memind/mindspace-html-download-links.mjs

import fs from 'node:fs';
import path from 'node:path';
import { buildPublicUrl } from './user-publish.mjs';

export const DOWNLOADABLE_FILE_PATTERN = /\.(?:docx?|pdf|xlsx?|pptx?|zip|csv|txt|md)$/i;

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

export function isRelativeDownloadReference(value) {
  const trimmed = String(value ?? '').trim();
  if (!trimmed) return false;
  const pathPart = trimmed.split('#')[0].split('?')[0];
  if (/^([a-z][a-z0-9+.-]*:|\/|#|data:|mailto:|javascript:)/i.test(pathPart)) return false;
  return DOWNLOADABLE_FILE_PATTERN.test(pathPart);
}

export function splitReferenceParts(relativeRef) {
  const value = String(relativeRef ?? '');
  const match = value.match(/^([^#?]+)(.*)$/);
  return {
    pathPart: match?.[1] ?? value,
    suffix: match?.[2] ?? '',
  };
}

export function resolveWorkspaceRelativeFilePath(htmlRelativePath, relativeRef) {
  const { pathPart } = splitReferenceParts(relativeRef);
  const ref = String(pathPart ?? '').replace(/^\/+/, '');
  if (!ref) return ref;
  if (/^(?:public|oa|private)\//.test(ref)) return ref;
  const htmlPath = String(htmlRelativePath ?? 'public/index.html').replace(/^\/+/, '');
  const htmlDir = path.posix.dirname(htmlPath);
  if (!htmlDir || htmlDir === '.') return `public/${ref}`;
  return path.posix.join(htmlDir, ref).replace(/\\/g, '/');
}

export function resolveSiblingDocxPath(htmlRelativePath) {
  const htmlPath = String(htmlRelativePath ?? '').replace(/^\/+/, '');
  if (!htmlPath.toLowerCase().endsWith('.html')) return null;
  return htmlPath.replace(/\.html$/i, '.docx');
}

function resolveExistingSiblingDocxPath(htmlRelativePath, publishDir) {
  const sibling = resolveSiblingDocxPath(htmlRelativePath);
  if (!sibling || !publishDir) return null;
  return workspaceFileExists(publishDir, sibling) ? sibling : null;
}

export function inferWorkspaceHtmlRelativePath({
  snapshotRelativePath = null,
  pageTitle = null,
  htmlContent = null,
  publishDir = null,
} = {}) {
  const fromSnapshot = String(snapshotRelativePath ?? '').trim();
  if (fromSnapshot) {
    const normalized = fromSnapshot.replace(/^\/+/, '');
    if (normalized.toLowerCase().startsWith('public/')) return normalized;
    if (normalized.toLowerCase().endsWith('.html')) return `public/${path.posix.basename(normalized)}`;
    return normalized;
  }

  const publicDir = publishDir ? path.join(publishDir, 'public') : null;
  if (!publicDir || !fs.existsSync(publicDir) || !htmlContent) {
    return 'public/index.html';
  }

  const normalizedContent = String(htmlContent);
  const title = String(pageTitle ?? '').trim();
  for (const name of fs.readdirSync(publicDir)) {
    if (!name.toLowerCase().endsWith('.html')) continue;
    const absolutePath = path.join(publicDir, name);
    try {
      const diskContent = fs.readFileSync(absolutePath, 'utf8');
      if (diskContent === normalizedContent) return `public/${name}`;
      if (title) {
        const titleMatch = diskContent.match(/<title[^>]*>([^<]+)<\/title>/i);
        if (titleMatch?.[1]?.trim() === title) return `public/${name}`;
      }
    } catch {
      // ignore unreadable files
    }
  }

  return 'public/index.html';
}

export function buildWorkspaceDownloadLinkIndex(assets = []) {
  const byBasename = new Map();
  const byPath = new Map();
  for (const asset of assets) {
    const id = asset.id;
    const filename = String(asset.original_filename ?? asset.originalFilename ?? '').replace(/\\/g, '/');
    if (!id || !filename) continue;
    byPath.set(filename, id);
    byBasename.set(path.posix.basename(filename), id);
    if (filename.startsWith('public/')) {
      byPath.set(filename.slice('public/'.length), id);
    } else if (!filename.includes('/')) {
      byPath.set(`public/${filename}`, id);
    }
  }
  return { byBasename, byPath };
}

