Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,1083 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import time
|
| 3 |
+
import hashlib
|
| 4 |
+
import threading
|
| 5 |
+
import asyncio
|
| 6 |
+
import re
|
| 7 |
+
from datetime import datetime, timedelta
|
| 8 |
+
from concurrent.futures import ThreadPoolExecutor
|
| 9 |
+
from typing import Dict, List, Optional, Any
|
| 10 |
+
import json
|
| 11 |
+
|
| 12 |
+
from fastapi import FastAPI, HTTPException, BackgroundTasks
|
| 13 |
+
from fastapi.middleware.cors import CORSMiddleware
|
| 14 |
+
from fastapi.responses import JSONResponse
|
| 15 |
+
from pydantic import BaseModel, Field
|
| 16 |
+
import torch
|
| 17 |
+
from transformers import M2M100ForConditionalGeneration, M2M100Tokenizer
|
| 18 |
+
import requests
|
| 19 |
+
|
| 20 |
+
# تنظیمات اولیه
|
| 21 |
+
MODEL_NAME = "facebook/m2m100_418M"
|
| 22 |
+
CACHE_EXPIRY = 60 * 60 # 60 دقیقه
|
| 23 |
+
MAX_CHUNK_SIZE = 350
|
| 24 |
+
MAX_WORKERS = 3
|
| 25 |
+
CLEANUP_INTERVAL = 300 # 5 دقیقه
|
| 26 |
+
|
| 27 |
+
# نگاشت زبانها
|
| 28 |
+
LANGUAGE_MAP = {
|
| 29 |
+
"English": "en",
|
| 30 |
+
"Persian": "fa",
|
| 31 |
+
"Farsi": "fa",
|
| 32 |
+
"Arabic": "ar",
|
| 33 |
+
"Spanish": "es",
|
| 34 |
+
"French": "fr",
|
| 35 |
+
"German": "de",
|
| 36 |
+
"Italian": "it",
|
| 37 |
+
"Portuguese": "pt",
|
| 38 |
+
"Russian": "ru",
|
| 39 |
+
"Chinese": "zh",
|
| 40 |
+
"Japanese": "ja",
|
| 41 |
+
"Korean": "ko",
|
| 42 |
+
"Hindi": "hi",
|
| 43 |
+
"Turkish": "tr",
|
| 44 |
+
"Dutch": "nl",
|
| 45 |
+
"Swedish": "sv",
|
| 46 |
+
"Norwegian": "no",
|
| 47 |
+
"Danish": "da",
|
| 48 |
+
"Finnish": "fi",
|
| 49 |
+
"Polish": "pl",
|
| 50 |
+
"Czech": "cs",
|
| 51 |
+
"Hungarian": "hu",
|
| 52 |
+
"Romanian": "ro",
|
| 53 |
+
"Greek": "el",
|
| 54 |
+
"Hebrew": "he",
|
| 55 |
+
"Thai": "th",
|
| 56 |
+
"Vietnamese": "vi",
|
| 57 |
+
"Indonesian": "id",
|
| 58 |
+
"Malay": "ms",
|
| 59 |
+
"Tamil": "ta",
|
| 60 |
+
"Bengali": "bn",
|
| 61 |
+
"Urdu": "ur",
|
| 62 |
+
"Ukrainian": "uk",
|
| 63 |
+
"Bulgarian": "bg",
|
| 64 |
+
"Croatian": "hr",
|
| 65 |
+
"Serbian": "sr",
|
| 66 |
+
"Slovak": "sk",
|
| 67 |
+
"Slovenian": "sl",
|
| 68 |
+
"Estonian": "et",
|
| 69 |
+
"Latvian": "lv",
|
| 70 |
+
"Lithuanian": "lt",
|
| 71 |
+
"Maltese": "mt",
|
| 72 |
+
"Catalan": "ca",
|
| 73 |
+
"Galician": "gl",
|
| 74 |
+
"Basque": "eu",
|
| 75 |
+
"Welsh": "cy",
|
| 76 |
+
"Irish": "ga",
|
| 77 |
+
"Scottish": "gd",
|
| 78 |
+
"Icelandic": "is",
|
| 79 |
+
"Albanian": "sq",
|
| 80 |
+
"Macedonian": "mk",
|
| 81 |
+
"Bosnian": "bs",
|
| 82 |
+
"Montenegrin": "cnr",
|
| 83 |
+
"Swahili": "sw",
|
| 84 |
+
"Amharic": "am",
|
| 85 |
+
"Yoruba": "yo",
|
| 86 |
+
"Igbo": "ig",
|
| 87 |
+
"Hausa": "ha",
|
| 88 |
+
"Somali": "so",
|
| 89 |
+
"Oromo": "om",
|
| 90 |
+
"Tigrinya": "ti",
|
| 91 |
+
"Afrikaans": "af",
|
| 92 |
+
"Zulu": "zu",
|
| 93 |
+
"Xhosa": "xh",
|
| 94 |
+
"Sotho": "st",
|
| 95 |
+
"Tswana": "tn",
|
| 96 |
+
"Tsonga": "ts",
|
| 97 |
+
"Venda": "ve",
|
| 98 |
+
"Ndebele": "nr",
|
| 99 |
+
"Gujarati": "gu",
|
| 100 |
+
"Punjabi": "pa",
|
| 101 |
+
"Telugu": "te",
|
| 102 |
+
"Kannada": "kn",
|
| 103 |
+
"Malayalam": "ml",
|
| 104 |
+
"Marathi": "mr",
|
| 105 |
+
"Nepali": "ne",
|
| 106 |
+
"Sinhala": "si",
|
| 107 |
+
"Burmese": "my",
|
| 108 |
+
"Khmer": "km",
|
| 109 |
+
"Lao": "lo",
|
| 110 |
+
"Mongolian": "mn",
|
| 111 |
+
"Kazakh": "kk",
|
| 112 |
+
"Uzbek": "uz",
|
| 113 |
+
"Tajik": "tg",
|
| 114 |
+
"Kyrgyz": "ky",
|
| 115 |
+
"Turkmen": "tk",
|
| 116 |
+
"Azerbaijani": "az",
|
| 117 |
+
"Georgian": "ka",
|
| 118 |
+
"Armenian": "hy"
|
| 119 |
+
}
|
| 120 |
+
|
| 121 |
+
# مدلهای Pydantic
|
| 122 |
+
class TranslationRequest(BaseModel):
|
| 123 |
+
text: str = Field(..., description="متن برای ترجمه")
|
| 124 |
+
source_lang: str = Field(..., description="زبان مبدا")
|
| 125 |
+
target_lang: str = Field(..., description="زبان مقصد")
|
| 126 |
+
auto_charge: bool = Field(default=False, description="کسر خودکار اعتبار")
|
| 127 |
+
|
| 128 |
+
class TranslationFormRequest(BaseModel):
|
| 129 |
+
text: str
|
| 130 |
+
source_lang: str
|
| 131 |
+
target_lang: str
|
| 132 |
+
|
| 133 |
+
class CompletionCheckRequest(BaseModel):
|
| 134 |
+
request_id: str
|
| 135 |
+
|
| 136 |
+
class StatusCheckRequest(BaseModel):
|
| 137 |
+
request_id: str
|
| 138 |
+
|
| 139 |
+
class AutoChargeStatusRequest(BaseModel):
|
| 140 |
+
request_id: str
|
| 141 |
+
|
| 142 |
+
# کلاس کش ترجمه
|
| 143 |
+
class TranslationCache:
|
| 144 |
+
def __init__(self, expiry_minutes: int = 60):
|
| 145 |
+
self.cache: Dict[str, Dict] = {}
|
| 146 |
+
self.expiry = expiry_minutes * 60
|
| 147 |
+
self.lock = threading.Lock()
|
| 148 |
+
|
| 149 |
+
def _generate_key(self, text: str, source_lang: str, target_lang: str) -> str:
|
| 150 |
+
content = f"{text}:{source_lang}:{target_lang}"
|
| 151 |
+
return hashlib.md5(content.encode()).hexdigest()
|
| 152 |
+
|
| 153 |
+
def get(self, text: str, source_lang: str, target_lang: str) -> Optional[str]:
|
| 154 |
+
key = self._generate_key(text, source_lang, target_lang)
|
| 155 |
+
with self.lock:
|
| 156 |
+
if key in self.cache:
|
| 157 |
+
entry = self.cache[key]
|
| 158 |
+
if time.time() - entry['timestamp'] < self.expiry:
|
| 159 |
+
return entry['translation']
|
| 160 |
+
else:
|
| 161 |
+
del self.cache[key]
|
| 162 |
+
return None
|
| 163 |
+
|
| 164 |
+
def set(self, text: str, source_lang: str, target_lang: str, translation: str):
|
| 165 |
+
key = self._generate_key(text, source_lang, target_lang)
|
| 166 |
+
with self.lock:
|
| 167 |
+
self.cache[key] = {
|
| 168 |
+
'translation': translation,
|
| 169 |
+
'timestamp': time.time()
|
| 170 |
+
}
|
| 171 |
+
|
| 172 |
+
def clear_expired(self):
|
| 173 |
+
current_time = time.time()
|
| 174 |
+
with self.lock:
|
| 175 |
+
expired_keys = [
|
| 176 |
+
key for key, entry in self.cache.items()
|
| 177 |
+
if current_time - entry['timestamp'] >= self.expiry
|
| 178 |
+
]
|
| 179 |
+
for key in expired_keys:
|
| 180 |
+
del self.cache[key]
|
| 181 |
+
|
| 182 |
+
def get_stats(self) -> Dict:
|
| 183 |
+
with self.lock:
|
| 184 |
+
return {
|
| 185 |
+
'cache_size': len(self.cache),
|
| 186 |
+
'total_entries': len(self.cache)
|
| 187 |
+
}
|
| 188 |
+
|
| 189 |
+
# کلاس تقسیم متن
|
| 190 |
+
class TextChunker:
|
| 191 |
+
def __init__(self, max_chunk_size: int = MAX_CHUNK_SIZE):
|
| 192 |
+
self.max_chunk_size = max_chunk_size
|
| 193 |
+
|
| 194 |
+
def chunk_text(self, text: str) -> List[str]:
|
| 195 |
+
if len(text) <= self.max_chunk_size:
|
| 196 |
+
return [text]
|
| 197 |
+
|
| 198 |
+
# تقسیم بر اساس پاراگراف
|
| 199 |
+
paragraphs = text.split('\n\n')
|
| 200 |
+
chunks = []
|
| 201 |
+
current_chunk = ""
|
| 202 |
+
|
| 203 |
+
for paragraph in paragraphs:
|
| 204 |
+
if len(current_chunk) + len(paragraph) <= self.max_chunk_size:
|
| 205 |
+
if current_chunk:
|
| 206 |
+
current_chunk += '\n\n' + paragraph
|
| 207 |
+
else:
|
| 208 |
+
current_chunk = paragraph
|
| 209 |
+
else:
|
| 210 |
+
if current_chunk:
|
| 211 |
+
chunks.append(current_chunk)
|
| 212 |
+
|
| 213 |
+
if len(paragraph) <= self.max_chunk_size:
|
| 214 |
+
current_chunk = paragraph
|
| 215 |
+
else:
|
| 216 |
+
# تقسیم پاراگراف طولانی
|
| 217 |
+
sub_chunks = self._split_long_paragraph(paragraph)
|
| 218 |
+
chunks.extend(sub_chunks[:-1])
|
| 219 |
+
current_chunk = sub_chunks[-1] if sub_chunks else ""
|
| 220 |
+
|
| 221 |
+
if current_chunk:
|
| 222 |
+
chunks.append(current_chunk)
|
| 223 |
+
|
| 224 |
+
return chunks
|
| 225 |
+
|
| 226 |
+
def _split_long_paragraph(self, text: str) -> List[str]:
|
| 227 |
+
# تقسیم بر اساس جملات
|
| 228 |
+
sentences = re.split(r'[.!?]+\s+', text)
|
| 229 |
+
chunks = []
|
| 230 |
+
current_chunk = ""
|
| 231 |
+
|
| 232 |
+
for sentence in sentences:
|
| 233 |
+
if len(current_chunk) + len(sentence) <= self.max_chunk_size:
|
| 234 |
+
if current_chunk:
|
| 235 |
+
current_chunk += '. ' + sentence
|
| 236 |
+
else:
|
| 237 |
+
current_chunk = sentence
|
| 238 |
+
else:
|
| 239 |
+
if current_chunk:
|
| 240 |
+
chunks.append(current_chunk)
|
| 241 |
+
|
| 242 |
+
if len(sentence) <= self.max_chunk_size:
|
| 243 |
+
current_chunk = sentence
|
| 244 |
+
else:
|
| 245 |
+
# تقسیم بر اساس کاما
|
| 246 |
+
comma_parts = sentence.split(', ')
|
| 247 |
+
for part in comma_parts:
|
| 248 |
+
if len(current_chunk) + len(part) <= self.max_chunk_size:
|
| 249 |
+
if current_chunk:
|
| 250 |
+
current_chunk += ', ' + part
|
| 251 |
+
else:
|
| 252 |
+
current_chunk = part
|
| 253 |
+
else:
|
| 254 |
+
if current_chunk:
|
| 255 |
+
chunks.append(current_chunk)
|
| 256 |
+
current_chunk = part
|
| 257 |
+
|
| 258 |
+
if current_chunk:
|
| 259 |
+
chunks.append(current_chunk)
|
| 260 |
+
|
| 261 |
+
return chunks
|
| 262 |
+
|
| 263 |
+
# صف ترجمه
|
| 264 |
+
class TranslationQueue:
|
| 265 |
+
def __init__(self, max_workers: int = MAX_WORKERS):
|
| 266 |
+
self.executor = ThreadPoolExecutor(max_workers=max_workers)
|
| 267 |
+
self.tasks: Dict[str, Dict] = {}
|
| 268 |
+
self.lock = threading.Lock()
|
| 269 |
+
|
| 270 |
+
def add_task(self, session_id: str, task_func, *args, **kwargs):
|
| 271 |
+
with self.lock:
|
| 272 |
+
future = self.executor.submit(task_func, *args, **kwargs)
|
| 273 |
+
self.tasks[session_id] = {
|
| 274 |
+
'future': future,
|
| 275 |
+
'start_time': time.time(),
|
| 276 |
+
'status': 'processing'
|
| 277 |
+
}
|
| 278 |
+
|
| 279 |
+
def get_task_status(self, session_id: str) -> Optional[Dict]:
|
| 280 |
+
with self.lock:
|
| 281 |
+
return self.tasks.get(session_id)
|
| 282 |
+
|
| 283 |
+
def remove_task(self, session_id: str):
|
| 284 |
+
with self.lock:
|
| 285 |
+
if session_id in self.tasks:
|
| 286 |
+
del self.tasks[session_id]
|
| 287 |
+
|
| 288 |
+
# کلاس اصلی مترجم
|
| 289 |
+
class MultilingualTranslator:
|
| 290 |
+
def __init__(self, cache_expiry_minutes: int = 60):
|
| 291 |
+
print("در حال بارگذاری مدل M2M100...")
