PDF content plus size scaling

[pachhaipurna]

@JorjMcKie ,

Hello, I am planning to scale a pdf page and content to specified page height and width,
Is it possible to scale pdf size as well as content, For some pdf size can be decreased whereas for some of the pdf size will e increased, I tried but I can't figure out.
If there is possible way please help me .

Also, Is it possible to reset he coordinates of pdf, Since for some pdf the origin is in different location.

Thank you.

[JorjMcKie]

You can use method Page.show_pdf_page() for this.
Assume w, h being the desired page width and height:

import fitz
doc = fitz.open()  # new PDF
src = fitz.open("input.pdf")

for srcpage in src:
    page = doc.new_page(width=w, height=h)
    page.show_page_pdf_page(page.rect, src, srcpage.number)

doc.ez_save("output.pdf")

The output pdf will show the original, scaled pages.
Please note:

1. If w, h are not the same aspect ratio (w/h != src-width / src-height), there will be empty stripes on the output pages.
2. If input pages are rotated, the code above will need some modification.
3. Links and annotations or fields will not be included in this type of copy. For links at least, there are ways to rescue them.

[pachhaipurna]

I am facing issue for note 1 Source page height width and the input provided by us are not matched so not working properly in that case how can we fixed .

…

On Fri, Jan 26, 2024 at 19:24 Jorj X. McKie ***@***.***> wrote: You can use method Page.show_pdf_page() for this. Assume w, h being the desired page width and height: import fitzdoc = fitz.open() # new PDFsrc = fitz.open("input.pdf") for srcpage in src: page = doc.new_page(width=w, height=h) page.show_page_pdf_page(page.rect, src, srcpage.number) doc.ez_save("output.pdf") The output pdf will show the original, scaled pages. Please note: 1. If w, h are not the same aspect ratio (w/h != src-width / src-height), there will be empty stripes on the output pages. 2. If input pages are rotated, the code above will need some modification. 3. Links and annotations or fields will not be included in this type of copy. For links at least, there are ways to rescue them. — Reply to this email directly, view it on GitHub <#3099 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AKJ2AQFOPWQJOXAU56S3RFTYQN747AVCNFSM6AAAAABCL2H4UWVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4DENJVGMZTM> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[pachhaipurna]

@JorjMcKie for case no 1 how can we fix the problem . Any idea please

[JorjMcKie]

It cannot be fixed if you insist on a fixed, file-wide w, h.
You must decide about another approach, like

• accept source page rectangle's aspect ratio but divide each edge by 2 or whatever
• accept that the source page is shown distorted (stretched or shrinked in one dimension)

[JorjMcKie]

Please don't be shy and consult the documentation of the method!

[pachhaipurna]

Could you give me some examples, So I would be more clear on this . Thanks

…

On Fri, Jan 26, 2024 at 20:43 Jorj X. McKie ***@***.***> wrote: It cannot be fixed if you insist on a fixed, file-wide w, h. You must decide about another approach, like - accept source page rectangle's aspect ratio but divide each edge by 2 or whatever - accept that the source page is shown distorted (stretched or shrinked in one dimension) — Reply to this email directly, view it on GitHub <#3099 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AKJ2AQHQYN47UN55XBTATSTYQOJE7AVCNFSM6AAAAABCL2H4UWVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4DENJWGEYTO> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[JorjMcKie]

page=doc.new_page(width=300,height=300)
page.show_pdf_page(page.rect,src,0,keep_proportion=False)
44
doc.ez_save("x.pdf")

Gives you this

from this

[pachhaipurna]

@JorjMcKie I tried this already, but it is not replacing properly, I am receiving this type of issue.

Here, from the image the blue color is the main pdf page rectangle and yellow color is the new pdf content that is replaced according to new size, the pdf is not for full page according to new pdf size.

The pdf coordinates for origin is in different position.

Can you please let me know how to fix this ?

[pachhaipurna]

The content is only added for yellow highlighted color only.

[JorjMcKie]

Can you let me have an example PDF and tell me what exactly you want to achieve with it please?

[JorjMcKie]

If input pages may be rotated, the code must indeed be adjusted. Ok, another try for what I understood:

1. you want to scale down pages by some factor
2. the width / height ratio must be retained
3. apart from that, pages must look the same
4. the CropBox must equal the MediaBox and start at (0,0)

Everything correct?

