} satisfies StatementRowCandidate;
|
||||
}
|
||||
|
||||
const matchedPhrase = (definition.matchers.allowedLabelPhrases ?? [])
|
||||
.map((phrase) => ({
|
||||
phrase,
|
||||
tokenCount: phraseTokens(phrase).length
|
||||
}))
|
||||
.filter(({ phrase }) => labelContainsPhrase(labelTokens, phrase))
|
||||
.sort((left, right) => right.tokenCount - left.tokenCount)[0];
|
||||
|
||||
if (!matchedPhrase) {
|
||||
return null;
|
||||
}
|
||||
|
||||
if (row.hasDimensions) {
|
||||
return null;
|
||||
}
|
||||
|
||||
return {
|
||||
row,
|
||||
matchKind: 'label_phrase',
|
||||
aliasRank: Number.MAX_SAFE_INTEGER,
|
||||
unit,
|
||||
labelTokenCount: labelTokens.length,
|
||||
matchedPhraseTokenCount: matchedPhrase.tokenCount
|
||||
} satisfies StatementRowCandidate;
|
||||
}
|
||||
|
||||
function classifyFactCandidate(
|
||||
fact: TaxonomyFactRow,
|
||||
definition: StandardTemplateRowDefinition
|
||||
) {
|
||||
if (!fact.isDimensionless) {
|
||||
return null;
|
||||
}
|
||||
|
||||
const localName = normalizeToken(fact.localName);
|
||||
if ((definition.matchers.excludeLocalNames ?? []).some((entry) => normalizeToken(entry) === localName)) {
|
||||
return null;
|
||||
}
|
||||
|
||||
const unit = inferUnit(fact.unit ?? null, definition.unit);
|
||||
if (!isUnitCompatible(definition.unit, unit)) {
|
||||
return null;
|
||||
}
|
||||
|
||||
if ((definition.matchers.exactLocalNames ?? []).some((entry) => normalizeToken(entry) === localName)) {
|
||||
return {
|
||||
fact,
|
||||
matchKind: 'exact_local_name',
|
||||
aliasRank: aliasRank(fact.localName, definition.matchers.exactLocalNames),
|
||||
unit
|
||||
} satisfies FactCandidate;
|
||||
}
|
||||
|
||||
if ((definition.matchers.secondaryLocalNames ?? []).some((entry) => normalizeToken(entry) === localName)) {
|
||||
return {
|
||||
fact,
|
||||
matchKind: 'secondary_local_name',
|
||||
aliasRank: aliasRank(fact.localName, definition.matchers.secondaryLocalNames),
|
||||
unit
|
||||
} satisfies FactCandidate;
|
||||
}
|
||||
|
||||
return null;
|
||||
}
|
||||
|
||||
export function factMatchesPeriod(fact: TaxonomyFactRow, period: FinancialStatementPeriod) {
|
||||
if (period.periodStart) {
|
||||
return fact.periodStart === period.periodStart && fact.periodEnd === period.periodEnd;
|
||||
@@ -367,390 +27,6 @@ export function factMatchesPeriod(fact: TaxonomyFactRow, period: FinancialStatem
|
||||
return (fact.periodInstant ?? fact.periodEnd) === period.periodEnd;
|
||||
}
|
||||
|
||||
function compareStatementRowCandidates(
|
||||
left: StatementRowCandidate,
|
||||
right: StatementRowCandidate,
|
||||
definition: StandardTemplateRowDefinition
|
||||
) {
|
||||
const matchDelta = matchRank(left.matchKind) - matchRank(right.matchKind);
|
||||
if (matchDelta !== 0) {
|
||||
return matchDelta;
|
||||
}
|
||||
|
||||
if (left.aliasRank !== right.aliasRank) {
|
||||
return left.aliasRank - right.aliasRank;
|
||||
}
|
||||
|
||||
if (left.row.hasDimensions !== right.row.hasDimensions) {
|
||||
return left.row.hasDimensions ? 1 : -1;
|
||||
}
|
||||
|
||||
if (definition.selectionPolicy === 'prefer_primary_statement_concept' && left.row.isExtension !== right.row.isExtension) {
|
||||
return left.row.isExtension ? 1 : -1;
|
||||
}
|
||||
|
||||
if (left.row.order !== right.row.order) {
|
||||
return left.row.order - right.row.order;
|
||||
}
|
||||
|
||||
if (left.matchedPhraseTokenCount !== right.matchedPhraseTokenCount) {
|
||||
return right.matchedPhraseTokenCount - left.matchedPhraseTokenCount;
|
||||
}
|
||||
|
||||
if (left.labelTokenCount !== right.labelTokenCount) {
|
||||
return left.labelTokenCount - right.labelTokenCount;
|
||||
}
|
||||
|
||||
return left.row.label.localeCompare(right.row.label);
|
||||
}
|
||||
|
||||
function compareFactCandidates(left: FactCandidate, right: FactCandidate) {
|
||||
const matchDelta = matchRank(left.matchKind) - matchRank(right.matchKind);
