20

app.py

@@ -1,8 +1,9 @@
 # app.py
 # PapuaTranslate — Flask + SQLAlchemy (Supabase/SQLite) + mT5/LoRA (lazy)
-# Fitur: ProxyFix, cookie Spaces, session permanent, /diag, preload, fallback, strip BOM JSON.
+# Fitur: ProxyFix, cookie Spaces, session permanent, /diag, preload, fallback,
+#        strip BOM JSON, sanitize adapter_config untuk kompatibilitas PEFT lama.
 
-import os, re, json, codecs, pathlib, logging, threading, traceback
+import os, re, json, codecs, pathlib, logging, threading, traceback, inspect
 from datetime import datetime, timezone, timedelta
 from functools import wraps
 from flask import Flask, render_template, request, redirect, url_for, session, jsonify, flash
@@ -15,16 +16,12 @@ log = logging.getLogger("papua-app")
 
 # ========== Flask ==========
 app = Flask(__name__, template_folder="frontend", static_folder="static")
-
-# Trust reverse proxy di HF supaya proto/host benar (perlu untuk cookie Secure)
 app.wsgi_app = ProxyFix(app.wsgi_app, x_for=1, x_proto=1, x_host=1)
-
-# Cookie/session aman untuk iframe (HF Spaces) dan tab langsung hf.space
 app.config.update(
     SECRET_KEY=os.getenv("SECRET_KEY", "dev-secret-change-me"),
     SESSION_COOKIE_NAME="hfspace_session",
-    SESSION_COOKIE_SAMESITE="None",   # penting: iframe = third-party context
-    SESSION_COOKIE_SECURE=True,       # wajib True jika SAMESITE=None
+    SESSION_COOKIE_SAMESITE="None",
+    SESSION_COOKIE_SECURE=True,
     SESSION_COOKIE_HTTPONLY=True,
     SESSION_COOKIE_PATH="/",
     PREFERRED_URL_SCHEME="https",
@@ -44,12 +41,10 @@ if not DATABASE_URL:
     DATABASE_URL = "sqlite:////tmp/app.db"
     log.warning("[DB] DATABASE_URL tidak diset; pakai SQLite /tmp/app.db")
 else:
-    # normalisasi skema ke psycopg2
     if DATABASE_URL.startswith("postgres://"):
         DATABASE_URL = DATABASE_URL.replace("postgres://", "postgresql+psycopg2://", 1)
     elif DATABASE_URL.startswith("postgresql://"):
         DATABASE_URL = DATABASE_URL.replace("postgresql://", "postgresql+psycopg2://", 1)
-    # tambahkan sslmode kalau belum ada
     if DATABASE_URL.startswith("postgresql+psycopg2") and "sslmode=" not in DATABASE_URL:
         sep = "&" if "?" in DATABASE_URL else "?"
         DATABASE_URL = f"{DATABASE_URL}{sep}sslmode=require"
@@ -95,7 +90,7 @@ def login_required(fn):
         return fn(*args, **kwargs)
     return _wrap
 
-# ========== Prenorm (heuristik ringan) ==========
+# ========== Prenorm ==========
 PAPUA_MAP = {
     r"\bsa\b": "saya", r"\bko\b": "kamu", r"\btra\b": "tidak", r"\bndak\b": "tidak",
     r"\bmo\b": "mau", r"\bpu\b": "punya", r"\bsu\b": "sudah", r"\bkong\b": "kemudian",
@@ -106,9 +101,9 @@ def prenorm(text: str) -> str:
     for pat, repl in PAPUA_MAP.items(): t = re.sub(pat, repl, t, flags=re.IGNORECASE)
     return t
 
-# ========== Model (lazy) + Strip BOM ==========
-BASE_MODEL_ID = os.getenv("BASE_MODEL_ID", "google/mt5-small")  # mulai kecil dulu untuk uji cepat
-ADAPTER_ID    = os.getenv("ADAPTER_ID", "")                     # kosongkan dulu; isi nanti
+# ========== Model (lazy) + Strip BOM + Sanitize adapter_config ==========
+BASE_MODEL_ID = os.getenv("BASE_MODEL_ID", "google/mt5-small")
+ADAPTER_ID    = os.getenv("ADAPTER_ID", "")
 DEVICE        = "cuda" if os.getenv("DEVICE", "cpu") == "cuda" else "cpu"
 
 TOK = None
@@ -118,7 +113,6 @@ _MODEL_READY = False
 _MODEL_ERROR = None
 
 def _strip_bom_in_dir(root_dir: str):
-    """Hapus BOM dari semua *.json (UTF-8-sig) agar tidak memicu JSONDecodeError."""
     root = pathlib.Path(root_dir)
     for p in root.rglob("*.json"):
         try:
@@ -128,11 +122,56 @@ def _strip_bom_in_dir(root_dir: str):
                 json.dump(data, f, ensure_ascii=False, indent=2)
             log.info(f"[BOM] stripped: {p}")
         except Exception as e:
-            # Jangan gagal hanya karena satu file
             log.warning(f"[BOM] skip {p}: {e}")
 
+def _sanitize_adapter_config(adapter_dir: str):
+    """
+    Buang kunci tidak dikenal oleh peft.LoraConfig saat ini.
+    Menangani error seperti: LoraConfig.__init__() got an unexpected keyword argument 'e_rank_pattern'
+    """
+    try:
+        from peft import LoraConfig  # versi yang sedang terpasang di runtime
+    except Exception as e:
+        log.warning(f"[SAN] Tidak bisa import LoraConfig: {e}")
+        return
+
+    sig = None
+    try:
+        sig = inspect.signature(LoraConfig.__init__)
+        allowed = set(p.name for p in sig.parameters.values())
+    except Exception as e:
+        log.warning(f"[SAN] Tidak bisa baca signature LoraConfig: {e}")
+        return
+
+    cfg_path = pathlib.Path(adapter_dir) / "adapter_config.json"
+    if not cfg_path.exists():
+        # beberapa repo pakai nama lain; normalnya adapter_config.json
+        for alt in ("adapter_config.json", "adapter_config_0.json", "config.json"):
+            candidate = pathlib.Path(adapter_dir) / alt
+            if candidate.exists():
+                cfg_path = candidate
+                break
+    if not cfg_path.exists():
+        log.warning(f"[SAN] adapter_config tidak ditemukan di {adapter_dir}")
+        return
+
+    try:
+        with codecs.open(cfg_path, "r", encoding="utf-8-sig") as f:
+            cfg = json.load(f)
+    except Exception as e:
+        log.warning(f"[SAN] gagal baca adapter_config: {e}")
+        return
+
+    # filter hanya key yang diizinkan oleh LoraConfig saat ini
+    cleaned = {k: v for k, v in cfg.items() if k in allowed}
+    dropped = [k for k in cfg.keys() if k not in allowed]
+    if dropped:
+        log.info(f"[SAN] drop fields tidak dikenal: {dropped}")
+        with open(cfg_path, "w", encoding="utf-8") as f:
+            json.dump(cleaned, f, ensure_ascii=False, indent=2)
+
 def _load_model():
-    """Download ke /tmp, strip BOM, lalu load dari path lokal."""
+    """Download ke /tmp, strip BOM, sanitize adapter_config, lalu load dari path lokal."""
     global TOK, MODEL, _MODEL_READY, _MODEL_ERROR
     try:
         log.info("[MODEL] downloading base=%s adapter=%s", BASE_MODEL_ID, ADAPTER_ID or "-")
@@ -154,8 +193,8 @@ def _load_model():
                 allow_patterns=None,
             )
             _strip_bom_in_dir(adapter_dir)
+            _sanitize_adapter_config(adapter_dir)
 
-        # Import di sini agar error versi lib ketangkep di /diag
         import torch
         from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
         from peft import PeftModel
@@ -194,12 +233,8 @@ def translate_with_model(text: str, max_new_tokens: int = 48) -> str:
     tok, m = get_model()
     if not _MODEL_READY or m is None:
         raise RuntimeError(f"Model not ready: {_MODEL_ERROR or 'unknown error'}")
-
-    # Guard panjang input supaya aman
     enc = tok([text], return_tensors="pt", truncation=True, max_length=256)
     enc = {k: v.to(DEVICE) for k, v in enc.items()}
-
-    log.info("[GEN] start (tok_len=%d)", int(enc["input_ids"].shape[-1]))
     out = m.generate(
         **enc,
         max_new_tokens=int(max_new_tokens),
@@ -208,10 +243,8 @@ def translate_with_model(text: str, max_new_tokens: int = 48) -> str:
         no_repeat_ngram_size=3,
         early_stopping=True,
     )
-    log.info("[GEN] done")
     return tok.decode(out[0], skip_special_tokens=True)
 
-# Preload supaya request pertama nggak lama/timeout
 def _preload_thread():
     try:
         _load_model()
@@ -239,8 +272,8 @@ def diag():
     try:
         import torch, transformers, peft
         torch_v = torch.__version__
-        tf_v = transformers.__version__
-        peft_v = peft.__version__
+        tf_v    = transformers.__version__
+        peft_v  = peft.__version__
     except Exception as e:
         torch_v = tf_v = peft_v = f"import error: {e}"
 
@@ -298,7 +331,7 @@ def register_get():
 @app.post("/register")
 def register_post():
     email = (request.form.get("email") or "").strip().lower()
-    pwd   = request.form.get("password") or ""
+    pwd   = (request.form.get("password") or "")
     if not email or not pwd:
         flash("Isi email dan password", "error"); return redirect(url_for("register_get"))
     with SessionLocal() as s:
@@ -344,18 +377,10 @@ def api_translate():
 
     try:
         clean = prenorm(text)
-
-        if FALLBACK_TRANSLATE:
-            mt = f"[FAKE] {clean}"
-        else:
-            mt = translate_with_model(clean, max_new_tokens=max_new)
-
-        # Simpan riwayat
+        mt = f"[FAKE] {clean}" if FALLBACK_TRANSLATE else translate_with_model(clean, max_new_tokens=max_new)
         with SessionLocal() as s:
             s.add(Translation(user_id=session["uid"], src=text, mt=mt))
             s.commit()
-
-        # Kompatibel dengan frontend kamu (pakai j.mt)
         return jsonify({"ok": True, "mt": mt})
     except Exception as e:
         log.error("[API] translate error: %s", e)