app.py.v1

# app.py
import gradio as gr
import pandas as pd
import requests
import io
import dask.dataframe as dd
from datasets import load_dataset, Image
from mlcroissant import Dataset as CroissantDataset
from huggingface_hub import get_token
import polars as pl
import warnings
import traceback

# 🤫 Let's ignore those pesky warnings, shall we?
warnings.filterwarnings("ignore")

# --- ⚙️ Configuration & Constants ---

# 🎨 Let's give our datasets some personality with emojis and names!
DATASET_CONFIG = {
    "caselaw": {
        "name": "common-pile/caselaw_access_project",
        "emoji": "⚖️",
        "search_col": "text",
        "methods": ["💨 API (requests)", "🧊 Dask", "🥐 Croissant"],
        "is_public": True,
    },
    "prompts": {
        "name": "fka/awesome-chatgpt-prompts",
        "emoji": "🤖",
        "search_col": ["act", "prompt"],
        "methods": ["🐼 Pandas", "💨 API (requests)", "🥐 Croissant"],
        "is_public": True,
    },
    "finance": {
        "name": "snorkelai/agent-finance-reasoning",
        "emoji": "💰",
        "search_col": ["question", "answer"],
        "methods": ["🐼 Pandas", "🧊 Polars", "💨 API (requests)", "🥐 Croissant"],
        "is_public": False,
    },
    "medical": {
        "name": "FreedomIntelligence/medical-o1-reasoning-SFT",
        "emoji": "🩺",
        "search_col": "conversations",
        "methods": ["🐼 Pandas", "🧊 Polars", "💨 API (requests)", "🥐 Croissant"],
        "is_public": False,
    },
    "inscene": {
        "name": "peteromallet/InScene-Dataset",
        "emoji": "🖼️",
        "search_col": "text",
        "methods": ["🤗 Datasets", "🐼 Pandas", "🧊 Polars", "💨 API (requests)", "🥐 Croissant"],
        "is_public": False,
    },
}

# --- 헬 Helpers & Utility Functions ---

def get_auth_headers():
    """🔑 Creates authorization headers if a Hugging Face token is available."""
    token = get_token()
    return {"Authorization": f"Bearer {token}"} if token else {}

def dataframe_to_outputs(df: pd.DataFrame):
    """
    📜 Takes a DataFrame and magically transforms it into various formats for your viewing pleasure.
    Like a data chameleon!
    """
    if df.empty:
        return "No results found. 🤷", None, None, "No results to copy."

    df_str = df.astype(str)
    markdown_output = df_str.to_markdown(index=False)

    csv_buffer = io.StringIO()
    df.to_csv(csv_buffer, index=False)
    csv_buffer.seek(0)
    
    excel_buffer = io.BytesIO()
    df.to_excel(excel_buffer, index=False, engine='openpyxl')
    excel_buffer.seek(0)

    tab_delimited_output = df.to_csv(sep='\t', index=False)

    return markdown_output, gr.File.from_bytes(csv_buffer.getvalue(), "results.csv"), gr.File.from_bytes(excel_buffer.getvalue(), "results.xlsx"), tab_delimited_output


def handle_error(e: Exception):
    """
    😱 Oh no! An error! This function catches it and displays it nicely.
    Because even errors deserve to look good.
    """
    error_message = f"🚨 An error occurred: {str(e)}\n\n"
    auth_tip = "🔑 For gated datasets, did you log in? Try `huggingface-cli login` in your terminal."
    full_trace = traceback.format_exc()
    print(full_trace) 
    
    if "401" in str(e) or "Gated" in str(e):
        error_message += auth_tip
        
    return (
        pd.DataFrame(),
        gr.Gallery(None, label="🖼️ Image Results"),
        f"```\n{error_message}\n\n{full_trace}\n```",
        None,
        None,
        error_message,
        f"```python\n# 🚨 Error during code generation:\n# {e}\n```"
    )

# --- 🎣 Data Fetching & Processing Functions ---

def fetch_data(dataset_key: str, access_method: str, query: str):
    """
    🚀 The main mission control function! It fetches, searches, and formats data.
    It's the brains of the operation.
    """
    try:
        config = DATASET_CONFIG[dataset_key]
        repo_id = config["name"]
        search_cols = [config["search_col"]] if isinstance(config["search_col"], str) else config["search_col"]
        df = pd.DataFrame()
        code_snippet = ""
        
        if "API" in access_method:
            url = f"https://datasets-server.huggingface.co/rows?dataset={repo_id}&config=default&split=train&offset=0&length=100"
            headers = get_auth_headers() if not config["is_public"] else {}
            response = requests.get(url, headers=headers)
            response.raise_for_status()
            data = response.json()
            df = pd.json_normalize(data['rows'], record_path='row', meta=['row_idx', 'truncated_cells'])
            df = df.drop(columns=['row_idx', 'truncated_cells'], errors='ignore')
            
            code_snippet = f"""
# 💻 Generated Code: API (requests)
import requests
import pandas as pd

# For gated datasets, get your token from https://huggingface.co/settings/tokens
# Make sure to `huggingface-cli login` first.
headers = {{"Authorization": "Bearer YOUR_HF_TOKEN"}}
url = "{url}"
response = requests.get(url, headers=headers) # Pass headers for gated datasets
data = response.json()
df = pd.json_normalize(data['rows'], record_path='row')
print(df.head())
"""

        elif "Pandas" in access_method:
            file_path = f"hf://datasets/{repo_id}/"
            if repo_id == "fka/awesome-chatgpt-prompts":
                file_path += "prompts.csv"
                df = pd.read_csv(file_path)
            else: 
                try:
                    df = pd.read_parquet(f"{file_path}data/train-00000-of-00001.parquet")
                except:
                     try:
                         df = pd.read_parquet(f"{file_path}train.parquet")
                     except:
                         df = pd.read_json(f"{file_path}medical_o1_sft.json")
            
            code_snippet = f"""
# 💻 Generated Code: Pandas
import pandas as pd

# Make sure to `huggingface-cli login` for gated datasets.
file_path = "{file_path}"
df = pd.{'read_csv' if '.csv' in file_path else ('read_json' if '.json' in file_path else 'read_parquet')}(file_path)
print(df.head())
"""
        
        elif "Polars" in access_method:
            file_path = f"hf://datasets/{repo_id}/"
            try:
                df = pl.read_parquet(f"{file_path}data/train-00000-of-00001.parquet").to_pandas()
            except:
                try:
                    df = pl.read_parquet(f"{file_path}train.parquet").to_pandas()
                except:
                    df = pl.read_json(f"{file_path}medical_o1_sft.json").to_pandas()
            
            code_snippet = f"""
# 💻 Generated Code: Polars
import polars as pl

# Make sure to `huggingface-cli login` for gated datasets.
file_path = "{'hf://datasets/' + repo_id + '/data/train-00000-of-00001.parquet'}"
df = pl.read_parquet(file_path)
print(df.head())
"""

        elif "Datasets" in access_method:
            ds = load_dataset(repo_id, split='train[:100]')
            df = ds.to_pandas()
            code_snippet = f"""
# 💻 Generated Code: Datasets
from datasets import load_dataset

# Make sure to `huggingface-cli login` for gated datasets.
ds = load_dataset("{repo_id}", split='train')
print(ds)
"""

        elif "Dask" in access_method:
            df = dd.read_json(f"hf://datasets/{repo_id}/**/*.jsonl.gz").head(100)
            code_snippet = f"""
# 💻 Generated Code: Dask
import dask.dataframe as dd

# Make sure to `huggingface-cli login` for gated datasets.
ddf = dd.read_json("hf://datasets/{repo_id}/**/*.jsonl.gz")
print(ddf.head())
"""
            
        elif "Croissant" in access_method:
            headers = get_auth_headers() if not config["is_public"] else {}
            jsonld_url = f"https://huggingface.co/api/datasets/{repo_id}/croissant"
            jsonld = requests.get(jsonld_url, headers=headers).json()
            ds = CroissantDataset(jsonld=jsonld)
            records = ds.records("default")
            data_rows = [row for _, row in zip(range(100), records)]
            df = pd.DataFrame(data_rows)
            code_snippet = f"""
# 💻 Generated Code: Croissant
import requests
from mlcroissant import Dataset as CroissantDataset
import pandas as pd

