ENH: Allow safe access to .book in ExcelWriter (#45687)

Author: rhshadrach

pandas/io/excel/_base.py

@@ -1048,6 +1048,12 @@ def engine(self) -> str:
  """Name of engine."""
  pass
 
+ @property
+ @abc.abstractmethod
+ def sheets(self) -> dict[str, Any]:
+ """Mapping of sheet names to sheet objects."""
+ pass
+
  @abc.abstractmethod
  def write_cells(
  self,
@@ -1112,7 +1118,6 @@ def __init__(
  self.handles = get_handle(
  path, mode, storage_options=storage_options, is_text=False
  )
- self.sheets: dict[str, Any] = {}
  self.cur_sheet = None
 
  if date_format is None:

pandas/io/excel/_odswriter.py

@@ -58,6 +58,17 @@ def __init__(
  self.book = OpenDocumentSpreadsheet(**engine_kwargs)
  self._style_dict: dict[str, str] = {}
 
+ @property
+ def sheets(self) -> dict[str, Any]:
+ """Mapping of sheet names to sheet objects."""
+ from odf.table import Table
+
+ result = {
+ sheet.getAttribute("name"): sheet
+ for sheet in self.book.getElementsByType(Table)
+ }
+ return result
+
  def save(self) -> None:
  """
  Save workbook to disk.
@@ -91,7 +102,7 @@ def write_cells(
  wks = self.sheets[sheet_name]
  else:
  wks = Table(name=sheet_name)
- self.sheets[sheet_name] = wks
+ self.book.spreadsheet.addElement(wks)
 
  if validate_freeze_panes(freeze_panes):
  freeze_panes = cast(Tuple[int, int], freeze_panes)

pandas/io/excel/_openpyxl.py

@@ -68,15 +68,19 @@ def __init__(
 
  self.book = load_workbook(self.handles.handle, **engine_kwargs)
  self.handles.handle.seek(0)
- self.sheets = {name: self.book[name] for name in self.book.sheetnames}
-
  else:
  # Create workbook object with default optimized_write=True.
  self.book = Workbook(**engine_kwargs)
 
  if self.book.worksheets:
  self.book.remove(self.book.worksheets[0])
 
+ @property
+ def sheets(self) -> dict[str, Any]:
+ """Mapping of sheet names to sheet objects."""
+ result = {name: self.book[name] for name in self.book.sheetnames}
+ return result
+
  def save(self) -> None:
  """
  Save workbook to disk.
@@ -440,7 +444,6 @@ def write_cells(
  target_index = self.book.index(old_wks)
  del self.book[sheet_name]
  wks = self.book.create_sheet(sheet_name, target_index)
- self.sheets[sheet_name] = wks
  elif self.if_sheet_exists == "error":
  raise ValueError(
  f"Sheet '{sheet_name}' already exists and "
@@ -458,7 +461,6 @@ def write_cells(
  else:
  wks = self.book.create_sheet()
  wks.title = sheet_name
- self.sheets[sheet_name] = wks
 
  if validate_freeze_panes(freeze_panes):
  freeze_panes = cast(Tuple[int, int], freeze_panes)

pandas/io/excel/_xlsxwriter.py

@@ -205,6 +205,11 @@ def __init__(
 
  self.book = Workbook(self.handles.handle, **engine_kwargs)
 
+ @property
+ def sheets(self) -> dict[str, Any]:
+ result = self.book.sheetnames
+ return result
+
  def save(self) -> None:
  """
  Save workbook to disk.
@@ -222,11 +227,9 @@ def write_cells(
  # Write the frame cells using xlsxwriter.
  sheet_name = self._get_sheet_name(sheet_name)
 
- if sheet_name in self.sheets:
- wks = self.sheets[sheet_name]
- else:
+ wks = self.book.get_worksheet_by_name(sheet_name)
+ if wks is None:
  wks = self.book.add_worksheet(sheet_name)
- self.sheets[sheet_name] = wks
 
  style_dict = {"null": None}
 

pandas/io/excel/_xlwt.py

@@ -63,6 +63,12 @@ def __init__(
  self.fm_datetime = xlwt.easyxf(num_format_str=self.datetime_format)
  self.fm_date = xlwt.easyxf(num_format_str=self.date_format)
 
+ @property
+ def sheets(self) -> dict[str, Any]:
+ """Mapping of sheet names to sheet objects."""
+ result = {sheet.name: sheet for sheet in self.book._Workbook__worksheets}
+ return result
+
  def save(self) -> None:
  """
  Save workbook to disk.

pandas/tests/io/excel/test_odswriter.py

@@ -56,3 +56,13 @@ def test_engine_kwargs(ext, engine_kwargs):
  else:
  with ExcelWriter(f, engine="odf", engine_kwargs=engine_kwargs) as _:
  pass
+
+
+def test_book_and_sheets_consistent(ext):
+ # GH#45687 - Ensure sheets is updated if user modifies book
+ with tm.ensure_clean(ext) as f:
+ with ExcelWriter(f) as writer:
+ assert writer.sheets == {}
+ table = odf.table.Table(name="test_name")
+ writer.book.spreadsheet.addElement(table)
+ assert writer.sheets == {"test_name": table}

pandas/tests/io/excel/test_openpyxl.py

@@ -379,3 +379,12 @@ def test_read_empty_with_blank_row(datapath, ext, read_only):
  result = pd.read_excel(wb, engine="openpyxl")
  expected = DataFrame()
  tm.assert_frame_equal(result, expected)
+
+
+def test_book_and_sheets_consistent(ext):
+ # GH#45687 - Ensure sheets is updated if user modifies book
+ with tm.ensure_clean(ext) as f:
+ with ExcelWriter(f, engine="openpyxl") as writer:
+ assert writer.sheets == {}
+ sheet = writer.book.create_sheet("test_name", 0)
+ assert writer.sheets == {"test_name": sheet}

pandas/tests/io/excel/test_writers.py

@@ -1271,10 +1271,12 @@ def test_register_writer(self):
  # some awkward mocking to test out dispatch and such actually works
  called_save = []
  called_write_cells = []
+ called_sheets = []
 
  class DummyClass(ExcelWriter):
  called_save = False
  called_write_cells = False
+ called_sheets = False
  supported_extensions = ["xlsx", "xls"]
  engine = "dummy"
 
@@ -1284,12 +1286,18 @@ def save(self):
  def write_cells(self, *args, **kwargs):
  called_write_cells.append(True)
 
+ @property
+ def sheets(self):
+ called_sheets.append(True)
+
  def check_called(func):
  func()
  assert len(called_save) >= 1
  assert len(called_write_cells) >= 1
+ assert len(called_sheets) == 0
  del called_save[:]
  del called_write_cells[:]
+ del called_sheets[:]
 
  with option_context("io.excel.xlsx.writer", "dummy"):
  path = "something.xlsx"

pandas/tests/io/excel/test_xlsxwriter.py

@@ -83,3 +83,12 @@ def test_engine_kwargs(ext, nan_inf_to_errors):
  with tm.ensure_clean(ext) as f:
  with ExcelWriter(f, engine="xlsxwriter", engine_kwargs=engine_kwargs) as writer:
  assert writer.book.nan_inf_to_errors == nan_inf_to_errors
+
+
+def test_book_and_sheets_consistent(ext):
+ # GH#45687 - Ensure sheets is updated if user modifies book
+ with tm.ensure_clean(ext) as f:
+ with ExcelWriter(f, engine="xlsxwriter") as writer:
+ assert writer.sheets == {}
+ sheet = writer.book.add_worksheet("test_name")
+ assert writer.sheets == {"test_name": sheet}

pandas/tests/io/excel/test_xlwt.py

@@ -125,3 +125,12 @@ def test_engine_kwargs(ext, style_compression):
  assert writer.book._Workbook__styles.style_compression == style_compression
  # xlwt won't allow us to close without writing something
  DataFrame().to_excel(writer)
+
+
+def test_book_and_sheets_consistent(ext):
+ # GH#45687 - Ensure sheets is updated if user modifies book
+ with tm.ensure_clean(ext) as f:
+ with ExcelWriter(f) as writer:
+ assert writer.sheets == {}
+ sheet = writer.book.add_sheet("test_name")
+ assert writer.sheets == {"test_name": sheet}