"""Logic for the **Debug Data** tab."""
from __future__ import annotations
from .state import app_state
__all__ = ["debug_data_structure"]
def debug_data_structure() -> str:
if app_state["clustered_df"] is None:
return "❌ No data loaded
"
df = app_state["clustered_df"]
n_rows = len(df)
n_cols = len(df.columns)
# Check for both naming patterns
has_fine_clusters = ("property_description_fine_cluster_id" in df.columns or
"fine_cluster_id" in df.columns)
has_coarse_clusters = ("property_description_coarse_cluster_id" in df.columns or
"coarse_cluster_id" in df.columns)
sample_rows = min(3, len(df))
sample_data = df.head(sample_rows).to_html(
escape=False,
classes="table table-striped",
table_id="debug-table",
)
html = f"""
🐛 Data Structure Debug Info
Basic Statistics
- Rows: {n_rows:,}
- Columns: {n_cols}
- Fine Clusters Available: {'✅ Yes' if has_fine_clusters else '❌ No'}
- Coarse Clusters Available: {'✅ Yes' if has_coarse_clusters else '❌ No'}
Available Columns
"""
for col in sorted(df.columns):
unique_values = df[col].nunique() if df[col].dtype == "object" else "N/A"
html += f"{col} - {df[col].dtype} (unique values: {unique_values})
Sample Data (First {sample_rows} rows)
{sample_data}
"""
return html