|
| 292 |
+
|
| 293 |
+
# بارگذاری مدل و توکنایزر
|
| 294 |
+
self.tokenizer = M2M100Tokenizer.from_pretrained(MODEL_NAME)
|
| 295 |
+
self.model = M2M100ForConditionalGeneration.from_pretrained(MODEL_NAME)
|
| 296 |
+
|
| 297 |
+
# تشخیص GPU
|
| 298 |
+
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 299 |
+
self.model.to(self.device)
|
| 300 |
+
print(f"مدل روی {self.device} بارگذاری شد")
|
| 301 |
+
|
| 302 |
+
# اجزای کمکی
|
| 303 |
+
self.cache = TranslationCache(cache_expiry_minutes)
|
| 304 |
+
self.chunker = TextChunker()
|
| 305 |
+
self.queue = TranslationQueue()
|
| 306 |
+
|
| 307 |
+
# ذخیره وضعیت ترجمهها
|
| 308 |
+
self.translation_sessions: Dict[str, Dict] = {}
|
| 309 |
+
self.completed_translations: Dict[str, Dict] = {}
|
| 310 |
+
self.translation_requests: Dict[str, Dict] = {}
|
| 311 |
+
|
| 312 |
+
# آمار
|
| 313 |
+
self.total_requests = 0
|
| 314 |
+
self.lock = threading.Lock()
|
| 315 |
+
|
| 316 |
+
def _normalize_language(self, lang: str) -> str:
|
| 317 |
+
"""تبدیل نام زبان به کد دوحرفی"""
|
| 318 |
+
if lang in LANGUAGE_MAP:
|
| 319 |
+
return LANGUAGE_MAP[lang]
|
| 320 |
+
elif lang.lower() in [v.lower() for v in LANGUAGE_MAP.values()]:
|
| 321 |
+
return lang.lower()
|
| 322 |
+
else:
|
| 323 |
+
raise ValueError(f"زبان پشتیبانی نمیشود: {lang}")
|
| 324 |
+
|
| 325 |
+
def translate_chunk(self, text: str, source_lang: str, target_lang: str) -> str:
|
| 326 |
+
"""ترجمه یک بخش از متن"""
|
| 327 |
+
try:
|
| 328 |
+
# تنظیم زبان مبدا
|
| 329 |
+
self.tokenizer.src_lang = source_lang
|
| 330 |
+
|
| 331 |
+
# کدگذاری متن
|
| 332 |
+
encoded = self.tokenizer(text, return_tensors="pt").to(self.device)
|
| 333 |
+
|
| 334 |
+
# تولید ترجمه
|
| 335 |
+
generated_tokens = self.model.generate(
|
| 336 |
+
**encoded,
|
| 337 |
+
forced_bos_token_id=self.tokenizer.get_lang_id(target_lang),
|
| 338 |
+
max_length=512,
|
| 339 |
+
num_beams=5,
|
| 340 |
+
early_stopping=True
|
| 341 |
+
)
|
| 342 |
+
|
| 343 |
+
# رمزگشایی نتیجه
|
| 344 |
+
translation = self.tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)[0]
|
| 345 |
+
|
| 346 |
+
return translation.strip()
|
| 347 |
+
|
| 348 |
+
except Exception as e:
|
| 349 |
+
print(f"خطا در ترجمه بخش: {str(e)}")
|
| 350 |
+
return text
|
| 351 |
+
|
| 352 |
+
def translate_text(self, text: str, source_lang: str, target_lang: str,
|
| 353 |
+
session_id: Optional[str] = None) -> Dict[str, Any]:
|
| 354 |
+
"""ترجمه متن کامل"""
|
| 355 |
+
start_time = time.time()
|
| 356 |
+
|
| 357 |
+
# تبدیل کدهای زبان
|
| 358 |
+
source_lang = self._normalize_language(source_lang)
|
| 359 |
+
target_lang = self._normalize_language(target_lang)
|
| 360 |
+
|
| 361 |
+
# بررسی کش
|
| 362 |
+
cached_result = self.cache.get(text, source_lang, target_lang)
|
| 363 |
+
if cached_result:
|
| 364 |
+
return {
|
| 365 |
+
'translated_text': cached_result,
|
| 366 |
+
'processing_time': 0,
|
| 367 |
+
'chunks_count': 1,
|
| 368 |
+
'from_cache': True
|
| 369 |
+
}
|
| 370 |
+
|
| 371 |
+
# تقسیم متن
|
| 372 |
+
chunks = self.chunker.chunk_text(text)
|
| 373 |
+
translated_chunks = []
|
| 374 |
+
|
| 375 |
+
# ذخیره وضعیت session
|
| 376 |
+
if session_id:
|
| 377 |
+
with self.lock:
|
| 378 |
+
self.translation_sessions[session_id] = {
|
| 379 |
+
'total_chunks': len(chunks),
|
| 380 |
+
'completed_chunks': 0,
|
| 381 |
+
'start_time': start_time,
|
| 382 |
+
'status': 'processing'
|
| 383 |
+
}
|
| 384 |
+
|
| 385 |
+
# ترجمه هر بخش
|
| 386 |
+
for i, chunk in enumerate(chunks):
|
| 387 |
+
translated_chunk = self.translate_chunk(chunk, source_lang, target_lang)
|
| 388 |
+
translated_chunks.append(translated_chunk)
|
| 389 |
+
|
| 390 |
+
# بروزرسانی پیشرفت
|
| 391 |
+
if session_id:
|
| 392 |
+
with self.lock:
|
| 393 |
+
if session_id in self.translation_sessions:
|
| 394 |
+
self.translation_sessions[session_id]['completed_chunks'] = i + 1
|
| 395 |
+
|
| 396 |
+
# ترکیب نتایج
|
| 397 |
+
final_translation = self.combine_translations(translated_chunks)
|
| 398 |
+
processing_time = time.time() - start_time
|
| 399 |
+
|
| 400 |
+
# ذخیره در کش
|
| 401 |
+
self.cache.set(text, source_lang, target_lang, final_translation)
|
| 402 |
+
|
| 403 |
+
# بروزرسانی وضعیت نهایی
|
| 404 |
+
if session_id:
|
| 405 |
+
with self.lock:
|
| 406 |
+
if session_id in self.translation_sessions:
|
| 407 |
+
self.translation_sessions[session_id].update({
|
| 408 |
+
'status': 'completed',
|
| 409 |
+
'end_time': time.time(),
|
| 410 |
+
'result': final_translation
|
| 411 |
+
})
|
| 412 |
+
|
| 413 |
+
# افزایش آمار
|
| 414 |
+
with self.lock:
|
| 415 |
+
self.total_requests += 1
|
| 416 |
+
|
| 417 |
+
result = {
|
| 418 |
+
'translated_text': final_translation,
|
| 419 |
+
'processing_time': processing_time,
|
| 420 |
+
'chunks_count': len(chunks),
|
| 421 |
+
'from_cache': False
|
| 422 |
+
}
|
| 423 |
+
|
| 424 |
+
return result
|
| 425 |
+
|
| 426 |
+
def combine_translations(self, chunks: List[str]) -> str:
|
| 427 |
+
"""ترکیب بخشهای ترجمه شده"""
|
| 428 |
+
return ' '.join(chunks)
|
| 429 |
+
|
| 430 |
+
def get_translation_progress(self, session_id: str) -> Optional[Dict]:
|
| 431 |
+
"""دریافت پیشرفت ترجمه"""
|
| 432 |
+
with self.lock:
|
| 433 |
+
if session_id not in self.translation_sessions:
|
| 434 |
+
return None
|
| 435 |
+
|
| 436 |
+
session = self.translation_sessions[session_id]
|
| 437 |
+
