Spaces:
Running
Running
13
Browse files
app.py
CHANGED
|
@@ -1,25 +1,24 @@
|
|
| 1 |
# app.py
|
| 2 |
-
|
|
|
|
| 3 |
from datetime import datetime, timezone
|
| 4 |
from functools import wraps
|
| 5 |
-
from typing import Optional, Tuple, List
|
| 6 |
from flask import Flask, render_template, request, redirect, url_for, session, jsonify, flash
|
| 7 |
|
|
|
|
| 8 |
logging.basicConfig(level=logging.INFO, format="%(asctime)s | %(levelname)s | %(message)s")
|
| 9 |
log = logging.getLogger("papua-app")
|
| 10 |
|
| 11 |
-
#
|
| 12 |
app = Flask(__name__, template_folder="templates", static_folder="static")
|
| 13 |
-
|
| 14 |
-
# cookie aman di HF (HTTPS). Untuk lokal bisa set False via ENV.
|
| 15 |
-
SESSION_SECURE = os.getenv("SESSION_COOKIE_SECURE", "true").lower() in ("1","true","yes")
|
| 16 |
app.config.update(
|
| 17 |
SECRET_KEY=os.getenv("SECRET_KEY", "dev-secret-change-me"),
|
| 18 |
SESSION_COOKIE_SAMESITE="Lax",
|
| 19 |
SESSION_COOKIE_SECURE=SESSION_SECURE,
|
| 20 |
)
|
| 21 |
|
| 22 |
-
#
|
| 23 |
from sqlalchemy import create_engine, Column, Integer, Text, DateTime, ForeignKey, func
|
| 24 |
from sqlalchemy.orm import declarative_base, sessionmaker, scoped_session, relationship
|
| 25 |
|
|
@@ -28,7 +27,7 @@ if not DATABASE_URL:
|
|
| 28 |
DATABASE_URL = "sqlite:////tmp/app.db"
|
| 29 |
log.warning("[DB] DATABASE_URL tidak diset; pakai SQLite /tmp/app.db")
|
| 30 |
else:
|
| 31 |
-
# normalisasi skema lama
|
| 32 |
if DATABASE_URL.startswith("postgres://"):
|
| 33 |
DATABASE_URL = DATABASE_URL.replace("postgres://", "postgresql+psycopg2://", 1)
|
| 34 |
elif DATABASE_URL.startswith("postgresql://"):
|
|
@@ -60,11 +59,11 @@ class Translation(Base):
|
|
| 60 |
|
| 61 |
try:
|
| 62 |
Base.metadata.create_all(engine)
|
| 63 |
-
log.info("[DB] Ready
|
| 64 |
except Exception as e:
|
| 65 |
log.exception("[DB] init error: %s", e)
|
| 66 |
|
| 67 |
-
#
|
| 68 |
from werkzeug.security import generate_password_hash, check_password_hash
|
| 69 |
def set_password(user: User, raw: str): user.pass_hash = generate_password_hash(raw)
|
| 70 |
def verify_password(user: User, raw: str) -> bool:
|
|
@@ -79,25 +78,28 @@ def login_required(fn):
|
|
| 79 |
return fn(*args, **kwargs)
|
| 80 |
return _wrap
|
| 81 |
|
| 82 |
-
#
|
| 83 |
PAPUA_MAP = {
|
| 84 |
r"\bsa\b": "saya", r"\bko\b": "kamu", r"\btra\b": "tidak", r"\bndak\b": "tidak",
|
| 85 |
r"\bmo\b": "mau", r"\bpu\b": "punya", r"\bsu\b": "sudah", r"\bkong\b": "kemudian",
|
| 86 |
}
|
| 87 |
def prenorm(text: str) -> str:
|
| 88 |
t = re.sub(r"\s+", " ", text.strip())
|
| 89 |
-
t = t.replace("β¦","...").replace("β","-").replace("β","-")
|
| 90 |
for pat, repl in PAPUA_MAP.items(): t = re.sub(pat, repl, t, flags=re.IGNORECASE)
|
| 91 |
return t
|
| 92 |
|
| 93 |
-
#
|
| 94 |
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
| 95 |
from peft import PeftModel
|
| 96 |
|
| 97 |
BASE_MODEL_ID = os.getenv("BASE_MODEL_ID", "amosnbn/cendol-mt5-base-inst")
|
| 98 |
