LlamaParse timing out on XLSX files. XLS versions work #466

Closed
opened 2026-02-16 00:17:56 -05:00 by yindo · 3 comments
Owner

Originally created by @alihussain5 on GitHub (May 6, 2025).

Originally assigned to: @pmrowla on GitHub.

Describe the bug
Hello, we've been using LlamaParse to parse XLSX files, however, as of ~1 month ago, we noticed that the XLSX files we were sending were timing out on LlamaParse. However, converting the same file to XLS works fine.

So, as a workaround, we've been converting all our XLSX to XLS before sending to LlamaParse.

The error on the Frontend is:

UNKNOWN_ERROR: DOCUMENT_CONVERSION_ERROR - Error while converting spreadsheet.

When clicking More details, the page hangs:

Image

Files

Job ID
92385c58-948b-4338-821e-e79742fcd9db

Client:
Please remove untested options:

  • API
  • Frontend (cloud.llamaindex.ai)
  • Typescript Library

Additional context
We use auto mode with Trigger Auto mode on Table

Originally created by @alihussain5 on GitHub (May 6, 2025). Originally assigned to: @pmrowla on GitHub. **Describe the bug** Hello, we've been using LlamaParse to parse XLSX files, however, as of ~1 month ago, we noticed that the XLSX files we were sending were timing out on LlamaParse. However, converting the same file to XLS works fine. So, as a workaround, we've been converting all our XLSX to XLS before sending to LlamaParse. The error on the Frontend is: ``` UNKNOWN_ERROR: DOCUMENT_CONVERSION_ERROR - Error while converting spreadsheet. ``` When clicking `More details`, the page hangs: <img width="702" alt="Image" src="https://github.com/user-attachments/assets/a5eba636-5e22-4b03-825b-3b98428b9b24" /> **Files** **Job ID** 92385c58-948b-4338-821e-e79742fcd9db **Client:** Please remove untested options: - API - Frontend (cloud.llamaindex.ai) - Typescript Library **Additional context** We use `auto mode` with `Trigger Auto mode on Table`
yindo added the bug label 2026-02-16 00:17:56 -05:00
yindo closed this issue 2026-02-16 00:17:56 -05:00
Author
Owner

@pmrowla commented on GitHub (May 7, 2025):

Sheet 2 in the XLSX contains over a million (mostly empty) rows which ends up causing the slow parsing and eventual timeout. Since XLS has a maximum size of 65536 rows, when you convert to XLS the spreadsheet is truncated to a size that gets parsed in a reasonable amount of time. Depending on how you are doing the conversion, it may also be truncating the empty rows and columns entirely, which would further improve the parsing time.

I'll continue looking into this on our end, there should be something we can do to improve detecting when the remainder of rows/columns in a worksheet are empty and then skip parsing them.

In the meantime, another potential workaround would be to truncate the extra empty rows and columns in your XLSX spreadsheets, which would let you keep the files in XLSX format and also result in LlamaParse being able to parse the documents without timing out.

@pmrowla commented on GitHub (May 7, 2025): Sheet 2 in the XLSX contains over a million (mostly empty) rows which ends up causing the slow parsing and eventual timeout. Since XLS has a maximum size of 65536 rows, when you convert to XLS the spreadsheet is truncated to a size that gets parsed in a reasonable amount of time. Depending on how you are doing the conversion, it may also be truncating the empty rows and columns entirely, which would further improve the parsing time. I'll continue looking into this on our end, there should be something we can do to improve detecting when the remainder of rows/columns in a worksheet are empty and then skip parsing them. In the meantime, another potential workaround would be to truncate the extra empty rows and columns in your XLSX spreadsheets, which would let you keep the files in XLSX format and also result in LlamaParse being able to parse the documents without timing out.
Author
Owner

@alihussain5 commented on GitHub (May 7, 2025):

Thank you @pmrowla! Silly debugging oversight on my part. I saw it work with XLS files and immediately stopped there.

I can confirm truncating the XLSX works. We're doing this with SheetJS' xlsx package by just reading and writing it as a buffer:

      const workbook = xlsx.read(data, {
        type: 'array',
      });
      const trimmedWorkbook = xlsx.write(workbook, {
        bookType: 'xlsx',
        type: 'buffer',
      }) as ArrayBuffer;

      return new Uint8Array(trimmedWorkbook);
@alihussain5 commented on GitHub (May 7, 2025): Thank you @pmrowla! Silly debugging oversight on my part. I saw it work with XLS files and immediately stopped there. I can confirm truncating the XLSX works. We're doing this with SheetJS' `xlsx` package by just reading and writing it as a buffer: ```javascript const workbook = xlsx.read(data, { type: 'array', }); const trimmedWorkbook = xlsx.write(workbook, { bookType: 'xlsx', type: 'buffer', }) as ArrayBuffer; return new Uint8Array(trimmedWorkbook); ```
Author
Owner

@pmrowla commented on GitHub (May 9, 2025):

A fix to handle this case on our end has been deployed on cloud.llamaindex.ai

@pmrowla commented on GitHub (May 9, 2025): A fix to handle this case on our end has been deployed on cloud.llamaindex.ai
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: run-llama/llama_cloud_services#466