superwiki 0.1.2 → 0.1.4

package/skills/sw-migrate/scripts/migrate.mjs

@@ -1,225 +1,467 @@
 #!/usr/bin/env node
 // Converts a table-based task index into Superwiki task files, driven by a mapping file.
 // Bulk, mechanical work only: what a table row and a detail section say goes into one task file,
-// links to tasks become wikilinks, consumed files move to an archive folder. Nothing is deleted.
-import { readFileSync, writeFileSync, existsSync, mkdirSync, readdirSync, renameSync, statSync } from 'node:fs';
-import { dirname, join, relative, resolve, posix } from 'node:path';
+// links to tasks become wikilinks, and the files that were consumed move to an archive folder.
+// Nothing is deleted.
+import { existsSync, mkdirSync, readFileSync, readdirSync, renameSync, rmdirSync, statSync, writeFileSync } from 'node:fs';
+import { dirname, join, posix, resolve } from 'node:path';
 
-const HELP = `migrate.mjs --mapping <file.json> [--docs <dir>] [--dry-run]
+const HELP = `migrate.mjs --inspect [--docs <dir>]
+migrate.mjs --mapping <file.json> [--docs <dir>] [--dry-run]
+
+--inspect prints the tables, columns, status values and task headings found under the docs
+folder, so the mapping can be written without reading the files.
 
 Mapping (paths relative to the docs folder):
 {
   "index": "index.md",                      file holding the task tables
   "columns": { "id": "ID", "title": "Task", "status": "Status", "deps": "Depends on",
                "milestone": "Target", "priority": "Order", "started": "Start", "finished": "End" },
-  "sections": { "Sources": "Source", "Notes": "Note" },   table columns kept as body sections
+  "fields": { "review": "Review" },         other columns kept as frontmatter fields
+  "sections": { "Sources": "Source", "Notes": "Note" },   other columns kept as body sections
   "status": { "Not started": "todo", "In progress": "in-progress", "Done": "done" },
   "softPrefix": "~",                        marks a soft dependency in the deps column
   "details": { "dir": "tasks", "heading": "## " },        per-task detail sections: "<heading><ID>"
   "log": { "file": "changelog.md", "columns": { "date": "Date", "id": "Task", "text": "Change" } },
-  "archive": "legacy"                       consumed files are moved here
-}
-Only "index", "columns.id", "columns.title", "columns.status" and "status" are required.`;
-
-const argv = process.argv.slice(2);
-const opt = n => { const i = argv.indexOf(n); return i >= 0 ? argv[i + 1] : undefined; };
-if (argv.includes('--help') || !opt('--mapping')) { console.log(HELP); process.exit(argv.includes('--help') ? 0 : 2); }
-const dry = argv.includes('--dry-run');
-const docs = resolve(opt('--docs') || 'docs');
-const map = JSON.parse(readFileSync(resolve(opt('--mapping')), 'utf8'));
-const archive = map.archive || 'legacy';
-const soft = map.softPrefix ?? '~';
+  "archiveAlso": ["plans/old"],             other files or folders to move to the archive
+  "archive": "legacy"                       where consumed files go
+}
+Required: "index", "columns.id", "columns.title", "columns.status", "status".
+Status keys are written without decoration: a cell showing "✅ Done" is matched by "Done".`;
+
 const ID = /[A-Z][A-Z0-9]*-\d+/;
-const fail = msg => { console.error(msg); process.exit(2); };
+const VAULT_FOLDERS = ['wiki', 'tasks', 'plans'];
+const CORE_FIELDS = ['type', 'id', 'title', 'status', 'deps', 'soft_deps', 'milestone', 'priority', 'started', 'finished'];
+const MARKDOWN_LINK = /\[([^\[\]]*)\]\(([^)\s]+)\)( ✅)?/g;
+const LIST_LIMIT = 10;
 
-const indexPath = join(docs, map.index);
-if (!existsSync(indexPath)) fail(`no ${map.index} in ${docs}`);
-// Empty folders left by an aborted run are not a vault and not an archive.
-const hasFiles = d => existsSync(d) && readdirSync(d, { recursive: true, withFileTypes: true }).some(e => e.isFile());
-if (existsSync(join(docs, '.sw', 'config.json'))) fail('docs/.sw/config.json exists: this folder is already a Superwiki vault');
-if (hasFiles(join(docs, archive))) fail(`docs/${archive} has files in it: choose another "archive" name or remove it`);
+class UsageError extends Error {}
 
-// ---------- Markdown tables ----------
-const splitRow = line => line.trim().replace(/^\|/, '').replace(/\|$/, '').split(/(?<!\\)\|/).map(c => c.trim().replace(/\\\|/g, '|'));
+// ---------- Markdown ----------
+
+const splitRow = line =>
+  line.trim().replace(/^\|/, '').replace(/\|$/, '').split(/(?<!\\)\|/).map(cell => cell.trim().replace(/\\\|/g, '|'));
+
+// Every table in the text as { line, head, rows }, each row an object keyed by column name.
 function tables(text) {
   const lines = text.split(/\r?\n/);
-  const out = [];
+  const found = [];
   for (let i = 0; i + 1 < lines.length; i++) {
-    if (!lines[i].trim().startsWith('|') || !/^\s*\|?[\s:|-]+\|[\s:|-]*$/.test(lines[i + 1])) continue;
+    const isHeader = lines[i].trim().startsWith('|') && /^\s*\|?[\s:|-]+\|[\s:|-]*$/.test(lines[i + 1]);
+    if (!isHeader) continue;
     const head = splitRow(lines[i]);
     const rows = [];
     let j = i + 2;
     for (; j < lines.length && lines[j].trim().startsWith('|'); j++) {
       const cells = splitRow(lines[j]);
-      rows.push(Object.fromEntries(head.map((h, k) => [h, cells[k] ?? ''])));
+      rows.push(Object.fromEntries(head.map((name, k) => [name, cells[k] ?? ''])));
     }
-    out.push({ head, rows });
+    found.push({ line: i + 1, head, rows });
     i = j - 1;
   }
-  return out;
+  return found;
+}
+
+// A cell as a reader sees it: no HTML, link labels instead of links, no emphasis marks.
+const plain = cell =>
+  cell.replace(/<[^>]+>/g, '').replace(/\[([^\[\]]*)\]\([^)]*\)/g, '$1').replace(/[`*]/g, '').trim();
+
+// A status cell without the check mark some indexes decorate finished rows with.
+const statusText = cell => plain(cell).replace(/✅/g, '').trim();
+
+// Single-quoted YAML: its only escape is a doubled quote, so values with " or \ survive as written.
+const yaml = value => (/^[\w./-]*$/.test(value) ? value : `'${value.replace(/'/g, "''")}'`);
+
+// ---------- Files ----------
+
+// Every file under dir as a path relative to it, skipping dot entries and node_modules.
+function walk(dir, rel = '') {
+  const files = [];
+  for (const entry of readdirSync(dir, { withFileTypes: true })) {
+    if (entry.name.startsWith('.') || entry.name === 'node_modules') continue;
+    const path = rel ? `${rel}/${entry.name}` : entry.name;
+    if (entry.isDirectory()) files.push(...walk(join(dir, entry.name), path));
+    else files.push(path);
+  }
+  return files;
+}
+
+const isMarkdown = path => path.endsWith('.md');
+const read = (docs, path) => readFileSync(join(docs, path), 'utf8');
+
+// ---------- Inspect ----------
+
+// A table is about tasks when one of its columns holds a task id in most rows.
+function taskColumn(table) {
+  return table.head.find(name => table.rows.filter(row => ID.test(plain(row[name]))).length >= table.rows.length * 0.7);
+}
+
+// The distinct values of each column that has only a few, with their counts: status-like columns.
+function fewValuedColumns(table) {
+  const lines = [];
+  for (const name of table.head) {
+    const counts = new Map();
+    for (const row of table.rows) {
+      const value = statusText(row[name]);
+      counts.set(value, (counts.get(value) || 0) + 1);
+    }
+    const values = [...counts.keys()].filter(Boolean);
+    const words = values.some(value => !/^[\d.\-—–]+$/.test(value)); // not just dates, numbers or dashes
+    const repeats = Math.max(...counts.values()) >= 2;
+    const statusLike = counts.size >= 2 && counts.size <= 6 && words && repeats && values.every(value => value.length <= 24);
+    if (statusLike) lines.push(`    values of "${name}": ${[...counts].map(([value, n]) => `${value || '(empty)'} ${n}`).join(', ')}`);
+  }
+  return lines;
+}
+
+// Prints what a mapping needs to know: the tables that list tasks, their columns, the values of
+// status-like columns, and the files that hold per-task headings. Other tables are only counted.
