run-llama/llama_cloud_services
1
0
Fork 0
mirror of https://github.com/run-llama/llama_cloud_services.git synced 2026-08-26 16:56:36 -04:00
Code Issues 352 Packages Projects Releases 10 Wiki Activity

Support for understanding tables spanning multiple pages #189

New Issue
Closed
opened 2026-02-16 00:17:05 -05:00 by yindo · 7 comments
yindo commented 2026-02-16 00:17:05 -05:00
Owner
Copy Link
Copy Source

Originally created by @galvangoh on GitHub (Jul 15, 2024).

It is mentioned in the LlamaParse UI that parsing instructions are only applied to each page of the document. It would be great if LlamaParse could provide additional support for understanding tables that span multiple pages.

I am currently facing an issue with LlamaParse while parsing a table where its content spans multiple pages. The table contains itemized line items from an invoice, each following a structured pattern. However, at the bottom of the page, this pattern gets interrupted and continues on the next page. Since LlamaParse processes each page independently, it fails to recognize the continuation of the pattern and interprets it as a new one. Consequently, this sometimes leads to the removal of texts in the table.

Thank you.

Originally created by @galvangoh on GitHub (Jul 15, 2024). It is mentioned in the LlamaParse UI that parsing instructions are only applied to each page of the document. It would be great if LlamaParse could provide additional support for understanding tables that span multiple pages. I am currently facing an issue with LlamaParse while parsing a table where its content spans multiple pages. The table contains itemized line items from an invoice, each following a structured pattern. However, at the bottom of the page, this pattern gets interrupted and continues on the next page. Since LlamaParse processes each page independently, it fails to recognize the continuation of the pattern and interprets it as a new one. Consequently, this sometimes leads to the removal of texts in the table. Thank you.
yindo added the enhancement label 2026-02-16 00:17:05 -05:00
yindo closed this issue 2026-02-16 00:17:05 -05:00
yindo commented 2026-02-16 00:17:05 -05:00
Author
Owner
Copy Link
Copy Source

@tkcoding commented on GitHub (Jul 16, 2024):

@galvangoh
This is interesting , perhaps you can show some example.
The quick way to go around this is you can use parsing_instructions to specifically tell table that have continuation on multiple pages should consolidate into one table (I guess you will get like two tables in different pages with the same title).

If you are using this table and store the table into a vector DB , i suggest you can ask llama-parse with parsing_instruction to describe the table content .

Hope this helps for a quick "fix" here.

@tkcoding commented on GitHub (Jul 16, 2024): @galvangoh This is interesting , perhaps you can show some example. The quick way to go around this is you can use parsing_instructions to specifically tell table that have continuation on multiple pages should consolidate into one table (I guess you will get like two tables in different pages with the same title). If you are using this table and store the table into a vector DB , i suggest you can ask llama-parse with parsing_instruction to describe the table content . Hope this helps for a quick "fix" here.
yindo commented 2026-02-16 00:17:05 -05:00
Author
Owner
Copy Link
Copy Source

@galvangoh commented on GitHub (Jul 16, 2024):

@tkcoding thanks for the reply.

The quick way to go around this is you can use parsing_instructions to specifically tell table that have continuation on multiple pages should consolidate into one table (I guess you will get like two tables in different pages with the same title).

I have tried to specify it in parsing_instruction but since its being instructed to the page individually, the parsing result still contain disjoint tables. I think LlamaParse sees pages of document as just pages instead of page 1 of x, page 2 of x...

If you are using this table and store the table into a vector DB , i suggest you can ask llama-parse with parsing_instruction to describe the table content .

This is possible through MarkdownElementNodeParser. A param called summary_query_str accepts some prompts. The default prompt has been working fine for me. Hence did not tinkered with it much further.

@galvangoh commented on GitHub (Jul 16, 2024): @tkcoding thanks for the reply. >The quick way to go around this is you can use parsing_instructions to specifically tell table that have continuation on multiple pages should consolidate into one table (I guess you will get like two tables in different pages with the same title). I have tried to specify it in `parsing_instruction` but since its being instructed to the page individually, the parsing result still contain disjoint tables. I think LlamaParse sees pages of document as just pages instead of page 1 of x, page 2 of x... > If you are using this table and store the table into a vector DB , i suggest you can ask llama-parse with parsing_instruction to describe the table content . This is possible through `MarkdownElementNodeParser`. A param called `summary_query_str` accepts some prompts. The default prompt has been working fine for me. Hence did not tinkered with it much further.
yindo commented 2026-02-16 00:17:05 -05:00
Author
Owner
Copy Link
Copy Source

@BinaryBrain commented on GitHub (Jul 16, 2024):

This is in our backlog and will be provided in a future release.

@BinaryBrain commented on GitHub (Jul 16, 2024): This is in our backlog and will be provided in a future release.
yindo commented 2026-02-16 00:17:05 -05:00
Author
Owner
Copy Link
Copy Source

@NG-201 commented on GitHub (Jul 19, 2024):

This is in our backlog and will be provided in a future release.

Great news. When should we expect the next release?

@NG-201 commented on GitHub (Jul 19, 2024): > This is in our backlog and will be provided in a future release. Great news. When should we expect the next release?
yindo commented 2026-02-16 00:17:05 -05:00
Author
Owner
Copy Link
Copy Source

@BinaryBrain commented on GitHub (Jul 22, 2024):

We release several times a week but this feature will take a bit of time to implement. It's not an easy one since we're currently working page by page. But it has been asked by many people.

@BinaryBrain commented on GitHub (Jul 22, 2024): We release several times a week but this feature will take a bit of time to implement. It's not an easy one since we're currently working page by page. But it has been asked by many people.
yindo commented 2026-02-16 00:17:05 -05:00
Author
Owner
Copy Link
Copy Source

@gpric024 commented on GitHub (Mar 6, 2025):

Is this feature still being considered? I tried today with premium parsing and some advances settings and I am still unable to have the tables be joined.

@gpric024 commented on GitHub (Mar 6, 2025): Is this feature still being considered? I tried today with premium parsing and some advances settings and I am still unable to have the tables be joined.
yindo commented 2026-02-16 00:17:05 -05:00
Author
Owner
Copy Link
Copy Source

@BinaryBrain commented on GitHub (Mar 7, 2025):

Hi @gpric024,
It's in production. You can use the Continuous Mode by passing: continuous_mode=True.
Cost is 30 credits per page.

@BinaryBrain commented on GitHub (Mar 7, 2025): Hi @gpric024, It's in production. You can use the Continuous Mode by passing: `continuous_mode=True`. Cost is 30 credits per page.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
LlamaCloud
LlamaExtract
LlamaParse
bug
client
complex layout
documentation
duplicate
enhancement
figure
font
header footer
help wanted
ocr
parser
pdf_debug
pdf_debug
pull-request
Mirrored from GitHub Pull Request
 python-lib
question
server_issue
server_issue
ui
No Label enhancement
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: run-llama/llama_cloud_services#189
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?