Spendlight/scripts/profile-samples.mjs

#!/usr/bin/env node
// Describe statement files without revealing their contents.
//
//   node scripts/profile-samples.mjs [folder]      (default: samples)
//
// Prints, for each file: its layout, header labels, and value *shapes* only.
// Every letter becomes A/a and every digit 9, so "KROGER #123" reads
// "AAAAAA #999" and "-54.20" reads "-99.99". Rows made only of labels (no
// digits at all, such as a CSV header) are shown as they are. This is what
// Claude reads to build importers; the raw files never leave this server.

import { readdirSync, readFileSync, statSync } from 'node:fs';
import { basename, extname, join, relative } from 'node:path';

const root = process.argv[2] ?? 'samples';
const SKIP = new Set(['README.md', '.DS_Store']);

const mask = (s) => s.replace(/[A-Z]/g, 'A').replace(/[a-z]/g, 'a').replace(/[0-9]/g, '9').replace(/[^\x20-\x7e]/g, '?');
const shape = (s) =>
  s.trim() === ''
    ? '(empty)'
    : s.trim().replace(/[0-9]+/g, '9').replace(/[A-Z]+/g, 'A').replace(/[a-z]+/g, 'a').replace(/[^\x20-\x7e]/g, '?');
const maskName = (name) => name.replace(/[0-9]{3,}/g, (d) => '9'.repeat(d.length));

function files(dir) {
  return readdirSync(dir).flatMap((name) => {
    if (SKIP.has(name) || name.startsWith('_') || name.startsWith('.')) return [];
    const path = join(dir, name);
    return statSync(path).isDirectory() ? files(path) : [path];
  });
}

/** A small CSV reader that understands quoted fields and embedded commas or newlines. */
function parseCsv(text, delimiter) {
  const rows = [];
  let row = [];
  let field = '';
  let quoted = false;
  for (let i = 0; i < text.length; i++) {
    const ch = text[i];
    if (quoted) {
      if (ch === '"' && text[i + 1] === '"') {
        field += '"';
        i++;
      } else if (ch === '"') quoted = false;
      else field += ch;
    } else if (ch === '"') quoted = true;
    else if (ch === delimiter) {
      row.push(field);
      field = '';
    } else if (ch === '\n' || ch === '\r') {
      if (ch === '\r' && text[i + 1] === '\n') i++;
      row.push(field);
      rows.push(row);
      row = [];
      field = '';
    } else field += ch;
  }
  if (field !== '' || row.length) {
    row.push(field);
    rows.push(row);
  }
  return rows;
}

const isLabelRow = (cells) => cells.filter((c) => c.trim()).length >= 2 && cells.every((c) => !/[0-9]/.test(c));
const showRow = (cells) => (isLabelRow(cells) ? cells.map((c) => c.trim()) : cells.map(mask));
const numeric = (s) => /^\s*[-+(]?\s*\$?\s*[0-9][0-9,]*(\.[0-9]+)?\s*\)?\s*(CR|DR)?\s*$/i.test(s);

function profileCsv(text) {
  const firstLines = text.split(/\r?\n/).slice(0, 20).join('\n');
  const counts = { ',': 0, '\t': 0, ';': 0, '|': 0 };
  for (const ch of firstLines) if (ch in counts) counts[ch]++;
  const delimiter = Object.entries(counts).sort((a, b) => b[1] - a[1])[0][0];
  const rows = parseCsv(text, delimiter).filter((r) => r.some((c) => c.trim() !== ''));
  const out = [];

  out.push(`- Delimiter: ${JSON.stringify(delimiter)} · rows (non-blank): ${rows.length}`);
  const widths = {};
  for (const r of rows) widths[r.length] = (widths[r.length] ?? 0) + 1;
  out.push(`- Columns per row: ${Object.entries(widths).map(([w, n]) => `${w} cols × ${n}`).join(', ')}`);

  const headerIndex = rows.findIndex((r, i) => isLabelRow(r) && rows.slice(i + 1, i + 4).some((n) => n.length === r.length));
  out.push(`- Header row: ${headerIndex >= 0 ? `row ${headerIndex + 1}` : 'none found'}`);

