Update data_clean_simple.py

data_clean_simple.py

@@ -101,23 +101,26 @@ def clean_data(file_path, progress_callback=None):
         
         # Generate suggestions for missing or weird values
         if null_count > 0 or empty_str_count > 0 or pattern_matches > 0:
-            # Get examples for analysis (both good and bad examples)
-            # Get non-null, non-empty examples
-            good_examples = df[col][df[col].notnull() & (df[col] != '')].drop_duplicates().sample(n=min(5, len(df)), random_state=1)
-
-            # Get bad examples
-            bad_examples = df[col][df[col].isna() | (df[col] == '') | df[col].astype(str).str.contains(r'none|null|n/a|na|\?+missing|unknown', 
-                                                                                                       na=False, case=False, regex=True)].sample(
-                                                                                                           min(5, df[col].isna().sum()), random_state=1)
-            # Combine good and bad examples
-            examples = good_examples + bad_examples
-
-            if examples is not None:
-                suggestion = suggest_fill_strategies(col, examples)
-                suggestions_log.append({
-                    'col': col,
-                    'suggestion': suggestion
-                })
+            # Get non-null, non-empty examples for analysis - handle empty dataframes
+            try:
+                good_df = df[col][
+                df[col].notnull() &
+                (df[col].astype(str) != "") &
+                ~df[col].astype(str).contains(r'none|null|n/a|na|\?+|missing|unknown',
+                                             na=False, case=False, regex=True)
+                ]
+
+                if len(good_df) > 0:
+                    sample_size = min(5, len(good_df))
+                    good_examples = good_df.drop_duplicates().sample(n=sample_size, random_state=1).tolist()
+                else:
+                    good_examples = []
+            except:
+                good_examples = []
+
+            # Generate suggestions
+            suggestion = suggest_fill_strategies(col, good_examples)
+            suggestions_log.append((col, suggestion))
 
     # Final progress update
     if progress_callback: