Spaces:
Running
Running
15
Browse files
app.py
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
# app.py
|
| 2 |
-
# PapuaTranslate β Flask + SQLAlchemy
|
| 3 |
import os, re, logging, threading
|
| 4 |
from datetime import datetime, timezone
|
| 5 |
from functools import wraps
|
|
@@ -9,9 +9,10 @@ from flask import Flask, render_template, request, redirect, url_for, session, j
|
|
| 9 |
logging.basicConfig(level=logging.INFO, format="%(asctime)s | %(levelname)s | %(message)s")
|
| 10 |
log = logging.getLogger("papua-app")
|
| 11 |
|
| 12 |
-
# ===== Flask
|
| 13 |
-
|
| 14 |
-
|
|
|
|
| 15 |
app.config.update(
|
| 16 |
SECRET_KEY=os.getenv("SECRET_KEY", "dev-secret-change-me"),
|
| 17 |
SESSION_COOKIE_SAMESITE="Lax",
|
|
@@ -27,7 +28,7 @@ if not DATABASE_URL:
|
|
| 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
|
| 31 |
if DATABASE_URL.startswith("postgres://"):
|
| 32 |
DATABASE_URL = DATABASE_URL.replace("postgres://", "postgresql+psycopg2://", 1)
|
| 33 |
elif DATABASE_URL.startswith("postgresql://"):
|
|
@@ -85,7 +86,7 @@ PAPUA_MAP = {
|
|
| 85 |
}
|
| 86 |
def prenorm(text: str) -> str:
|
| 87 |
t = re.sub(r"\s+", " ", text.strip())
|
| 88 |
-
t = t.replace("β¦",
|
| 89 |
for pat, repl in PAPUA_MAP.items(): t = re.sub(pat, repl, t, flags=re.IGNORECASE)
|
| 90 |
return t
|
| 91 |
|
|
@@ -93,8 +94,9 @@ def prenorm(text: str) -> str:
|
|
| 93 |
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
| 94 |
from peft import PeftModel
|
| 95 |
|
| 96 |
-
|
| 97 |
-
|
|
|
|
| 98 |
DEVICE = "cuda" if os.getenv("DEVICE", "cpu") == "cuda" else "cpu"
|
| 99 |
|
| 100 |
TOK = None
|
|
@@ -103,10 +105,10 @@ _MODEL_LOCK = threading.Lock()
|
|
| 103 |
|
| 104 |
def _load_model():
|
| 105 |
global TOK, MODEL
|
| 106 |
-
log.info("[MODEL] loading base=%s adapter=%s", BASE_MODEL_ID, ADAPTER_ID)
|
| 107 |
TOK = AutoTokenizer.from_pretrained(BASE_MODEL_ID)
|
| 108 |
base = AutoModelForSeq2SeqLM.from_pretrained(BASE_MODEL_ID)
|
| 109 |
-
MODEL = PeftModel.from_pretrained(base, ADAPTER_ID)
|
| 110 |
MODEL.eval().to(DEVICE)
|
| 111 |
log.info("[MODEL] ready on %s", DEVICE)
|
| 112 |
|
|
@@ -131,7 +133,7 @@ def translate_with_model(text: str, max_new_tokens: int = 48) -> str:
|
|
| 131 |
)
|
| 132 |
return tok.decode(outputs[0], skip_special_tokens=True)
|
| 133 |
|
| 134 |
-
# =====
|
| 135 |
@app.before_request
|
| 136 |
def _log_req():
|
| 137 |
if request.path not in ("/health", "/ping", "/favicon.ico"):
|
|
@@ -157,13 +159,11 @@ 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 |
|
|
@@ -176,12 +176,10 @@ 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
|
|
@@ -202,7 +200,6 @@ def index():
|
|
| 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")
|
|
|
|
| 1 |
# app.py
|
| 2 |
+
# PapuaTranslate β Flask + SQLAlchemy (Supabase/SQLite) + mT5-LoRA (lazy)
|
| 3 |
import os, re, logging, threading
|
| 4 |
from datetime import datetime, timezone
|
| 5 |
from functools import wraps
|
|
|
|
| 9 |
logging.basicConfig(level=logging.INFO, format="%(asctime)s | %(levelname)s | %(message)s")
|
| 10 |
log = logging.getLogger("papua-app")
|
| 11 |
|
| 12 |
+
# ===== Flask =====
|
