Adding form fields to a PDF

[meghanaviyyapu]

Is there a method in PyMuPDF to add form fields to PDF

[JorjMcKie]

Sure - here is an example script: widgettest.zip

[JorjMcKie]

No

[JorjMcKie]

If you want all that, then why don't you simply insert that page in another PDF!

[meghanaviyyapu]

I want to tag all the elements in a PDF so I was thinking of extracting all elements from an existing PDF and create a new PDF with elements and their tags.

[JorjMcKie]

Ah, I see. Sometimes MS Word or LibreOffice do a good job when opening a PDF. Then you can work with the resulting office document ...

[meghanaviyyapu]

Is there a way to extract text and its (x,y) co-ordinates?

[JorjMcKie]

Is there a way to extract text and its (x,y) co-ordinates?

You don't know page.get_text()? For all method variants "words", "blocks", "dict", "rawdict" you get the full bbox coordinates.

[meghanaviyyapu]

Yeah I know that but instead of that I wanted to know if there is any other method.

[JorjMcKie]

well there also is page.get_texttrace()- comparable with "rawdict" but with even more information

[meghanaviyyapu]

Can you let me know based on what properties page.extract_text() extracts text from PDF without depending on content stream?

[JorjMcKie]

The correct method name is page.get_text().
It reads out information stored in the TextPage object. This is a wrapper of MuPDF's fz_stext_page. When MuPDF code creation this textpage structure, it of course reads the page's /Contents objects for a PDF - or the corresponding information for other document types.

[meghanaviyyapu]

Using page.get_text(), PDF text is being rendered in ASCII format but when the text is extracted from the content stream(using xref of /Contents), all the contents are not in ASCII format. Is there a way to get the output identical to page.get_text() even when the text is extracted from the stream?

[JorjMcKie]

Is there a way to get the output identical to page.get_text() ...

No:

1. The text is extracted as unicode, not just ASCII.
2. The /Contents object almost never contains text in directly readable form. Only the most simple fonts (e.g. Base-14) will be like this. Instead you will see glyph numbers (in 4-digit hex format). They are used to communicate with the font program and cannot be backtranslated to original text (by you or by me).

Didn't we discuss this already earlier in another post?

[meghanaviyyapu]

Yeah we had a discussion but I had a doubt how page.get_text() is extracting the content in ASCII or unicode format because as you mentioned backtranslation is not possible using the text present in /Contents object.