current_time = time.time()
|
| 438 |
+
elapsed_time = current_time - session['start_time']
|
| 439 |
+
|
| 440 |
+
if session['status'] == 'completed':
|
| 441 |
+
return {
|
| 442 |
+
'progress': 100,
|
| 443 |
+
'completed_chunks': session['completed_chunks'],
|
| 444 |
+
'total_chunks': session['total_chunks'],
|
| 445 |
+
'elapsed_time': elapsed_time,
|
| 446 |
+
'status': 'completed',
|
| 447 |
+
'result': session.get('result')
|
| 448 |
+
}
|
| 449 |
+
|
| 450 |
+
progress = (session['completed_chunks'] / session['total_chunks']) * 100
|
| 451 |
+
|
| 452 |
+
# تخمین زمان باقیمانده
|
| 453 |
+
if session['completed_chunks'] > 0:
|
| 454 |
+
avg_time_per_chunk = elapsed_time / session['completed_chunks']
|
| 455 |
+
remaining_chunks = session['total_chunks'] - session['completed_chunks']
|
| 456 |
+
estimated_remaining = avg_time_per_chunk * remaining_chunks
|
| 457 |
+
else:
|
| 458 |
+
estimated_remaining = None
|
| 459 |
+
|
| 460 |
+
return {
|
| 461 |
+
'progress': progress,
|
| 462 |
+
'completed_chunks': session['completed_chunks'],
|
| 463 |
+
'total_chunks': session['total_chunks'],
|
| 464 |
+
'elapsed_time': elapsed_time,
|
| 465 |
+
'estimated_remaining': estimated_remaining,
|
| 466 |
+
'status': 'processing'
|
| 467 |
+
}
|
| 468 |
+
|
| 469 |
+
async def translate_text_async(self, text: str, source_lang: str, target_lang: str) -> Dict[str, Any]:
|
| 470 |
+
"""نسخه آسنکرون ترجمه"""
|
| 471 |
+
loop = asyncio.get_event_loop()
|
| 472 |
+
with ThreadPoolExecutor() as executor:
|
| 473 |
+
result = await loop.run_in_executor(
|
| 474 |
+
executor,
|
| 475 |
+
self.translate_text,
|
| 476 |
+
text, source_lang, target_lang
|
| 477 |
+
)
|
| 478 |
+
return result
|
| 479 |
+
|
| 480 |
+
# توابع مربوط به WordPress
|
| 481 |
+
def notify_wordpress_completion_and_charge(request_id: str, character_count: int,
|
| 482 |
+
translation_length: int, source_lang: str,
|
| 483 |
+
target_lang: str):
|
| 484 |
+
"""اطلاعرسانی تکمیل به WordPress و کسر اعتبار"""
|
| 485 |
+
try:
|
| 486 |
+
wordpress_url = os.getenv("WORDPRESS_NOTIFICATION_URL")
|
| 487 |
+
if not wordpress_url:
|
| 488 |
+
print("URL وردپرس تنظیم نشده است")
|
| 489 |
+
return False
|
| 490 |
+
|
| 491 |
+
payload = {
|
| 492 |
+
'request_id': request_id,
|
| 493 |
+
'character_count': character_count,
|
| 494 |
+
'translation_length': translation_length,
|
| 495 |
+
'source_lang': source_lang,
|
| 496 |
+
'target_lang': target_lang,
|
| 497 |
+
'completed_at': datetime.now().isoformat()
|
| 498 |
+
}
|
| 499 |
+
|
| 500 |
+
response = requests.post(wordpress_url, json=payload, timeout=30)
|
| 501 |
+
|
| 502 |
+
if response.status_code == 200:
|
| 503 |
+
# ثبت فلگ کسر خودکار
|
| 504 |
+
if request_id in translator.translation_requests:
|
| 505 |
+
translator.translation_requests[request_id]['auto_charged'] = True
|
| 506 |
+
print(f"اطلاعرسانی موفق برای درخواست {request_id}")
|
| 507 |
+
return True
|
| 508 |
+
else:
|
| 509 |
+
print(f"خطا در اطلاعرسانی: {response.status_code}")
|
| 510 |
+
return False
|
| 511 |
+
|
| 512 |
+
except Exception as e:
|
| 513 |
+
print(f"خطا در اطلاعرسانی به WordPress: {str(e)}")
|
| 514 |
+
return False
|
| 515 |
+
|
| 516 |
+
def process_heavy_translation_background(text: str, source_lang: str, target_lang: str,
|
| 517 |
+
request_id: str, auto_charge: bool = False):
|
| 518 |
+
"""پردازش ترجمه سنگین در پسزمینه"""
|
| 519 |
+
try:
|
| 520 |
+
# ثبت درخواست
|
| 521 |
+
translator.translation_requests[request_id] = {
|
| 522 |
+
'text': text,
|
| 523 |
+
'source_lang': source_lang,
|
| 524 |
+
'target_lang': target_lang,
|
| 525 |
+
'start_time': time.time(),
|
| 526 |
+
'status': 'processing',
|
| 527 |
+
'auto_charge': auto_charge,
|
| 528 |
+
'auto_charged': False
|
| 529 |
+
}
|
| 530 |
+
|
| 531 |
+
# انجام ترجمه
|
| 532 |
+
result = translator.translate_text(text, source_lang, target_lang, request_id)
|
| 533 |
+
|
| 534 |
+
# ذخیره نتیجه
|
| 535 |
+
translator.completed_translations[request_id] = {
|
| 536 |
+
'result': result,
|
| 537 |
+
'completed_at': time.time(),
|
| 538 |
+
'character_count': len(text),
|
| 539 |
+
'translation_length': len(result['translated_text'])
|
| 540 |
+
}
|
| 541 |
+
|
| 542 |
+
# اطلاعرسانی به WordPress در صورت نیاز
|
| 543 |
+
if auto_charge:
|
| 544 |
+
notify_wordpress_completion_and_charge(
|
| 545 |
+
request_id,
|
| 546 |
+
len(text),
|
| 547 |
+
len(result['translated_text']),
|
| 548 |
+
source_lang,
|
| 549 |
+
target_lang
|
| 550 |
+
)
|
| 551 |
+
|
| 552 |
+
print(f"ترجمه پسزمینه {request_id} تکمیل شد")
|
| 553 |
+
|
| 554 |
+
except Exception as e:
|
| 555 |
+
print(f"خطا در ترجمه پسزمینه {request_id}: {str(e)}")
|
| 556 |
+
if request_id in translator.translation_requests:
|
| 557 |
+
translator.translation_requests[request_id]['status'] = 'failed'
|
| 558 |
+
translator.translation_requests[request_id]['error'] = str(e)
|
| 559 |
+
|
| 560 |
+
def perform_translation_internal(text: str, source_lang: str, target_lang: str) -> Dict[str, Any]:
|
| 561 |
+
"""تابع کمکی برای انجام ترجمه"""
|
| 562 |
+
try:
|
| 563 |
+
result = translator.translate_text(text, source_lang, target_lang)
|
| 564 |
+
return result
|
| 565 |
+
except Exception as e:
|
| 566 |
+
raise HTTPException(status_code=500, detail=f"خطا در ترجمه: {str(e)}")
|
| 567 |
+
|
| 568 |
+
# ایجاد مترجم
|
| 569 |
+
print("در حال راهاندازی مترجم...")
