Merge / convert multiple PDF files into one PDF

Question

How could I merge / convert multiple PDF files into one large PDF file?

I tried the following, but the content of the target file was not as expected:

convert file1.pdf file2.pdf merged.pdf

I need a very simple/basic command line (CLI) solution. Best would be if I could pipe the output of the merge / convert straight into pdf2ps ( as originally attempted in my previously asked question here: Linux piping ( convert -> pdf2ps -> lp) ).

Solution 1

I'm sorry, I managed to find the answer myself using google and a bit of luck : )

For those interested;

I installed the pdftk (pdf toolkit) on our debian server, and using the following command I achieved desired output:

pdftk file1.pdf file2.pdf cat output output.pdf

OR

gs -q -sPAPERSIZE=letter -dNOPAUSE -dBATCH -sDEVICE=pdfwrite -sOutputFile=output.pdf file1.pdf file2.pdf file3.pdf ...

This in turn can be piped directly into pdf2ps.

Solution 2

Considering that pdfunite is part of poppler it has a higher chance to be installed, usage is also simpler than pdftk:

pdfunite in-1.pdf in-2.pdf in-n.pdf out.pdf

Just make sure you remember to provide out.pdf, or else it will overwrite the last file in your command

Solution 3

Try the good ghostscript:

gs -dBATCH -dNOPAUSE -q -sDEVICE=pdfwrite -sOutputFile=merged.pdf mine1.pdf mine2.pdf

or even this way for an improved version for low resolution PDFs (thanks to Adriano for pointing this out):

gs -dBATCH -dNOPAUSE -q -sDEVICE=pdfwrite -dPDFSETTINGS=/prepress -sOutputFile=merged.pdf mine1.pdf mine2.pdf

In both cases the ouput resolution is much higher and better than this way using convert:

convert -density 300x300 -quality 100 mine1.pdf mine2.pdf merged.pdf

In this way you wouldn't need to install anything else, just work with what you already have installed in your system (at least both come by default in my box).

UPDATE: first of all thanks for all your nice comments!! just a tip that may work for you guys, after googleing, I found a superb trick to shrink the size of PDFs, I reduced with it one PDF of 300 MB to just 15 MB with an acceptable resolution! and all of this with the good ghostscript, here it is:

gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 -dPDFSETTINGS=/default -dNOPAUSE -dQUIET -dBATCH -dDetectDuplicateImages -dCompressFonts=true -r150 -sOutputFile=output.pdf input.pdf

cheers!!

Solution 4

This is the easiest solution if you have multiple files and do not want to type in the names one by one:

qpdf --empty --pages *.pdf -- out.pdf

Solution 5

Also pdfjoin a.pdf b.pdf will create a new b-joined.pdf with the contents of a.pdf and b.pdf

View more solutions

Comments

[sabujp]

ymmv, but this doesn't seem to have as good of a resolution in the output file as pdfunite and it also results in a file size larger than the output from pdfunite

[Franck Dernoncourt]

related: linux command merge pdf files with numerical sort

[Clément]

Whenever links are preserved or not by those solutions is discussed in this post. If you want to preserve the links (probably along with other annotations), use pdftk if want a command-line interface, pdfsam if you want graphical user interface, sejda if you want a web interface.

[Alexis Wilke]

The convert command line is from ImageMagick and it converts the PDF to an image before doing whatever else it will be doing.

[Nate Kohl]

Using ghostscript also might work: gs -q -sPAPERSIZE=letter -dNOPAUSE -dBATCH -sDEVICE=pdfwrite -sOutputFile=out.pdf in1.pdf in2.pdf in3.pdf ...

[alcohol]

I'd have to look more into the possible options/flags (because I don't want output in a file), but probably yes. Thank you for the suggestion.

[Thomas]

It is worth to mention that pdftk can merge encrypted pdfs while pdfunite cant

[taxilian]

Poppler is also very fast, from my tests

[Danilo Bargen]

It is fast, but it seems to break hyperlinks. See blog.dbrgn.ch/2013/8/14/merge-multiple-pdfs

[mlissner]

Just make sure you remember to provide out.pdf, or else it will overwrite the last file in your command, sigh.

[Jocelyn delalande]

package for pdfunite is poppler-utils in debian but may not be present in old debian releases.

[Torben]

Cannot recommend this. The size of the the resulting PDF is far too big. For example: Pdfunite gives me a 75MB file while Ghostscript packs everything into 1MB.

[Adriano P]

Nice tip, gs runs very fast and it compresses a lot. However, the quality improved a lot after I used this param: -dPDFSETTINGS=/prepress

[Aaron R.]

Hmmm... @Torben I just packed 300+ pdfs (total of 13MB) into a single PDF using this utility and I got a 12MB file at the end. Maybe it was the version you used? I am on OpenSUSE 12.2, using pdfunite version 0.20.0.

[Torben]

@Aaaron My comment was a bit misleading. What I meant is that pdfunite doesn't optimize the filesize. For example: 10 similar pdfs (slides for a presentation) of 1MB result in a ~10MB pdf when I use pdfunite. With ghostscript the resulting pdf is <1MB.

[Kiran K Telukunta]

gives better resolution with pdftk compare to convert in default options.

[ahrobins]

In my case pdfunite did not produce a usable PDF file. When I loaded it with evince I got lots of errors. The gs solution worked.

[Ben Ruijl]

This is not lossless.

[arielnmz]

You can convert -compress lossless sub1.pdf sub2.pdf sub3.pdf merged.pdf, but the resulting file size's could be way too big. I'd suggest convert -compress jpeg -quality 90 sub1.pdf sub2.pdf sub3.pdf merged.pdf instead.

[r_31415]

I found that -dPDFSETTINGS=/prepress has the very nice effect of rotating pages that are too wide and force annoying horizontal scroll bars.

[Julia Ebert]

This involves converting everything to raster images, it seems, which is definitely not the best, especially when dealing with text-based PDFs.

[bright-star]

This is nice and succinct, but breaks hyperlinks.

[mdrozdziel]

pdfjoin (pdflatex) fails with files with lots of pages. Failed to merge to 1k pages files.

[lepe]

You can use: pdfunite *.pdf out.pdf assuming no other pdf exists in that directory and their order is preserved by "*". If its not preserved, using ranges: filename_{0..9}.pdf solves it.

[Nathan Tuggy]

Given how similar this looks to the original question, it seems like this should have been a comment, not an answer. With a bit more rep, you will be able to post comments. Until then, please do not use answers as a workaround.

[peterh]

@Silfheed No, it answers the question! Although the answer maybe should have more elaborated.

[ryantm]

pdfunite worked much worse for my use case than pdftk. I was trying to combine together copies of a particular form into the same pdf. In pdfunite they were linked, in pdftk they were separated and could be filled out individually.

[MariuszS]

@Torben The file is smaller with gs because with default settings image has reduced quality (screen-view-only quality, 72 dpi images). With -dPDFSETTINGS=/printer file size is almost identical (high quality, 300 dpi images).

[Torben]

@MariuszS Even though you might be right for some special cases, you're wrong with you're general assumption. I just merged 133MB of PDFs (84 files exported from Inkscape), which contain bitmaps, vectorgraphics and text, into one PDF with a size of 1.6MB. I used /prepress for that and /printer even reduced the size to 1.3MB. Even though I zoomed in and printed a part, I couldn't find any visible difference between the single PDF and the merged version. I'm pretty sure Ghostscript compares the merged PDFs and stores shared content just for one time.

[Torben]

Add the following line to your .bash_profile and you have a nice shortcut: pdfmerge() { gs -dBATCH -dNOPAUSE -q -sDEVICE=pdfwrite -dPDFSETTINGS=/prepress -sOutputFile=$@ ; } This saves you some typing, if you have to use the command a lot. The usage looks like this: pdfmerge merged.pdf mine1.pdf mine2.pdf

[Michal Gonda]

I tried to find description for -dBATCH flag but couldn't. Even man gs doesn't say anything. But great and without any additional programs!

[jmiserez]

pdftk file1.pdf file2.pdf cat output out.pdf will output the merged file as out.pdf

[user829755]

almost a copy of what the OP has described as not working

[Sablefoste]

Since Imagemagick is based upon Ghostscript, if you already have it, you can use it as well; convert file1.pdf file2.pdf outputfile.pdf.

[markgalassi]

Do not use convert for postscript or PDF files unless you go from vector to raster and never go back. It is hard to overstate what a bad idea this is.

[markgalassi]

Do not use convert for postscript or PDF files unless you go from vector to raster and never go back. It is hard to overstate what a bad idea this is.

[markgalassi]

Do not use convert for postscript or PDF files unless you go from vector to raster and never go back. It is hard to overstate what a bad idea this is.

[firegurafiku]

What is the point of using $(ls *.pdf) in place of simple wildcard *.pdf?

[sabujp]

pdfjoin breaks annotations or additional non graphics items

[a coder]

pdftk is not available for EL7 systems due to missing dependency libgcj.

[Paul Rougieux]

@DaniloBargen pdfunite doesn't break external hyperlinks. I have merged documents with links. Links were kept functional. But pdfunite might break internal hyperlinks as indicated in the blog you mentioned.

[Mahendar Patel]

@NateKohn @IcyFlame Unfortunately it did not work for me - said Unknown device: pdfwriter. Like you I preferred that approach because I had it installed already. pdftk worked like a charm though.

[gaoithe]

pdfunite worked well for me. preserved resolution of original pdfs. pdf size simple addition of original pdfs. convert -compress lossless did the same as convert, resolution lost and file size much increased. pdfunite version 0.22.1 vs ImageMagick 6.7.8-9. 69k + 130k pdf => 198k with pdfunite, => 771k with convert. text+gfx(originally odt) + gfx(pdf print to pdf) pdf.

[v_2e]

The "URW Palladio L" font became invisible after pdfjoin'ing the pages.

[Charles-Édouard Coste]

If you don't want to type each file name : pdftk ``ls *.pdf`` cat output out.pdf (with just one backticks instead of two, but I didn't manage to render it with the stackoverflow markdown parser)

[JAB]

@RobertSmith Interestingly, when I tried the first command it rotated some pages (that seemed to be the same proportions as the others) while the /prepress version did no rotation.

[Zelphir Kaltstahl]

@Charles-EdouardCoste You can use two backticks as surrounding backticks if you want to use backticks inside. Or three, if you want to use double backticks. Not sure if it goes even higher. But backticks is not recommended anymore in Bash, so better would be to use $(ls *.pdf)

[Zelphir Kaltstahl]

@alcohol What does the cat in this command (not in general) do? I was able to merge two pdfs without that and don't see any issues with the resulting pdf.

[Calaf]

pdfunite usually works well, but if it says "Unimplemented Feature: Could not merge encrypted files ", pdfjoin is a nice alternative. For whatever reason, pdfjoin doesn't complain of encryption.

[alcohol]

@Zelphir cat is the operation here (that pdftk performs), not the shell command.

[Bhoom Suktitipat]

Works for me natively (probably) without any installation on MacOs Sierra (OsX).

[salotz]

Doesn't work with a pdf I have gives Unimplemented Feature: Could not merge encrypted files ('MR1418_introduction.pdf'). But pdftk was able to handle it, albeit admonishing me for not having some passwords you don't need.

[z0r]

pdftk is neat because you can easily select page ranges to merge: pdftk A=file1.pdf B=file2.pdf cat A1-3 B1 output out.pdf

[awelkie]

I like this solution because it keeps section headers that I can use to jump around with my pdf software

[Campa]

Incredibly much faster than convert, and the resolution does not get worse. TOP

[enkiv2]

"Argument list too long" indicates that you're going over the shell's allocated buffer size for the environment -- it's not actually a limitation of the tool. In such a case, switching to Python may be overkill, since you can just batch: find input -name *.pdf | xargs -P1 -n500 sh -c 'pdfunite "$@" output-date +%s.pdf' && pdfunite output-*.pdf output.pdf (This will create batches of 500 files processed serially, make the resulting temporary files sort in the right order, and produce an appropriate output file; you'll need to clean up the temporary files after)

[Mingwei Zhang]

For my usage, pdftk somehow just stuck and not producing anything, while gs gives perfect result. convert's resolution download is true (with default setting).

[Nostradamnit]

Still works to perfection. I didn't see any size change either; 9.2 mb of files made a single 9.2mb file. Didn't have any hyperlinks, so can't comment on that aspect.

[Diego 72]

Since GCJ is deprecated link), pdftk is as well (most distros phased it out already). A nice alternative is pdfunite and the other poppler-utils.

[palik]

v2.08 of pdfjoin doesn't proper work for me. One of the input PDF-files contains filled forms. But no inputs appear in the result PDF-file.

[bjd2385]

Just put the following in a bash script: pdfunite $@ out.pdf

[Antek]

gs for joining documents apparently created under Windows gave me this error Missing glyph CID=48, glyph=0030 in the font EAAAAB+Tahoma,Bold . The output PDF may fail with some viewers. and indeed, on the default Ubuntu viewer I couldn't browse it

[Egel]

Additionally with reference to @firegurafiku answer, with ls *.pdf wildcard you lose a control over the order of merged files. In an example, the following list: 1.pdf, 2.pdf, 3.pdf, ..., 10.pdf, ..., 100.pdf will actually be merged like 1.pdf, 10.pdf, 100.pdf, 2.pdf, 3.pdf (due to default Linux way of ordering files - here you have more details about this problem - stackoverflow.com/q/22948042/1977012).

[Pysis]

Beautiful answer. I was having a ton of frustration ever since I heard Mac's Preview application could help, but I barely got it to work the first time, and none the second time I needed it. The biggest trouble was it not being determinant with its actions. I used to be able to drag entire PDF files over, but tried invidiual thumbnails, and also had saving issues between a new copy and at all! So this solution was a nice break. The only additional task I had was throwing in a GIF image at the end using Preview since Ghostscript could not handle that. Also installed from Homebrew!

