Deleting last page without knowing its number

[IgorCF-IS]

Hi!
Thanks so much for the great work on the tool, it’s incredibly helpful!
I have a question about version 3.7. The new clean and draw features allow deleting pages from a PDF, which is great. However, I was wondering if there’s an easier way to delete the last page without knowing its exact number. For example, something like:
SumatraPDF clean input.pdf output.pdf 1-(N-1)
Even better, it would be great to have a delete option where you can specify pages directly, like:
SumatraPDF delete input.pdf output.pdf 1,N - to remove the first and last pages.
Is there an alternative approach for this? Or at least a way to retrieve the total number of pages from the command line, so I could calculate N-1 and use it with the clean command?

[GitHubRulesOK]

Hmm Mutool/SumatraPDF Run is fairly "easy" for such a task as it has the concept countPages as soon as opened and thus you can script around that

if (debug) print("file is open entering page parser");
var pages = parsePages(args["p"], pdf.countPages());          // 1-based
for (var i = 0; i < pages.length; i++) {
    var page = pdf.loadPage(pages[i]-1); // convert to 0-based

however the character N denotes the last page. If no pages are specified, then all pages will be included. thus in theory if you say N it "should" remove the last ?
Just tried that and page 3 of 3 is gone so to remove last 2 you would need to run N then N on the 1st result ?

[IgorCF-IS]

That would work, I guess. How do I get numpages in a CMD?

[GitHubRulesOK]

LATER EDIT THIS HAS NOW BEEN FIXED
sorry have been testing why it cannot work and this is another issue with SumatraPDF compared to MuPDF Mutool that means you cannot use CMD!

@kjk SumatraPDF outputs an odd response to CMD queries (console shows one output for reading but not redirection or filtering) thus SumatraPDF seems to reply with a cryptic

OpenEmbeddedFilesArchive: opened archive

Rather than the normal console output from "mutool" info output.pdf >info.txt which is parsable e.g

"mutool" info input.pdf |find "Pages:"
Pages: 3

[IgorCF-IS]

Is the MUTOOL the source for SumatraPDF? Maybe I should use it.

[kjk]

@GitHubRulesOK I need to know the exact steps to reproduce this. "CMD queries" is too vague.

The OpenEmbeddedFilesArchive: opened archive is SumatraPDF's logging.

Generally speaking, SumatraPDF.exe is a windows subsystem application, not a console application, like mutool.exe. Handling input / output (stdout, stderr, stdin, console output) is always going to be a hack. I'm sure I can improve it compared to what it is today but I need a precise description of the bug, as an issue, one issue per unique problem.

[GitHubRulesOK]

LATER EDIT THIS WAS FIXED in #5677
@kjk any console command normally uses 2 basic output channels that can be filtered
stdout — standard output (file descriptor 1)
stderr — standard error (file descriptor 2)
either can be used by other command or combined "1>&2"
however according to my untrustworthy assistant (copilot)

SumatraPDF uses WriteConsoleW() (or a similar Win32 console function) to emit diagnostic messages.
These functions bypass:, stdout, stderr. the shell’s redirection system They write directly to the console screen buffer.
SumatraPDF.exe info input.pdf > test.txt does not output the screen text as reusable text

[IgorCF-IS]

Maybe I’m overcomplicating this, but here’s what I’m trying to do.
I have a PDF with 4 labels per page, and I need to convert it into a file with 1 label per page. That part works fine using:
SumatraPDF.exe poster -x 2 -y 2 input.pdf output.pdf
The issue is that the last page of the original PDF might not contain all 4 labels. When that happens, the output ends up with 1 to 3 blank pages at the end. I want to eliminate blank pages.
I can check whether a page is blank using:
SumatraPDF.exe convert -o output.txt output.pdf -1
This converts the last page into a text file. If output.txt is empty, then the last page is blank.
What I’d like to do next is delete that last page and repeat the check until the last page is no longer blank. This is where I’m stuck, specifically with removing the last page.
Alternatively, I could approach it from the beginning by scanning pages one by one until I find the first empty page.
SumatraPDF.exe convert -o output.txt output.pdf 1
/% check output, if not empty %/
SumatraPDF.exe convert -o output.txt output.pdf 2
At that point, I would know the page number and could delete it using clean, and continue from there.
My question is, how do I know when to stop? What’s the indication that there are no more pages left to process? Error message?

[GitHubRulesOK]

OK I think I have a workaround it is slow as it has to work forwards and set to 1000 as that is not the problem until the info command can be made to work otherwise. LATER EDIT See the more optimised reverse method below

@echo off
set "SumatraPDF=C:\Program Files\SumatraPDF\SumatraPDF.exe"
"%SumatraPDF%" poster -x 2 -y 2 "%~1" "%temp%\temp.pdf"
set "lastgood=0"
for /L %%c in (1,1,1000) do (
    "%SumatraPDF%" convert -o page.txt "%temp%\temp.pdf" %%c
    for %%A in ("page.txt") do if %%~zA==0 (
        rem page is blank so stop scanning
        goto :done
    )
    rem page is NOT blank so remember this page number
    set "lastgood=%%c"
)
:done
echo Finished scanning. Last non-blank page = %lastgood%
"%SumatraPDF%" clean "%temp%\temp.pdf" output.pdf 1-%lastgood%
pause