|
||||
if (matchDelta !== 0) {
|
||||
return matchDelta;
|
||||
}
|
||||
|
||||
if (left.aliasRank !== right.aliasRank) {
|
||||
return left.aliasRank - right.aliasRank;
|
||||
}
|
||||
|
||||
return left.fact.qname.localeCompare(right.fact.qname);
|
||||
}
|
||||
|
||||
function compareResolvedCandidates(
|
||||
left: ResolvedCandidate,
|
||||
right: ResolvedCandidate,
|
||||
definition: StandardTemplateRowDefinition
|
||||
) {
|
||||
const matchDelta = matchRank(left.matchKind) - matchRank(right.matchKind);
|
||||
if (matchDelta !== 0) {
|
||||
return matchDelta;
|
||||
}
|
||||
|
||||
if (left.aliasRank !== right.aliasRank) {
|
||||
return left.aliasRank - right.aliasRank;
|
||||
}
|
||||
|
||||
if (left.sourceType === 'row' && right.sourceType === 'row') {
|
||||
return compareStatementRowCandidates(left, right, definition);
|
||||
}
|
||||
|
||||
if (left.sourceType === 'fact' && right.sourceType === 'fact') {
|
||||
return compareFactCandidates(left, right);
|
||||
}
|
||||
|
||||
if (left.sourceType === 'row' && right.sourceType === 'fact') {
|
||||
return left.row.hasDimensions ? 1 : -1;
|
||||
}
|
||||
|
||||
if (left.sourceType === 'fact' && right.sourceType === 'row') {
|
||||
return right.row.hasDimensions ? -1 : 1;
|
||||
}
|
||||
|
||||
return 0;
|
||||
}
|
||||
|
||||
function buildTemplateRow(
|
||||
definition: StandardTemplateRowDefinition,
|
||||
candidates: StatementRowCandidate[],
|
||||
factCandidates: FactCandidate[],
|
||||
periods: FinancialStatementPeriod[]
|
||||
) {
|
||||
const sourceConcepts = new Set<string>();
|
||||
const sourceRowKeys = new Set<string>();
|
||||
const sourceFactIds = new Set<number>();
|
||||
const matchedRowKeys = new Set<string>();
|
||||
const values: Record<string, number | null> = Object.fromEntries(periods.map((period) => [period.id, null]));
|
||||
const resolvedSourceRowKeys: Record<string, string | null> = Object.fromEntries(periods.map((period) => [period.id, null]));
|
||||
const metadata: InternalRowMetadata = {
|
||||
derivedRoleByPeriod: Object.fromEntries(periods.map((period) => [period.id, null]))
|
||||
};
|
||||
let unit = definition.unit;
|
||||
let hasDimensions = false;
|
||||
|
||||
for (const period of periods) {
|
||||
const resolvedCandidates = resolvedCandidatesForPeriod({
|
||||
definition,
|
||||
candidates,
|
||||
factCandidates,
|
||||
period
|
||||
});
|
||||
|
||||
if (resolvedCandidates.length === 0) {
|
||||
continue;
|
||||
}
|
||||
|
||||
if (definition.key === 'depreciation_and_amortization') {
|
||||
metadata.derivedRoleByPeriod[period.id] = resolvedCandidates.some((candidate) => {
|
||||
const localName = candidate.sourceType === 'row'
|
||||
? candidate.row.localName
|
||||
: candidate.fact.localName;
|
||||
return normalizeToken(localName) === normalizeToken('CostOfGoodsAndServicesSoldDepreciationAndAmortization');
|
||||
})
|
||||
? 'expense'
|
||||
: 'addback';
|
||||
}
|
||||
|
||||
values[period.id] = definition.selectionPolicy === 'aggregate_multiple_components'
|
||||
? sumValues(resolvedCandidates.map((candidate) => {
|
||||
if (candidate.sourceType === 'row') {
|
||||
return applySignTransform(candidate.row.values[period.id] ?? null, definition.signTransform);
|
||||
}
|
||||
|
||||
return applySignTransform(candidate.fact.value ?? null, definition.signTransform);
|
||||
}))
|
||||
: (() => {
|
||||
const resolvedCandidate = resolvedCandidates[0]!;
|
||||
if (resolvedCandidate.sourceType === 'row') {
|
||||
return applySignTransform(resolvedCandidate.row.values[period.id] ?? null, definition.signTransform);
|
||||
}
|
||||
|
||||
return applySignTransform(resolvedCandidate.fact.value ?? null, definition.signTransform);
|
||||
})();
|
||||
resolvedSourceRowKeys[period.id] = resolvedCandidates.length === 1
|
||||
? (resolvedCandidates[0]!.sourceType === 'row'
|
||||