[pachhaipurna]

@JorjMcKie Yes, totally correct.

[pachhaipurna]

@JorjMcKie Is there any process for this?
Thank you.

[JorjMcKie]

Ok, good! Alright then.
I am calling the scale factor scale. After the following code snippet, pages are resized, have rotation 0 and CropBox = MediaBox.

src = fitz.open("input.pdf")
doc = fitz.open()
for srcpage in src:
    srect = srcpage.rect
    w, h = srect.width, srect.height
    w *= scale
    h *= scale
    rot = srcpage.rotation
    srcpage.set_rotation(0)
    page = doc.new_page(width=w, height=h)
    page.show_pdf_page(page.rect, src, srcpage.number, rotate=-rot)
doc.ez_save("output.pdf")

[pachhaipurna]

@JorjMcKie

Thank you so much for your response, It is working correctly, But I have 1 issue.

1.)Once the file height and width are scaled and saved, I need to replace this pdf page with some target page While replacing the annotations must be kept, I already have that one, But once replaced using show_pdf_page the annotations are moved to a different position. Any reason for that?

If I am not wrong annotations are independent of pdf contents.
Any idea on this?

Thank you.

[JorjMcKie]

By design, '.show_pdf_page()` cannot copy annotations or fields.
The main reason being, that the source page is converted to a non-page object, a so-called "Form XObject". This PDF object type cannot have annotations, links or fields - and nothing can be done about this.

With one possible part exception: hyper-links. The script above is basically equivalent to applying a suitably computed Matrix. So hyper-links that point to external resources could be re-created on the target page, IAW GoTo, and Named links will never work.

If you find a way to also recreate annotation on the target, you must of course apply the appropriate scaling, complete transformation. For example, if the source page rectangle is srect and that of the (scaled) target is trect then you can make a matrix mat = srect.torect(trect). Then the rectangle of an annotation on the target page is annot.rect * mat.

[pachhaipurna]

As per my understanding target page have markups (funky type annotations) from bluebeam software , even though I tried to import manually to the file that we have just generated that annotation are moved in different position .Any reason for that ? As per I understand when we use show_pdf_page it must just replace the target page content . How it is related to the content .

…

On Mon, Jan 29, 2024 at 2:14 Jorj X. McKie ***@***.***> wrote: By design, '.show_pdf_page()` *cannot* copy annotations or fields. The main reason being, that the source page is converted to a non-page object, a so-called "Form XObject". This PDF object type cannot have annotations, links or fields - and nothing can be done about this. With one possible part exception: hyper-links. The script above is basically equivalent to applying a suitably computed Matrix. So hyper-links that point to *external* resources could be re-created on the target page, IAW GoTo, and Named links will never work. If you find a way to also recreate annotation on the target, you must of course apply the appropriate scaling, complete transformation. For example, if the source page rectangle is srect and that of the (scaled) target is trect then you can make a matrix mat = srect.torect(trect). Then the rectangle of an annotation on the target page is annot.rect * mat. — Reply to this email directly, view it on GitHub <#3099 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AKJ2AQA3GMPILWXL6HAGSDLYQ2BPDAVCNFSM6AAAAABCL2H4UWVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4DENZTGIZDS> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[JorjMcKie]

see my comment here.

[pachhaipurna]

If we apply transformation for the annotation for funky type that are not supported by this library , May be it will loose its attribute while doing transformation. Any idea to control this ? On Mon, Jan 29, 2024 at 2:35 Purna Pachhai ***@***.***> wrote:

…

As per my understanding target page have markups (funky type annotations) from bluebeam software , even though I tried to import manually to the file that we have just generated that annotation are moved in different position .Any reason for that ? As per I understand when we use show_pdf_page it must just replace the target page content . How it is related to the content . On Mon, Jan 29, 2024 at 2:14 Jorj X. McKie ***@***.***> wrote: > By design, '.show_pdf_page()` *cannot* copy annotations or fields. > The main reason being, that the source page is converted to a non-page > object, a so-called "Form XObject". This PDF object type cannot have > annotations, links or fields - and nothing can be done about this. > > With one possible part exception: hyper-links. The script above is > basically equivalent to applying a suitably computed Matrix. So > hyper-links that point to *external* resources could be re-created on > the target page, IAW GoTo, and Named links will never work. > > If you find a way to also recreate annotation on the target, you must of > course apply the appropriate scaling, complete transformation. For example, > if the source page rectangle is srect and that of the (scaled) target is > trect then you can make a matrix mat = srect.torect(trect). Then the > rectangle of an annotation on the target page is annot.rect * mat. > > — > Reply to this email directly, view it on GitHub > <#3099 (reply in thread)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AKJ2AQA3GMPILWXL6HAGSDLYQ2BPDAVCNFSM6AAAAABCL2H4UWVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4DENZTGIZDS> > . > You are receiving this because you authored the thread.Message ID: > ***@***.***> >

[JorjMcKie]

see my comment here.

