Annot Border and Rect shift. #4512

ibbit · 2025-05-20T12:45:12Z

ibbit
May 20, 2025

Hello everyone,

I wanted to quickly ask why PyMuPDF behaves the way it does in the following example.

I added a rectangle as a note using my editor (to define a zone on my floor plan). Now I want to use PyMuPDF to add my "content" and save it. The problem I observe is that after calling annot.update(), the boundaries somehow change.

See following images:
Rectangle (created in the editor)

After I run:

doc_new = pymupdf.open(pth2)
p1 = doc_new[0]
for annot in p1.annots():
    if annot.type[0] in [4]:
        annot.update()

Image 2: (updated rectangle)
When I open the PDF, you can see that the "selection" frame does not match the one in the first image.

Additional:
Image 3:
Shows the behavior when I simply stretch the rectangle to the right in the editor. Then the "border" just jumps back.

It's hard for me to explain this more precisely.

Am I missing something about how to properly use the annot.update() method?

But I also observe:

doc = pymupdf.open(pth1)
p1 = doc[0]
rect = pymupdf.Rect([10, 10, 100, 100])
new_rect = p1.add_rect_annot()
print("rect:", new_rect.rect)

terminal printout:

rect: Rect(9.0, 9.0, 101.0, 101.0)

But my definition is 10,10,100,100

Maybe some can drop a comment to understand this circumstance here and thx for your help!

BTW: I have the same behavior with PDFXchange and Ocular.

greets: ibbit

JorjMcKie · 2025-05-22T09:33:27Z

JorjMcKie
May 22, 2025
Maintainer

When updating an annotation with a tool that is different from the tool that created it, errors or unexpected results are to be expected in general - see here.

Independently from this, there is some inconsistency in PyMuPDF's behavior as you have noticed:
The base library automatically adds a /RD (= "rectangle difference") item to Square and Circle annotations which is used to compute the inner annotation rectangle - i.e. the one resulting when disregarding the line width.
What we should be doing in PyMuPDF is modifying the provided Annot.rect such that your intended rectangle results after taking /RD into account.
Either that or offer specifying the /RD values in the API and leave it to the user's discretion.

In any case, your creator software makes the annotation in a way such that everything is inside the specified rectangle, whereas the PyMuPDF update changes the rectangle such that 50% of the line width area is outside ...

Can you please give me the annotation object details as created by your initial editor? Probably an out like doc.xref_object(annot.xref) would be sufficient.

0 replies

ibbit · 2025-05-25T07:25:58Z

ibbit
May 25, 2025
Author

I did run this:

doc = pymupdf.open(pth)
p1 = doc[0]
for annot in p1.annots():
    print(annot.rect)
    print(annot.border)
    print(doc.xref_object(annot.xref))
    annot.update()

Terminal printout:

Rect(72.0, 72.0, 216.0, 216.0)
{'width': 10.0, 'dashes': (), 'style': None, 'clouds': -1}
<<
  /AP <<
    /N 41 0 R
  >>
  /BS <<
    /W 10
  >>
  /C [ 1 1 0 ]
  /CreationDate (D:20250525091919+02'00')
  /F 4
  /IC [ 1 0 0 ]
  /M (D:20250525092003+02'00')
  /NM (bdb27af4-ccc2-4f87-86f37f5ab15b55e4)
  /P 35 0 R
  /RC (<?xml version="1.0"?><body xmlns="http://www.w3.org/1999/xhtml" xmlns:xfa="http://www.xfa.org/schema/xfa-data/1.0/" xfa:APIVersion="Acrobat:10.1.5" xfa:spec="2.1" style="text-align:left;font-family:Liberation Sans;font-size:12pt;font-weight:normal;font-style:normal;text-decoration:none;color:#000000;"><p></p></body>)
  /Rect [ 72 626 216 770 ]
  /Subj (Rectangle)
  /Subtype /Square
  /T (w7e)
  /Type /Annot
>>

6 replies

ibbit May 25, 2025
Author

Thx for your help!
btw, I've read the doc's for the Page.annot() method:
You cannot safely update annotations from within this generator. This is because most annotation updates require reloading the page via page = doc.reload_page(page). To circumvent this restriction, make a list of annotations xref numbers first and then iterate over these numbers:

I'm generally manipulating shapes annotations like "lines, polylines, squares, polygons, circles..."
My project is quite far at this point and I didn't encounter any problems (or I haven't noticed them).
Is there a list of what type of annots need a page = doc.reload_page(page)?

Thank you!

JorjMcKie May 25, 2025
Maintainer

No. You only ever ever need to reload a page if you depend on results of previous changes while still working with the same page.
As soon as you have left a page (i.e. loaded another one) there is no need for this.

ibbit Jun 14, 2025
Author

May I ask a following annot.update() question without opening a new discussion.
I have a freeText annot and I want to unify the appearance. My PDF Editor uses Liberation Mono font and when i call the xref object i have the font reference for the Default Apearance: /DA (0 1 0 rg /F1 1.382 Tf)
Due to appearance changes i run:
annot.update()
after that I call the xref object again and I have this "/DA": /DA (0 1 0 rg /Helv 1.382 Tf)
Despite the situation that the richtext is active i the annot has now the appearance font of Helvetica and as soon I double click it to change the content it renders the the Liberation Mono.
Is there a viable solution to avoid the "force" reset of the Font reference?

