Neon-Desk/lib/server/taxonomy/xbrl-parser.ts

import type { FinancialStatementKind } from '@/lib/types';
import type { TaxonomyContext, TaxonomyFact, TaxonomyNamespaceMap, TaxonomyUnit } from '@/lib/server/taxonomy/types';

function decodeXmlEntities(value: string) {
  return value
    .replace(/&/gi, '&')
    .replace(/&lt;/gi, '<')
    .replace(/&gt;/gi, '>')
    .replace(/&quot;/gi, '"')
    .replace(/&#39;/gi, "'")
    .replace(/&#160;|&nbsp;/gi, ' ')
    .replace(/&#x([0-9a-f]+);/gi, (_match, hex) => {
      const parsed = Number.parseInt(hex, 16);
      if (!Number.isFinite(parsed)) {
        return ' ';
      }

      try {
        return String.fromCodePoint(parsed);
      } catch {
        return ' ';
      }
    })
    .replace(/&#([0-9]+);/g, (_match, numeric) => {
      const parsed = Number.parseInt(numeric, 10);
      if (!Number.isFinite(parsed)) {
        return ' ';
      }

      try {
        return String.fromCodePoint(parsed);
      } catch {
        return ' ';
      }
    });
}

function parseNumber(value: string) {
  const trimmed = value.trim();
  if (!trimmed) {
    return null;
  }

  if (/^--+$/.test(trimmed)) {
    return null;
  }

  const negative = trimmed.startsWith('(') && trimmed.endsWith(')');
  const normalized = trimmed
    .replace(/<[^>]+>/g, ' ')
    .replace(/[,$\s]/g, '')
    .replace(/[()]/g, '')
    .replace(/\u2212/g, '-');

  if (!normalized) {
    return null;
  }

  const parsed = Number.parseFloat(normalized);
  if (!Number.isFinite(parsed)) {
    return null;
  }

  return negative ? -Math.abs(parsed) : parsed;
}

function parseNamespaceMapFromDocument(raw: string): TaxonomyNamespaceMap {
  const map: TaxonomyNamespaceMap = {};
  const rootStart = raw.match(/<[^>]*xbrl[^>]*>/i)?.[0] ?? raw.slice(0, 1200);

  for (const match of rootStart.matchAll(/xmlns:([a-zA-Z0-9_\-]+)=["']([^"']+)["']/g)) {
    const prefix = (match[1] ?? '').trim();
    const uri = (match[2] ?? '').trim();

    if (!prefix || !uri) {
      continue;
    }

    map[prefix] = uri;
  }

  return map;
}

function parseContexts(raw: string): Record<string, TaxonomyContext> {
  const contexts: Record<string, TaxonomyContext> = {};
  const contextPattern = /<(?:[a-z0-9_\-]+:)?context\b[^>]*\bid=["']([^"']+)["'][^>]*>([\s\S]*?)<\/(?:[a-z0-9_\-]+:)?context>/gi;

  for (const match of raw.matchAll(contextPattern)) {
    const contextId = (match[1] ?? '').trim();
    const block = match[2] ?? '';
    if (!contextId) {
      continue;
    }

    const periodStart = block.match(/<(?:[a-z0-9_\-]+:)?startDate>([^<]+)<\/(?:[a-z0-9_\-]+:)?startDate>/i)?.[1]?.trim() ?? null;
    const periodEnd = block.match(/<(?:[a-z0-9_\-]+:)?endDate>([^<]+)<\/(?:[a-z0-9_\-]+:)?endDate>/i)?.[1]?.trim() ?? null;
    const periodInstant = block.match(/<(?:[a-z0-9_\-]+:)?instant>([^<]+)<\/(?:[a-z0-9_\-]+:)?instant>/i)?.[1]?.trim() ?? null;

    const dimensions: Array<{ axis: string; member: string }> = [];
    const dimPattern = /<(?:[a-z0-9_\-]+:)?explicitMember\b[^>]*\bdimension=["']([^"']+)["'][^>]*>([^<]+)<\/(?:[a-z0-9_\-]+:)?explicitMember>/gi;
    for (const dimMatch of block.matchAll(dimPattern)) {
      const axis = decodeXmlEntities((dimMatch[1] ?? '').trim());
      const member = decodeXmlEntities((dimMatch[2] ?? '').trim());
      if (!axis || !member) {
        continue;
      }

      dimensions.push({ axis, member });
    }

    contexts[contextId] = {
      id: contextId,
      periodStart,
      periodEnd,
      periodInstant,
      dimensions
    };
  }