ADAPTER_ID = os.getenv("ADAPTER_ID", "amosnbn/papua-lora-ckpt-168")
|
| 99 |
-
DEVICE = "cuda" if os.getenv("DEVICE","cpu") == "cuda" else "cpu"
|
| 100 |
-
|
|
|
|
|
|
|
|
|
|
| 101 |
|
| 102 |
def _load_model():
|
| 103 |
global TOK, MODEL
|
|
@@ -112,19 +114,24 @@ def get_model():
|
|
| 112 |
global MODEL
|
| 113 |
if MODEL is None:
|
| 114 |
with _MODEL_LOCK:
|
| 115 |
-
if MODEL is None:
|
|
|
|
| 116 |
return TOK, MODEL
|
| 117 |
|
| 118 |
def translate_with_model(text: str, max_new_tokens: int = 48) -> str:
|
| 119 |
tok, m = get_model()
|
| 120 |
inputs = tok([text], return_tensors="pt").to(DEVICE)
|
| 121 |
outputs = m.generate(
|
| 122 |
-
**inputs,
|
| 123 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 124 |
)
|
| 125 |
return tok.decode(outputs[0], skip_special_tokens=True)
|
| 126 |
|
| 127 |
-
#
|
| 128 |
@app.before_request
|
| 129 |
def _log_req():
|
| 130 |
if request.path not in ("/health", "/ping", "/favicon.ico"):
|
|
@@ -132,48 +139,58 @@ def _log_req():
|
|
| 132 |
|
| 133 |
@app.errorhandler(Exception)
|
| 134 |
def _err(e):
|
| 135 |
-
# supaya 500 kelihatan jelas di Logs HF
|
| 136 |
log.exception("Unhandled error")
|
| 137 |
return "Internal Server Error", 500
|
| 138 |
|
| 139 |
-
#
|
| 140 |
-
@app.get("/health")
|
| 141 |
-
|
|
|
|
|
|
|
| 142 |
|
| 143 |
@app.get("/login")
|
| 144 |
-
def login_get():
|
|
|
|
| 145 |
|
| 146 |
@app.post("/login")
|
| 147 |
def login_post():
|
| 148 |
email = (request.form.get("email") or "").strip().lower()
|
| 149 |
pwd = request.form.get("password") or ""
|
| 150 |
if not email or not pwd:
|
| 151 |
-
flash("Isi email dan password", "error")
|
|
|
|
| 152 |
with SessionLocal() as s:
|
| 153 |
u = s.query(User).filter_by(email=email).first()
|
| 154 |
if not u or not verify_password(u, pwd):
|
| 155 |
-
flash("Email atau password salah", "error")
|
|
|
|
| 156 |
session["uid"], session["email"] = u.id, u.email
|
| 157 |
-
|
| 158 |
|
| 159 |
@app.get("/register")
|
| 160 |
-
def register_get():
|
|
|
|
| 161 |
|
| 162 |
@app.post("/register")
|
| 163 |
def register_post():
|
| 164 |
email = (request.form.get("email") or "").strip().lower()
|
| 165 |
pwd = request.form.get("password") or ""
|
| 166 |
if not email or not pwd:
|
| 167 |
-
flash("Isi email dan password", "error")
|
|
|
|
| 168 |
with SessionLocal() as s:
|
| 169 |
if s.query(User).filter_by(email=email).first():
|
| 170 |
-
flash("Email sudah terdaftar", "error")
|
| 171 |
-
|
|
|
|
|
|
|
| 172 |
session["uid"], session["email"] = u.id, u.email
|
| 173 |
return redirect(url_for("index"))
|
| 174 |
|
| 175 |
@app.get("/logout")
|
| 176 |
-
def logout():
|
|
|
|
|
|
|
| 177 |
|
| 178 |
@app.get("/")
|
| 179 |
@login_required
|
|
@@ -181,33 +198,36 @@ def index():
|
|
| 181 |
with SessionLocal() as s:
|
| 182 |
uid = session["uid"]
|
| 183 |
items = (s.query(Translation)
|
| 184 |
-
|
| 185 |
-
|
|
|
|
| 186 |
data = [{"src": it.src, "mt": it.mt, "created_at": it.created_at} for it in items]
|
| 187 |
-
#
|
| 188 |
-
return render_template("index.html",