+function inspect(docs) {
+  const out = [];
+  let otherTables = 0;
+  for (const path of walk(docs).filter(isMarkdown)) {
+    if (statSync(join(docs, path)).size > 5e6) continue;
+    const text = read(docs, path);
+    const all = tables(text).filter(table => table.rows.length >= 3);
+    const taskTables = all.filter(taskColumn);
+    otherTables += all.length - taskTables.length;
+    const headings = text.split(/\r?\n/).filter(line => new RegExp(`^#{1,4} ${ID.source}\\s*$`).test(line));
+    if (!taskTables.length && headings.length < 2) continue;
+    out.push(`${path} (${Math.round(text.length / 1024)} KB)`);
+    for (const table of taskTables) {
+      out.push(`  line ${table.line}: ${table.rows.length} rows, ids in "${taskColumn(table)}" | ${table.head.join(' | ')}`);
+      if (table.rows.length >= 5) out.push(...fewValuedColumns(table));
+    }
+    if (headings.length >= 2) {
+      const mark = headings[0].match(/^#+ /)[0];
+      out.push(`  ${headings.length} task headings like "${headings[0]}" (details heading: "${mark}")`);
+    }
+  }
+  if (otherTables) out.push(`${otherTables} other tables without task ids are not shown`);
+  for (const folder of VAULT_FOLDERS) {
+    if (!existsSync(join(docs, folder))) continue;
+    out.push(`note: docs/${folder}/ exists (${walk(join(docs, folder)).length} files). Superwiki uses that folder: what the mapping does not consume must go into "archiveAlso".`);
+  }
+  console.log(out.length ? out.join('\n') : 'no task tables or task headings found');
 }
 
-// ---------- Links ----------
-const consumed = new Set([map.index]); // files whose task anchors turn into wikilinks
-const detailDir = map.details?.dir;
-if (detailDir && existsSync(join(docs, detailDir))) {
-  for (const n of readdirSync(join(docs, detailDir))) if (n.endsWith('.md')) consumed.add(posix.join(detailDir, n));
+// ---------- Conversion ----------
+
+function readMapping(path) {
+  const map = JSON.parse(readFileSync(path, 'utf8'));
+  const missing = ['index', 'status'].filter(key => !map[key]).concat(['id', 'title', 'status'].filter(key => !map.columns?.[key]).map(key => `columns.${key}`));
+  if (missing.length) throw new UsageError(`the mapping lacks: ${missing.join(', ')}`);
+  const reserved = Object.keys(map.fields || {}).filter(name => CORE_FIELDS.includes(name));
+  if (reserved.length) throw new UsageError(`"fields" may not redefine: ${reserved.join(', ')}`);
+  return { archive: 'legacy', softPrefix: '~', fields: {}, sections: {}, archiveAlso: [], ...map };
+}
+
+// The files that leave their place: the index, the detail files, the log source and anything
+// named in archiveAlso. Links into them are re-pointed, and the files themselves are archived.
+function filesToArchive(docs, map, problems) {
+  const moved = new Set([map.index]);
+  if (map.log?.file) moved.add(map.log.file);
+  const detailDir = map.details?.dir;
+  if (detailDir && existsSync(join(docs, detailDir))) {
+    for (const name of readdirSync(join(docs, detailDir))) if (isMarkdown(name)) moved.add(posix.join(detailDir, name));
+  }
+  for (const path of map.archiveAlso) {
+    const full = join(docs, path);
+    if (!existsSync(full)) problems.push(`archiveAlso: docs/${path} does not exist`);
+    else if (statSync(full).isDirectory()) for (const file of walk(full)) moved.add(posix.join(path, file));
+    else moved.add(path);
+  }
+  return moved;
 }
-if (map.log?.file) consumed.add(map.log.file);
 
-const stats = { taskLinks: 0, relinked: 0, deadLinks: [] };
-let ids = new Set();
+function readTasks(docs, map, problems) {
+  const col = map.columns;
+  const rows = tables(read(docs, map.index))
+    .filter(table => [col.id, col.title, col.status].every(name => table.head.includes(name)))
+    .flatMap(table => table.rows);
+  if (!rows.length) throw new UsageError(`no table in ${map.index} has the columns "${col.id}", "${col.title}", "${col.status}"`);
 
-// from: the file the text was in. to: the file it will be in. Both relative to docs.
-function rewrite(text, from, to) {
-  const fromDir = posix.dirname(from);
-  const toDir = posix.dirname(to);
-  const decode = s => { try { return decodeURI(s); } catch { return s; } };
-  // Row anchors only existed as link targets for the old index; in a foreign file they are not ours to remove.
-  return (from === to ? text : text.replace(/<a id="[^"]*"><\/a>/g, ''))
-    .replace(/\[([^\[\]]*)\]\(([^)\s]+)\)( ✅)?/g, (all, label, href, tick) => {
+  const softMark = map.softPrefix.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
+  const depPattern = new RegExp(`(${softMark})?\\[?(${ID.source})`, 'g');
+  const tasks = [];
+  const ids = new Set();
+  for (const row of rows) {
+    const id = (plain(row[col.id]).match(ID) || [])[0];
+    if (!id) { problems.push(`row without an id: ${row[col.id].slice(0, 60)}`); continue; }
+    if (ids.has(id)) { problems.push(`duplicate id ${id}`); continue; }
+    ids.add(id);
+    const shown = statusText(row[col.status]);
+    const key = Object.keys(map.status).find(name => name.toLowerCase() === shown.toLowerCase());
+    if (!key) problems.push(`${id}: status "${shown}" is not in the mapping`);
+    const deps = [];
+    const softDeps = [];
+    for (const match of (col.deps ? row[col.deps] : '').matchAll(depPattern)) {
+      const list = match[1] ? softDeps : deps;
+      if (!list.includes(match[2])) list.push(match[2]);
+    }
+    const cell = name => (col[name] ? plain(row[col[name]] ?? '') : '');
+    const extra = {};
+    for (const [field, column] of Object.entries(map.fields)) {
+      const value = plain(row[column] ?? '');
+      if (value && !/^[—–-]$/.test(value)) extra[field] = value;
+    }
+    tasks.push({
+      id, row, deps, softDeps, extra,
+      title: cell('title'), status: key ? map.status[key] : 'todo',
+      milestone: cell('milestone'), priority: cell('priority'), started: cell('started'), finished: cell('finished'),
+    });
+  }
+  return { tasks, ids };
+}
+
+// Detail sections by task id, and the headings that have no row in the index.
+function readDetails(docs, map, moved, ids) {
+  const details = new Map();
+  const stray = [];
+  const dir = map.details?.dir;
+  if (!dir) return { details, stray };
+  const mark = map.details.heading || '## ';
+  for (const file of [...moved].filter(path => path.startsWith(`${dir}/`) && isMarkdown(path))) {
+    let current = null;
+    for (const line of read(docs, file).split(/\r?\n/)) {
+      const id = line.startsWith(mark) ? (line.slice(mark.length).trim().match(new RegExp(`^${ID.source}$`)) || [])[0] : null;
+      if (id) {
+        current = { file, lines: [] };
+        if (ids.has(id)) details.set(id, current);
+        else stray.push(`${file}: ${id}`);
+      } else if (line.startsWith(mark)) {
+        current = null; // a same-level heading that is not a task ends the section
+      } else if (current) {
+        current.lines.push(line);
+      }
+    }
+  }
+  return { details, stray };
+}
+
+// The link from a folder to a target, keeping the trailing slash a link to a folder was written with.
+const relativeHref = (fromDir, target, written) =>
+  encodeURI(posix.relative(fromDir, target)) + (written.endsWith('/') ? '/' : '');
+
+// Rewrites the links of a text that lived in `from` and will live in `to` (paths relative to docs).
+// A link to a task in an archived file becomes a wikilink; any other link into an archived file
+// follows the file to the archive; a relative link is re-based when the text changes folder.
+function makeRewriter({ moved, ids, archive, stats }) {
+  const decode = text => { try { return decodeURI(text); } catch { return text; } };
+  return function rewrite(text, from, to) {
+    const fromDir = posix.dirname(from);
+    const toDir = posix.dirname(to);
+    // Row anchors existed only as link targets for the old index. A file that stays where it is keeps its own.
+    const body = from === to ? text : text.replace(/<a id="[^"]*"><\/a>/g, '');
+    return body.replace(MARKDOWN_LINK, (all, label, href, tick) => {
       if (/^[a-z][a-z0-9+.-]*:/i.test(href)) return all;
       const [path, anchor = ''] = href.split('#');
       const target = path ? posix.normalize(posix.join(fromDir, decode(path))) : from;
-      if (consumed.has(target)) {
+      const suffix = `${anchor ? `#${anchor}` : ''})${tick ?? ''}`;
+      if (moved.has(target)) {
         const id = (anchor.match(new RegExp(`^${ID.source}$`, 'i')) || [])[0]?.toUpperCase();
-        if (id && ids.has(id)) { stats.taskLinks++; return label.trim() === id ? `[[${id}]]` : `[[${id}|${label}]]`; }
-        stats.deadLinks.push(`${to}: ${all.slice(0, 80)}`);
-        return `[${label}](${encodeURI(posix.relative(toDir, posix.join(archive, target)))}${anchor ? `#${anchor}` : ''})${tick ?? ''}`;
+        if (id && ids.has(id)) {
+          stats.taskLinks++;
+          return label.trim() === id ? `[[${id}]]` : `[[${id}|${label}]]`;
+        }
+        stats.archiveLinks.push(`${to}: ${all.slice(0, 80)}`);
+        return `[${label}](${encodeURI(posix.relative(toDir, posix.join(archive, target)))}${suffix}`;
       }
       if (!path || fromDir === toDir) return all;
-      stats.relinked++;
-      return `[${label}](${encodeURI(posix.relative(toDir, target))}${anchor ? `#${anchor}` : ''})${tick ?? ''}`;
+      stats.rebased++;
+      return `[${label}](${relativeHref(toDir, target, path)}${suffix}`;
     });
+  };
 }
 
-// ---------- Read the index ----------
-const col = map.columns;
-const indexText = readFileSync(indexPath, 'utf8');
-const rows = tables(indexText).filter(t => [col.id, col.title, col.status].every(h => t.head.includes(h))).flatMap(t => t.rows);
-if (!rows.length) fail(`no table in ${map.index} has the columns "${col.id}", "${col.title}", "${col.status}"`);
-
-const plain = s => s.replace(/<[^>]+>/g, '').replace(/\[([^\[\]]*)\]\([^)]*\)/g, '$1').replace(/[`*]/g, '').trim();
-const statusOf = cell => {
-  const text = plain(cell).replace(/✅/g, '').trim();
-  const hit = Object.keys(map.status).find(k => k.toLowerCase() === text.toLowerCase());
-  return hit ? map.status[hit] : null;
-};
-
-const tasks = [];
-const problems = [];
-for (const r of rows) {
-  const id = (plain(r[col.id]).match(ID) || [])[0];
-  if (!id) { problems.push(`row without an id: ${r[col.id].slice(0, 60)}`); continue; }
-  if (ids.has(id)) { problems.push(`duplicate id ${id}`); continue; }
-  ids.add(id);
-  const status = statusOf(r[col.status]);
-  if (!status) problems.push(`${id}: status "${plain(r[col.status])}" is not in the mapping`);
-  const deps = [];
-  const softDeps = [];
-  for (const m of (col.deps ? r[col.deps] : '').matchAll(new RegExp(`(${soft.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')})?\\[?(${ID.source})`, 'g'))) {
-    const list = m[1] ? softDeps : deps;
-    if (!list.includes(m[2])) list.push(m[2]);
-  }
-  const get = k => (col[k] ? plain(r[col[k]] ?? '') : '');
-  tasks.push({ id, row: r, title: get('title'), status: status || 'todo', deps, softDeps, milestone: get('milestone'), priority: get('priority'), started: get('started'), finished: get('finished') });
-}
-
-// ---------- Detail sections ----------
-const details = new Map();
-const strayDetails = [];
-if (detailDir) {
-  const mark = map.details.heading || '## ';
-  for (const file of [...consumed].filter(f => f.startsWith(`${detailDir}/`))) {
-    let cur = null;
-    for (const line of readFileSync(join(docs, file), 'utf8').split(/\r?\n/)) {
-      const id = line.startsWith(mark) ? (line.slice(mark.length).trim().match(new RegExp(`^${ID.source}$`)) || [])[0] : null;
-      if (id) { cur = { file, lines: [] }; (ids.has(id) ? details.set(id, cur) : strayDetails.push(`${file}: ${id}`)); continue; }
-      if (line.startsWith(mark)) cur = null; // a same-level heading that is not a task ends the section
-      if (cur) cur.lines.push(line);
-    }
-  }
+// An archived file keeps its text, but it sits one folder deeper: its relative links to files
+// that stayed behind are re-based so the archive remains readable. Links between archived files
+// need no change, because the archive keeps their layout.
+function rebaseArchived(text, path, moved, archive) {
+  const fromDir = posix.dirname(path);
+  const toDir = posix.dirname(posix.join(archive, path));
+  return text.replace(MARKDOWN_LINK, (all, label, href, tick) => {
+    const [target, anchor = ''] = href.split('#');
+    if (!target || /^[a-z][a-z0-9+.-]*:/i.test(href)) return all;
+    let decoded = target;
+    try { decoded = decodeURI(target); } catch { /* keep the raw path */ }
+    const resolved = posix.normalize(posix.join(fromDir, decoded));
+    if (moved.has(resolved)) return all;
+    return `[${label}](${relativeHref(toDir, resolved, target)}${anchor ? `#${anchor}` : ''})${tick ?? ''}`;
+  });
 }
 
-// ---------- Write ----------
-// Single-quoted YAML: the only escape is a doubled quote, so titles with " or \\ survive as written.
-const yaml = v => (/^[\w./-]*$/.test(v) ? v : `'${v.replace(/'/g, "''")}'`);
-const out = new Map();
-for (const t of tasks) {
-  const to = `tasks/${t.id}.md`;
-  const fm = ['---', 'type: task', `id: ${t.id}`, `title: ${yaml(t.title)}`, `status: ${t.status}`, `deps: [${t.deps.join(', ')}]`];
-  if (t.softDeps.length) fm.push(`soft_deps: [${t.softDeps.join(', ')}]`);
-  if (t.milestone) fm.push(`milestone: ${yaml(t.milestone)}`);
-  if (/^\d+$/.test(t.priority)) fm.push(`priority: ${t.priority}`);
-  fm.push(`started: ${t.started}`.trimEnd(), `finished: ${t.finished}`.trimEnd(), '---', '');
+function taskFile(task, detail, map, rewrite) {
+  const path = `tasks/${task.id}.md`;
+  const frontmatter = ['type: task', `id: ${task.id}`, `title: ${yaml(task.title)}`, `status: ${task.status}`, `deps: [${task.deps.join(', ')}]`];
+  if (task.softDeps.length) frontmatter.push(`soft_deps: [${task.softDeps.join(', ')}]`);
+  if (task.milestone) frontmatter.push(`milestone: ${yaml(task.milestone)}`);
+  if (/^\d+$/.test(task.priority)) frontmatter.push(`priority: ${task.priority}`);
+  for (const [field, value] of Object.entries(task.extra)) frontmatter.push(`${field}: ${yaml(value)}`);
+  frontmatter.push(`started: ${task.started}`.trimEnd(), `finished: ${task.finished}`.trimEnd());
+
   const body = [];
-  const d = details.get(t.id);
-  if (d) body.push(rewrite(d.lines.join('\n').trim(), d.file, to), '');
-  for (const [name, column] of Object.entries(map.sections || {})) {
-    const cell = (t.row[column] || '').trim();
-    if (cell) body.push(`## ${name}`, '', rewrite(cell, map.index, to), '');
-  }
-  out.set(to, fm.join('\n') + '\n' + body.join('\n').replace(/\n{3,}/g, '\n\n').trimEnd() + '\n');
-}
-
-let logEntries = 0;
-if (map.log?.file && existsSync(join(docs, map.log.file))) {
-  const lc = map.log.columns;
-  const entries = tables(readFileSync(join(docs, map.log.file), 'utf8'))
-    .filter(t => [lc.date, lc.text].every(h => t.head.includes(h))).flatMap(t => t.rows)
-    .filter(r => /^\d{4}-\d{2}-\d{2}/.test(plain(r[lc.date])));
-  // Stable sort keeps same-day rows in reverse source order: sources list newest first, the log is oldest first.
-  entries.reverse().sort((a, b) => plain(a[lc.date]).localeCompare(plain(b[lc.date])));
-  logEntries = entries.length;
-  const text = entries.map(r => `## [${plain(r[lc.date]).slice(0, 10)}] task | ${lc.id ? plain(r[lc.id]) || 'general' : 'general'}\n\n${rewrite(r[lc.text], map.log.file, 'log.md')}\n`).join('\n');
-  out.set('log.md', `# Log\n\nAppend-only. Entry format: \`## [YYYY-MM-DD] kind | title\`.\n\n${text}`);
-}
-
-// Other markdown files that pointed at tasks in the consumed files.
-const others = [];
-(function walk(dir, rel) {
-  for (const e of readdirSync(dir, { withFileTypes: true })) {
-    if (e.name.startsWith('.') || e.name === 'node_modules') continue;
-    const p = rel ? `${rel}/${e.name}` : e.name;
-    if (e.isDirectory()) walk(join(dir, e.name), p);
-    else if (e.name.endsWith('.md') && !consumed.has(p) && statSync(join(dir, e.name)).size < 2e6) others.push(p);
-  }
-})(docs, '');
-let touchedOthers = 0;
-for (const p of others) {
-  const before = readFileSync(join(docs, p), 'utf8');
-  const after = rewrite(before, p, p);
-  if (after === before) continue;
-  touchedOthers++;
-  out.set(p, after);
-}
-
-const counts = {};
-for (const t of tasks) counts[t.status] = (counts[t.status] || 0) + 1;
-const areas = [...new Set(tasks.map(t => t.id.slice(0, t.id.lastIndexOf('-'))))];
-const unknownDeps = tasks.flatMap(t => [...t.deps, ...t.softDeps].filter(d => !ids.has(d)).map(d => `${t.id} -> ${d}`));
-
-if (!dry) {
-  for (const f of consumed) {
-    if (!existsSync(join(docs, f))) continue;
-    mkdirSync(dirname(join(docs, archive, f)), { recursive: true });
-    renameSync(join(docs, f), join(docs, archive, f));
-  }
-  for (const [p, text] of out) {
-    mkdirSync(dirname(join(docs, p)), { recursive: true });
-    writeFileSync(join(docs, p), text);
-  }
-}
-
-const show = (title, list, max = 10) => list.length && console.log(`\n${title} (${list.length}):\n${list.slice(0, max).map(s => `  ${s}`).join('\n')}${list.length > max ? `\n  ... ${list.length - max} more` : ''}`);
-console.log(`${dry ? 'DRY RUN, nothing written\n' : ''}tasks: ${tasks.length}  ${Object.entries(counts).map(([k, v]) => `${k} ${v}`).join('  ')}`);
-console.log(`areas: ${areas.join(', ')}`);
-console.log(`with detail section: ${[...details.keys()].length}  without: ${tasks.length - details.size}`);
-console.log(`task links turned into wikilinks: ${stats.taskLinks}  relative links re-based: ${stats.relinked}  other files updated: ${touchedOthers}`);
-if (map.log?.file) console.log(`log entries: ${logEntries}`);
-console.log(`archived to docs/${archive}/: ${[...consumed].join(', ')}`);
-show('problems', problems);
-show('dependencies on unknown ids', unknownDeps);
-show('detail sections without a table row (left in the archive)', strayDetails);
-show(`links into archived files that are not task links (now point into ${archive}/)`, stats.deadLinks);
-console.log(`\nnext: node <sw-init>/scripts/init.mjs --tasks --areas "${areas.map(a => `${a}=${a}`).join(',')}"  then  node docs/.sw/sw.mjs lint`);
+  if (detail) body.push(rewrite(detail.lines.join('\n').trim(), detail.file, path), '');
+  for (const [heading, column] of Object.entries(map.sections)) {
+    const cell = (task.row[column] || '').trim();
+    if (cell) body.push(`## ${heading}`, '', rewrite(cell, map.index, path), '');
+  }
+  const text = body.join('\n').replace(/\n{3,}/g, '\n\n').trimEnd();
+  return [path, `---\n${frontmatter.join('\n')}\n---\n\n${text}\n`];
+}
+
+function logFile(docs, map, rewrite) {
+  if (!map.log?.file || !existsSync(join(docs, map.log.file))) return null;
+  const col = map.log.columns;
+  const rows = tables(read(docs, map.log.file))
+    .filter(table => [col.date, col.text].every(name => table.head.includes(name)))
+    .flatMap(table => table.rows)
+    .filter(row => /^\d{4}-\d{2}-\d{2}/.test(plain(row[col.date])));
+  // Sources list newest first and the log is oldest first; a stable sort after reversing keeps
+  // same-day rows in the order they happened.