| 13 |
+
# >> PAKAI folder 'frontend' (sesuai struktur kamu)
|
| 14 |
+
app = Flask(__name__, template_folder="frontend", static_folder="static")
|
| 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",
|
|
|
|
| 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 β psycopg2
|
| 32 |
if DATABASE_URL.startswith("postgres://"):
|
| 33 |
DATABASE_URL = DATABASE_URL.replace("postgres://", "postgresql+psycopg2://", 1)
|
| 34 |
elif DATABASE_URL.startswith("postgresql://"):
|
|
|
|
| 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 |
|
|
|
|
| 94 |
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
| 95 |
from peft import PeftModel
|
| 96 |
|
| 97 |
+
# Untuk uji UI cepat gunakan mt5-small & kosongkan adapter, lalu ganti ke model kamu
|
| 98 |
+
BASE_MODEL_ID = os.getenv("BASE_MODEL_ID", "google/mt5-small")
|
| 99 |
+
ADAPTER_ID = os.getenv("ADAPTER_ID", "")
|
| 100 |
DEVICE = "cuda" if os.getenv("DEVICE", "cpu") == "cuda" else "cpu"
|
| 101 |
|
| 102 |
TOK = None
|
|
|
|
| 105 |
|
| 106 |
def _load_model():
|
| 107 |
global TOK, MODEL
|
| 108 |
+
log.info("[MODEL] loading base=%s adapter=%s", BASE_MODEL_ID, ADAPTER_ID or "-")
|
| 109 |
TOK = AutoTokenizer.from_pretrained(BASE_MODEL_ID)
|
| 110 |
base = AutoModelForSeq2SeqLM.from_pretrained(BASE_MODEL_ID)
|
| 111 |
+
MODEL = PeftModel.from_pretrained(base, ADAPTER_ID) if ADAPTER_ID else base
|
| 112 |
MODEL.eval().to(DEVICE)
|
| 113 |
log.info("[MODEL] ready on %s", DEVICE)
|
| 114 |
|
|
|
|
| 133 |
)
|
| 134 |
return tok.decode(outputs[0], skip_special_tokens=True)
|
| 135 |
|
| 136 |
+
# ===== Utils / logging =====
|
| 137 |
@app.before_request
|
| 138 |
def _log_req():
|
| 139 |
if request.path not in ("/health", "/ping", "/favicon.ico"):
|
|
|
|
| 159 |
email = (request.form.get("email") or "").strip().lower()
|
| 160 |
pwd = request.form.get("password") or ""
|
| 161 |
if not email or not pwd:
|
| 162 |
+
flash("Isi email dan password", "error"); return redirect(url_for("login_get"))
|
|
|
|
| 163 |
with SessionLocal() as s:
|
| 164 |
u = s.query(User).filter_by(email=email).first()
|
| 165 |
if not u or not verify_password(u, pwd):
|
| 166 |
+
flash("Email atau password salah", "error"); return redirect(url_for("login_get"))
|
|
|
|
| 167 |
session["uid"], session["email"] = u.id, u.email
|
| 168 |
return redirect(url_for("index"))
|
| 169 |
|
|
|
|
| 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"); return redirect(url_for("register_get"))
|
|
|
|
| 180 |
with SessionLocal() as s:
|
| 181 |
if s.query(User).filter_by(email=email).first():
|
| 182 |
+
flash("Email sudah terdaftar", "error"); return redirect(url_for("register_get"))
|
|
|
|
| 183 |
u = User(email=email); set_password(u, pwd)
|
| 184 |
s.add(u); s.commit()
|
| 185 |
session["uid"], session["email"] = u.id, u.email
|
|
|
|
| 200 |
.order_by(Translation.id.desc())
|
| 201 |
.limit(10).all())
|
| 202 |
data = [{"src": it.src, "mt": it.mt, "created_at": it.created_at} for it in items]
|
|
|
|
| 203 |
return render_template("index.html", user=session.get("email"), data=data, device=DEVICE)
|
| 204 |
|
| 205 |
@app.get("/about")
|