Adjust the positioning of PDF content

[JackieFei]

Is there any way to adjust the positioning of PDF content?

• without converting to image/svg format for text editing later.
• without rewriting text to avoid missing any attribute.

For example:

PDF has text "Hello World" at (top, left), then want to move this text to (top+10, left+10).

Thanks in advanced.

[JorjMcKie]

Is there any way to adjust the positioning of PDF content?

You mean some specific text only, and not the whole page content, right?
That won't really work: you have to find the place, where that text piece is defined in the page /Contents and then modify its text matrix Tm.
Problems are manifold:

• In its simplest (!!!) form, the text object may look like this:

BT
/helv 11 Tf
1 0 0 1 100 742 Tm  % = point (100,100) on a A4 page
[(Hello Earth!)] TJ
ET

or like this (text encoded before TJ is "Hello Mars!" - undetectable in any practicable way):

BT
/F0 11 Tf
1 0 0 1 100 692 Tm  % = point (100, 150)
[<00290046004D004D00500001002E0042005300540002>] TJ
ET

• then, the text could be multi-line - with or without its own Tm
• the Tm may not contain the starting position so clearly like here: previous matrices may exist (somewhere!), that scale or otherwise modify the page geometry. Those would have to be found and concatenated to our Tm delivering Point(100, 100) in the end
• then the text may not even be on the page's /Contents, but in some Form XObject ...

So, the short answer to your question is "no".

[JackieFei]

Thanks for the explaination about the specific text only.

Is it possible about the whole page contents moved from (0, 0) to (10, 10)?

maybe put/merge a whole text page into another empty page at location (10, 10)

[JorjMcKie]

Is it possible about the whole page contents moved from (0, 0) to (10, 10)?

Yes, that works using a clean approach:

1. make a new page in a new pdf
2. define the rectangle in which to show the original page: rect = fitz.Rect(10, 10, newpage.rect.x1, newpage.rect.y1). Or, the same but more obvious: rect = newpage.rect + (10, 10, 0, 0)
3. define the rectangle of the source page, which we want to show: this starts from its (0,0) and ends 10 points less in each dimension clip = fitz.Rect(0, 0, oldpage.rect.x1 - 10, oldpage.rect.y1 - 10). If we start the display shifted by displacement (10, 10), we must of course cut those values off of the rectangle's width and height. Again: clip = oldpage.rect + (0, 0, -10, -10) also works.
4. then newpage.show_pdf_page(rect, oldpdf, oldpage.number, clip=clip).

Note: rect and clip do not have to have identical dimensions. The source is always scaled to fill the target rect - keeping aspect ratios by default (can be switched off, too).

[JackieFei]

Does the show_pdf_page generate a vector image in newpage? Is it editable by pdf editor (text mode)?

[JorjMcKie]

Does the show_pdf_page generate a vector image in newpage?

Yes!

Is it editable by pdf editor (text mode)?

I don't know pdf editor.

[JorjMcKie]

Another, less clean alternative: Change the page's mediabox by low-level code.

>>> page.xref
4
>>> doc.xref_get_key(4, "MediaBox")
('array', '[0 0 595 842]')
>>> doc.xref_set_key(4,"MediaBox", "[-10 10 585 852]")  # effectively add (-10, 10, -10, 10)

Example text search shows the shift:

>>> doc1[0].search_for("hello world!")
[Rect(100.0, 88.17500305175781, 157.45298767089844, 103.28900146484375)]
>>> doc2[0].search_for("hello world!")
[Rect(110.0, 98.17500305175781, 167.45298767089844, 113.28900146484375)]
>>>

The top-left and bottom-right coordinates both are 10 points larger.