Thx

ibbit Jun 14, 2025
Author

Additional info:
page.get_fonts()

[
    (99, 'ttf', 'Type0', 'Liberation Mono Regular', 'monr', 'Identity-H'),
    (50, 'cid', 'Type0', 'Swiss721BT-Roman', 'F1', 'Identity-H'),
    (51, 'cid', 'Type0', 'Swiss721BT-Bold', 'F2', 'Identity-H'),
    (52, 'cid', 'Type0', 'Arial-BoldMT', 'F3', 'Identity-H'),
    (53, 'cid', 'Type0', 'ArialMT', 'F4', 'Identity-H'),
    (54, 'n/a', 'Type1', 'Times-Roman', 'tiro', 'WinAnsiEncoding')
]

my Changes manually in the xref object: from -> to
0 0 0 rg /F1 12 Tf
0 0 0 rg /monr 12 Tf

run update.annot()

<<
  /Type /Annot
  /Subtype /FreeText
  /Rect [ 78.588329 361.28303 89.52763 375.16127 ]
  /CL [ 0 841.89 82.82655 375.16127 ]
  /BS <<
    /W 0
  >>
  /DA (0 0 0 rg /Helv 1.152 Tf)
  /P 15 0 R
  /F 4
  /RC (<?xml version="1.0"?><body xmlns="http://www.w3.org/1999/xhtml" xmlns:xfa="http://www.xfa.org/schema/xfa-data/1.0/" xfa:APIVersion="Acrobat:10.1.5" 
xfa:spec="2.1" 
style="text-align:right;font-family:LiberationMono;font-size:1.152pt;font-weight:normal;font-style:normal;text-decoration:none;color:#9A0089;text-valign:top;"><p 
style="font-size:1.843pt;color:#FF6820;"><span>Test</span><span style="font-size:1.152pt;color:#9A0089;">here</span></p></body>)
  /DS (text-align:right;font-family:LiberationMono;font-size:1.152pt;color:#9A0089;)
  /Q 0
  /RD [ 0 0 0 0 ]
  /AP <<
    /N 116 0 R
  >>
  /NM (fitz-A2)
  /BM /Multiply
  /C [ 0 .576 0 ]
  /Subj (Typewriter)
  /Contents (Testhere)
  /T (w7e)
  /CreationDate (D:20250614113644+02'00')
  /M (D:20250614113737+02'00')
>>

JorjMcKie Jun 15, 2025
Maintainer

No, I'm afraid there is not. The appearance logic is fully (and deliberately) under the control of MuPDF. When MuPDF encounters the existence of an /RC key (rich text), it will use its own logic for choosing fonts.

As a general policy, MuPDF will never look around for fonts installed in the OS / platform. Only its own fonts (usually from NOTO) and any fonts present (and properly referenced!) in the PDF are possible candidates.

Fonts used in annotations (and widgets) are not the same fonts used by text on the page. So, for "font-family:LiberationMono" the question is where (within the PDF!) this family is defined, such that MuPDF has a chance to locate and process it.

ibbit · 2025-06-28T12:54:45Z

ibbit
Jun 28, 2025
Author

To my main Question about the Border/Corner width...

The method annot.update() is adding a /RD [left top right bottom] element.
So if my border widht is 8, annot.update() adds a /RD[4 4 4 4]
My solution atm is simply to remove the entry from the annot xref dictionary:

doc = pymupdf.open("tests/test.pdf")
page = doc[0]
xrefs = [annot.xref for annot in page.annots()]

for xref in xrefs:
    annot = page.load_annot(xref)
    colors = annot.colors
    colors["stroke"] = [0, 1, 0]
    annot.set_colors(colors)
    annot.update()

    xref_body = doc.xref_object(xref, compressed=True)
    rd_el = _get_key("/RD", xref_body)
    if rd_el:
        xref_body = xref_body.replace(rd_el, "")
    doc.update_object(xref, xref_body)


doc.save("tests/out.pdf")
doc.close()

So far this works for me.

0 replies

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Annot Border and Rect shift. #4512

Uh oh!

{{title}}

Uh oh!

Replies: 3 comments 6 replies

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Select a reply

Uh oh!

Annot Border and Rect shift. #4512

Uh oh!

ibbit May 20, 2025

Replies: 3 comments · 6 replies

Uh oh!

JorjMcKie May 22, 2025 Maintainer

Uh oh!

ibbit May 25, 2025 Author

Uh oh!

ibbit May 25, 2025 Author

Uh oh!

JorjMcKie May 25, 2025 Maintainer

Uh oh!

ibbit Jun 14, 2025 Author

Uh oh!

ibbit Jun 14, 2025 Author

Uh oh!

JorjMcKie Jun 15, 2025 Maintainer

Uh oh!

ibbit Jun 28, 2025 Author

ibbit
May 20, 2025

Replies: 3 comments 6 replies

JorjMcKie
May 22, 2025
Maintainer

ibbit
May 25, 2025
Author

ibbit May 25, 2025
Author

JorjMcKie May 25, 2025
Maintainer

ibbit Jun 14, 2025
Author

ibbit Jun 14, 2025
Author

JorjMcKie Jun 15, 2025
Maintainer

ibbit
Jun 28, 2025
Author