[pachhaipurna]

@JorjMcKie I am extremely sorry,I could not understand.

My intention is.

1.) Source page (Generated using scale factor with new height and width). No, annotations.
2) The target page contains annotations having the same height and width as the scaled pdf page.

  Need to replace target page content only from the source (Scaled page) , I am receiving the output that annotations are moved in different positions, and doing this logic in the previous (another program) using show_pdf_page working properly.

AS I understand show_pdf_page just replaces content irrespective of annotations.

[pachhaipurna]

@JorjMcKie So sorry for asking lots, I have used mat = srect.torect(trect) as per your suggestion,
However when I try to replace target page with transformed coordinates of source page, The original coordinates are saved in the output file.

Here is my code

` mat = src_page.rect.torect(target_page.rect)

target_page.show_pdf_page(target_page.rect, from_train9_to_new, 0, rotate=0)

from_train9_to_new.ez_save("output.pdf")`

Do we need to do rotation or not ?

[JorjMcKie]

Method page.show_pdf_page() replaces nothing on the target page. The source page content is written above the target content.
Existing annotations on the target should not be affected because they are not part of the page itself, but live in a layer above that. So they remain simply visible.

[JorjMcKie]

I thought we had been talking about ways to let source annotations survive???

[pachhaipurna]

When I checked source as well as target page what I found is , In source page the cordinates top left , bottom left, top right and bottom right are not same as target page . As Target page contains markups , When I try to replace target page with source page May be due to coordinates it replaces the position of markups. Although the orientation seems same between source and target page and width height is also same between two pages the top left coordinates starting are different . Target page coordinates is (0,0) Source page coordinates is (0,something) So, if I can reset the coordinates of source page keeping page orientation same as target page it can be fixed . So, I need to know if it’s possible to reset pdf coordinates keeping orientation same and save it . And just replace to target page . Can you help me on this ?

…

On Tue, Jan 30, 2024 at 23:37 Jorj X. McKie ***@***.***> wrote: I thought we had been talking about ways to let *source* annotations survive??? — Reply to this email directly, view it on GitHub <#3099 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AKJ2AQDQAUQHTL6MD3MTL3LYREAT7AVCNFSM6AAAAABCL2H4UWVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4DGMBWGYYTC> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[pachhaipurna]

@JorjMcKie I solved this issue like this,

`source_pdf = fitz.open(src_pdf)
src_page = source_pdf [0]
rot = src_page.rotation
src_page.set_rotation(0)

##TGT PAGE
tgt_pdf= fitz.open(tgt_pdf)

if not(src_page .is_wrapped):
    src_page .wrap_contents()
##Clean the pdf contents of page. !!
src_page .clean_contents(sanitize=True)

src_page .draw_rect(source_page8.rect,fill=(1,1,1))
src_page .show_pdf_page(src_page .rect, tgt_pdf, src_page .number, rotate = rot)
src_page .set_rotation(rot)

source_pdf.ez_save("output.pdf")`

I am not sure if this is correct way to replace source page without changing orientation, even though the starting coordinates between those 2 pdfs are different. I want confirm few things.

1.) clean_contents() will clean all pdf contents before replacing by target pdf.

IS this correct way to replace source page by target page, so when replacing annotations in source page are not rotated and orientation is same,

[JorjMcKie]

Well, you never supplied your example PDF's. Therefore, the unprotected geometry changes in the target page couldn't ever be detected.
Wrapping the target contents (or cleaning contents) definitely is a good idea. Not necessary for the source.
Otherwise, congratulations!

[pachhaipurna]

@JorjMcKie Thank you so much for your help.
It really works for me and reduce Man hour from 70 percent, I really appreciate for your library.
You are really great. Hope for other help in future also.

Thank you once again.

[pachhaipurna]

I am closing this discussion for now because my issue is solved for now.
Thank you.