in chunking/chunkers/spreadsheet_chunker.py [0:0]
def _clean_markdown_table(self, table_str):
"""
Cleans up a Markdown table string by removing excessive whitespace from each cell.
Args:
table_str (str): The Markdown table string to be cleaned.
Returns:
str: The cleaned Markdown table string with reduced whitespace.
"""
cleaned_lines = []
lines = table_str.splitlines()
for line in lines:
if set(line.strip()) <= set('-| '):
cleaned_lines.append(line)
continue
cells = line.split('|')
stripped_cells = [cell.strip() for cell in cells[1:-1]]
cleaned_line = '| ' + ' | '.join(stripped_cells) + ' |'
cleaned_lines.append(cleaned_line)
return '\n'.join(cleaned_lines)