Funksiyalar

Fayllar va extraction

Presigned/multipart upload, integrity tekshiruvi, preview, extraction va cleanup.

Upload protokoli

Client filename, size, MIME va SHA-256 bilan upload intent yaratadi.
API private object key va 300 soniyalik upload URL/proxy qaytaradi.
Katta faylda multipart part URL’lari olinadi va ETaglar bilan finalize qilinadi.
Client complete endpointni chaqiradi.
Server HEAD/stream orqali size, digest, MIME va magic bytesni tekshiradi.
Verified state va extraction job bir izchil DB operatsiyasida yaratiladi.

Format va chegaralar

Parametr	Qiymat
Maksimal fayl	50 MiB
Filename	255 baytgacha
Upload URL TTL	300 soniya
Integrity	Base64 SHA-256 + obyekt size
Rasmlar	PNG, JPEG, WebP
Hujjatlar	PDF, DOCX, plain text, Markdown, JSON, CSV
Kod	Keng tarqalgan source extensionlari
Arxiv	ZIP, bounded extraction

Xavfsiz extraction

PDF child process 30 soniyadan keyin kill qilinadi va GOMEMLIMIT=256MiB bilan cheklanadi.
DOCX/ZIP entry soni, uncompressed size va path traversal bo‘yicha bounded.
Declared MIME extension va magic-byte sniffing bilan solishtiriladi.
Extraction natijasi chunklarga yoziladi; job retry olsa, fencing eski workerning write’ini rad etadi.

Preview, download va delete

Preview va download avval user ownershipni tekshiradi, keyin private obyektga vaqtinchalik kirish beradi. O‘chirish ikki bosqichli: DB yozuvi deleting bo‘ladi, worker S3 obyekt/chunklarni tozalaydi va deleted holatiga o‘tkazadi.

Delete state machine

ready | failed → deleting → deleted
                      └─ retryable cleanup job

Single-part upload misoli

Intent → upload → complete

SHA256_BASE64=$(openssl dgst -sha256 -binary report.pdf | openssl base64 -A)

curl -X POST "$APP_URL/api/files/presign" \
  -H "Content-Type: application/json" -H "X-CSRF-Token: $CSRF" \
  -b cookies.txt \
  -d "{\"filename\":\"report.pdf\",\"size\":1048576,\"mime_type\":\"application/pdf\",\"sha256\":\"$SHA256_BASE64\"}"

# Javobdagi upload URL bilan PUT qiling, so‘ng:
curl -X POST "$APP_URL/api/files/$FILE_ID/complete" \
  -H "X-CSRF-Token: $CSRF" -b cookies.txt