|
| 189 |
-
user=session.get("email"),
|
| 190 |
-
data=data,
|
| 191 |
-
device=DEVICE)
|
| 192 |
|
| 193 |
@app.get("/about")
|
| 194 |
-
def about_page():
|
|
|
|
| 195 |
|
| 196 |
@app.post("/translate")
|
| 197 |
def api_translate():
|
| 198 |
-
if not session.get("uid"):
|
|
|
|
| 199 |
payload = request.get_json(silent=True) or {}
|
| 200 |
text = (payload.get("text") or "").strip()
|
| 201 |
max_new = int(payload.get("max_new_tokens", 48))
|
| 202 |
-
if not text:
|
|
|
|
| 203 |
try:
|
| 204 |
-
clean = prenorm(text)
|
|
|
|
| 205 |
with SessionLocal() as s:
|
| 206 |
-
s.add(Translation(user_id=session["uid"], src=text, mt=mt))
|
|
|
|
| 207 |
return jsonify({"mt": mt})
|
| 208 |
-
except Exception
|
| 209 |
log.exception("translate error")
|
| 210 |
return jsonify({"error": "server error"}), 500
|
| 211 |
|
| 212 |
if __name__ == "__main__":
|
| 213 |
-
app.run(host="0.0.0.0", port=int(os.getenv("PORT","7860")), debug=False)
|
|
|
|
| 1 |
# app.py
|
| 2 |
+
# PapuaTranslate β Flask + SQLAlchemy + Supabase + mT5-LoRA (lazy load)
|
| 3 |
+
import os, re, logging, threading
|
| 4 |
from datetime import datetime, timezone
|
| 5 |
from functools import wraps
|
|
|
|
| 6 |
from flask import Flask, render_template, request, redirect, url_for, session, jsonify, flash
|
| 7 |
|
| 8 |
+
# ===== Logging =====
|
| 9 |
logging.basicConfig(level=logging.INFO, format="%(asctime)s | %(levelname)s | %(message)s")
|
| 10 |
log = logging.getLogger("papua-app")
|
| 11 |
|
| 12 |
+
# ===== Flask (templates di folder 'templates') =====
|
| 13 |
app = Flask(__name__, template_folder="templates", static_folder="static")
|
| 14 |
+
SESSION_SECURE = os.getenv("SESSION_COOKIE_SECURE", "true").lower() in ("1", "true", "yes")
|
|
|
|
|
|
|
| 15 |
app.config.update(
|
| 16 |
SECRET_KEY=os.getenv("SECRET_KEY", "dev-secret-change-me"),
|
| 17 |
SESSION_COOKIE_SAMESITE="Lax",
|
| 18 |
SESSION_COOKIE_SECURE=SESSION_SECURE,
|
| 19 |
)
|
| 20 |
|
| 21 |
+
# ===== DB: SQLAlchemy (Supabase Postgres / SQLite fallback) =====
|
| 22 |
from sqlalchemy import create_engine, Column, Integer, Text, DateTime, ForeignKey, func
|
| 23 |
from sqlalchemy.orm import declarative_base, sessionmaker, scoped_session, relationship
|
| 24 |
|
|
|
|
| 27 |
DATABASE_URL = "sqlite:////tmp/app.db"
|
| 28 |
log.warning("[DB] DATABASE_URL tidak diset; pakai SQLite /tmp/app.db")
|
| 29 |
else:
|
| 30 |
+
# normalisasi skema lama β psycopg2 driver
|
| 31 |
if DATABASE_URL.startswith("postgres://"):
|
| 32 |
DATABASE_URL = DATABASE_URL.replace("postgres://", "postgresql+psycopg2://", 1)
|
| 33 |
elif DATABASE_URL.startswith("postgresql://"):
|
|
|
|
| 59 |
|
| 60 |
try:
|
| 61 |
Base.metadata.create_all(engine)
|
| 62 |
+
log.info("[DB] Ready")
|
| 63 |
except Exception as e:
|
| 64 |
log.exception("[DB] init error: %s", e)
|
| 65 |
|
| 66 |
+
# ===== Auth helpers =====
|
| 67 |
from werkzeug.security import generate_password_hash, check_password_hash
|
| 68 |