  out.push('', '```text');
  rows.slice(0, 12).forEach((r, i) => out.push(`${String(i + 1).padStart(3)}: ${showRow(r).join(' | ')}`));
  if (rows.length > 15) out.push('    …');
  rows.slice(Math.max(12, rows.length - 3)).forEach((r, i, arr) => out.push(`${String(rows.length - arr.length + i + 1).padStart(3)}: ${showRow(r).join(' | ')}`));
  out.push('```');

  if (headerIndex >= 0) {
    const header = rows[headerIndex];
    const data = rows.slice(headerIndex + 1).filter((r) => r.length === header.length);
    out.push('', `Columns (${data.length} data rows):`, '');
    header.forEach((name, col) => {
      const values = data.map((r) => r[col] ?? '');
      const shapes = {};
      for (const v of values) shapes[shape(v)] = (shapes[shape(v)] ?? 0) + 1;
      const top = Object.entries(shapes)
        .sort((a, b) => b[1] - a[1])
        .slice(0, 4)
        .map(([s, n]) => `\`${s}\` ×${n}`)
        .join(', ');
      let extra = '';
      const nums = values.filter((v) => numeric(v));
      if (nums.length >= values.length * 0.6 && nums.length > 0) {
        const neg = nums.filter((v) => /^\s*-|\(|DR/i.test(v)).length;
        extra = ` · numeric: ${nums.length - neg} positive, ${neg} negative`;
      } else {
        const distinct = new Set(values.map((v) => v.trim()).filter(Boolean)).size;
        extra = ` · ${distinct} distinct values`;
        if (/^\d{1,2}\/\d{1,2}\/\d{2,4}$/.test(values.find((v) => v.trim()) ?? '')) {
          const firsts = values.map((v) => Number(v.split('/')[0])).filter(Number.isFinite);
          const seconds = values.map((v) => Number(v.split('/')[1])).filter(Number.isFinite);
          extra += ` · first part max ${Math.max(...firsts)}, second part max ${Math.max(...seconds)}`;
        }
      }
      out.push(`- **${name.trim() || `(column ${col + 1})`}**: ${top}${extra}`);
    });
  }
  return out;
}

function profileOfx(text) {
  const out = [];
  const headerLines = text.split(/\r?\n/).filter((l) => /^[A-Z]+:[A-Z0-9]+$/.test(l.trim()));
  if (headerLines.length) out.push(`- OFX header: ${headerLines.join(' ')}`);
  const tags = {};
  const examples = {};
  for (const m of text.matchAll(/<([A-Z0-9.]+)>([^<\r\n]*)/g)) {
    tags[m[1]] = (tags[m[1]] ?? 0) + 1;
    if (m[2].trim() && !examples[m[1]]) examples[m[1]] = mask(m[2].trim());
  }
  out.push(`- Transactions (STMTTRN): ${tags.STMTTRN ?? 0}`);
  out.push('', 'Tags (count, masked example):', '');
  for (const [tag, n] of Object.entries(tags)) out.push(`- ${tag} ×${n}${examples[tag] ? ` · \`${examples[tag]}\`` : ''}`);
  return out;
}

const list = files(root);
console.log(`# Sample profiles\n\n${list.length} file(s) under ${root}/. Values are masked: letters → A/a, digits → 9.\n`);
for (const path of list) {
  const buffer = readFileSync(path);
  const bom = buffer[0] === 0xef && buffer[1] === 0xbb && buffer[2] === 0xbf;
  const text = buffer.toString('utf8').replace(/^/, '');
  const ext = extname(path).toLowerCase();
  const rel = relative(root, path);
  console.log(`## ${maskName(rel)}\n`);
  console.log(`- Size: ${buffer.length} bytes · extension: ${ext || '(none)'} · BOM: ${bom ? 'yes' : 'no'} · line endings: ${text.includes('\r\n') ? 'CRLF' : 'LF'} · non-ASCII bytes: ${/[^\x00-\x7f]/.test(text) ? 'yes' : 'no'}`);
  const looksOfx = /<OFX>|OFXHEADER/i.test(text.slice(0, 2000));
  const lines = looksOfx ? profileOfx(text) : profileCsv(text);
  console.log(lines.join('\n') + '\n');
}
void basename;