memind/mindspace-page-purge.mjs

import fs from 'node:fs/promises';
import path from 'node:path';
import {
  isRelativeDownloadReference,
  resolveWorkspaceRelativeFilePath,
  splitReferenceParts,
} from './mindspace-html-download-links.mjs';
import { workspaceThumbnailRelativePath } from './mindspace-workspace-thumbnails.mjs';

const PRIVATE_ASSET_URL_PATTERN =
  /(?:https?:\/\/[^/]+)?\/api\/mindspace\/v1\/assets\/([a-z0-9-]+)\/download(?:\?[^"'<>\\\s)]*)?/gi;

const URL_ATTR_PATTERN = /(\b(?:href|src)\s*=\s*["'])([^"']+)(["'])/gi;

export function extractAssetIdsFromHtml(html) {
  return [
    ...new Set(
      [...String(html ?? '').matchAll(PRIVATE_ASSET_URL_PATTERN)]
        .map((match) => match[1])
        .filter(Boolean),
    ),
  ];
}

export function extractRelativeDownloadRefs(html) {
  const refs = new Set();
  for (const match of String(html ?? '').matchAll(URL_ATTR_PATTERN)) {
    const url = match[2];
    if (isRelativeDownloadReference(url)) {
      refs.add(splitReferenceParts(url).pathPart);
    }
  }
  return [...refs];
}

export function collectWorkspacePurgeTargets(htmlRelativePath, html) {
  const targets = new Set();
  const normalizedHtmlPath = String(htmlRelativePath ?? '').replace(/^\/+/, '');
  if (normalizedHtmlPath) {
    targets.add(normalizedHtmlPath);
    if (normalizedHtmlPath.toLowerCase().endsWith('.html')) {
      targets.add(workspaceThumbnailRelativePath(normalizedHtmlPath));
      const pngThumb = normalizedHtmlPath.replace(/\.html$/i, '.thumbnail.png');
      if (pngThumb !== normalizedHtmlPath) targets.add(pngThumb);
    }
  }
  for (const ref of extractRelativeDownloadRefs(html)) {
    targets.add(resolveWorkspaceRelativeFilePath(normalizedHtmlPath || 'public/index.html', ref));
  }
  return [...targets].filter(Boolean);
}

export async function purgeWorkspacePageArtifacts({
  publishDir,
  htmlRelativePath = null,
  html = '',
} = {}) {
  if (!publishDir) return { removed: [], skipped: [] };
  const resolvedRoot = path.resolve(publishDir);
  const removed = [];
  const skipped = [];

  for (const relativeTarget of collectWorkspacePurgeTargets(htmlRelativePath, html)) {
    const absolutePath = path.resolve(publishDir, ...relativeTarget.split('/'));
    if (
      absolutePath !== resolvedRoot &&
      !absolutePath.startsWith(`${resolvedRoot}${path.sep}`)
    ) {
      skipped.push(relativeTarget);
      continue;
    }
    try {
      await fs.unlink(absolutePath);
      removed.push(relativeTarget);
    } catch (error) {
      if (error?.code !== 'ENOENT') skipped.push(relativeTarget);
    }
  }

  return { removed, skipped };
}

export const pagePurgeInternals = {
  extractAssetIdsFromHtml,
  extractRelativeDownloadRefs,
  collectWorkspacePurgeTargets,
};