? resolvedCandidates[0]!.row.key
|
||||
: resolvedCandidates[0]!.fact.conceptKey ?? null)
|
||||
: null;
|
||||
|
||||
for (const resolvedCandidate of resolvedCandidates) {
|
||||
unit = resolvedCandidate.unit;
|
||||
|
||||
if (resolvedCandidate.sourceType === 'row') {
|
||||
hasDimensions = hasDimensions || resolvedCandidate.row.hasDimensions;
|
||||
matchedRowKeys.add(resolvedCandidate.row.key);
|
||||
sourceConcepts.add(resolvedCandidate.row.qname);
|
||||
sourceRowKeys.add(resolvedCandidate.row.key);
|
||||
for (const factId of resolvedCandidate.row.sourceFactIds) {
|
||||
sourceFactIds.add(factId);
|
||||
}
|
||||
continue;
|
||||
}
|
||||
|
||||
sourceConcepts.add(resolvedCandidate.fact.qname);
|
||||
sourceRowKeys.add(resolvedCandidate.fact.conceptKey);
|
||||
sourceFactIds.add(resolvedCandidate.fact.id);
|
||||
}
|
||||
}
|
||||
|
||||
return {
|
||||
row: {
|
||||
key: definition.key,
|
||||
label: definition.label,
|
||||
category: definition.category,
|
||||
templateSection: definition.category,
|
||||
order: definition.order,
|
||||
unit,
|
||||
values,
|
||||
sourceConcepts: [...sourceConcepts].sort((left, right) => left.localeCompare(right)),
|
||||
sourceRowKeys: [...sourceRowKeys].sort((left, right) => left.localeCompare(right)),
|
||||
sourceFactIds: [...sourceFactIds].sort((left, right) => left - right),
|
||||
formulaKey: null,
|
||||
hasDimensions,
|
||||
resolvedSourceRowKeys
|
||||
} satisfies StandardizedFinancialRow,
|
||||
matchedRowKeys,
|
||||
metadata
|
||||
};
|
||||
}
|
||||
|
||||
function computeFormulaValue(
|
||||
formula: TemplateFormula,
|
||||
rowsByKey: Map<string, StandardizedFinancialRow>,
|
||||
periodId: string
|
||||
) {
|
||||
switch (formula.kind) {
|
||||
case 'sum':
|
||||
return sumValues(
|
||||
formula.sourceKeys.map((key) => valueOrNull(rowsByKey.get(key)?.values ?? {}, periodId)),
|
||||
formula.treatNullAsZero ?? false
|
||||
);
|
||||
case 'subtract':
|
||||
return subtractValues(
|
||||
valueOrNull(rowsByKey.get(formula.left)?.values ?? {}, periodId),
|
||||
valueOrNull(rowsByKey.get(formula.right)?.values ?? {}, periodId)
|
||||
);
|
||||
case 'divide':
|
||||
return divideValues(
|
||||
valueOrNull(rowsByKey.get(formula.numerator)?.values ?? {}, periodId),
|
||||
valueOrNull(rowsByKey.get(formula.denominator)?.values ?? {}, periodId)
|
||||
);
|
||||
}
|
||||
}
|
||||
|
||||
function rowValueForPeriod(
|
||||
rowsByKey: Map<string, StandardizedFinancialRow>,
|
||||
key: string,
|
||||
periodId: string
|
||||
) {
|
||||
return valueOrNull(rowsByKey.get(key)?.values ?? {}, periodId);
|
||||
}
|
||||
|
||||
function computeOperatingIncomeFallbackValue(
|
||||
rowsByKey: Map<string, StandardizedFinancialRow>,
|
||||
rowMetadataByKey: Map<string, InternalRowMetadata>,
|
||||
periodId: string
|
||||
) {
|
||||
const grossProfit = rowValueForPeriod(rowsByKey, 'gross_profit', periodId);
|
||||
const sellingGeneralAndAdministrative = rowValueForPeriod(rowsByKey, 'selling_general_and_administrative', periodId);
|
||||
const researchAndDevelopment = rowValueForPeriod(rowsByKey, 'research_and_development', periodId) ?? 0;
|
||||
const depreciationAndAmortization = rowValueForPeriod(rowsByKey, 'depreciation_and_amortization', periodId);
|
||||
const depreciationRole = rowMetadataByKey.get('depreciation_and_amortization')?.derivedRoleByPeriod[periodId] ?? null;
|
||||
|
||||
if (
|
||||
depreciationRole === 'expense'
|
||||
&& grossProfit !== null
|
||||
&& sellingGeneralAndAdministrative !== null
|
||||
&& depreciationAndAmortization !== null
|
||||
) {
|
||||
return grossProfit - sellingGeneralAndAdministrative - researchAndDevelopment - depreciationAndAmortization;
|
||||
}
|
||||
|
||||
const pretaxIncome = rowValueForPeriod(rowsByKey, 'pretax_income', periodId);
|
||||
if (pretaxIncome === null) {
|
||||
return null;
|
||||
}
|
||||
|
||||
