add 2 endpoint

app.py

@@ -1,10 +1,20 @@
 import gradio as gr
 from sentence_transformers import SentenceTransformer
+from supabase import create_client
+import os
+from dotenv import load_dotenv
+
+load_dotenv()
 
 # Load the model globally to ensure it's only loaded once
 # This is crucial for performance and resource management
 model = SentenceTransformer("sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2")
 
+# === Supabase ===
+supabase_url = os.getenv("SUPABASE_URL")
+supabase_key = os.getenv("SUPABASE_KEY")
+supabase = create_client(supabase_url, supabase_key)
+
 def get_embedding(text: str):
     """
     Menghasilkan embedding vector dari teks menggunakan model SentenceTransformer.
@@ -17,22 +27,38 @@ def get_embedding(text: str):
     embedding = model.encode(text).tolist()
     return embedding
 
+def search_kbli(text: str):
+    if not text:
+        return {"embedding": [], "results": []}
+
+    embedding = get_embedding(text)  # <-- pakai fungsi lama
+
+    response = supabase.rpc(
+        "search_kbli",
+        {"query_embedding": embedding, "match_count": 5}
+    ).execute()
+
+    results = response.data if response.data else []
+
+    return {"embedding": embedding, "results": results}
+    
 # Gradio Interface
 # `fn` adalah fungsi yang akan dieksekusi
 # `inputs` adalah jenis input (dalam hal ini, sebuah teks)
 # `outputs` adalah jenis output (dalam hal ini, sebuah JSON)
 # `api_name` adalah nama endpoint API
-demo = gr.Interface(
-    fn=get_embedding,
-    inputs=gr.Textbox(label="Masukkan teks"),
-    outputs=gr.JSON(label="Embedding Vector"),
-    title="Multilingual Embedding API",
-    description="Masukkan teks untuk mendapatkan vektor embedding menggunakan model paraphrase-multilingual-MiniLM-L12-v2.",
-    examples=["Halo dunia!", "Apa kabar hari ini?", "Selamat pagi!"],
-    allow_flagging="never",
-    api_name="get_embedding"
-)
-
-# Launch the app
+with gr.Blocks() as demo:
+    gr.Markdown("## Semantic KBLI Search")
+
+    with gr.Tab("Embedding Only"):
+        inp1 = gr.Textbox(label="Masukkan teks")
+        out1 = gr.JSON(label="Embedding Vector")
+        inp1.submit(get_embedding, inp1, out1, api_name="get_embedding")
+
+    with gr.Tab("Search KBLI"):
+        inp2 = gr.Textbox(label="Masukkan teks")
+        out2 = gr.JSON(label="Hasil KBLI (Embedding + Match)")
+        inp2.submit(search_kbli, inp2, out2, api_name="search_kbli")
+
 if __name__ == "__main__":
     demo.queue().launch()