Create repair-transcript-images.js

Author: Musiker15

scripts/repair-transcript-images.js

@@ -0,0 +1,174 @@
+#!/usr/bin/env node
+/**
+ * repair-transcript-images.js — one-off repair for already-stored transcripts.
+ *
+ * Background: older transcripts embedded message image/file attachments using the
+ * raw Discord CDN URL (cdn.discordapp.com/attachments/…). Those URLs are signed
+ * and expire ~24h after generation, so every such image is now broken. The bot
+ * already uploaded a permanent copy of each attachment to this server
+ * (attachments/<uuid>.<ext>), it was just never referenced by the HTML.
+ *
+ * This script rewrites the stored transcript.html files: it replaces every dead
+ * Discord CDN attachment URL with the matching local copy, addressed by a RELATIVE
+ * path (attachments/<uuid>.<ext>) so it resolves under both a custom domain and
+ * the main site.
+ *
+ * Mapping: old transcripts don't store the Discord attachment id, so files are
+ * matched by ORIGINAL FILENAME (present both in the Discord URL and in
+ * ticketbot_attachments.original_name). Unique names → exact. Several files
+ * sharing a name within one transcript (e.g. multiple "image.png") → matched
+ * positionally in document order (best effort; logged as ambiguous).
+ *
+ * SAFE BY DEFAULT: dry-run unless --apply is passed. The rewrite is idempotent
+ * (it only touches cdn.discordapp.com/media.discordapp.net attachment URLs that
+ * map to a stored file), so re-running causes no further changes.
+ *
+ * Deployed with the repo at /opt/msk-shop/scripts/. Reads DB_* from the env but
+ * does NOT load dotenv and requires mysql2 — run it like cleanup.js:
+ *
+ *   set -a; . /opt/msk-shop/.env.local; set +a; \
+ *     NODE_PATH=/opt/msk-shop/node_modules /usr/bin/node \
+ *     /opt/msk-shop/scripts/repair-transcript-images.js            # dry-run
+ *
+ *   …same… /opt/msk-shop/scripts/repair-transcript-images.js --apply   # write
+ *
+ * Recommended: snapshot the transcripts dir first, e.g.
+ *   tar czf /root/transcripts-backup-$(date +%F).tgz -C /var/www/html transcripts
+ *
+ * Flags:
+ *   --apply           actually write the files (default: dry-run, no writes)
+ *   --guild <id>      only process transcripts of this guild_id
+ *   --limit <n>       process at most n transcripts (useful for a test run)
+ */
+
+const { readFile, writeFile } = require('fs/promises');
+const path  = require('path');
+const mysql = require('mysql2/promise');
+
+// Discord attachment URLs as embedded in the transcript HTML (raw, inside a
+// src="" / href="" attribute). Both CDN hosts are matched defensively.
+const DISCORD_URL_RE =
+  /(["'])(https:\/\/(?:cdn\.discordapp\.com|media\.discordapp\.net)\/attachments\/\d+\/\d+\/[^"']+)\1/g;
+
+function parseArgs(argv) {
+  const args = { apply: false, guild: null, limit: null };
+  for (let i = 0; i < argv.length; i++) {
+    const a = argv[i];
+    if (a === '--apply')      args.apply = true;
+    else if (a === '--guild') args.guild = argv[++i] ?? null;
+    else if (a === '--limit') args.limit = Number(argv[++i] ?? '') || null;
+  }
+  return args;
+}
+
+/** Filename segment of a Discord attachment URL, query stripped + percent-decoded. */
+function filenameFromUrl(url) {
+  const noQuery = url.split('?')[0];
+  const last    = noQuery.substring(noQuery.lastIndexOf('/') + 1);
+  try { return decodeURIComponent(last); } catch { return last; }
+}
+
+async function main() {
+  const args = parseArgs(process.argv.slice(2));
+  const mode = args.apply ? 'APPLY' : 'DRY-RUN';
+
+  const pool = mysql.createPool({
+    host:     process.env.DB_HOST     ?? 'localhost',
+    port:     Number(process.env.DB_PORT ?? 3306),
+    user:     process.env.DB_USER     ?? '',