# For gated datasets, get your token from https://huggingface.co/settings/tokens
headers = {{"Authorization": "Bearer YOUR_HF_TOKEN"}}
jsonld_url = "{jsonld_url}"
jsonld = requests.get(jsonld_url, headers=headers).json()
ds = CroissantDataset(jsonld=jsonld)
records = ds.records("default") # This is a generator

# To preview data:
preview_rows = [row for _, row in zip(range(100), records)]
df = pd.DataFrame(preview_rows)
print(df.head())
"""

        # --- 🔍 Universal Search Logic ---
        if query and not df.empty:
            if dataset_key == 'medical':
                 df = df[df['conversations'].apply(lambda x: isinstance(x, list) and len(x) > 1 and query.lower() in str(x[1].get('value', '')).lower())]
            else:
                combined_mask = pd.Series([False] * len(df))
                for col in search_cols:
                    if col in df.columns and pd.api.types.is_string_dtype(df[col]):
                        combined_mask |= df[col].str.contains(query, case=False, na=False)
                df = df[combined_mask]

        # --- 🖼️ Special Image Handling ---
        gallery_output = None
        if dataset_key == 'inscene' and not df.empty:
            gallery_data = []
            for _, row in df.iterrows():
                if isinstance(row.get('image'), Image.Image):
                    gallery_data.append((row['image'], row.get('text', '')))
            gallery_output = gr.Gallery(gallery_data, label="🖼️ Image Results", height=400)

        md, csv, xlsx, tab = dataframe_to_outputs(df)
        return df, gallery_output, md, csv, xlsx, tab, code_snippet

    except Exception as e:
        return handle_error(e)


# --- 🖼️ UI Generation ---

def create_dataset_tab(dataset_key: str):
    """
    🏗️ This function builds a whole tab in our UI for a single dataset.
    It's like a little construction worker for Gradio interfaces.
    """
    config = DATASET_CONFIG[dataset_key]
    
    with gr.Tab(f"{config['emoji']} {dataset_key.capitalize()}"):
        gr.Markdown(f"## {config['emoji']} Query the `{config['name']}` Dataset")
        if not config['is_public']:
            gr.Markdown("**Note:** This is a gated dataset. Please log in via `huggingface-cli login` in your terminal first.")
        
        with gr.Row():
            access_method = gr.Radio(config['methods'], label="🔑 Access Method", value=config['methods'][0])
            query = gr.Textbox(label="🔍 Search Query", placeholder="Enter a keyword to search...")
        
        fetch_button = gr.Button("🚀 Go Fetch!")

        df_output = gr.DataFrame(label="📊 Results DataFrame", interactive=False, wrap=True)
        gallery_output = gr.Gallery(visible=(dataset_key == 'inscene'), label="🖼️ Image Results")

        with gr.Accordion("📂 View/Export Full Results", open=False):
            markdown_output = gr.Markdown(label="📝 Markdown View")
            with gr.Row():
                csv_output = gr.File(label="⬇️ Download CSV")
                xlsx_output = gr.File(label="⬇️ Download XLSX")
            # CHANGED: Removed the language parameter entirely for maximum compatibility.
            copy_output = gr.Code(label="📋 Copy-Paste (Tab-Delimited)")
        
        code_output = gr.Code(label="💻 Python Code Snippet", language="python")
        
        fetch_button.click(
            fn=fetch_data,
            inputs=[gr.State(dataset_key), access_method, query],
            outputs=[df_output, gallery_output, markdown_output, csv_output, xlsx_output, copy_output, code_output]
        )

# --- 🚀 Main App ---
with gr.Blocks(theme=gr.themes.Soft(), title="Hugging Face Dataset Explorer") as demo:
    gr.Markdown("# 🤗 Hugging Face Dataset Explorer")
    gr.Markdown(
        "Select a dataset, choose an access method, type a query, and see the results instantly. "
        "The app demonstrates various ways to access and search Hugging Face datasets and generates the code for you!"
    )

    with gr.Tabs():
        for key in DATASET_CONFIG.keys():
            create_dataset_tab(key)

if __name__ == "__main__":
    demo.launch(debug=True)