+  rows.reverse().sort((a, b) => plain(a[col.date]).localeCompare(plain(b[col.date])));
+  const entries = rows.map(row => {
+    const title = (col.id && plain(row[col.id])) || 'general';
+    return `## [${plain(row[col.date]).slice(0, 10)}] task | ${title}\n\n${rewrite(row[col.text], map.log.file, 'log.md')}\n`;
+  });
+  return { count: entries.length, text: `# Log\n\nAppend-only. Entry format: \`## [YYYY-MM-DD] kind | title\`.\n\n${entries.join('\n')}` };
+}
+
+// Works out everything the conversion will write and move, without touching the disk.
+function plan(docs, map) {
+  if (!existsSync(join(docs, map.index))) throw new UsageError(`no ${map.index} in ${docs}`);
+  if (existsSync(join(docs, '.sw', 'config.json'))) throw new UsageError('docs/.sw/config.json exists: this folder is already a Superwiki vault');
+  const archiveDir = join(docs, map.archive);
+  // Empty folders left by an undone attempt are not an archive.
+  if (existsSync(archiveDir) && walk(archiveDir).length) throw new UsageError(`docs/${map.archive} has files in it: choose another "archive" name or remove it`);
+
+  const problems = [];
+  const moved = filesToArchive(docs, map, problems);
+  const { tasks, ids } = readTasks(docs, map, problems);
+  const { details, stray } = readDetails(docs, map, moved, ids);
+  const stats = { taskLinks: 0, rebased: 0, archiveLinks: [] };
+  const rewrite = makeRewriter({ moved, ids, archive: map.archive, stats });
+
+  const writes = new Map();
+  for (const task of tasks) writes.set(...taskFile(task, details.get(task.id), map, rewrite));
+  const log = logFile(docs, map, rewrite);
+  if (log) writes.set('log.md', log.text);
+
+  // Files that stay where they are but pointed at something that moves.
+  const updated = [];
+  for (const path of walk(docs).filter(file => isMarkdown(file) && !moved.has(file))) {
+    if (statSync(join(docs, path)).size > 2e6) continue;
+    const before = read(docs, path);
+    const after = rewrite(before, path, path);
+    if (after !== before) { writes.set(path, after); updated.push(path); }
+  }
+
+  for (const path of [...moved].filter(file => isMarkdown(file) && existsSync(join(docs, file)))) {
+    const before = read(docs, path);
+    const after = rebaseArchived(before, path, moved, map.archive);
+    if (after !== before) writes.set(posix.join(map.archive, path), after);
+  }
+
+  // Superwiki owns wiki/, tasks/ and plans/. A file already in one of them would be read as a
+  // wiki page, task or plan after the conversion, and lint would reject it.
+  const foreign = VAULT_FOLDERS.flatMap(folder => (existsSync(join(docs, folder)) ? walk(join(docs, folder)).map(file => `${folder}/${file}`) : []))
+    .filter(path => isMarkdown(path) && !moved.has(path));
+  for (const path of foreign) problems.push(`docs/${path} is in a folder Superwiki uses: add it (or its folder) to "archiveAlso", or move it elsewhere first`);
+
+  for (const task of tasks) {
+    for (const dep of [...task.deps, ...task.softDeps]) if (!ids.has(dep)) problems.push(`${task.id} depends on ${dep}, which is not in the index`);
+  }
+
+  const counts = {};
+  for (const task of tasks) counts[task.status] = (counts[task.status] || 0) + 1;
+  const areas = [...new Set(tasks.map(task => task.id.slice(0, task.id.lastIndexOf('-'))))];
+  return { map, moved, writes, problems, stray, stats, updated, counts, areas, taskCount: tasks.length, detailCount: details.size, logCount: log?.count ?? null };
+}
+
+// A folder that held only archived files would be left behind empty. Remove those, deepest first;
+// rmdir refuses a folder that still has something in it, which is the check.
+function removeEmptiedFolders(docs, moved) {
+  const folders = new Set();
+  for (const path of moved) {
+    for (let dir = posix.dirname(path); dir !== '.'; dir = posix.dirname(dir)) folders.add(dir);
+  }
+  for (const dir of [...folders].sort((a, b) => b.length - a.length)) {
+    try { rmdirSync(join(docs, dir)); } catch { /* not empty, or already gone */ }
+  }
+}
+
+function apply(docs, result) {
+  const { map, moved, writes } = result;
+  for (const path of moved) {
+    if (!existsSync(join(docs, path))) continue;
+    const target = join(docs, map.archive, path);
+    mkdirSync(dirname(target), { recursive: true });
+    renameSync(join(docs, path), target);
+  }
+  removeEmptiedFolders(docs, moved);
+  for (const [path, text] of writes) {
+    mkdirSync(dirname(join(docs, path)), { recursive: true });
+    writeFileSync(join(docs, path), text);
+  }
+}
+
+function report(result, dryRun) {
+  const { map, counts, areas, stats } = result;
+  const list = (title, items) => {
+    if (!items.length) return;
+    const more = items.length > LIST_LIMIT ? `\n  ... ${items.length - LIST_LIMIT} more` : '';
+    console.log(`\n${title} (${items.length}):\n${items.slice(0, LIST_LIMIT).map(item => `  ${item}`).join('\n')}${more}`);
+  };
+  if (dryRun) console.log('DRY RUN, nothing written');
+  console.log(`tasks: ${result.taskCount}  ${Object.entries(counts).map(([status, n]) => `${status} ${n}`).join('  ')}`);
+  console.log(`areas: ${areas.join(', ')}`);
+  console.log(`with detail section: ${result.detailCount}  without: ${result.taskCount - result.detailCount}`);
+  console.log(`task links turned into wikilinks: ${stats.taskLinks}  relative links re-based: ${stats.rebased}  other files updated: ${result.updated.length}`);
+  if (result.logCount !== null) console.log(`log entries: ${result.logCount} -> docs/log.md`);
+  console.log(`archived to docs/${map.archive}/: ${result.moved.size} files`);
+
+  list('PROBLEMS, fix these in the mapping or the files and run again', result.problems);
+  if (!result.problems.length) console.log('\nno problems');
+  console.log('\nFor information; none of these blocks the conversion:');
+  list('other files whose task links were rewritten', result.updated);
+  list('detail sections without a table row (they stay in the archive)', result.stray);
+  list(`links into archived files that are not task links (now point into ${map.archive}/)`, stats.archiveLinks);
+  console.log(`\nnext: node <init-dir>/scripts/init.mjs --tasks --areas "${areas.map(area => `${area}=${area}`).join(',')}"  then  node docs/.sw/sw.mjs status  and  lint`);
+  console.log('`status` splits "todo" into ready and blocked; their sum is the todo count above.');
+}
+
+// ---------- Entry point ----------
+
+function parseArgs(argv) {
+  const options = { docs: 'docs', mapping: null, inspect: false, dryRun: false, help: false };
+  for (let i = 0; i < argv.length; i++) {
+    if (argv[i] === '--docs') options.docs = argv[++i];
+    else if (argv[i] === '--mapping') options.mapping = argv[++i];
+    else if (argv[i] === '--inspect') options.inspect = true;
+    else if (argv[i] === '--dry-run') options.dryRun = true;
+    else if (argv[i] === '--help') options.help = true;
+    else throw new UsageError(`unknown argument: ${argv[i]}`);
+  }
+  return options;
+}
+
+function main(argv) {
+  try {
+    const options = parseArgs(argv);
+    if (options.help) { console.log(HELP); return 0; }
+    const docs = resolve(options.docs || 'docs');
+    if (!existsSync(docs)) throw new UsageError(`no docs folder at ${docs}`);
+    if (options.inspect) { inspect(docs); return 0; }
+    if (!options.mapping) throw new UsageError(HELP);
+    const result = plan(docs, readMapping(resolve(options.mapping)));
+    if (!options.dryRun) apply(docs, result);
+    report(result, options.dryRun);
+    return 0;
+  } catch (error) {
+    if (!(error instanceof UsageError)) throw error;
+    console.error(error.message);
+    return 2;
+  }
+}
+
+process.exitCode = main(process.argv.slice(2));