const interestExpense = rowValueForPeriod(rowsByKey, 'interest_expense', periodId) ?? 0;
|
||||
const interestIncome = rowValueForPeriod(rowsByKey, 'interest_income', periodId) ?? 0;
|
||||
const otherNonOperatingIncome = rowValueForPeriod(rowsByKey, 'other_non_operating_income', periodId) ?? 0;
|
||||
|
||||
return pretaxIncome + interestExpense - interestIncome - otherNonOperatingIncome;
|
||||
}
|
||||
|
||||
function computeFallbackValueForDefinition(
|
||||
definition: StandardTemplateRowDefinition,
|
||||
rowsByKey: Map<string, StandardizedFinancialRow>,
|
||||
rowMetadataByKey: Map<string, InternalRowMetadata>,
|
||||
periodId: string
|
||||
) {
|
||||
if (definition.key === 'operating_income') {
|
||||
return computeOperatingIncomeFallbackValue(rowsByKey, rowMetadataByKey, periodId);
|
||||
}
|
||||
|
||||
if (!definition.fallbackFormula) {
|
||||
return null;
|
||||
}
|
||||
|
||||
return computeFormulaValue(definition.fallbackFormula, rowsByKey, periodId);
|
||||
}
|
||||
|
||||
function applyFormulas(
|
||||
rowsByKey: Map<string, StandardizedFinancialRow>,
|
||||
rowMetadataByKey: Map<string, InternalRowMetadata>,
|
||||
definitions: StandardTemplateRowDefinition[],
|
||||
periods: FinancialStatementPeriod[]
|
||||
) {
|
||||
for (let pass = 0; pass < definitions.length; pass += 1) {
|
||||
let changed = false;
|
||||
|
||||
for (const definition of definitions) {
|
||||
if (!definition.fallbackFormula && definition.key !== 'operating_income') {
|
||||
continue;
|
||||
}
|
||||
|
||||
const target = rowsByKey.get(definition.key);
|
||||
if (!target) {
|
||||
continue;
|
||||
}
|
||||
|
||||
let usedFormula = target.formulaKey !== null;
|
||||
for (const period of periods) {
|
||||
if (definition.selectionPolicy !== 'formula_only' && target.values[period.id] !== null) {
|
||||
continue;
|
||||
}
|
||||
|
||||
const computed = computeFallbackValueForDefinition(definition, rowsByKey, rowMetadataByKey, period.id);
|
||||
if (computed === null) {
|
||||
continue;
|
||||
}
|
||||
|
||||
target.values[period.id] = applySignTransform(computed, definition.signTransform);
|
||||
target.resolvedSourceRowKeys[period.id] = null;
|
||||
usedFormula = true;
|
||||
changed = true;
|
||||
}
|
||||
|
||||
if (usedFormula) {
|
||||
target.formulaKey = definition.key;
|
||||
}
|
||||
}
|
||||
|
||||
if (!changed) {
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
export function buildStandardizedRows(input: {
|
||||
rows: TaxonomyStatementRow[];
|
||||
statement: Extract<FinancialStatementKind, 'income' | 'balance' | 'cash_flow'>;
|
||||
periods: FinancialStatementPeriod[];
|
||||
facts: TaxonomyFactRow[];
|
||||
}) {
|
||||
const definitions = STANDARD_FINANCIAL_TEMPLATES[input.statement];
|
||||
const rowsByKey = new Map<string, StandardizedFinancialRow>();
|
||||
const rowMetadataByKey = new Map<string, InternalRowMetadata>();
|
||||
const matchedRowKeys = new Set<string>();
|
||||
|
||||
for (const definition of definitions) {
|
||||
const candidates = input.rows
|
||||
.map((row) => classifyStatementRowCandidate(row, definition))
|
||||
.filter((candidate): candidate is StatementRowCandidate => candidate !== null);
|
||||
const factCandidates = input.facts
|
||||
.map((fact) => classifyFactCandidate(fact, definition))
|
||||
.filter((candidate): candidate is FactCandidate => candidate !== null);
|
||||
const templateRow = buildTemplateRow(definition, candidates, factCandidates, input.periods);
|
||||
|
||||
for (const rowKey of templateRow.matchedRowKeys) {
|
||||
matchedRowKeys.add(rowKey);
|
||||
}
|
||||
|
||||
const hasAnyValue = Object.values(templateRow.row.values).some((value) => value !== null);
|
||||
if (hasAnyValue || definition.fallbackFormula || definition.key === 'operating_income') {
|
||||
rowsByKey.set(definition.key, templateRow.row);
|
||||
rowMetadataByKey.set(definition.key, templateRow.metadata);
