Update app.py

app.py

@@ -2,16 +2,17 @@ from transformers import BlenderbotTokenizer, BlenderbotForConditionalGeneration
 import torch
 import gradio as gr
 
-
 # PersistDataset -----
 import os
 import csv
-import gradio as gr
 from gradio import inputs, outputs
 import huggingface_hub
 from huggingface_hub import Repository, hf_hub_download, upload_file
 from datetime import datetime
 
+from typing import List, Dict
+import httpx
+import pandas as pd
 
 # -------------------------------------------- For Memory - you will need to set up a dataset and HF_TOKEN ---------
 UseMemory=True
@@ -37,7 +38,60 @@ if UseMemory:
       local_dir="data", clone_from=DATASET_REPO_URL, use_auth_token=HF_TOKEN
       )
 
-       
+async def get_splits(dataset_name: str) -> Dict[str, List[Dict]]:
+    URL = f"https://datasets-server.huggingface.co/splits?dataset={dataset_name}"
+    async with httpx.AsyncClient() as session:
+        response = await session.get(URL)
+        return response.json()
+
+async def get_valid_datasets() -> Dict[str, List[str]]:
+    URL = f"https://datasets-server.huggingface.co/valid"
+    async with httpx.AsyncClient() as session:
+        response = await session.get(URL)
+        datasets = response.json()["valid"]
+        return gr.Dropdown.update(choices=datasets, value="kelm")
+        # The one to watch: https://huggingface.co/rungalileo
+        # rungalileo/medical_transcription_40
+
+async def get_first_rows(dataset: str, config: str, split: str) -> Dict[str, Dict[str, List[Dict]]]:
+    URL = f"https://datasets-server.huggingface.co/first-rows?dataset={dataset}&config={config}&split={split}"
+    async with httpx.AsyncClient() as session:
+        response = await session.get(URL)
+        print(URL)
+        gr.Markdown(URL)
+        return response.json()
+
+def get_df_from_rows(api_output):
+    return pd.DataFrame([row["row"] for row in api_output["rows"]])
+
+async def update_configs(dataset_name: str):
+    splits = await get_splits(dataset_name)
+    all_configs = sorted(set([s["config"] for s in splits["splits"]]))
+    return (gr.Dropdown.update(choices=all_configs, value=all_configs[0]),
+            splits)
+
+async def update_splits(config_name: str, state: gr.State):
+    splits_for_config = sorted(set([s["split"] for s in state["splits"] if s["config"] == config_name]))
+    dataset_name = state["splits"][0]["dataset"]
+    dataset = await update_dataset(splits_for_config[0], config_name, dataset_name)
+    return (gr.Dropdown.update(choices=splits_for_config, value=splits_for_config[0]), dataset)
+
+async def update_dataset(split_name: str, config_name: str, dataset_name: str):
+    rows = await get_first_rows(dataset_name, config_name, split_name)
+    df = get_df_from_rows(rows)
+    return df
+
+# Guido von Roissum: https://www.youtube.com/watch?v=-DVyjdw4t9I
+async def update_URL(dataset: str, config: str, split: str) -> str:
+    URL = f"https://datasets-server.huggingface.co/first-rows?dataset={dataset}&config={config}&split={split}"
+    URL = f"https://huggingface.co/datasets/{split}"
+    return (URL)
+   
+async def openurl(URL: str) -> str:
+    html = f"<a href={URL} target=_blank>{URL}</a>"
+    return (html)
+
+
 def store_message(name: str, message: str):
     if name and message:
         with open(DATA_FILE, "a") as csvfile: