extracted excel data in text_as_html have negative values

[pradyrk]

Describe the bug
I have a xlsx , read using partition_xlsx and parsing the text_as_html , I can find negative values whereas excel doesnt have any negative value

To Reproduce
from unstructured.partition.xlsx import partition_xlsx

elements = partition_xlsx(filename="excelfilepath")
print(elements[0].metadata.text_as_html)

Expected behavior
Values should be positive and exact value needs to be extracted

Screenshots

Environment Info
Databricks cluster - 16.1 ML Runtime - Complete details - https://docs.databricks.com/aws/en/release-notes/runtime/16.1ml

Additional context
file is sensitive , wouldnt be able to share the actual file , providing any direction to resolve this can help

[pradyrk]

If we notice 98 is the value in the sheet was extracted as -97.90044 , decimal value is still fine but negative value is not acceptable

[aswinjoseroy]

Maybe try reading the excel through a normal library such as pandas once to verify if the actual data is not different to what's being displayed. Excel can be weird with it's data

[pradyrk]

thats my bad , its not a parser issue, in financial terms "(10k)" means(value within bracket) , the values are negative , the LLM conversion is the cause . Closing the ticket, thanks