update url uploads

This commit is contained in:
2026-07-12 18:46:09 +02:00
parent 06ced46db1
commit 6adfb55581
4 changed files with 589 additions and 17 deletions

View File

@@ -1,4 +1,5 @@
import { promises as fs } from "fs";
import { spawn as _spawnRaw } from 'child_process';
import db from '../../sql.mjs';
import lib from '../../lib.mjs';
import cfg from '../../config.mjs';
@@ -6,6 +7,90 @@ import { applyWordFilter } from '../../wordfilter.mjs';
import queue from '../../queue.mjs';
import path from "path";
// ──────────────────────────────────────────────────────────────────────
// In-memory job progress map (keyed by jobId string)
// Entries: { stage, percent, speed, eta, done, error }
// Auto-cleaned 10 min after completion.
// ──────────────────────────────────────────────────────────────────────
const progressMap = new Map();
const setProgress = (jobId, patch) => {
const existing = progressMap.get(jobId) || { stage: 'queued', percent: 0, speed: null, eta: null, done: false, error: null };
progressMap.set(jobId, { ...existing, ...patch });
};
const cleanupJob = (jobId) => {
setTimeout(() => progressMap.delete(jobId), 10 * 60 * 1000);
};
/**
* Like queue.spawn() but streams stderr lines to a callback for live progress,
* while still resolving with { stdout, stderr } when the process exits.
* Splits on both \n and \r so yt-dlp's carriage-return progress works.
*/
const spawnWithProgress = (cmd, args, onLine) => {
return new Promise((resolve, reject) => {
const child = _spawnRaw(cmd, args);
const stdoutChunks = [];
const stderrChunks = [];
let stderrBuf = '';
if (child.stdout) child.stdout.on('data', d => stdoutChunks.push(d));
if (child.stderr) {
child.stderr.on('data', chunk => {
stderrChunks.push(chunk);
stderrBuf += chunk.toString();
// Split on \r or \n (yt-dlp uses \r to overwrite progress in-place)
const parts = stderrBuf.split(/[\r\n]/);
// Keep the last (potentially incomplete) segment in the buffer
stderrBuf = parts.pop() ?? '';
for (const line of parts) {
// Strip ANSI escape codes
const clean = line.replace(/\x1b\[[0-9;]*m/g, '').trim();
if (clean && onLine) onLine(clean);
}
});
}
child.on('close', code => {
// Process any remaining buffered content
if (stderrBuf.trim() && onLine) {
const clean = stderrBuf.replace(/\x1b\[[0-9;]*m/g, '').trim();
if (clean) onLine(clean);
}
const stdout = Buffer.concat(stdoutChunks).toString();
const stderr = Buffer.concat(stderrChunks).toString();
if (code !== 0) {
const err = new Error(`Command '${cmd} ${args.join(' ')}' failed with code ${code}`);
err.stderr = stderr;
err.stdout = stdout;
return reject(err);
}
resolve({ stdout, stderr });
});
child.on('error', err => {
err.stderr = Buffer.concat(stderrChunks).toString();
err.stdout = Buffer.concat(stdoutChunks).toString();
reject(err);
});
});
};
/** Parse a yt-dlp stderr line and return a progress patch object or null. */
const parseYtdlpLine = (line) => {
// [download] 47.3% of ~ 58.23MiB at 3.22MiB/s ETA 00:13
const dlMatch = line.match(/\[download\]\s+([\d.]+)%.*?at\s+([\d.]+\s*\S+\/s)(?:.*?ETA\s+(\S+))?/);
if (dlMatch) {
return { stage: 'downloading', percent: parseFloat(dlMatch[1]), speed: dlMatch[2] || null, eta: dlMatch[3] || null };
}
// [download] Destination: ...