|
||||
}
|
||||
}
|
||||
|
||||
applyFormulas(rowsByKey, rowMetadataByKey, definitions, input.periods);
|
||||
|
||||
const templateRows = definitions
|
||||
.filter((definition) => definition.includeInOutput !== false)
|
||||
.map((definition) => rowsByKey.get(definition.key))
|
||||
.filter((row): row is StandardizedFinancialRow => row !== undefined);
|
||||
const coveredTemplateSourceRowKeys = new Set(templateRows.flatMap((row) => row.sourceRowKeys));
|
||||
|
||||
const unmatchedRows = input.rows
|
||||
.filter((row) => !matchedRowKeys.has(row.key))
|
||||
.filter((row) => !(row.hasDimensions && coveredTemplateSourceRowKeys.has(row.key)))
|
||||
.map((row) => ({
|
||||
key: `other:${row.key}`,
|
||||
label: row.label,
|
||||
category: 'other',
|
||||
templateSection: 'other',
|
||||
order: 10_000 + row.order,
|
||||
unit: inferUnit(Object.values(row.units)[0] ?? null, 'currency'),
|
||||
values: { ...row.values },
|
||||
sourceConcepts: [row.qname],
|
||||
sourceRowKeys: [row.key],
|
||||
sourceFactIds: [...row.sourceFactIds],
|
||||
formulaKey: null,
|
||||
hasDimensions: row.hasDimensions,
|
||||
resolvedSourceRowKeys: Object.fromEntries(
|
||||
input.periods.map((period) => [period.id, period.id in row.values ? row.key : null])
|
||||
)
|
||||
} satisfies StandardizedFinancialRow));
|
||||
|
||||
return [...templateRows, ...unmatchedRows].sort((left, right) => {
|
||||
if (left.order !== right.order) {
|
||||
return left.order - right.order;
|
||||
}
|
||||
|
||||
return left.label.localeCompare(right.label);
|
||||
});
|
||||
}
|
||||
|
||||
export function buildDimensionBreakdown(
|
||||
facts: TaxonomyFactRow[],
|
||||
periods: FinancialStatementPeriod[],
|
||||
|
||||
@@ -1,320 +0,0 @@
|
||||
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
|
||||
};
|
||||
}
|
||||
@@ -1,400 +0,0 @@
|
||||
import type { Filing, FinancialStatementKind, TaxonomyStatementRow } from '@/lib/types';
|
||||
import type { TaxonomyConcept, TaxonomyFact, TaxonomyPresentationConcept } from '@/lib/server/taxonomy/types';
|
||||
import type { FilingTaxonomyPeriod } from '@/lib/server/repos/filing-taxonomy';
|
||||
import { classifyStatementRole, conceptStatementFallback } from '@/lib/server/taxonomy/classifiers';
|
||||
|
||||
function compactAccessionNumber(value: string) {
|
||||
return value.replace(/-/g, '');
|
||||
}
|
||||
|
||||
function isUsGaapNamespace(namespaceUri: string) {
|
||||
return /fasb\.org\/us-gaap/i.test(namespaceUri) || /us-gaap/i.test(namespaceUri);
|
||||
}
|
||||
|
||||
function splitConceptKey(conceptKey: string) {
|
||||
const index = conceptKey.lastIndexOf('#');
|
||||
if (index < 0) {
|
||||
return {
|
||||
namespaceUri: 'urn:unknown',
|
||||
localName: conceptKey
|
||||
};
|
||||
}
|
||||
|
||||
return {
|
||||
namespaceUri: conceptKey.slice(0, index),
|
||||
localName: conceptKey.slice(index + 1)
|
||||
};
|
||||
}
|
||||
|
||||
function localNameToLabel(localName: string) {
|
||||
return localName
|
||||
.replace(/([a-z0-9])([A-Z])/g, '$1 $2')
|
||||
.replace(/([A-Z]+)([A-Z][a-z])/g, '$1 $2')
|
||||
.replace(/_/g, ' ')
|
||||
.trim();
|
||||
}
|
||||
|
||||
function createStatementRecord<T>(factory: () => T): Record<FinancialStatementKind, T> {
|
||||
return {
|
||||
income: factory(),
|
||||
balance: factory(),
|
||||
cash_flow: factory(),
|
||||
equity: factory(),
|
||||
comprehensive_income: factory()
|
||||
};
|
||||
}
|
||||
|
||||
function periodSignature(fact: TaxonomyFact) {
|
||||
const start = fact.periodStart ?? '';
|
||||
const end = fact.periodEnd ?? '';
|
||||
const instant = fact.periodInstant ?? '';
|
||||
return `start:${start}|end:${end}|instant:${instant}`;
|
||||
}
|
||||
|
||||
function periodDate(fact: TaxonomyFact, fallbackDate: string) {
|
||||
return fact.periodEnd ?? fact.periodInstant ?? fallbackDate;
|
||||
}
|
||||
|
||||
