Is it possible to get the xref for a block?

[phper5]

1 My real intention is to delete a piece of text. I get the text by for block in page.get_text("dict", flags=extr_flags)["blocks"]: But how can I delete it? can I get the xref for the block?

2 I noticed there is a code https://github.com/pymupdf/PyMuPDF-Utilities/tree/master/font-replacement
in this code it can remove text written with one of the fonts.
this is a solution, I saved all the text which use the same font, and delete all the text which use this font. and then rewrite the text I do not want to delete. It will works. But I have to delete a lot text, and then write it back Is there a better way?

3 Follow the code above, it checked the font name to location in the stream, Can I locate the word in the stream directly? How? I found someone user code like this stream = doc.xref_stream(xref).replace(b'seard',b'kjhkj') but it not work for me.

so how can i do if i just want to delete a few word in the pdf

thanks

[JorjMcKie]

Arranging text to blocks is happening inside MuPDF - it is done with some heuristics based on coordinates, fontsizes and so on.
There is nothing "physical" in the document representing a block.

To delete text, you can either follow the pattern of font replacement as you already observed.
Or use redaction annotations. Then MuPDF code will physically remove text for which you have provided a rectangle. You then have a number of options to fill that gap with new stuff: text, images, ...

Forget your option 3: this requires a lot of translation work (unicode -> glyph -> position info ...) already implemented by MuPDF logic in redactions.

[phper5]

thanks a lot.
It helped me a lot