|
| 570 |
+
translator = MultilingualTranslator(60)
|
| 571 |
+
|
| 572 |
+
# تابع پاکسازی
|
| 573 |
+
def cleanup_old_data():
|
| 574 |
+
"""پاکسازی دادههای قدیمی"""
|
| 575 |
+
while True:
|
| 576 |
+
try:
|
| 577 |
+
current_time = time.time()
|
| 578 |
+
|
| 579 |
+
# پاکسازی کش
|
| 580 |
+
translator.cache.clear_expired()
|
| 581 |
+
|
| 582 |
+
# پاکسازی درخواستهای قدیمی (بیش از 2 ساعت)
|
| 583 |
+
expired_requests = []
|
| 584 |
+
for req_id, req_data in translator.translation_requests.items():
|
| 585 |
+
if current_time - req_data['start_time'] > 7200: # 2 ساعت
|
| 586 |
+
expired_requests.append(req_id)
|
| 587 |
+
|
| 588 |
+
for req_id in expired_requests:
|
| 589 |
+
translator.translation_requests.pop(req_id, None)
|
| 590 |
+
translator.completed_translations.pop(req_id, None)
|
| 591 |
+
|
| 592 |
+
# پاکسازی session های قدیمی
|
| 593 |
+
expired_sessions = []
|
| 594 |
+
for session_id, session_data in translator.translation_sessions.items():
|
| 595 |
+
if current_time - session_data['start_time'] > 3600: # 1 ساعت
|
| 596 |
+
expired_sessions.append(session_id)
|
| 597 |
+
|
| 598 |
+
for session_id in expired_sessions:
|
| 599 |
+
translator.translation_sessions.pop(session_id, None)
|
| 600 |
+
|
| 601 |
+
print(f"پاکسازی انجام شد: {len(expired_requests)} درخواست و {len(expired_sessions)} session حذف شد")
|
| 602 |
+
|
| 603 |
+
except Exception as e:
|
| 604 |
+
print(f"خطا در پاکسازی: {str(e)}")
|
| 605 |
+
|
| 606 |
+
time.sleep(CLEANUP_INTERVAL)
|
| 607 |
+
|
| 608 |
+
# راهاندازی FastAPI
|
| 609 |
+
app = FastAPI(
|
| 610 |
+
title="سرویس ترجمه چندزبانه M2M100",
|
| 611 |
+
description="API ترجمه مبتنی بر مدل M2M100 فیسبوک",
|
| 612 |
+
version="1.0.0"
|
| 613 |
+
)
|
| 614 |
+
|
| 615 |
+
# تنظیم CORS
|
| 616 |
+
app.add_middleware(
|
| 617 |
+
CORSMiddleware,
|
| 618 |
+
allow_origins=["*"],
|
| 619 |
+
allow_credentials=True,
|
| 620 |
+
allow_methods=["*"],
|
| 621 |
+
allow_headers=["*"],
|
| 622 |
+
)
|
| 623 |
+
|
| 624 |
+
# شروع نخ پاکسازی
|
| 625 |
+
cleanup_thread = threading.Thread(target=cleanup_old_data, daemon=True)
|
| 626 |
+
cleanup_thread.start()
|
| 627 |
+
|
| 628 |
+
# Endpoints
|
| 629 |
+
|
| 630 |
+
@app.get("/")
|
| 631 |
+
async def root():
|
| 632 |
+
"""صفحه اصلی API"""
|
| 633 |
+
return {
|
| 634 |
+
"message": "سرویس ترجمه چندزبانه M2M100",
|
| 635 |
+
"model": MODEL_NAME,
|
| 636 |
+
"device": str(translator.device),
|
| 637 |
+
"features": [
|
| 638 |
+
"multilingual_translation",
|
| 639 |
+
"text_chunking",
|
| 640 |
+
"translation_cache",
|
| 641 |
+
"background_processing",
|
| 642 |
+
"progress_tracking",
|
| 643 |
+
"wordpress_integration"
|
| 644 |
+
],
|
| 645 |
+
"supported_languages": len(LANGUAGE_MAP),
|
| 646 |
+
"endpoints": {
|
| 647 |
+
"/": "صفحه اصلی",
|
| 648 |
+
"/api/translate": "ترجمه همزمان",
|
| 649 |
+
"/api/translate/form": "ترجمه از فرم",
|
| 650 |
+
"/api/languages": "لیست زبانها",
|
| 651 |
+
"/api/health": "وضعیت سلامت",
|
| 652 |
+
"/api/progress/{session_id}": "پیگیری پیشرفت",
|
| 653 |
+
"/api/status/{session_id}": "وضعیت کلی",
|
| 654 |
+
"/api/server-status": "وضعیت سرور",
|
| 655 |
+
"/api/check-completion": "بررسی تکمیل",
|
| 656 |
+
"/api/check-translation-status": "وضعیت ترجمه",
|
| 657 |
+
"/api/check-auto-charge-status": "وضعیت کسر خودکار"
|
| 658 |
+
}
|
| 659 |
+
}
|
| 660 |
+
|
| 661 |
+
@app.post("/api/translate")
|
| 662 |
+
async def translate_text_api(request: TranslationRequest):
|
| 663 |
+
"""ترجمه همزمان متن"""
|
| 664 |
+
try:
|
| 665 |
+
result = perform_translation_internal(
|
| 666 |
+
request.text,
|
| 667 |
+
request.source_lang,
|
| 668 |
+
request.target_lang
|
| 669 |
+
)
|
| 670 |
+
|
| 671 |
+
return {
|
| 672 |
+
"success": True,
|
| 673 |
+
"translated_text": result['translated_text'],
|
| 674 |
+
"processing_time": result['processing_time'],
|
| 675 |
+
"chunks_count": result['chunks_count'],
|
| 676 |
+
"from_cache": result.get('from_cache', False),
|
| 677 |
+
"character_count": len(request.text),
|
| 678 |
+
"translation_length": len(result['translated_text'])
|
| 679 |
+
}
|
| 680 |
+
|
| 681 |
+
except Exception as e:
|
| 682 |
+
raise HTTPException(status_code=500, detail=str(e))
|
| 683 |
+
|
| 684 |
+
@app.post("/api/translate/form")
|
| 685 |
+
async def translate_form_api(request: TranslationFormRequest):
|
| 686 |
+
"""ترجمه از فرم (با احتمال استفاده از کش)"""
|
| 687 |
+
try:
|
| 688 |
+
result = perform_translation_internal(
|
| 689 |
+
request.text,
|
| 690 |
+
request.source_lang,
|
| 691 |
+
request.target_lang
|
| 692 |
+
)
|
| 693 |
+
|
| 694 |
+
return {
|
| 695 |
+
"success": True,
|
| 696 |
+
"translated_text": result['translated_text'],
|
| 697 |
+
"processing_time": result['processing_time'],
|
| 698 |
+
"chunks_count": result['chunks_count'],
|
| 699 |
+
"from_cache": result.get('from_cache', False)
|
| 700 |
+
}
|