def set_password(user: User, raw: str): user.pass_hash = generate_password_hash(raw)
|
| 69 |
def verify_password(user: User, raw: str) -> bool:
|
|
|
|
| 78 |
return fn(*args, **kwargs)
|
| 79 |
return _wrap
|
| 80 |
|
| 81 |
+
# ===== Prenorm (heuristik ringan) =====
|
| 82 |
PAPUA_MAP = {
|
| 83 |
r"\bsa\b": "saya", r"\bko\b": "kamu", r"\btra\b": "tidak", r"\bndak\b": "tidak",
|
| 84 |
r"\bmo\b": "mau", r"\bpu\b": "punya", r"\bsu\b": "sudah", r"\bkong\b": "kemudian",
|
| 85 |
}
|
| 86 |
def prenorm(text: str) -> str:
|
| 87 |
t = re.sub(r"\s+", " ", text.strip())
|
| 88 |
+
t = t.replace("β¦", "...").replace("β", "-").replace("β", "-")
|
| 89 |
for pat, repl in PAPUA_MAP.items(): t = re.sub(pat, repl, t, flags=re.IGNORECASE)
|
| 90 |
return t
|
| 91 |
|
| 92 |
+
# ===== Model (lazy-load LoRA) =====
|
| 93 |
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
| 94 |
from peft import PeftModel
|
| 95 |
|
| 96 |
BASE_MODEL_ID = os.getenv("BASE_MODEL_ID", "amosnbn/cendol-mt5-base-inst")
|
| 97 |
ADAPTER_ID = os.getenv("ADAPTER_ID", "amosnbn/papua-lora-ckpt-168")
|
| 98 |
+
DEVICE = "cuda" if os.getenv("DEVICE", "cpu") == "cuda" else "cpu"
|
| 99 |
+
|
| 100 |
+
TOK = None
|
| 101 |
+
MODEL = None
|
| 102 |
+
_MODEL_LOCK = threading.Lock()
|
| 103 |
|
| 104 |
def _load_model():
|
| 105 |
global TOK, MODEL
|
|
|
|
| 114 |
global MODEL
|
| 115 |
if MODEL is None:
|
| 116 |
with _MODEL_LOCK:
|
| 117 |
+
if MODEL is None:
|
| 118 |
+
_load_model()
|
| 119 |
return TOK, MODEL
|
| 120 |
|
| 121 |
def translate_with_model(text: str, max_new_tokens: int = 48) -> str:
|
| 122 |
tok, m = get_model()
|
| 123 |
inputs = tok([text], return_tensors="pt").to(DEVICE)
|
| 124 |
outputs = m.generate(
|
| 125 |
+
**inputs,
|
| 126 |
+
max_new_tokens=max_new_tokens,
|
| 127 |
+
num_beams=4,
|
| 128 |
+
length_penalty=0.9,
|
| 129 |
+
no_repeat_ngram_size=3,
|
| 130 |
+
early_stopping=True,
|
| 131 |
)
|
| 132 |
return tok.decode(outputs[0], skip_special_tokens=True)
|
| 133 |
|
| 134 |
+
# ===== Little utils / logging =====
|
| 135 |
@app.before_request
|
| 136 |
def _log_req():
|
| 137 |
if request.path not in ("/health", "/ping", "/favicon.ico"):
|
|
|
|
| 139 |
|
| 140 |
@app.errorhandler(Exception)
|
| 141 |
def _err(e):
|
|
|
|
| 142 |
log.exception("Unhandled error")
|
| 143 |
return "Internal Server Error", 500
|
| 144 |
|
| 145 |
+
# ===== Routes =====
|
| 146 |
+
@app.get("/health")
|
| 147 |
+
@app.get("/ping")
|
| 148 |
+
def health():
|
| 149 |
+
return jsonify({"ok": True, "time": datetime.now(timezone.utc).isoformat()})
|
| 150 |
|
| 151 |
@app.get("/login")
|
| 152 |
+
def login_get():
|
| 153 |
+
return render_template("login.html")
|
| 154 |
|
| 155 |
@app.post("/login")
|
| 156 |
def login_post():
|
| 157 |
email = (request.form.get("email") or "").strip().lower()
|
| 158 |
pwd = request.form.get("password") or ""
|
| 159 |
if not email or not pwd:
|
| 160 |
+
flash("Isi email dan password", "error")
|
| 161 |
+
return redirect(url_for("login_get"))
|
| 162 |