export function lookupAssetIdForWorkspacePath(index, workspaceRelativePath) {
  const normalized = String(workspaceRelativePath ?? '').replace(/\\/g, '/');
  if (!normalized || !index) return null;
  return (
    index.byPath.get(normalized)
    ?? index.byPath.get(`public/${normalized}`)
    ?? index.byBasename.get(path.posix.basename(normalized))
    ?? null
  );
}

export function buildAssetDownloadUrl(assetId) {
  return `/api/mindspace/v1/assets/${encodeURIComponent(assetId)}/download`;
}

export function buildPublicWorkspaceFileUrl(publicBaseUrl, publishKey, workspaceRelativePath) {
  const clean = String(workspaceRelativePath ?? '').replace(/^\/+/, '');
  return buildPublicUrl(publicBaseUrl, publishKey, clean);
}

function workspaceFileExists(publishDir, workspaceRelativePath) {
  if (!publishDir || !workspaceRelativePath) return false;
  const absolutePath = path.join(publishDir, ...String(workspaceRelativePath).split('/'));
  try {
    return fs.existsSync(absolutePath) && fs.statSync(absolutePath).isFile();
  } catch {
    return false;
  }
}

function collectDownloadWorkspaceCandidates(htmlRelativePath, relativeRef, publishDir) {
  const workspacePath = resolveWorkspaceRelativeFilePath(htmlRelativePath, relativeRef);
  const candidates = new Set([
    workspacePath,
    path.posix.join('public', path.posix.basename(workspacePath)),
  ]);
  const siblingDocx = resolveSiblingDocxPath(htmlRelativePath);
  if (siblingDocx) {
    candidates.add(siblingDocx);
    candidates.add(path.posix.join('public', path.posix.basename(siblingDocx)));
  }
  return [...candidates];
}

export function resolveDownloadTargetUrl({
  relativeRef,
  htmlRelativePath = 'public/index.html',
  publishDir = null,
  linkIndex = null,
  publicBaseUrl = '',
  publishKey = null,
  preferAssetDownload = true,
}) {
  const { suffix } = splitReferenceParts(relativeRef);
  const candidates = collectDownloadWorkspaceCandidates(htmlRelativePath, relativeRef, publishDir);

  for (const candidate of candidates) {
    const assetId = lookupAssetIdForWorkspacePath(linkIndex, candidate);
    if (preferAssetDownload && assetId) {
      return `${buildAssetDownloadUrl(assetId)}${suffix}`;
    }
  }

  if (publishDir && publishKey) {
    for (const candidate of candidates) {
      if (workspaceFileExists(publishDir, candidate)) {
        return `${buildPublicWorkspaceFileUrl(publicBaseUrl, publishKey, candidate)}${suffix}`;
      }
    }
  }

  for (const candidate of candidates) {
    const assetId = lookupAssetIdForWorkspacePath(linkIndex, candidate);
    if (assetId) {
      return `${buildAssetDownloadUrl(assetId)}${suffix}`;
    }
  }

  return null;
}

export function rewriteRelativeDownloadLinks(html, options = {}) {
  const source = String(html ?? '');
  if (!source) return { html: source, count: 0 };

  let count = 0;
  const rewritten = source.replace(URL_ATTR_PATTERN, (full, prefix, url, suffix) => {
    if (!isRelativeDownloadReference(url)) return full;
    const resolved = resolveDownloadTargetUrl({ ...options, relativeRef: url });
    if (!resolved) return full;
    count += 1;
    return `${prefix}${resolved}${suffix}`;
  });

  return { html: rewritten, count };
}

export async function loadWorkspaceDownloadLinkIndex(pool, userId) {
  const [rows] = await pool.query(
    `SELECT id, original_filename
     FROM h5_assets
     WHERE user_id = ? AND status <> 'deleted'`,
    [userId],
  );
  return buildWorkspaceDownloadLinkIndex(rows);
}

export async function prepareHtmlDownloadLinks(pool, userId, html, options = {}) {
  const linkIndex = options.linkIndex ?? (await loadWorkspaceDownloadLinkIndex(pool, userId));
  return rewriteRelativeDownloadLinks(html, { ...options, linkIndex });
}

export const downloadLinkInternals = {
  isRelativeDownloadReference,
  resolveWorkspaceRelativeFilePath,
  resolveSiblingDocxPath,
  inferWorkspaceHtmlRelativePath,
  buildWorkspaceDownloadLinkIndex,
  lookupAssetIdForWorkspacePath,
  resolveDownloadTargetUrl,
  rewriteRelativeDownloadLinks,
};