| 701 |
+
|
| 702 |
+
except Exception as e:
|
| 703 |
+
raise HTTPException(status_code=500, detail=str(e))
|
| 704 |
+
|
| 705 |
+
@app.get("/api/languages")
|
| 706 |
+
async def get_supported_languages():
|
| 707 |
+
"""دریافت لیست زبانهای پشتیبانی شده"""
|
| 708 |
+
return {
|
| 709 |
+
"success": True,
|
| 710 |
+
"languages": LANGUAGE_MAP,
|
| 711 |
+
"total_count": len(LANGUAGE_MAP)
|
| 712 |
+
}
|
| 713 |
+
|
| 714 |
+
@app.get("/api/health")
|
| 715 |
+
async def health_check():
|
| 716 |
+
"""بررسی سلامت سرویس"""
|
| 717 |
+
cache_stats = translator.cache.get_stats()
|
| 718 |
+
|
| 719 |
+
return {
|
| 720 |
+
"status": "healthy",
|
| 721 |
+
"model": MODEL_NAME,
|
| 722 |
+
"device": str(translator.device),
|
| 723 |
+
"gpu_available": torch.cuda.is_available(),
|
| 724 |
+
"cache_size": cache_stats['cache_size'],
|
| 725 |
+
"total_requests": translator.total_requests,
|
| 726 |
+
"active_sessions": len(translator.translation_sessions),
|
| 727 |
+
"completed_translations": len(translator.completed_translations),
|
| 728 |
+
"version": "1.0.0",
|
| 729 |
+
"timestamp": datetime.now().isoformat()
|
| 730 |
+
}
|
| 731 |
+
|
| 732 |
+
@app.get("/api/progress/{session_id}")
|
| 733 |
+
async def get_translation_progress(session_id: str):
|
| 734 |
+
"""دریافت پیشرفت ترجمه"""
|
| 735 |
+
progress = translator.get_translation_progress(session_id)
|
| 736 |
+
|
| 737 |
+
if progress is None:
|
| 738 |
+
raise HTTPException(status_code=404, detail="Session پیدا نشد یا تکمیل شده است")
|
| 739 |
+
|
| 740 |
+
return {
|
| 741 |
+
"success": True,
|
| 742 |
+
"session_id": session_id,
|
| 743 |
+
**progress
|
| 744 |
+
}
|
| 745 |
+
|
| 746 |
+
@app.get("/api/status/{session_id}")
|
| 747 |
+
async def get_translation_status(session_id: str):
|
| 748 |
+
"""دریافت وضعیت کلی ترجمه"""
|
| 749 |
+
progress = translator.get_translation_progress(session_id)
|
| 750 |
+
|
| 751 |
+
if progress is None:
|
| 752 |
+
# بررسی در ترجمههای تکمیل شده
|
| 753 |
+
if session_id in translator.completed_translations:
|
| 754 |
+
completed = translator.completed_translations[session_id]
|
| 755 |
+
return {
|
| 756 |
+
"success": True,
|
| 757 |
+
"status": "completed",
|
| 758 |
+
"result": completed['result'],
|
| 759 |
+
"completed_at": completed['completed_at']
|
| 760 |
+
}
|
| 761 |
+
else:
|
| 762 |
+
raise HTTPException(status_code=404, detail="Session پیدا نشد")
|
| 763 |
+
|
| 764 |
+
return {
|
| 765 |
+
"success": True,
|
| 766 |
+
"session_id": session_id,
|
| 767 |
+
**progress
|
| 768 |
+
}
|
| 769 |
+
|
| 770 |
+
@app.get("/api/server-status")
|
| 771 |
+
async def get_server_status():
|
| 772 |
+
"""دریافت وضعیت کلی سرور"""
|
| 773 |
+
active_sessions = len(translator.translation_sessions)
|
| 774 |
+
background_tasks = len(translator.translation_requests)
|
| 775 |
+
completed_count = len(translator.completed_translations)
|
| 776 |
+
|
| 777 |
+
# شمارش وظایف در حال پردازش
|
| 778 |
+
processing_count = sum(1 for req in translator.translation_requests.values()
|
| 779 |
+
if req.get('status') == 'processing')
|
| 780 |
+
|
| 781 |
+
return {
|
| 782 |
+
"success": True,
|
| 783 |
+
"server_status": "running",
|
| 784 |
+
"active_sessions": active_sessions,
|
| 785 |
+
"background_tasks": background_tasks,
|
| 786 |
+
"processing_tasks": processing_count,
|
| 787 |
+
"completed_translations": completed_count,
|
| 788 |
+
"total_requests": translator.total_requests,
|
| 789 |
+
"uptime": time.time(),
|
| 790 |
+
"message": f"سرور فعال - {active_sessions} session فعال، {processing_count} در حال پردازش"
|
| 791 |
+
}
|
| 792 |
+
|
| 793 |
+
@app.post("/api/check-completion")
|
| 794 |
+
async def check_completion(request: CompletionCheckRequest):
|
| 795 |
+
"""بررسی تکمیل ترجمه با request_id"""
|
| 796 |
+
request_id = request.request_id
|
| 797 |
+
|
| 798 |
+
# بررسی در ترجمههای تکمیل شده
|
| 799 |
+
if request_id in translator.completed_translations:
|
| 800 |
+
completed = translator.completed_translations[request_id]
|
| 801 |
+
return {
|
| 802 |
+
"success": True,
|
| 803 |
+
"completed": True,
|
| 804 |
+
"completed_at": completed['completed_at'],
|
| 805 |
+
"processing_time": completed.get('processing_time', 0)
|
| 806 |
+
}
|
| 807 |
+
|
| 808 |
+
# بررسی در درخواستهای در حال پردازش
|
| 809 |
+
if request_id in translator.translation_requests:
|
| 810 |
+
req_data = translator.translation_requests[request_id]
|
| 811 |
+
if req_data.get('status') == 'processing':
|
| 812 |
+
return {
|
| 813 |
+
"success": True,
|
| 814 |
+
"completed": False,
|
| 815 |
+
"status": "در حال پردازش",
|
| 816 |
+
"elapsed_time": time.time() - req_data['start_time']
|
| 817 |
+
}
|
| 818 |
+
elif req_data.get('status') == 'failed':
|
| 819 |
+
return {
|
| 820 |
+
"success": False,
|
| 821 |
+
"completed": False,
|
| 822 |
+
"status": "ناموفق",
|
| 823 |
+
"error": req_data.get('error', 'خطای ناشناخته')
|
| 824 |
+
}
|
| 825 |
+
|
| 826 |
+
# درخواست پیدا نشد
|
| 827 |
+
return {
|
| 828 |
+
"success": False,
|
| 829 |
+
"completed": False,
|
| 830 |
+
"status": "درخواست پیدا نشد"
|
| 831 |
+
}
|
| 832 |
+
|
| 833 |
+
@app.post("/api/check-translation-status")
|
| 834 |
+
async def check_translation_status(request: StatusCheckRequest):
|
| 835 |
+
"""بررسی وضعیت و نتیجه نهایی ترجمه"""
|
| 836 |
+
request_id = request.request_id
|
| 837 |
+
|
| 838 |
+
# بررسی در ترجمههای تکمیل شده
|
| 839 |
+
if request_id in translator.completed_translations:
|
| 840 |
+
completed = translator.completed_translations[request_id]
|
| 841 |
+
req_data = translator.translation_requests.get(request_id, {})
|
| 842 |
+
|
| 843 |
+
return {
|
| 844 |
+
"success": True,
|
| 845 |
+
"status": "completed",
|
| 846 |
+
"translated_text": completed['result']['translated_text'],
|
| 847 |
+
"processing_time": completed['result']['processing_time'],
|
| 848 |
+
"chunks_count": completed['result']['chunks_count'],
|
| 849 |
+
"character_count": completed['character_count'],
|
| 850 |
+
"translation_length": completed['translation_length'],
|
| 851 |
+
"completed_at": completed['completed_at'],
|
| 852 |
+
"source_lang": req_data.get('source_lang'),
|
| 853 |
+
"target_lang": req_data.get('target_lang')
|
| 854 |
+
}
|
| 855 |
+
|
| 856 |
+
# بررسی در درخواستهای در حال پردازش
|
| 857 |
+
if request_id in translator.translation_requests:
|
| 858 |
+
req_data = translator.translation_requests[request_id]
|
| 859 |
+
elapsed_time = time.time() - req_data['start_time']
|
| 860 |
+
|
| 861 |
+
if req_data.get('status') == 'processing':
|
| 862 |
+
# تخمین پیشرفت بر اساس طول متن
|
| 863 |
+
text_length = len(req_data.get('text', ''))
|
| 864 |
+
chunks_estimate = max(1, text_length // MAX_CHUNK_SIZE)
|
| 865 |
+
|
| 866 |
+
# تخمین پیشرفت (این تخمینی است)
|
| 867 |
+
progress_estimate = min(90, (elapsed_time / 10) * 100) # حداکثر 90% تا زمان تکمیل
|
| 868 |
+
|
| 869 |
+
return {
|
| 870 |
+
"success": True,
|
| 871 |
+
"status": "processing",
|
| 872 |
+
"progress": progress_estimate,
|
| 873 |
+
"elapsed_time": elapsed_time,
|
| 874 |
+
"estimated_chunks": chunks_estimate,
|
| 875 |
+
"message": "در حال پردازش ترجمه..."
|
| 876 |
+
}
|
| 877 |
+
elif req_data.get('status') == 'failed':
|
| 878 |
+
return {
|
| 879 |
+
"success": False,
|
| 880 |
+
"status": "failed",
|
| 881 |
+
"error": req_data.get('error', 'خطای ناشناخته'),
|
| 882 |
+
"elapsed_time": elapsed_time
|
| 883 |
+
}
|
| 884 |
+
|
| 885 |
+
# درخواست پیدا نشد
|
| 886 |
+
return {
|
| 887 |
+
"success": False,
|
| 888 |
+
"status": "not_found",
|
| 889 |
+
"message": "درخواست ترجمه پیدا نشد"
|
| 890 |
+
}
|
| 891 |
+
|
| 892 |
+
@app.post("/api/check-auto-charge-status")
|
| 893 |
+
async def check_auto_charge_status(request: AutoChargeStatusRequest):
|
| 894 |
+
"""بررسی وضعیت کسر اعتبار خودکار"""
|
| 895 |
+
request_id = request.request_id
|
| 896 |
+
|
| 897 |
+
if request_id not in translator.translation_requests:
|
| 898 |
+
return {
|
| 899 |
+
"success": False,
|
| 900 |
+
"message": "درخواست پیدا نشد"
|
| 901 |
+
}
|
| 902 |
+
|
| 903 |
+
req_data = translator.translation_requests[request_id]
|
| 904 |
+
|
| 905 |
+
return {
|
| 906 |
+
"success": True,
|
| 907 |
+
"request_id": request_id,
|
| 908 |
+
"auto_charge_enabled": req_data.get('auto_charge', False),
|
| 909 |
+
"auto_charged": req_data.get('auto_charged', False),
|
| 910 |
+
"status": req_data.get('status', 'unknown')
|
| 911 |
+
}
|
| 912 |
+
|
| 913 |
+
@app.post("/api/translate/heavy")
|
| 914 |
+
async def translate_heavy_text(request: TranslationRequest, background_tasks: BackgroundTasks):
|
| 915 |
+
"""ترجمه متنهای سنگین در پسزمینه"""
|
| 916 |
+
import uuid
|
| 917 |
+
|
| 918 |
+
# تولید شناسه منحصر به فرد
|
| 919 |
+
request_id = str(uuid.uuid4())
|
| 920 |
+
|
| 921 |
+
# اضافه کردن به وظایف پسزمینه
|
| 922 |
+
background_tasks.add_task(
|
| 923 |
+
process_heavy_translation_background,
|
| 924 |
+
request.text,
|
| 925 |
+
request.source_lang,
|
| 926 |
+
request.target_lang,
|
| 927 |
+
request_id,
|
| 928 |
+
request.auto_charge
|
| 929 |
+
)
|
| 930 |
+
|
| 931 |
+
return {
|
| 932 |
+
"success": True,
|
| 933 |
+
"message": "ترجمه در پسزمینه آغاز شد",
|
| 934 |
+
"request_id": request_id,
|
| 935 |
+
"estimated_time": len(request.text) // 100, # تخمین بر اساس طول متن
|
| 936 |
+
"auto_charge": request.auto_charge
|
| 937 |
+
}
|
| 938 |
+
|
| 939 |
+
@app.post("/api/translate/session")
|
| 940 |
+
async def translate_with_session(request: TranslationRequest):
|
| 941 |
+
"""ترجمه با session برای پیگیری پیشرفت"""
|
| 942 |
+
import uuid
|
| 943 |
+
|
| 944 |
+
session_id = str(uuid.uuid4())
|
| 945 |
+
|
| 946 |
+
try:
|
| 947 |
+
# شروع ترجمه با session_id
|
| 948 |
+
result = translator.translate_text(
|
| 949 |
+
