Upload 4 files

Dockerfile

@@ -12,14 +12,15 @@ COPY . .
 RUN mkdir -p /app/model_cache && chmod -R 777 /app
 
 RUN pip install --upgrade pip setuptools wheel
-RUN pip install numpy==1.24.4
 RUN pip install --no-cache-dir -r requirements.txt
 
 ENV FLASK_APP=app.py
 ENV FLASK_ENV=production
 ENV TRANSFORMERS_CACHE=/app/model_cache
 ENV TORCH_CACHE=/app/model_cache
+ENV PYTHONUNBUFFERED=1
+ENV TOKENIZERS_PARALLELISM=false
 
 EXPOSE 7860
 
-CMD ["gunicorn", "--workers", "1", "--threads", "2", "--bind", "0.0.0.0:7860", "--timeout", "300", "--preload", "app:app"]
+CMD ["gunicorn", "--workers", "1", "--threads", "2", "--bind", "0.0.0.0:7860", "--timeout", "120", "--preload", "app:app"]

app.py

@@ -1,280 +1,205 @@
-import os
-import re
-import logging
-from flask import Flask, request, jsonify, Response
-from flask_cors import CORS
-from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
-import torch
-import psycopg2
-import json
-
-app = Flask(__name__)
-CORS(app)
-
-# إعدادات التسجيل
-logging.basicConfig(level=logging.INFO)
-logger = logging.getLogger(__name__)
-
-# إعدادات النموذج
-MODEL_NAME = "tscholak/cxmefzzi"  # نموذج أصغر
-SUPABASE_DB_URL = os.getenv('SUPABASE_DB_URL')
-
-# تهيئة مجلد الذاكرة المؤقتة
-os.makedirs("model_cache", exist_ok=True)
-os.environ["TRANSFORMERS_CACHE"] = "model_cache"
-os.environ["TORCH_CACHE"] = "model_cache"
-
-tokenizer = None
-model = None
-
-def initialize():
-    global tokenizer, model
-    try:
-        logger.info("جاري تحميل النموذج بتهيئة منخفضة الذاكرة...")
-        tokenizer = AutoTokenizer.from_pretrained(
-            MODEL_NAME,
-            cache_dir="model_cache",
-            local_files_only=False
-        )
-        model = AutoModelForSeq2SeqLM.from_pretrained(
-            MODEL_NAME,
-            cache_dir="model_cache",
-            device_map="auto",
-            torch_dtype=torch.float16,
-            low_cpu_mem_usage=True
-        )
-        model.eval()
-        logger.info("تم تحميل النموذج بنجاح")
-    except Exception as e:
-        logger.error(f"فشل في تحميل النموذج: {str(e)}")
-        raise
-
-initialize()
-# --- سكيمة قاعدة البيانات ---
-DB_SCHEMA = """
-CREATE TABLE public.profiles (
-  id uuid NOT NULL,
-  updated_at timestamp with time zone,
-  username text UNIQUE CHECK (char_length(username) >= 3),
-  full_name text,
-  avatar_url text,
-  website text,
-  cam_mac text UNIQUE,
-  fcm_token text,
-  notification_enabled boolean DEFAULT true,
-  CONSTRAINT profiles_pkey PRIMARY KEY (id),
-  CONSTRAINT profiles_id_fkey FOREIGN KEY (id) REFERENCES auth.users(id)
-);
-
-CREATE TABLE public.place (
-  id bigint GENERATED ALWAYS AS IDENTITY NOT NULL,
-  created_at timestamp with time zone DEFAULT (now() AT TIME ZONE 'utc'::text),
-  name text,
-  CONSTRAINT place_pkey PRIMARY KEY (id)
-);
-
-CREATE TABLE public.user_place (
-  id bigint GENERATED ALWAYS AS IDENTITY NOT NULL,
-  created_at timestamp with time zone NOT NULL DEFAULT now(),
-  place_id bigint,
-  user_cam_mac text,
-  CONSTRAINT user_place_pkey PRIMARY KEY (id),
-  CONSTRAINT user_place_place_id_fkey FOREIGN KEY (place_id) REFERENCES public.place(id),
-  CONSTRAINT user_place_user_cam_mac_fkey FOREIGN KEY (user_cam_mac) REFERENCES public.profiles(cam_mac)
-);
-
-CREATE TABLE public.data (
-  id bigint GENERATED ALWAYS AS IDENTITY NOT NULL,
-  created_at timestamp without time zone,
-  caption text,
-  image_url text,
-  latitude double precision DEFAULT '36.1833854'::double precision,
-  longitude double precision DEFAULT '37.1309255'::double precision,
-  user_place_id bigint,
-  cam_mac text,
-  CONSTRAINT data_pkey PRIMARY KEY (id),
-  CONSTRAINT data_user_place_id_fkey FOREIGN KEY (user_place_id) REFERENCES public.user_place(id)
-);
-
-CREATE TABLE public.biodata (
-  id bigint GENERATED ALWAYS AS IDENTITY NOT NULL,
-  created_at timestamp with time zone NOT NULL DEFAULT now(),
-  mac_address text,
-  acceleration_x double precision,
-  acceleration_y double precision,
-  acceleration_z double precision,
-  gyro_x double precision,
-  gyro_y double precision,
-  gyro_z double precision,
-  temperature double precision,
-  CONSTRAINT biodata_pkey PRIMARY KEY (id),
-  CONSTRAINT biodata_mac_address_fkey FOREIGN KEY (mac_address) REFERENCES public.profiles(cam_mac)
-);
-
-CREATE TABLE public.notification (
-  id bigint GENERATED ALWAYS AS IDENTITY NOT NULL,
-  created_at timestamp without time zone NOT NULL DEFAULT now(),
-  user_cam_mac text,
-  title text,
-  message text,
-  is_read boolean,
-  acceleration_x double precision,
-  acceleration_y double precision,
-  acceleration_z double precision,
-  gyro_x double precision,
-  gyro_y double precision,
-  gyro_z double precision,
-  CONSTRAINT notification_pkey PRIMARY KEY (id),
-  CONSTRAINT notification_user_cam_mac_fkey FOREIGN KEY (user_cam_mac) REFERENCES public.profiles(cam_mac)
-);
-
-CREATE TABLE public.flag (
-  id bigint GENERATED ALWAYS AS IDENTITY NOT NULL,
-  flag smallint,
-  user_mac_address text,
-  CONSTRAINT flag_pkey PRIMARY KEY (id),
-  CONSTRAINT flag_user_mac_address_fkey FOREIGN KEY (user_mac_address) REFERENCES public.profiles(cam_mac)
-);
-""".strip()
-
-# --- الاتصال بقاعدة البيانات ---
-def get_db_connection():
-    try:
-        return psycopg2.connect(SUPABASE_DB_URL)
-    except Exception as err:
-        logger.error(f"Database connection error: {err}")
-        return None
-
-# --- التحقق من صحة cam_mac ---
-def validate_cam_mac(cam_mac):
-    conn = get_db_connection()
-    if not conn:
-        return False
-        
-    try:
-        cursor = conn.cursor()
-        cursor.execute("SELECT 1 FROM profiles WHERE cam_mac = %s;", (cam_mac,))
-        return cursor.fetchone() is not None
-    except Exception as e:
-        logger.error(f"Validation error: {e}")
-        return False
-    finally:
-        if conn:
-            conn.close()
-
-@app.route('/api/query', methods=['POST'])
-def handle_query():
-    if tokenizer is None or model is None:
-        return jsonify({"error": "النموذج غير محمل، يرجى المحاولة لاحقاً"}), 503
-        
-    try:
-        data = request.get_json()
-        if not data or 'text' not in data or 'cam_mac' not in data:
-            return jsonify({"error": "يرجى إرسال 'text' و 'cam_mac'"}), 400
-
-        natural_query = data['text']
-        cam_mac = data['cam_mac']
-        logger.info(f"استعلام من {cam_mac}: {natural_query}")
-        
-        if not validate_cam_mac(cam_mac):
-            return jsonify({"error": "عنوان MAC غير صالح"}), 403
-
-        prompt = f"""
-        ### Postgres SQL table definitions
-        {DB_SCHEMA}
-
-        ### Rules:
-        - Always filter by cam_mac = '{cam_mac}'
-        - Use only SELECT statements
-        - Use proper JOINs
-        - Use table aliases when helpful
-        - The output must contain only the SQL query
-
-        ### User question: {natural_query}
-
-        ### SQL query:
-        SELECT
-        """.strip()
-
-        inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
-        
-        with torch.no_grad():
-            outputs = model.generate(**inputs, max_length=256)
-            
-        sql = tokenizer.decode(outputs[0], skip_special_tokens=True)
-
-        # تنظيف الناتج
-        sql = re.sub(r"^```sql\s*", "", sql, flags=re.IGNORECASE)
-        sql = re.sub(r"\s*```$", "", sql)
-        sql = re.sub(r"^SQL:\s*", "", sql, flags=re.IGNORECASE)
-
-        if not sql.upper().startswith("SELECT"):
-            sql = "SELECT " + sql.split("SELECT")[-1] if "SELECT" in sql else f"SELECT * FROM ({sql}) AS subquery"
-
-        if not sql.endswith(";"):
-            sql += ";"
-
-        logger.info(f"استعلام SQL المولد: {sql}")
-
-        if not sql.upper().strip().startswith("SELECT"):
-            return jsonify({"error": "يُسمح فقط باستعلامات SELECT"}), 403
-
-        conn = get_db_connection()
-        if not conn:
-            return jsonify({"error": "فشل الاتصال بقاعدة البيانات"}), 500