+    password: process.env.DB_PASSWORD ?? '',
+    database: process.env.DB_NAME     ?? '',
+  });
+
+  console.log(`[repair] Starting (${mode}) at ${new Date().toISOString()}`);
+
+  const where  = ['has_attachments = 1'];
+  const params = [];
+  if (args.guild) { where.push('guild_id = ?'); params.push(args.guild); }
+  let sql = `SELECT id, file_path FROM ticketbot_transcripts WHERE ${where.join(' AND ')} ORDER BY created_at ASC`;
+  if (args.limit) sql += ` LIMIT ${args.limit}`;
+
+  const [transcripts] = await pool.execute(sql, params);
+  console.log(`[repair] ${transcripts.length} transcript(s) with attachments to inspect`);
+
+  let filesChanged = 0, urlsReplaced = 0, urlsUnmatched = 0, ambiguous = 0, errors = 0;
+
+  for (const t of transcripts) {
+    try {
+      const [atts] = await pool.execute(
+        `SELECT original_name, file_path FROM ticketbot_attachments WHERE transcript_id = ?`,
+        [t.id],
+      );
+      if (atts.length === 0) continue;
+
+      // Group local copies by original filename → list of relative URLs.
+      const groups = new Map();   // name → ["attachments/<uuid>.<ext>", …]
+      for (const a of atts) {
+        const rel = `attachments/${path.basename(a.file_path)}`;
+        if (!groups.has(a.original_name)) groups.set(a.original_name, []);
+        groups.get(a.original_name).push(rel);
+      }
+
+      let html;
+      try {
+        html = await readFile(t.file_path, 'utf-8');
+      } catch (err) {
+        console.warn(`[repair] ${t.id}: cannot read ${t.file_path} (${err.code || err.message}) — skipped`);
+        errors++;
+        continue;
+      }
+
+      const cursor = new Map();   // name → next index into its group
+      let replacedHere = 0, unmatchedHere = 0;
+
+      const out = html.replace(DISCORD_URL_RE, (full, quote, url) => {
+        const name = filenameFromUrl(url);
+        const list = groups.get(name);
+        const idx  = cursor.get(name) ?? 0;
+        if (list && idx < list.length) {
+          cursor.set(name, idx + 1);
+          replacedHere++;
+          return `${quote}${list[idx]}${quote}`;
+        }
+        unmatchedHere++;
+        return full;   // no stored copy for this name → leave the (dead) link
+      });
+
+      // Flag transcripts where a name maps to >1 local file (positional guess).
+      for (const [, list] of groups) if (list.length > 1) { ambiguous++; break; }
+
+      urlsReplaced  += replacedHere;
+      urlsUnmatched += unmatchedHere;
+
+      if (out !== html) {
+        filesChanged++;
+        if (args.apply) {
+          await writeFile(t.file_path, out, 'utf-8');
+          console.log(`[repair] ${t.id}: replaced ${replacedHere} link(s)` + (unmatchedHere ? `, ${unmatchedHere} unmatched` : ''));
+        } else {
+          console.log(`[repair] ${t.id}: WOULD replace ${replacedHere} link(s)` + (unmatchedHere ? `, ${unmatchedHere} unmatched` : ''));
+        }
+      } else if (unmatchedHere) {
+        console.log(`[repair] ${t.id}: 0 replaced, ${unmatchedHere} Discord link(s) without a stored copy`);
+      }
+    } catch (err) {
+      console.error(`[repair] ${t.id}: error — ${err.message}`);
+      errors++;
+    }
+  }
+
+  await pool.end();
+  console.log(
+    `[repair] Done (${mode}). Files ${args.apply ? 'changed' : 'to change'}: ${filesChanged}, ` +
+    `links replaced: ${urlsReplaced}, unmatched: ${urlsUnmatched}, ` +
+    `transcripts with duplicate-name guesses: ${ambiguous}, errors: ${errors}`,
+  );
+  if (!args.apply && filesChanged > 0) {
+    console.log('[repair] This was a DRY-RUN — re-run with --apply to write the changes.');
+  }
+}
+
+main().catch(err => {
+  console.error('[repair] Fatal error:', err);
+  process.exit(1);
+});