function parseEpoch(value: string | null) {
|
||||
if (!value) {
|
||||
return Number.NaN;
|
||||
}
|
||||
|
||||
return Date.parse(value);
|
||||
}
|
||||
|
||||
function sortPeriods(periods: FilingTaxonomyPeriod[]) {
|
||||
return [...periods].sort((left, right) => {
|
||||
const leftDate = parseEpoch(left.periodEnd ?? left.filingDate);
|
||||
const rightDate = parseEpoch(right.periodEnd ?? right.filingDate);
|
||||
|
||||
if (Number.isFinite(leftDate) && Number.isFinite(rightDate) && leftDate !== rightDate) {
|
||||
return leftDate - rightDate;
|
||||
}
|
||||
|
||||
return left.id.localeCompare(right.id);
|
||||
});
|
||||
}
|
||||
|
||||
function pickPreferredFact<T extends TaxonomyFact>(facts: T[]) {
|
||||
if (facts.length === 0) {
|
||||
return null;
|
||||
}
|
||||
|
||||
const ordered = [...facts].sort((left, right) => {
|
||||
const leftScore = left.isDimensionless ? 1 : 0;
|
||||
const rightScore = right.isDimensionless ? 1 : 0;
|
||||
if (leftScore !== rightScore) {
|
||||
return rightScore - leftScore;
|
||||
}
|
||||
|
||||
const leftDate = parseEpoch(left.periodEnd ?? left.periodInstant);
|
||||
const rightDate = parseEpoch(right.periodEnd ?? right.periodInstant);
|
||||
if (Number.isFinite(leftDate) && Number.isFinite(rightDate) && leftDate !== rightDate) {
|
||||
return rightDate - leftDate;
|
||||
}
|
||||
|
||||
return Math.abs(right.value) - Math.abs(left.value);
|
||||
});
|
||||
|
||||
return ordered[0] ?? null;
|
||||
}
|
||||
|
||||
export function materializeTaxonomyStatements(input: {
|
||||
filingId: number;
|
||||
accessionNumber: string;
|
||||
filingDate: string;
|
||||
filingType: '10-K' | '10-Q';
|
||||
facts: TaxonomyFact[];
|
||||
presentation: TaxonomyPresentationConcept[];
|
||||
labelByConcept: Map<string, string>;
|
||||
}) {
|
||||
const periodBySignature = new Map<string, FilingTaxonomyPeriod>();
|
||||
const compactAccession = compactAccessionNumber(input.accessionNumber);
|
||||
|
||||
for (const fact of input.facts) {
|
||||
const signature = periodSignature(fact);
|
||||
if (periodBySignature.has(signature)) {
|
||||
continue;
|
||||
}
|
||||
|
||||
const date = periodDate(fact, input.filingDate);
|
||||
const id = `${date}-${compactAccession}-${periodBySignature.size + 1}`;
|
||||
|
||||
periodBySignature.set(signature, {
|
||||
id,
|
||||
filingId: input.filingId,
|
||||
accessionNumber: input.accessionNumber,
|
||||
filingDate: input.filingDate,
|
||||
periodStart: fact.periodStart,
|
||||
periodEnd: fact.periodEnd ?? fact.periodInstant ?? input.filingDate,
|
||||
filingType: input.filingType,
|
||||
periodLabel: fact.periodInstant && !fact.periodStart
|
||||
? 'Instant'
|
||||
: fact.periodStart && fact.periodEnd
|
||||
? `${fact.periodStart} to ${fact.periodEnd}`
|
||||
: 'Filing Period'
|
||||
});
|
||||
}
|
||||
|
||||
const periods = sortPeriods([...periodBySignature.values()]);
|
||||
const periodIdBySignature = new Map<string, string>(
|
||||
[...periodBySignature.entries()].map(([signature, period]) => [signature, period.id])
|
||||
);
|
||||
|
||||
const presentationByConcept = new Map<string, TaxonomyPresentationConcept[]>();
|
||||
for (const node of input.presentation) {
|
||||
const existing = presentationByConcept.get(node.conceptKey);
|
||||
if (existing) {
|
||||
existing.push(node);
|
||||
} else {
|
||||
presentationByConcept.set(node.conceptKey, [node]);
|
||||
}
|
||||
}
|
||||
|
||||
const enrichedFacts = input.facts.map((fact, index) => {
|
||||
const nodes = presentationByConcept.get(fact.conceptKey) ?? [];
|
||||
const bestNode = nodes[0] ?? null;
|
||||
const statementKind = bestNode
|
||||
? classifyStatementRole(bestNode.roleUri)
|
||||
: conceptStatementFallback(fact.localName);
|
||||
|
||||
return {
|
||||
...fact,
|
||||
__sourceFactId: index + 1,
|
||||
statement_kind: statementKind,
|
||||
role_uri: bestNode?.roleUri ?? null
|
||||
};
|
||||
});
|
||||
|
||||
const rowsByStatement = createStatementRecord<TaxonomyStatementRow[]>(() => []);
|
||||
const conceptByKey = new Map<string, TaxonomyConcept>();
|
||||
const groupedByStatement = createStatementRecord<Map<string, typeof enrichedFacts>>(() => new Map());
|
||||
|
||||
for (const fact of enrichedFacts) {
|
||||
if (!fact.statement_kind) {
|
||||
continue;
|
||||
}
|
||||
|
||||
const group = groupedByStatement[fact.statement_kind].get(fact.conceptKey);
|
||||
if (group) {
|
||||
group.push(fact);
|
||||
} else {
|
||||
groupedByStatement[fact.statement_kind].set(fact.conceptKey, [fact]);
|
||||
}
|
||||
}
|
||||
|
||||
for (const statement of Object.keys(rowsByStatement) as FinancialStatementKind[]) {
|
||||
const conceptKeys = new Set<string>();
|
||||
|
||||
for (const node of input.presentation) {
|
||||
if (classifyStatementRole(node.roleUri) !== statement) {
|
||||
continue;
|
||||
}
|
||||
|
||||
conceptKeys.add(node.conceptKey);
|
||||
}
|
||||
|
||||
for (const conceptKey of groupedByStatement[statement].keys()) {
|
||||
conceptKeys.add(conceptKey);
|
||||
}
|
||||
|
||||
const orderedConcepts = [...conceptKeys]
|
||||
.map((conceptKey) => {
|
||||
const presentationNodes = input.presentation.filter(
|
||||
(node) => node.conceptKey === conceptKey && classifyStatementRole(node.roleUri) === statement
|
||||
);
|
||||
const presentationOrder = presentationNodes.length > 0
|
||||
? Math.min(...presentationNodes.map((node) => node.order))
|
||||
: Number.MAX_SAFE_INTEGER;
|
||||
const presentationDepth = presentationNodes.length > 0
|
||||
? Math.min(...presentationNodes.map((node) => node.depth))
|
||||
: 0;
|
||||
const roleUri = presentationNodes[0]?.roleUri ?? null;
|
||||
const parentConceptKey = presentationNodes[0]?.parentConceptKey ?? null;
|
||||
return {
|
||||
conceptKey,
|
||||
presentationOrder,
|
||||
presentationDepth,
|
||||
roleUri,
|
||||
parentConceptKey
|
||||
};
|
||||
})
|
||||
.sort((left, right) => {
|
||||
if (left.presentationOrder !== right.presentationOrder) {
|
||||
return left.presentationOrder - right.presentationOrder;
|
||||
}
|
||||
|
||||
return left.conceptKey.localeCompare(right.conceptKey);
|
||||
});
|
||||
|
||||
for (const orderedConcept of orderedConcepts) {
|
||||
const facts = groupedByStatement[statement].get(orderedConcept.conceptKey) ?? [];
|
||||
const { namespaceUri, localName } = splitConceptKey(orderedConcept.conceptKey);
|
||||
const qname = facts[0]?.qname ?? `unknown:${localName}`;
|
||||
const label = input.labelByConcept.get(orderedConcept.conceptKey) ?? localNameToLabel(localName);
|
||||
const values: Record<string, number | null> = {};
|
||||
const units: Record<string, string | null> = {};
|
||||
|
||||
const factGroups = new Map<string, typeof facts>();
|
||||
for (const fact of facts) {
|
||||
const signature = periodSignature(fact);
|
||||
const group = factGroups.get(signature);
|
||||
if (group) {
|
||||
group.push(fact);
|
||||
} else {
|
||||
factGroups.set(signature, [fact]);
|
||||
}
|
||||
}
|
||||
|
||||
const sourceFactIds: number[] = [];
|
||||
let hasDimensions = false;
|
||||
for (const [signature, group] of factGroups.entries()) {
|
||||
const periodId = periodIdBySignature.get(signature);
|
||||
if (!periodId) {
|
||||
continue;
|
||||
}
|
||||
|
||||
const preferred = pickPreferredFact(group);
|
||||
if (!preferred) {
|
||||
continue;
|
||||
}
|
||||
|
||||
values[periodId] = preferred.value;
|
||||
units[periodId] = preferred.unit;
|
||||
const sourceFactId = (preferred as { __sourceFactId?: number }).__sourceFactId;
|
||||
if (typeof sourceFactId === 'number') {
|
||||
sourceFactIds.push(sourceFactId);
|
||||
}
|
||||
|
||||
if (group.some((entry) => !entry.isDimensionless)) {
|
||||
hasDimensions = true;
|
||||
}
|
||||
}
|
||||
|
||||
if (Object.keys(values).length === 0) {
|
||||
continue;
|