request.text,
|
| 950 |
+
request.source_lang,
|
| 951 |
+
request.target_lang,
|
| 952 |
+
session_id
|
| 953 |
+
)
|
| 954 |
+
|
| 955 |
+
return {
|
| 956 |
+
"success": True,
|
| 957 |
+
"session_id": session_id,
|
| 958 |
+
"translated_text": result['translated_text'],
|
| 959 |
+
"processing_time": result['processing_time'],
|
| 960 |
+
"chunks_count": result['chunks_count'],
|
| 961 |
+
"from_cache": result.get('from_cache', False)
|
| 962 |
+
}
|
| 963 |
+
|
| 964 |
+
except Exception as e:
|
| 965 |
+
raise HTTPException(status_code=500, detail=str(e))
|
| 966 |
+
|
| 967 |
+
@app.get("/api/cache/stats")
|
| 968 |
+
async def get_cache_stats():
|
| 969 |
+
"""آمار کش ترجمه"""
|
| 970 |
+
stats = translator.cache.get_stats()
|
| 971 |
+
|
| 972 |
+
return {
|
| 973 |
+
"success": True,
|
| 974 |
+
"cache_stats": stats,
|
| 975 |
+
"expiry_minutes": 60
|
| 976 |
+
}
|
| 977 |
+
|
| 978 |
+
@app.post("/api/cache/clear")
|
| 979 |
+
async def clear_cache():
|
| 980 |
+
"""پاک کردن کش (فقط برای ��دیران)"""
|
| 981 |
+
try:
|
| 982 |
+
with translator.cache.lock:
|
| 983 |
+
translator.cache.cache.clear()
|
| 984 |
+
|
| 985 |
+
return {
|
| 986 |
+
"success": True,
|
| 987 |
+
"message": "کش پاک شد"
|
| 988 |
+
}
|
| 989 |
+
except Exception as e:
|
| 990 |
+
raise HTTPException(status_code=500, detail=f"خطا در پاک کردن کش: {str(e)}")
|
| 991 |
+
|
| 992 |
+
@app.get("/api/stats")
|
| 993 |
+
async def get_api_stats():
|
| 994 |
+
"""آمار کلی API"""
|
| 995 |
+
return {
|
| 996 |
+
"success": True,
|
| 997 |
+
"total_requests": translator.total_requests,
|
| 998 |
+
"active_sessions": len(translator.translation_sessions),
|
| 999 |
+
"background_tasks": len(translator.translation_requests),
|
| 1000 |
+
"completed_translations": len(translator.completed_translations),
|
| 1001 |
+
"cache_size": translator.cache.get_stats()['cache_size'],
|
| 1002 |
+
"supported_languages": len(LANGUAGE_MAP),
|
| 1003 |
+
"model_info": {
|
| 1004 |
+
"name": MODEL_NAME,
|
| 1005 |
+
"device": str(translator.device),
|
| 1006 |
+
"gpu_available": torch.cuda.is_available()
|
| 1007 |
+
}
|
| 1008 |
+
}
|
| 1009 |
+
|
| 1010 |
+
@app.post("/api/webhook/wordpress")
|
| 1011 |
+
async def wordpress_webhook(data: dict):
|
| 1012 |
+
"""Webhook برای دریافت اطلاعات از WordPress"""
|
| 1013 |
+
try:
|
| 1014 |
+
# پردازش دادههای دریافتی از WordPress
|
| 1015 |
+
request_id = data.get('request_id')
|
| 1016 |
+
action = data.get('action')
|
| 1017 |
+
|
| 1018 |
+
if action == 'translation_request':
|
| 1019 |
+
# درخواست ترجمه از WordPress
|
| 1020 |
+
text = data.get('text')
|
| 1021 |
+
source_lang = data.get('source_lang')
|
| 1022 |
+
target_lang = data.get('target_lang')
|
| 1023 |
+
auto_charge = data.get('auto_charge', False)
|
| 1024 |
+
|
| 1025 |
+
if not all([text, source_lang, target_lang, request_id]):
|
| 1026 |
+
raise HTTPException(status_code=400, detail="دادههای ناقص")
|
| 1027 |
+
|
| 1028 |
+
# شروع ترجمه در پسزمینه
|
| 1029 |
+
background_tasks = BackgroundTasks()
|
| 1030 |
+
background_tasks.add_task(
|
| 1031 |
+
process_heavy_translation_background,
|
| 1032 |
+
text, source_lang, target_lang, request_id, auto_charge
|
| 1033 |
+
)
|
| 1034 |
+
|
| 1035 |
+
return {
|
| 1036 |
+
"success": True,
|
| 1037 |
+
"message": "ترجمه آغاز شد",
|
| 1038 |
+
"request_id": request_id
|
| 1039 |
+
}
|
| 1040 |
+
|
| 1041 |
+
elif action == 'status_check':
|
| 1042 |
+
# بررسی وضعیت
|
| 1043 |
+
if request_id in translator.completed_translations:
|
| 1044 |
+
completed = translator.completed_translations[request_id]
|
| 1045 |
+
return {
|
| 1046 |
+
"success": True,
|
| 1047 |
+
"status": "completed",
|
| 1048 |
+
"result": completed['result']
|
| 1049 |
+
}
|
| 1050 |
+
elif request_id in translator.translation_requests:
|
| 1051 |
+
return {
|
| 1052 |
+
"success": True,
|
| 1053 |
+
"status": "processing"
|
| 1054 |
+
}
|
| 1055 |
+
else:
|
| 1056 |
+
return {
|
| 1057 |
+
"success": False,
|
| 1058 |
+
"status": "not_found"
|
| 1059 |
+
}
|
| 1060 |
+
|
| 1061 |
+
else:
|
| 1062 |
+
raise HTTPException(status_code=400, detail="عمل نامعتبر")
|
| 1063 |
+
|
| 1064 |
+
except Exception as e:
|
| 1065 |
+
raise HTTPException(status_code=500, detail=f"خطا در webhook: {str(e)}")
|
| 1066 |
+
|
| 1067 |
+
# تابع راهاندازی (برای Hugging Face Spaces)
|
| 1068 |
+
if __name__ == "__main__":
|
| 1069 |
+
import uvicorn
|
| 1070 |
+
|
| 1071 |
+
port = int(os.getenv("PORT", 7860)) # پورت پیشفرض Hugging Face Spaces
|
| 1072 |
+
|
| 1073 |
+
print(f"راهاندازی سرور روی پورت {port}")
|
| 1074 |
+
print(f"مدل: {MODEL_NAME}")
|
| 1075 |
+
print(f"دستگاه: {translator.device}")
|
| 1076 |
+
print(f"زبانهای پشتیبانی شده: {len(LANGUAGE_MAP)}")
|
| 1077 |
+
|
| 1078 |
+
uvicorn.run(
|
| 1079 |
+
app,
|
| 1080 |
+
host="0.0.0.0",
|
| 1081 |
+
port=port,
|
| 1082 |
+
log_level="info"
|
| 1083 |
+
)
|