-
-        cursor = None
-        try:
-            cursor = conn.cursor()
-            cursor.execute(sql)
-            columns = [desc[0] for desc in cursor.description]
-            rows = cursor.fetchall()
-            data = [dict(zip(columns, row)) for row in rows]
-
-            response_data = {
-                "data": data,
-                "generated_sql": sql
-            }
-            
-            response_json = json.dumps(response_data, ensure_ascii=False)
-            
-            return Response(
-                response_json,
-                status=200,
-                mimetype='application/json; charset=utf-8'
-            )
-            
-        except Exception as e:
-            logger.error(f"خطأ في تنفيذ SQL: {e}")
-            return jsonify({
-                "error": str(e),
-                "generated_sql": sql
-            }), 500
-        finally:
-            if cursor:
-                cursor.close()
-            if conn:
-                conn.close()
-
-    except Exception as e:
-        logger.error(f"خطأ في التوليد: {str(e)}")
-        return jsonify({"error": "فشل في توليد الاستعلام"}), 500
-
-@app.route('/health')
-def health_check():
-    return jsonify({
-        "status": "healthy",
-        "model_loaded": model is not None,
-        "db_connection": get_db_connection() is not None
-    })
-
-@app.route('/')
-def home():
-    return """
-    <h1>Text2SQL API</h1>
-    <p>استخدم <code>/api/query</code> مع POST {"text": "سؤالك", "cam_mac": "عنوان MAC"}</p>
-    <p>تحقق من حالة الخدمة: <a href="/health">/health</a></p>
-    """
-
-if __name__ == '__main__':
+import os
+import re
+import logging
+from flask import Flask, request, jsonify, Response
+from flask_cors import CORS
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+import torch
+import psycopg2
+import json
+
+app = Flask(__name__)
+CORS(app)
+
+# إعدادات التسجيل
+logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger(__name__)
+
+# إعدادات النموذج
+MODEL_NAME = "tscholak/cxmefzzi"
+SUPABASE_DB_URL = os.getenv('SUPABASE_DB_URL')
+
+# تهيئة مجلد الذاكرة المؤقتة
+os.makedirs("model_cache", exist_ok=True)
+os.environ["TRANSFORMERS_CACHE"] = "model_cache"
+os.environ["TORCH_CACHE"] = "model_cache"
+
+tokenizer = None
+model = None
+
+def initialize():
+    global tokenizer, model
+    try:
+        logger.info("🚀 جاري تحميل النموذج...")
+        tokenizer = AutoTokenizer.from_pretrained(
+            MODEL_NAME,
+            cache_dir="model_cache",
+            local_files_only=False
+        )
+        model = AutoModelForSeq2SeqLM.from_pretrained(
+            MODEL_NAME,
+            cache_dir="model_cache",
+            device_map="auto",
+            torch_dtype=torch.float16,
+            low_cpu_mem_usage=True
+        )
+        model.eval()
+        logger.info("✅ تم تحميل النموذج بنجاح")
+    except Exception as e:
+        logger.error(f"❌ فشل في تحميل النموذج: {str(e)}")
+        raise
+
+initialize()
+
+# سكيما قاعدة البيانات
+DB_SCHEMA = """
+CREATE TABLE public.profiles (
+  id uuid NOT NULL,
+  cam_mac text UNIQUE,
+  ...
+);
+CREATE TABLE public.data (
+  id bigint GENERATED ALWAYS AS IDENTITY,
+  cam_mac text,
+  ...
+);
+""".strip()
+
+def get_db_connection():
+    try:
+        conn = psycopg2.connect(SUPABASE_DB_URL)
+        logger.info("✅ تم الاتصال بقاعدة البيانات")
+        return conn
+    except Exception as err:
+        logger.error(f"❌ خطأ في الاتصال بقاعدة البيانات: {err}")
+        return None
+
+def validate_cam_mac(cam_mac):
+    conn = get_db_connection()
+    if not conn:
+        return False
+        
+    try:
+        cursor = conn.cursor()
+        cursor.execute("SELECT 1 FROM profiles WHERE cam_mac = %s;", (cam_mac,))
+        return cursor.fetchone() is not None
+    except Exception as e:
+        logger.error(f"❌ خطأ في التحقق من cam_mac: {e}")
+        return False
+    finally:
+        if conn:
+            conn.close()
+
+@app.route('/api/query', methods=['POST'])
+def handle_query():
+    try:
+        data = request.get_json()
+        logger.info(f"📩 بيانات الطلب: {data}")
+        
+        if not data or 'text' not in data or 'cam_mac' not in data:
+            logger.error("❌ بيانات ناقصة في الطلب")
+            return jsonify({"error": "يرجى إرسال 'text' و 'cam_mac'"}), 400
+
+        if not validate_cam_mac(data['cam_mac']):
+            logger.error(f"❌ cam_mac غير صالح: {data['cam_mac']}")
+            return jsonify({"error": "عنوان MAC غير صالح"}), 403
+
+        prompt = f"""
+        ### هيكل قاعدة البيانات:
+        {DB_SCHEMA}
+
+        ### السؤال:
+        {data['text']}
+
+        ### قم بإنشاء استعلام SQL بحيث:
+        - يستخدم SELECT فقط
+        - يتضمن تصفية حسب cam_mac = '{data['cam_mac']}'
+        - يكون صالحًا لـ PostgreSQL
+
+        SQL:
+        """
+
+        inputs = tokenizer(prompt, return_tensors="pt", truncation=True, max_length=512).to(model.device)
+        
+        with torch.no_grad():
+            outputs = model.generate(
+                **inputs,
+                max_length=256,
+                num_beams=4,
+                early_stopping=True
+            )
+            
+        sql = tokenizer.decode(outputs[0], skip_special_tokens=True)
+        logger.info(f"⚡ الاستعلام المولد: {sql}")
+
+        # تنظيف الاستعلام
+        sql = re.sub(r"^```sql\s*", "", sql, flags=re.IGNORECASE)
+        sql = re.sub(r"\s*```$", "", sql).strip()
+        
+        if not sql.upper().startswith("SELECT"):
+            sql = f"SELECT {sql}"
+        if not sql.endswith(";"):
+            sql += ";"
+
+        conn = get_db_connection()
+        if not conn:
+            return jsonify({"error": "فشل الاتصال بقاعدة البيانات"}), 500
+
+        try:
+            cursor = conn.cursor()
+            cursor.execute(sql)
+            
+            if cursor.description:
+                columns = [desc[0] for desc in cursor.description]
+                rows = cursor.fetchall()
+                data = [dict(zip(columns, row)) for row in rows]
+                
+                response = {
+                    "data": data,
+                    "sql": sql,
+                    "timestamp": datetime.now().isoformat()
+                }
+            else:
+                conn.commit()
+                response = {
+                    "message": "تم تنفيذ الاستعلام بنجاح",
+                    "sql": sql
+                }
+                
+            return jsonify(response)
+            
+        except Exception as e:
+            logger.error(f"❌ خطأ في تنفيذ SQL: {e}\nالاستعلام: {sql}")
+            return jsonify({
+                "error": "خطأ في تنفيذ الاستعلام",
+                "sql": sql,
+                "details": str(e)
+            }), 500
+            
+        finally:
+            if conn:
+                conn.close()
+                
+    except Exception as e:
+        logger.error(f"❌ خطأ غير متوقع: {str(e)}", exc_info=True)
+        return jsonify({"error": "فشل في معالجة الطل��"}), 500
+
+@app.route('/health')
+def health_check():
+    return jsonify({
+        "status": "healthy",
+        "model_loaded": model is not None,
+        "db_connection": get_db_connection() is not None,
+        "timestamp": datetime.now().isoformat()
+    })
+
+@app.route('/')
+def home():
+    return """
+    <h1>Text2SQL API</h1>
+    <p>استخدم <code>/api/query</code> مع POST</p>
+    <p>تحقق من الحالة: <a href="/health">/health</a></p>
+    """
+
+if __name__ == '__main__':
     app.run(host='0.0.0.0', port=7860)

dockerignore

@@ -0,0 +1,9 @@
+__pycache__
+*.pyc
+*.pyo
+*.pyd
+.DS_Store
+.env
+venv/
+model_cache/
+*.log

requirements.txt

@@ -1,9 +1,9 @@
-numpy==1.24.4
-torch==2.0.1
-transformers==4.30.2
 flask==2.3.2
 flask-cors==3.0.10
+torch==2.0.1
+transformers==4.30.2
 psycopg2-binary==2.9.7
 gunicorn==20.1.0
 accelerate==0.21.0
-python-dotenv==1.0.0
+python-dotenv==1.0.0
+numpy==1.24.4