||||
}
|
||||
|
||||
const row: TaxonomyStatementRow = {
|
||||
key: orderedConcept.conceptKey,
|
||||
label,
|
||||
conceptKey: orderedConcept.conceptKey,
|
||||
qname,
|
||||
namespaceUri,
|
||||
localName,
|
||||
isExtension: !isUsGaapNamespace(namespaceUri),
|
||||
statement,
|
||||
roleUri: orderedConcept.roleUri,
|
||||
order: Number.isFinite(orderedConcept.presentationOrder)
|
||||
? orderedConcept.presentationOrder
|
||||
: rowsByStatement[statement].length + 1,
|
||||
depth: orderedConcept.presentationDepth,
|
||||
parentKey: orderedConcept.parentConceptKey,
|
||||
values,
|
||||
units,
|
||||
hasDimensions,
|
||||
sourceFactIds
|
||||
};
|
||||
|
||||
rowsByStatement[statement].push(row);
|
||||
|
||||
if (!conceptByKey.has(orderedConcept.conceptKey)) {
|
||||
conceptByKey.set(orderedConcept.conceptKey, {
|
||||
concept_key: orderedConcept.conceptKey,
|
||||
qname,
|
||||
namespace_uri: namespaceUri,
|
||||
local_name: localName,
|
||||
label,
|
||||
is_extension: !isUsGaapNamespace(namespaceUri),
|
||||
balance: null,
|
||||
period_type: null,
|
||||
data_type: null,
|
||||
statement_kind: statement,
|
||||
role_uri: orderedConcept.roleUri,
|
||||
authoritative_concept_key: null,
|
||||
mapping_method: null,
|
||||
surface_key: null,
|
||||
detail_parent_surface_key: null,
|
||||
kpi_key: null,
|
||||
residual_flag: false,
|
||||
presentation_order: row.order,
|
||||
presentation_depth: row.depth,
|
||||
parent_concept_key: row.parentKey,
|
||||
is_abstract: /abstract/i.test(localName)
|
||||
});
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
for (const fact of enrichedFacts) {
|
||||
if (conceptByKey.has(fact.conceptKey)) {
|
||||
continue;
|
||||
}
|
||||
|
||||
conceptByKey.set(fact.conceptKey, {
|
||||
concept_key: fact.conceptKey,
|
||||
qname: fact.qname,
|
||||
namespace_uri: fact.namespaceUri,
|
||||
local_name: fact.localName,
|
||||
label: input.labelByConcept.get(fact.conceptKey) ?? localNameToLabel(fact.localName),
|
||||
is_extension: !isUsGaapNamespace(fact.namespaceUri),
|
||||
balance: null,
|
||||
period_type: null,
|
||||
data_type: fact.dataType,
|
||||
statement_kind: fact.statement_kind,
|
||||
role_uri: fact.role_uri,
|
||||
authoritative_concept_key: null,
|
||||
mapping_method: null,
|
||||
surface_key: null,
|
||||
detail_parent_surface_key: null,
|
||||
kpi_key: null,
|
||||
residual_flag: false,
|
||||
presentation_order: null,
|
||||
presentation_depth: null,
|
||||
parent_concept_key: null,
|
||||
is_abstract: /abstract/i.test(fact.localName)
|
||||
});
|
||||
}
|
||||
|
||||
const concepts = [...conceptByKey.values()];
|
||||
const factRows = enrichedFacts.map((fact) => ({
|
||||
concept_key: fact.conceptKey,
|
||||
qname: fact.qname,
|
||||
namespace_uri: fact.namespaceUri,
|
||||
local_name: fact.localName,
|
||||
data_type: fact.dataType,
|
||||
statement_kind: fact.statement_kind,
|
||||
role_uri: fact.role_uri,
|
||||
authoritative_concept_key: null,
|
||||
mapping_method: null,
|
||||
surface_key: null,
|
||||
detail_parent_surface_key: null,
|
||||
kpi_key: null,
|
||||
residual_flag: false,
|
||||
context_id: fact.contextId,
|
||||
unit: fact.unit,
|
||||
decimals: fact.decimals,
|
||||
precision: fact.precision,
|
||||
nil: fact.nil,
|
||||
value_num: fact.value,
|
||||
period_start: fact.periodStart,
|
||||
period_end: fact.periodEnd,
|
||||
period_instant: fact.periodInstant,
|
||||
dimensions: fact.dimensions,
|
||||
is_dimensionless: fact.isDimensionless,
|
||||
source_file: fact.sourceFile,
|
||||
}));
|
||||
|
||||
const dimensionsCount = enrichedFacts.reduce((total, fact) => {
|
||||
return total + fact.dimensions.length;
|
||||
}, 0);
|
||||
|
||||
return {
|
||||
periods,
|
||||
statement_rows: rowsByStatement,
|
||||
concepts,
|
||||
facts: factRows,
|
||||
dimensionsCount
|
||||
};
|
||||
}
|
||||
Reference in New Issue
Block a user