if (line.includes('[download] Destination:')) return { stage: 'downloading' };
if (line.includes('[Merger]') || line.includes('[ffmpeg]')) return { stage: 'processing', percent: 99 };
if (line.includes('[ExtractAudio]')) return { stage: 'extracting', percent: 99 };
return null;
};
// Native multipart form data parser
const parseMultipart = (buffer, boundary) => {
const parts = {};
@@ -80,6 +165,15 @@ const collectBody = (req) => {
export default router => {
router.group(/^\/api\/v2/, group => {
// ── GET /api/v2/upload-url/progress/:jobId ──────────────────────────────
group.get(/\/upload-url\/progress\/(?<jobId>[a-zA-Z0-9_-]+)$/, lib.loggedin, (req, res) => {
const jobId = req.params?.jobId || (req.url?.pathname || req.url || '').split('/').pop();
const state = progressMap.get(jobId);
res.setHeader?.('Cache-Control', 'no-store');
if (!state) return res.json({ success: false, msg: 'Job not found' }, 404);
return res.json({ success: true, ...state });
});
const saveComment = async (itemid, userid, content) => {
if (!content || !content.trim()) return;
try {
@@ -345,9 +439,14 @@ export default router => {
};
// Return immediately to avoid proxy timeouts
// Generate a client-side trackable job ID
const jobId = await queue.genuuid();
setProgress(jobId, { stage: 'queued', percent: 0, speed: null, eta: null, done: false, error: null });
res.json({
success: true,
pending: true,
jobId,
msg: 'URL processing started in background. You will receive a notification when it is finished.'
});
@@ -392,7 +491,7 @@ export default router => {
try {
const proxyArgs = (cfg.main.socks && cfg.main.socks !== 'undefined') ? ['--proxy', cfg.main.socks] : [];
const ytdlpArgs = ['--js-runtimes', 'node', '--geo-bypass', '--extractor-args', 'youtube:player-client=ios,web'];
const ytdlpArgs = ['--js-runtimes', 'node', '--geo-bypass', '--extractor-args', 'youtube:player-client=ios,web', '--newline', '--no-colors'];
let maxfilesize = cfg.main.maxfilesize;
if (session.admin) maxfilesize = Math.floor(maxfilesize * cfg.main.adminmultiplier);
@@ -403,9 +502,10 @@ export default router => {
let source;
console.log(`[UPLOAD-URL-ASYNC] Starting Stage 1 (constrained) download for ${url} (user: ${session.user})`);
setProgress(jobId, { stage: 'downloading', percent: 0 });
try {
source = (await queue.spawn('yt-dlp', [
source = (await spawnWithProgress('yt-dlp', [
...proxyArgs, ...ytdlpArgs,
'-f', 'bv*[height<=1080]+ba/b[height<=1080] / wv*+ba/w',
url,
@@ -414,22 +514,30 @@ export default router => {
'-o', path.join(cfg.paths.tmp, `${uuid}.%(ext)s`),
'--print', 'after_move:filepath',
'--merge-output-format', 'mp4'
])).stdout.trim().split('\n').map(l => l.trim()).filter(l => l.length > 0).pop();
], (line) => {
const patch = parseYtdlpLine(line);
if (patch) setProgress(jobId, patch);
})).stdout.trim().split('\n').map(l => l.trim()).filter(l => l.length > 0).pop();
} catch (err) {
console.warn(`[UPLOAD-URL-ASYNC] Stage 1 failed: ${err.message}`);
if (isInstagram) throw new Error(sanitizeError(err));
setProgress(jobId, { stage: 'downloading', percent: 0, speed: null, eta: null });
try {
source = (await queue.spawn('yt-dlp', [
source = (await spawnWithProgress('yt-dlp', [
...proxyArgs, ...ytdlpArgs,
url,
'--max-filesize', `${maxfilesize / 1024}k`,
'-o', path.join(cfg.paths.tmp, `${uuid}.%(ext)s`),
'--print', 'after_move:filepath'