[GitHubRulesOK]

@IgorCF-IS
The latest SumatraPDF PRE-release e now has a working info output, so this is a rough example of use to count pages before and after "clean" but shows the methodology

Page 16 is blank
Checking page 15...
Page 15 is blank
Checking page 14...
Page 14 is blank
Checking page 13...
Page 13 is blank
Checking page 12...
Page 12 is NOT blank
Finished scanning. NON Blank page at end: 12
Keeping pages 1-12
attempting "C:\Program Files\SumatraPDF\SumatraPDF.exe" clean "C:\Users\WDAGUtilityAccount\AppData\Local\Temp\temp.pdf" output.pdf 1-12
 12
Pages detected:  12
Done.
Press any key to continue . . .

you can remove many of the lines (especially echo's) I added while testing so replace ( echo %%A & set pages=%%A ) without the echo ... & eg just ( set pages=%%A )

@echo off
setlocal enabledelayedexpansion

set "SumatraPDF=C:\Program Files\SumatraPDF\SumatraPDF.exe"
"%SumatraPDF%" poster -x 2 -y 2 "%~1" "%temp%\temp.pdf"

"%SumatraPDF%" info "%temp%\temp.pdf" | find "Pages:" > "%temp%\temp.txt"
for /f "tokens=2 delims=:" %%A in ('type "%temp%\temp.txt"') do ( echo %%A & set pages=%%A )
echo Pages detected: %pages%

rem --- Loop backwards ---
set "lastblank=0"

for /l %%P in (%pages%,-1,1) do (
    echo Checking page %%P...
    "%SumatraPDF%" convert -o "%temp%\page.txt" "%temp%\temp.pdf" %%P
    for %%A in ("%temp%\page.txt") do (
        if %%~zA==0 ( echo Page %%P is blank & set /a lastblank+=1
        ) else ( echo Page %%P is NOT blank 
          goto :done
        )
    )
)

:done
echo Finished scanning. NON Blank page at end: !lastblank!
set /a keepTo=pages-lastblank & if !keepTo! lss 1 set keepTo=1
echo Keeping pages 1-!keepTo!
echo attempting "%SumatraPDF%" clean "%temp%\temp.pdf" output.pdf 1-!keepTo!
"%SumatraPDF%" clean "%temp%\temp.pdf" output.pdf 1-!keepTo!

REM lets see how many pages in output
"%SumatraPDF%" info output.pdf | find "Pages:" > "%temp%\temp.txt"
for /f "tokens=2 delims=:" %%A in ('type "%temp%\temp.txt"') do ( echo %%A & set pages=%%A )
echo Pages detected: %pages%

echo Done.
pause

[IgorCF-IS]

That’s great to hear! After upgrading to the latest pre-release, the info feature is now working as expected, and I can successfully retrieve the page count.
I’ll go ahead and wrap everything in a PowerShell script since I need a simple GUI to select multiple files and display a progress bar.
Thanks again, everyone, for your help!

[GitHubRulesOK]

We can use SumatraPDF RUN to make it far more internal

rem a basic count all pages is to echo "var doc = Document.openDocument(scriptArgs[0]); print(doc.countPages());" > PageCount.js

set "SumatraPDF=C:\Program Files\SumatraPDF\SumatraPDF.exe"
"%SumatraPDF%" poster -x 2 -y 2 "%~1" "%temp%\temp.pdf"

REM we use the internal ability to read pages with or without text
echo var d=Document.openDocument^(scriptArgs^[0^]^);var n=d.countPages^(^);for^(var i=n-1;i^>=0;i--^){var p=d.loadPage(i);var t=p.toStructuredText^(^).asText^(^);if^(t^&^&t.trim^(^).length^>0^){print^(i+1^);quit^(^);}}print^(0^); > lasttext.js

for /f "usebackq delims=" %%A in (`""%SumatraPDF%" run lasttext.js "%temp%\temp.pdf""`) do set LastText=%%A
echo LastText = %LastText%

"%SumatraPDF%" clean "%temp%\temp.pdf" "%~dpn1-noblanks.pdf" 1-%LastText%

So if you save the script as lasttext.js

var d=Document.openDocument(scriptArgs[0]);var n=d.countPages();for(var i=n-1;i>=0;i--){var p=d.loadPage(i);
var t=p.toStructuredText().asText();if(t&&t.trim().length>0){print(i+1);quit();}}print(0); 

you can reuse it fastest in a cmd like this where you ether drop a file on the cmd or run with a folder of inputs etc.

set "SumatraPDF=C:\Program Files\SumatraPDF\SumatraPDF.exe"

"%SumatraPDF%" poster -x 2 -y 2 "%~1" "%temp%\temp.pdf"
for /f "usebackq delims=" %%A in (`""%SumatraPDF%" run lasttext.js "%temp%\temp.pdf""`) do set LastText=%%A
"%SumatraPDF%" clean "%temp%\temp.pdf" "%~dpn1-noblanks.pdf" 1-%LastText%

[GitHubRulesOK]

Hmm note there is another change in the wings to improve tool ability that may use a different filename.exe ?
Just change the top line with the exe name.
But the principle is basically the same. It is possible for the run script to save the tempfile to a given name at that last number but then it would much more complex than simply reporting the last page number.