mirrors/docling
1
0
Fork 0
mirror of https://github.com/DS4SD/docling.git synced 2025-07-31 22:44:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

reformatted the code

Browse Source 
Signed-off-by: Peter Staar <taa@zurich.ibm.com>
This commit is contained in:
Peter Staar 2024-11-19 11:22:29 +01:00
parent f837105a09
commit d23aea981d
1 changed files with 31 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

59
docling/backend/msexcel_backend.py
View File

@ -177,8 +177,8 @@ class MsExcelDocumentBackend(DeclarativeDocumentBackend):
""" """
Find all compact rectangular data tables in a sheet. Find all compact rectangular data tables in a sheet.
""" """
#_log.info("find_data_tables") # _log.info("find_data_tables")
tables = [] # List to store found tables tables = [] # List to store found tables
visited: set[Tuple[int, int]] = set() # Track already visited cells visited: set[Tuple[int, int]] = set() # Track already visited cells
@ -199,7 +199,7 @@ class MsExcelDocumentBackend(DeclarativeDocumentBackend):
tables.append(table_bounds) tables.append(table_bounds)
return tables return tables
def _find_table_bounds( def _find_table_bounds(
self, self,
sheet: Worksheet, sheet: Worksheet,
@ -217,25 +217,29 @@ class MsExcelDocumentBackend(DeclarativeDocumentBackend):
max_row = self._find_table_bottom(sheet, start_row, start_col) max_row = self._find_table_bottom(sheet, start_row, start_col)
max_col = self._find_table_right(sheet, start_row, start_col) max_col = self._find_table_right(sheet, start_row, start_col)
# Collect the data within the bounds # Collect the data within the bounds
data = [] data = []
visited_cells = set() visited_cells = set()
for ri in range(start_row, max_row + 1): for ri in range(start_row, max_row + 1):
for rj in range(start_col, max_col + 1): for rj in range(start_col, max_col + 1):
cell = sheet.cell(row=ri + 1, column=rj + 1) # 1-based indexing cell = sheet.cell(row=ri + 1, column=rj + 1) # 1-based indexing
# Check if the cell belongs to a merged range # Check if the cell belongs to a merged range
row_span = 1 row_span = 1
col_span = 1 col_span = 1
#_log.info(sheet.merged_cells.ranges) # _log.info(sheet.merged_cells.ranges)
for merged_range in sheet.merged_cells.ranges: for merged_range in sheet.merged_cells.ranges:
if merged_range.min_row<=ri+1 and ri+1<=merged_range.max_row and \ if (
merged_range.min_col<=rj+1 and rj+1<=merged_range.max_col: merged_range.min_row <= ri + 1
and ri + 1 <= merged_range.max_row
and merged_range.min_col <= rj + 1
and rj + 1 <= merged_range.max_col
):
row_span = merged_range.max_row - merged_range.min_row + 1 row_span = merged_range.max_row - merged_range.min_row + 1
col_span = merged_range.max_col - merged_range.min_col + 1 col_span = merged_range.max_col - merged_range.min_col + 1
break break
@ -243,15 +247,15 @@ class MsExcelDocumentBackend(DeclarativeDocumentBackend):
if (ri, rj) not in visited_cells: if (ri, rj) not in visited_cells:
data.append( data.append(
ExcelCell( ExcelCell(
row = ri - start_row, row=ri - start_row,
col = rj - start_col, col=rj - start_col,
text=str(cell.value), text=str(cell.value),
row_span=row_span, row_span=row_span,
col_span=col_span, col_span=col_span,
) )
) )
# _log.info(f"cell: {ri}, {rj} -> {ri - start_row}, {rj - start_col}, {row_span}, {col_span}: {str(cell.value)}") # _log.info(f"cell: {ri}, {rj} -> {ri - start_row}, {rj - start_col}, {row_span}, {col_span}: {str(cell.value)}")
# Mark all cells in the span as visited # Mark all cells in the span as visited
for span_row in range(ri, ri + row_span): for span_row in range(ri, ri + row_span):
for span_col in range(rj, rj + col_span): for span_col in range(rj, rj + col_span):
@ -266,41 +270,41 @@ class MsExcelDocumentBackend(DeclarativeDocumentBackend):
visited_cells, visited_cells,
) )
def _find_table_bottom(self, sheet: Worksheet, start_row:int, start_col:int): def _find_table_bottom(self, sheet: Worksheet, start_row: int, start_col: int):
"""Function to find the bottom boundary of the table""" """Function to find the bottom boundary of the table"""
max_row = start_row max_row = start_row
while max_row < sheet.max_row - 1: while max_row < sheet.max_row - 1:
# Get the cell value or check if it is part of a merged cell # Get the cell value or check if it is part of a merged cell
cell = sheet.cell(row=max_row + 2, column=start_col + 1) cell = sheet.cell(row=max_row + 2, column=start_col + 1)
# Check if the cell is part of a merged range # Check if the cell is part of a merged range
merged_range = next( merged_range = next(
(mr for mr in sheet.merged_cells.ranges if cell.coordinate in mr), (mr for mr in sheet.merged_cells.ranges if cell.coordinate in mr),
None, None,
) )
if cell.value is None and not merged_range: if cell.value is None and not merged_range:
break # Stop if the cell is empty and not merged break # Stop if the cell is empty and not merged
# Expand max_row to include the merged range if applicable # Expand max_row to include the merged range if applicable
if merged_range: if merged_range:
max_row = max(max_row, merged_range.max_row-1) max_row = max(max_row, merged_range.max_row - 1)
else: else:
max_row += 1 max_row += 1
return max_row return max_row
def _find_table_right(self, sheet: Worksheet, start_row:int, start_col:int): def _find_table_right(self, sheet: Worksheet, start_row: int, start_col: int):
"""Function to find the right boundary of the table""" """Function to find the right boundary of the table"""
max_col = start_col max_col = start_col
while max_col < sheet.max_column - 1: while max_col < sheet.max_column - 1:
# Get the cell value or check if it is part of a merged cell # Get the cell value or check if it is part of a merged cell
cell = sheet.cell(row=start_row + 1, column=max_col + 2) cell = sheet.cell(row=start_row + 1, column=max_col + 2)
# Check if the cell is part of a merged range # Check if the cell is part of a merged range
merged_range = next( merged_range = next(
(mr for mr in sheet.merged_cells.ranges if cell.coordinate in mr), (mr for mr in sheet.merged_cells.ranges if cell.coordinate in mr),
@ -309,16 +313,15 @@ class MsExcelDocumentBackend(DeclarativeDocumentBackend):
if cell.value is None and not merged_range: if cell.value is None and not merged_range:
break # Stop if the cell is empty and not merged break # Stop if the cell is empty and not merged
# Expand max_col to include the merged range if applicable # Expand max_col to include the merged range if applicable
if merged_range: if merged_range:
max_col = max(max_col, merged_range.max_col-1) max_col = max(max_col, merged_range.max_col - 1)
else: else:
max_col += 1 max_col += 1
return max_col return max_col
def _find_images_in_sheet( def _find_images_in_sheet(
self, doc: DoclingDocument, sheet: Worksheet self, doc: DoclingDocument, sheet: Worksheet
) -> DoclingDocument: ) -> DoclingDocument: