Neon-Desk/lib/server/financials/kpi-dimensions.ts

import type {
  FinancialStatementPeriod,
  StructuredKpiRow,
  TaxonomyFactRow
} from '@/lib/types';
import type { KpiDefinition } from '@/lib/server/financials/kpi-registry';
import { factMatchesPeriod } from '@/lib/server/financials/standardize';

function normalizeSegmentToken(value: string) {
  return value.trim().toLowerCase().replace(/[^a-z0-9]+/g, '_').replace(/^_+|_+$/g, '');
}

function humanizeMember(value: string) {
  const source = value.split(':').pop() ?? value;
  return source
    .replace(/Member$/i, '')
    .replace(/([a-z])([A-Z])/g, '$1 $2')
    .replace(/_/g, ' ')
    .trim();
}

function factMatchesDefinition(fact: TaxonomyFactRow, definition: KpiDefinition) {
  if (definition.preferredConceptNames && !definition.preferredConceptNames.includes(fact.localName)) {
    return false;
  }

  if (!definition.preferredAxisIncludes || definition.preferredAxisIncludes.length === 0) {
    return fact.dimensions.length > 0;
  }

  return fact.dimensions.some((dimension) => {
    const axisMatch = definition.preferredAxisIncludes?.some((token) => dimension.axis.toLowerCase().includes(token.toLowerCase())) ?? false;
    const memberMatch = definition.preferredMemberIncludes && definition.preferredMemberIncludes.length > 0
      ? definition.preferredMemberIncludes.some((token) => dimension.member.toLowerCase().includes(token.toLowerCase()))
      : true;
    return axisMatch && memberMatch;
  });
}

function categoryForDefinition(definition: KpiDefinition, axis: string) {
  if (definition.key === 'segment_revenue' && /geo|country|region|area/i.test(axis)) {
    return 'geographic_mix';
  }

  return definition.category;
}

export function extractStructuredKpisFromDimensions(input: {
  facts: TaxonomyFactRow[];
  periods: FinancialStatementPeriod[];
  definitions: KpiDefinition[];
}) {
  const rowMap = new Map<string, StructuredKpiRow>();
  const orderByKey = new Map<string, number>();

  input.definitions.forEach((definition, index) => {
    orderByKey.set(definition.key, (index + 1) * 10);
  });

  for (const definition of input.definitions) {
    for (const fact of input.facts) {
      if (fact.dimensions.length === 0 || !factMatchesDefinition(fact, definition)) {
        continue;
      }

      const matchedPeriod = input.periods.find((period) => period.filingId === fact.filingId && factMatchesPeriod(fact, period));
      if (!matchedPeriod) {
        continue;
      }

      for (const dimension of fact.dimensions) {
        const axis = dimension.axis;
        const member = dimension.member;
        const normalizedAxis = normalizeSegmentToken(axis);
        const normalizedMember = normalizeSegmentToken(member);
        const key = `${definition.key}__${normalizedAxis}__${normalizedMember}`;
        const labelSuffix = humanizeMember(member);
        const existing = rowMap.get(key);

        if (existing) {
          existing.values[matchedPeriod.id] = fact.value;
          if (!existing.sourceConcepts.includes(fact.qname)) {
            existing.sourceConcepts.push(fact.qname);
          }
          if (!existing.sourceFactIds.includes(fact.id)) {
            existing.sourceFactIds.push(fact.id);
          }
          continue;
        }

        rowMap.set(key, {
          key,
          label: `${definition.label} - ${labelSuffix}`,
          category: categoryForDefinition(definition, axis),
          unit: definition.unit,
          order: orderByKey.get(definition.key) ?? 999,
          segment: labelSuffix || null,
          axis,
          member,
          values: { [matchedPeriod.id]: fact.value },
          sourceConcepts: [fact.qname],
          sourceFactIds: [fact.id],
          provenanceType: 'taxonomy',
          hasDimensions: true
        });
      }
    }
  }

  const rows = [...rowMap.values()].sort((left, right) => {
    if (left.order !== right.order) {
      return left.order - right.order;
    }

    return left.label.localeCompare(right.label);
  });

  const marginRows = new Map<string, StructuredKpiRow>();
  for (const row of rows.filter((entry) => entry.category === 'segment_profit')) {
    const revenueKey = row.key.replace(/^segment_profit__/, 'segment_revenue__');
    const revenueRow = rowMap.get(revenueKey);
    if (!revenueRow) {
      continue;
    }

    const values: Record<string, number | null> = {};
    for (const period of input.periods) {
      const revenue = revenueRow.values[period.id] ?? null;
      const profit = row.values[period.id] ?? null;
      values[period.id] = revenue === null || profit === null || revenue === 0
        ? null
        : profit / revenue;
    }

    marginRows.set(row.key.replace(/^segment_profit__/, 'segment_margin__'), {
      key: row.key.replace(/^segment_profit__/, 'segment_margin__'),
      label: row.label.replace(/^Segment Profit/, 'Segment Margin'),
      category: 'segment_margin',
      unit: 'percent',
      order: 25,
      segment: row.segment,
      axis: row.axis,
      member: row.member,
      values,
      sourceConcepts: [...new Set([...row.sourceConcepts, ...revenueRow.sourceConcepts])],
      sourceFactIds: [...new Set([...row.sourceFactIds, ...revenueRow.sourceFactIds])],
      provenanceType: 'taxonomy',
      hasDimensions: true
    });
  }

  return [...rows, ...marginRows.values()].sort((left, right) => {
    if (left.order !== right.order) {
      return left.order - right.order;
    }

    return left.label.localeCompare(right.label);
  });
}