Fresh push: Dataset Insight Portal with Parquet files via LFS

all_minimal_dataset_cards.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5adb59f94fb6f08f5c0859e21e55ed56ec40f40d9cde349427bf24065e775d60
+size 17318878

all_rich_dataset_cards.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:94ac600eb5100aa7acaeeec3d05becbee7ac11eba9595a0f9e38286879285349
+size 5475858

app.py

@@ -0,0 +1,107 @@
+import gradio as gr
+import polars as pl
+
+# Paths or HF Hub URLs for Parquet files
+RICH_PARQUET_PATH = "all_rich_dataset_cards.parquet"
+MISSING_PARQUET_PATH = "all_minimal_dataset_cards.parquet"
+
+ROWS_PER_PAGE = 50
+
+# Lazy load datasets
+lazy_rich = pl.scan_parquet(RICH_PARQUET_PATH)
+lazy_missing = pl.scan_parquet(MISSING_PARQUET_PATH)
+
+current_lazy_df = lazy_missing  # Default dataset
+
+# Helper function to fetch a page
+def get_page(lazy_df: pl.LazyFrame, page: int, column: str = None, query: str = ""):
+    filtered_df = lazy_df
+    if column and query:
+        query_lower = query.lower().strip()
+        # Case-insensitive search
+        filtered_df = filtered_df.with_columns([
+            pl.col(column).cast(pl.Utf8).str.to_lowercase().alias(column)
+        ]).filter(pl.col(column).str.contains(query_lower, literal=False))
+    start = page * ROWS_PER_PAGE
+    page_df = filtered_df.slice(start, ROWS_PER_PAGE).collect().to_pandas()
+    total_rows = filtered_df.collect().height
+    total_pages = (total_rows - 1) // ROWS_PER_PAGE + 1
+    return page_df, total_pages
+
+# Initialize first page
+initial_df, total_pages = get_page(current_lazy_df, 0)
+columns = list(initial_df.columns)
+
+with gr.Blocks() as demo:
+    gr.Markdown("## Dataset Insight Portal")
+
+    # Dataset selection
+    dataset_select = gr.Dropdown(
+        choices=["DatasetCards rich in information", "DatasetCards missing information"],
+        value="DatasetCards missing information",
+        label="Select Dataset"
+    )
+
+    # Pagination controls
+    with gr.Row():
+        prev_btn = gr.Button("Previous", elem_id="small-btn")
+        next_btn = gr.Button("Next", elem_id="small-btn")
+        page_number = gr.Number(value=0, label="Page", precision=0)
+        total_pages_display = gr.Label(value=f"Total Pages: {total_pages}")
+
+    # Data table
+    data_table = gr.Dataframe(
+        value=initial_df, headers=columns, datatype="str",
+        interactive=False, row_count=ROWS_PER_PAGE
+    )
+
+    # Column search
+    with gr.Row():
+        col_dropdown = gr.Dropdown(choices=columns, label="Column")
+        search_text = gr.Textbox(label="Search")
+        search_btn = gr.Button("Search", elem_id="small-btn")
+        reset_btn = gr.Button("Reset", elem_id="small-btn")
+
+    # --- Functions ---
+    def load_dataset(dataset_choice):
+        global current_lazy_df
+        current_lazy_df = lazy_rich if dataset_choice == "DatasetCards rich in information" else lazy_missing
+        initial_df, total_pages = get_page(current_lazy_df, 0)
+        columns = list(initial_df.columns)
+        return (
+            gr.update(value=initial_df, headers=columns),
+            f"Total Pages: {total_pages}",
+            0,
+            gr.update(choices=columns, value=columns[0])
+        )
+
+    def next_page_func(page, column, query):
+        page += 1
+        page_df, total_pages = get_page(current_lazy_df, page, column, query)
+        if page >= total_pages:
+            page = total_pages - 1
+            page_df, total_pages = get_page(current_lazy_df, page, column, query)
+        return page_df, f"Total Pages: {total_pages}", page
+
+    def prev_page_func(page, column, query):
+        page -= 1
+        page = max(0, page)
+        page_df, total_pages = get_page(current_lazy_df, page, column, query)
+        return page_df, f"Total Pages: {total_pages}", page
+
+    def search_func(column, query):
+        page_df, total_pages = get_page(current_lazy_df, 0, column, query)
+        return page_df, f"Total Pages: {total_pages}", 0
+
+    def reset_func():
+        page_df, total_pages = get_page(current_lazy_df, 0)
+        return page_df, f"Total Pages: {total_pages}", 0
+
+    # --- Event Listeners ---
+    dataset_select.change(load_dataset, dataset_select, [data_table, total_pages_display, page_number, col_dropdown])
+    next_btn.click(next_page_func, [page_number, col_dropdown, search_text], [data_table, total_pages_display, page_number])
+    prev_btn.click(prev_page_func, [page_number, col_dropdown, search_text], [data_table, total_pages_display, page_number])
+    search_btn.click(search_func, [col_dropdown, search_text], [data_table, total_pages_display, page_number])
+    reset_btn.click(reset_func, [], [data_table, total_pages_display, page_number])
+
+demo.launch()