I want to use get_tables, which contains this method

[juzj]

My requirement is to extract the text in pdf. If it is the deletion or skip of the table, save the result as txt

But the latest version doesn't have this feature, so I can use that method instead

[JorjMcKie]

I do not fully understand:
Do you mean to add a feature that automatically extracts a table from the text of a document page?

[juzj]

I want to delete all the table data from the pdf

[juzj]

But now without the get_tables method, I can't tell if the current data is a table

[juzj]

I will cycle through all the pages, if there is tabular data to skip without processing, what do I need to do?

[JorjMcKie]

I think I understand now.
Your request implies the need for a capability to understand what a table is in the first place.
This is a highly complex task - for which there exists no general solution - nowhere in the software market. There are however some approaches that come close to doing this, which employ advanced Artificial Intelligence and Machine Learning features.
While PyMuPDF's extraction features deliver a large set of information on a document page's contents, it will remain your task to programatically interpret these data.
So you must use your knowledge about a given page to determine a table's whereabouts (like certain keywords in its header text or any footer lines, etc.) and then exclude the respective text from your extraction.

[JorjMcKie]

Let me first move your post to the "Discussions" tab - where it belongs.

[JorjMcKie]

You seem to be lucky:
The tables on your example page are all wrapped by rectangle drawings. PyMuPDF has a way to extract vector graphics - use Page.get_drawings(). Please do read the documentation.
The method returns a list of dictionaries, which each contain atomic draw commands (like draw a line, draw a curve, etc.), plus a rectangle that encapsulates those draw commands.
So I suggest you extract the drawings of your page and confirm that you will see the 4 table rectangles being represented by a drawing dictionary each. Do a few experimenting here.
Once confirmed, check whether your extracted text is contained in one of these rectangles and if so skip it.

[juzj]

Thank you. I'll try.

[JorjMcKie]

Please do not hesitate to ask if you need help.

[Nicolas2912]

I have encountered a related problem. I'm using get_drawings() to extract vector graphics from a PDF page, but the function also returns tables, which I want to ignore. Is there a recommended way to distinguish between actual drawings and tables in the output? Any suggestions on filtering or post-processing the extracted data would be appreciated!

[JorjMcKie]

You could use page.cluster_drawings(). This works on top of .get_drawings() trying to find interconnected / neighboring / overlapping vector graphics and returns a list of rectangles that wrap more or less isolated vector graphics.
This could save you a lot programming effort.

Once you have this rectangle list, execute .find_tables() and check each table.bbox. If contained in one of the above clusters, you have what you are looking for.