])).stdout.trim().split('\n').map(l => l.trim()).filter(l => l.length > 0).pop();
], (line) => {
const patch = parseYtdlpLine(line);
if (patch) setProgress(jobId, patch);
})).stdout.trim().split('\n').map(l => l.trim()).filter(l => l.length > 0).pop();
} catch (err2) {
console.warn(`[UPLOAD-URL-ASYNC] Stage 2 failed: ${err2.message}`);
console.log(`[UPLOAD-URL-ASYNC] Starting Stage 3 (curl) fallback for ${url}`);
setProgress(jobId, { stage: 'downloading', percent: 0, speed: null, eta: null });
const fallbackTmp = path.join(cfg.paths.tmp, `${uuid}.tmp`);
let referer = url;
try {
@@ -474,6 +582,8 @@ export default router => {
if (!source || source.match(/larger than/)) throw new Error('File too large or download failed');
setProgress(jobId, { stage: 'analyzing', percent: 100, speed: null, eta: null });
const { stat } = await import('fs/promises');
const size = (await stat(source)).size;
if (size > maxfilesize) {
@@ -515,7 +625,7 @@ export default router => {
const repostSum = await queue.checkrepostsum(checksum);
if (repostSum) {
await fs.unlink(source).catch(() => {});
await db`INSERT INTO notifications (user_id, type, reference_id, item_id, data) VALUES (${session.id}, 'upload_error', 0, ${repostSum}, ${db.json({ url, msg: 'Duplicate detected (Checksum)' })})`;
await db`INSERT INTO notifications (user_id, type, reference_id, item_id, data) VALUES (${session.id}, 'upload_error', 0, ${repostSum}, ${db.json({ jobId, url, msg: 'Duplicate detected (Checksum)' })})`;
return;
}
}
@@ -527,7 +637,7 @@ export default router => {
const phashMatch = await queue.checkrepostphash(phash);
if (phashMatch) {
await fs.unlink(source).catch(() => {});
await db`INSERT INTO notifications (user_id, type, reference_id, item_id, data) VALUES (${session.id}, 'upload_error', 0, ${phashMatch}, ${db.json({ url, msg: 'Visual duplicate detected (PHash)' })})`;
await db`INSERT INTO notifications (user_id, type, reference_id, item_id, data) VALUES (${session.id}, 'upload_error', 0, ${phashMatch}, ${db.json({ jobId, url, msg: 'Visual duplicate detected (PHash)' })})`;
return;
}
}
@@ -548,6 +658,7 @@ export default router => {
} catch (e) { }
}
}
setProgress(jobId, { stage: 'saving', percent: 100 });
if (!linkedToExistingUrl) await fs.copyFile(source, path.join(destDir, filename));
await fs.unlink(source).catch(() => { });
@@ -640,12 +751,16 @@ export default router => {
}
// Completion notification
await db`INSERT INTO notifications (user_id, type, reference_id, item_id) VALUES (${session.id}, 'upload_success', 0, ${itemid})`;
setProgress(jobId, { stage: 'done', percent: 100, done: true, error: null });
cleanupJob(jobId);
await db`INSERT INTO notifications (user_id, type, reference_id, item_id, data) VALUES (${session.id}, 'upload_success', 0, ${itemid}, ${db.json({ jobId })})`;
} catch (err) {
console.error('[UPLOAD-URL-ASYNC] Final Error:', err);
// Error notification
await db`INSERT INTO notifications (user_id, type, reference_id, item_id, data) VALUES (${session.id}, 'upload_error', 0, ${null}, ${db.json({ url, msg: sanitizeError(err) })})`;
setProgress(jobId, { stage: 'error', done: true, error: sanitizeError(err) });
cleanupJob(jobId);
await db`INSERT INTO notifications (user_id, type, reference_id, item_id, data) VALUES (${session.id}, 'upload_error', 0, ${null}, ${db.json({ jobId, url, msg: sanitizeError(err) })})`;
}
})();
}