Remove xlwings for hugging face

apps/dump_compare.py

@@ -4,8 +4,8 @@ import tempfile
 
 import pandas as pd
 import streamlit as st
-import xlwings as xw
-from pyxlsb import open_workbook
+
+# import xlwings as xw
 
 # === Core Logic ===
 
@@ -18,29 +18,33 @@ def find_header_row(df, keyword="Dist_Name"):
     raise ValueError(f"No row with '{keyword}' found.")
 
 
-def read_xlsb_with_pyxlsb(file, sheet):
-    rows = []
-    with open_workbook(file) as wb:
-        with wb.get_sheet(sheet) as s:
-            for row in s.rows():
-                rows.append([item.v for item in row])
-    return pd.DataFrame(rows)
+# import xlwings as xw
+# def read_sheet_fallback(file, sheet):
+#     try:
+#         app = xw.App(visible=False)
+#         book = app.books.open(file)
+#         sht = book.sheets[sheet]
+#         df = sht.used_range.options(pd.DataFrame, header=False, index=False).value
+#         book.close()
+#         app.quit()
+#         return df
+#     except Exception as e2:
+#         raise RuntimeError(f"xlwings failed: {e2}")
 
 
 def read_sheet_fallback(file, sheet):
     try:
-        return read_xlsb_with_pyxlsb(file, sheet)
-    except Exception:
-        try:
-            app = xw.App(visible=False)
-            book = app.books.open(file)
-            sht = book.sheets[sheet]
-            df = sht.used_range.options(pd.DataFrame, header=False, index=False).value
-            book.close()
-            app.quit()
-            return df
-        except Exception as e2:
-            raise RuntimeError(f"xlwings failed: {e2}")
+        # pandas can directly read Excel files
+        # sheet_name can be the sheet name (string) or sheet number (0-indexed)
+        df = pd.read_excel(file, sheet_name=sheet, header=None, engine="calamine")
+        return df
+    except FileNotFoundError:
+        raise FileNotFoundError(f"The file '{file}' was not found.")
+    except ValueError as e:
+        # This could happen if the sheet doesn't exist, or other pandas-related errors
+        raise ValueError(f"Error reading sheet '{sheet}' from '{file}': {e}")
+    except Exception as e:
+        raise RuntimeError(f"An unexpected error occurred: {e}")
 
 
 def load_clean_df(file, sheet):