  return contexts;
}

function parseUnits(raw: string): Record<string, TaxonomyUnit> {
  const units: Record<string, TaxonomyUnit> = {};
  const unitPattern = /<(?:[a-z0-9_\-]+:)?unit\b[^>]*\bid=["']([^"']+)["'][^>]*>([\s\S]*?)<\/(?:[a-z0-9_\-]+:)?unit>/gi;

  for (const match of raw.matchAll(unitPattern)) {
    const id = (match[1] ?? '').trim();
    const block = match[2] ?? '';
    if (!id) {
      continue;
    }

    const measures = [...block.matchAll(/<(?:[a-z0-9_\-]+:)?measure>([^<]+)<\/(?:[a-z0-9_\-]+:)?measure>/gi)]
      .map((entry) => decodeXmlEntities((entry[1] ?? '').trim()))
      .filter(Boolean);

    let measure: string | null = null;
    if (measures.length === 1) {
      measure = measures[0] ?? null;
    } else if (measures.length > 1) {
      measure = measures.join('/');
    }

    units[id] = {
      id,
      measure
    };
  }

  return units;
}

function classifyStatementKind(localName: string): FinancialStatementKind | null {
  const normalized = localName.toLowerCase();

  if (/cash|operatingactivities|investingactivities|financingactivities/.test(normalized)) {
    return 'cash_flow';
  }

  if (/equity|retainedearnings|additionalpaidincapital/.test(normalized)) {
    return 'equity';
  }

  if (/comprehensiveincome/.test(normalized)) {
    return 'comprehensive_income';
  }

  if (/asset|liabilit|debt/.test(normalized)) {
    return 'balance';
  }

  if (/revenue|income|profit|expense|costof/.test(normalized)) {
    return 'income';
  }

  return null;
}

function isXbrlInfrastructurePrefix(prefix: string) {
  const normalized = prefix.toLowerCase();
  return normalized === 'xbrli'
    || normalized === 'xlink'
    || normalized === 'link'
    || normalized === 'xbrldi'
    || normalized === 'xbrldt';
}

function localNameToKey(namespaceUri: string, localName: string) {
  return `${namespaceUri}#${localName}`;
}

export function parseXbrlInstance(
  raw: string,
  sourceFile: string | null
): {
  namespaces: TaxonomyNamespaceMap;
  contexts: Record<string, TaxonomyContext>;
  units: Record<string, TaxonomyUnit>;
  facts: TaxonomyFact[];
} {
  const namespaces = parseNamespaceMapFromDocument(raw);
  const contexts = parseContexts(raw);
  const units = parseUnits(raw);
  const facts: TaxonomyFact[] = [];

  const factPattern = /<([a-zA-Z0-9_\-]+):([a-zA-Z0-9_\-.]+)\b([^>]*\bcontextRef=["'][^"']+["'][^>]*)>([\s\S]*?)<\/\1:\2>/g;

  for (const match of raw.matchAll(factPattern)) {
    const prefix = (match[1] ?? '').trim();
    const localName = (match[2] ?? '').trim();
    const attrs = match[3] ?? '';
    const body = decodeXmlEntities((match[4] ?? '').trim());

    if (!prefix || !localName || isXbrlInfrastructurePrefix(prefix)) {
      continue;
    }

    const contextId = attrs.match(/\bcontextRef=["']([^"']+)["']/i)?.[1]?.trim() ?? '';
    if (!contextId) {
      continue;
    }

    const value = parseNumber(body);
    if (value === null) {
      continue;
    }

    const unitRef = attrs.match(/\bunitRef=["']([^"']+)["']/i)?.[1]?.trim() ?? null;
    const decimals = attrs.match(/\bdecimals=["']([^"']+)["']/i)?.[1]?.trim() ?? null;

    const namespaceUri = namespaces[prefix] ?? `urn:unknown:${prefix}`;
    const context = contexts[contextId];

    facts.push({
      conceptKey: localNameToKey(namespaceUri, localName),
      qname: `${prefix}:${localName}`,
      namespaceUri,
      localName,
      contextId,
      unit: unitRef && units[unitRef]?.measure ? units[unitRef]?.measure ?? unitRef : unitRef,
      decimals,
      value,
      periodStart: context?.periodStart ?? null,
      periodEnd: context?.periodEnd ?? null,
      periodInstant: context?.periodInstant ?? null,
      dimensions: context?.dimensions ?? [],
      isDimensionless: (context?.dimensions.length ?? 0) === 0,
      sourceFile,
    });
  }

  return {
    namespaces,
    contexts,
    units,
    facts
  };
}

export function conceptStatementFallback(localName: string) {
  return classifyStatementKind(localName);
}