Update app.py

app.py

@@ -2,60 +2,80 @@ from flask import Flask, render_template, request, jsonify
 from processor import DatasetCommandCenter
 import threading
 import uuid
+import os
 
 app = Flask(__name__)
-
-# In-memory storage for job status (Use Redis for production)
 jobs = {}
 
 @app.route('/')
 def index():
     return render_template('index.html')
 
-@app.route('/inspect', methods=['POST'])
-def inspect():
+# --- Step 1: Analyze Metadata (Configs/Splits) ---
+@app.route('/analyze_metadata', methods=['POST'])
+def analyze_metadata():
     data = request.json
-    token = data.get('token')
-    dataset_id = data.get('dataset_id')
-    split = data.get('split', 'train')
-    
-    center = DatasetCommandCenter(token=token)
-    result = center.inspect_dataset(dataset_id, split=split)
+    center = DatasetCommandCenter(token=data.get('token'))
+    result = center.get_dataset_metadata(data.get('dataset_id'))
     return jsonify(result)
 
+@app.route('/get_splits', methods=['POST'])
+def get_splits():
+    data = request.json
+    center = DatasetCommandCenter(token=data.get('token'))
+    result = center.get_splits_for_config(data.get('dataset_id'), data.get('config'))
+    return jsonify(result)
+
+# --- Step 2: Inspect Rows ---
+@app.route('/inspect_rows', methods=['POST'])
+def inspect_rows():
+    data = request.json
+    center = DatasetCommandCenter(token=data.get('token'))
+    result = center.inspect_dataset(
+        data.get('dataset_id'), 
+        data.get('config'), 
+        data.get('split')
+    )
+    return jsonify(result)
+
+# --- Step 3: Preview Transform ---
 @app.route('/preview', methods=['POST'])
 def preview():
     data = request.json
-    token = data.get('token')
-    dataset_id = data.get('dataset_id')
-    split = data.get('split', 'train')
-    recipe = data.get('recipe', {})
-    
-    center = DatasetCommandCenter(token=token)
+    center = DatasetCommandCenter(token=data.get('token'))
     try:
-        preview_rows = center.preview_transform(dataset_id, split, recipe)
-        return jsonify({"status": "success", "rows": preview_rows})
+        rows = center.preview_transform(
+            data.get('dataset_id'), 
+            data.get('config'), 
+            data.get('split'),
+            data.get('recipe', {})
+        )
+        return jsonify({"status": "success", "rows": rows})
     except Exception as e:
         return jsonify({"status": "error", "message": str(e)})
 
+# --- Step 4: Execute ---
 @app.route('/execute', methods=['POST'])
 def execute():
     data = request.json
     job_id = str(uuid.uuid4())
     
     token = data.get('token')
-    source_id = data.get('dataset_id')
-    target_id = data.get('target_id')
-    split = data.get('split', 'train')
-    recipe = data.get('recipe', {})
-    max_rows = data.get('max_rows')
+    center = DatasetCommandCenter(token=token)
+    
+    args = (
+        data.get('dataset_id'),
+        data.get('config'),
+        data.get('split'),
+        data.get('target_id'),
+        data.get('recipe', {}),
+        data.get('max_rows')
+    )
 
-    # Start background thread
     def task():
-        center = DatasetCommandCenter(token=token)
         try:
             jobs[job_id] = {"status": "running"}
-            res = center.process_and_push(source_id, split, target_id, recipe, max_rows)
+            res = center.process_and_push(*args)
             jobs[job_id] = {"status": "completed", "result": res}
         except Exception as e:
             jobs[job_id] = {"status": "failed", "error": str(e)}