with SessionLocal() as s:
|
| 163 |
u = s.query(User).filter_by(email=email).first()
|
| 164 |
if not u or not verify_password(u, pwd):
|
| 165 |
+
flash("Email atau password salah", "error")
|
| 166 |
+
return redirect(url_for("login_get"))
|
| 167 |
session["uid"], session["email"] = u.id, u.email
|
| 168 |
+
return redirect(url_for("index"))
|
| 169 |
|
| 170 |
@app.get("/register")
|
| 171 |
+
def register_get():
|
| 172 |
+
return render_template("register.html")
|
| 173 |
|
| 174 |
@app.post("/register")
|
| 175 |
def register_post():
|
| 176 |
email = (request.form.get("email") or "").strip().lower()
|
| 177 |
pwd = request.form.get("password") or ""
|
| 178 |
if not email or not pwd:
|
| 179 |
+
flash("Isi email dan password", "error")
|
| 180 |
+
return redirect(url_for("register_get"))
|
| 181 |
with SessionLocal() as s:
|
| 182 |
if s.query(User).filter_by(email=email).first():
|
| 183 |
+
flash("Email sudah terdaftar", "error")
|
| 184 |
+
return redirect(url_for("register_get"))
|
| 185 |
+
u = User(email=email); set_password(u, pwd)
|
| 186 |
+
s.add(u); s.commit()
|
| 187 |
session["uid"], session["email"] = u.id, u.email
|
| 188 |
return redirect(url_for("index"))
|
| 189 |
|
| 190 |
@app.get("/logout")
|
| 191 |
+
def logout():
|
| 192 |
+
session.clear()
|
| 193 |
+
return redirect(url_for("login_get"))
|
| 194 |
|
| 195 |
@app.get("/")
|
| 196 |
@login_required
|
|
|
|
| 198 |
with SessionLocal() as s:
|
| 199 |
uid = session["uid"]
|
| 200 |
items = (s.query(Translation)
|
| 201 |
+
.filter(Translation.user_id == uid)
|
| 202 |
+
.order_by(Translation.id.desc())
|
| 203 |
+
.limit(10).all())
|
| 204 |
data = [{"src": it.src, "mt": it.mt, "created_at": it.created_at} for it in items]
|
| 205 |
+
# index.html kamu memakai 'user' & 'data'
|
| 206 |
+
return render_template("index.html", user=session.get("email"), data=data, device=DEVICE)
|
|
|
|
|
|
|
|
|
|
| 207 |
|
| 208 |
@app.get("/about")
|
| 209 |
+
def about_page():
|
| 210 |
+
return render_template("about.html")
|
| 211 |
|
| 212 |
@app.post("/translate")
|
| 213 |
def api_translate():
|
| 214 |
+
if not session.get("uid"):
|
| 215 |
+
return jsonify({"error": "Unauthorized"}), 401
|
| 216 |
payload = request.get_json(silent=True) or {}
|
| 217 |
text = (payload.get("text") or "").strip()
|
| 218 |
max_new = int(payload.get("max_new_tokens", 48))
|
| 219 |
+
if not text:
|
| 220 |
+
return jsonify({"error": "Empty text"}), 400
|
| 221 |
try:
|
| 222 |
+
clean = prenorm(text)
|
| 223 |
+
mt = translate_with_model(clean, max_new_tokens=max_new)
|
| 224 |
with SessionLocal() as s:
|
| 225 |
+
s.add(Translation(user_id=session["uid"], src=text, mt=mt))
|
| 226 |
+
s.commit()
|
| 227 |
return jsonify({"mt": mt})
|
| 228 |
+
except Exception:
|
| 229 |
log.exception("translate error")
|
| 230 |
return jsonify({"error": "server error"}), 500
|
| 231 |
|
| 232 |
if __name__ == "__main__":
|
| 233 |
+
app.run(host="0.0.0.0", port=int(os.getenv("PORT", "7860")), debug=False)
|