[k.stm]

So much this. Parabola for instance doesn’t package pdftk anymore either because of its dependance on gcj, for which support has been dropped I believe. Despite searching for pdf manipulation tools via pacman -Ss pdf, I missed this. Thanks for this answer! I should receive way more upvotes, so it shows up right next to suggestions for pdfunite or pdftk.

[Tom Russell]

@AaronR Happy to report working well on openSuse Leap 42.3. Its speed is quite blazing, combining 100-odd pages in less than a second.

[supergra]

This is perfect. Using gs (all variants listed above), a simple merge of two PDFs, 2MB and 500Kb, was taking minutes to complete and resulting in a 40MB file! pdftools completes instantaneously with identical file size.

[Winny]

Breaks hyperlinks.

[Winny]

This preserves hyperlinks. Nice!

[adonese]

I'm just gonna run this command so that it will be saved to my fish history! I'll definitely need it in the future.

[Jimbo]

The version from the official website doesn't work. This answer does as of January 2019 (still official).

[mvreijn]

A note about PDF Forms. Thusfar, pdftk is the only tool I have tried that preserves my PDF forms exactly as they were (functional in any PDF reader plus Acrobat Reader).

[Admin]

@Winny, for me it didn't preserve hyperlinks. any idea why?

[Winny]

@EnanAjmain I used gs -dBATCH -dNOPAUSE -q -sDEVICE=pdfwrite -dPDFSETTINGS=/prepress -sOutputFile=output.pdf a.pdf b.pdf c.pdf This worked for me

[Admin]

I used it too. But in my case it didn't work. I then used an online service. Thanks for replying.

[MiniMax]

The gs -dBATCH -dNOPAUSE -q -sDEVICE=pdfwrite -dPDFSETTINGS=/prepress -sOutputFile=merged.pdf mine1.pdf mine2.pdf can be shortened to the gs -q -sDEVICE=pdfwrite -dPDFSETTINGS=/prepress -o merged.pdf mine1.pdf mine2.pdf. From Documentation: "As a convenient shorthand you can use the -o option followed by the output file specification as discussed above. The -o option also sets the -dBATCH and -dNOPAUSE options. This is intended to be a quick way to invoke ghostscript to convert one or more input files."

[Wallace Kelly]

On my fresh install of Linux Mint, this ran in the Terminal window without requiring any installs or path adjustments. Nice!

[Freedom_Ben]

pdftk is deprecated and no longer available in Fedora repos for example. If you can't use pdfunite cause encryption, then try pdf-stapler: github.com/hellerbarde/stapler

[JonShipman]

This also works well in windows with WSL. Install ghostscript from apt first.

[doak]

I was used to pdfunite, but it resulted today in a 850 MB PDF (original 24 MB). pdftk did it with 44 MB, gs with only 32 MB (both with top resolution as far as I can tell).

[David Granqvist]

qpdf seems to break hyperlinks in the document

[Skippy le Grand Gourou]

pdftk also allows to duplicate pages within a single document : pdftk input.pdf cat 1-10 10 10 10 10-20 output output.pdf would produce input.pdf with 5 times the page 10.

[Jonathan Holvey]

Although difficult to get your head around the complex options to start with, qpdf is a very handy and powerful tool. Online documentation is available here

[minexew]

It seems that page numbers from individual documents are also not preserved. (pdfunite version 0.73.0)

[ostrokach]

sudo apt-get install poppler-utils

[Yai0Phah]

Strange enough. After merging with gs, some of pages are rotated.

[Siwoku Adeola]

This worked perfectly and also gave a clearer merged document that the other commands I tried out. Thanks for the post.

[mario ruiz]

OMHO the best to tool to do these type of tasks

[alphaGeek]

gs -dBATCH -dNOPAUSE -q -sDEVICE=pdfwrite -dPDFSETTINGS=/prepress -sOutputFile=merged.pdf mine1.pdf mine2.pdf This worked perfectly on cygwin @Win 10 64 bit machine. Thanks

[Henrik Pingel]

pdfjam package doesn't include pdfjoin script anymore. You can find the script here

[Hashim Aziz]

Came here looking for a qpdf solution but didn't want to wade through the documentation yet again to figure it out, thank you.

[Gabriel Staples]

WATCH OUT! The last input parameter is where the program writes the output, so if you accidentally forget out.pdf and do pdfunite in1.pdf in2.pdf instead of pdfunite in1.pdf in2.pdf out.pdf, you'll have just accidentally written in1.pdf right on top of in2.pdf, ruining it! Don't do that.

[lmat - Reinstate Monica]

Using a shell wildcard is great as long as the order works for you! Check the order first with echo *.pdf | tr ' ' $'\n' or so!

[qdread]

@Winny I needed to add dPrinted=false to preserve hyperlinks. Otherwise it broke the links for all but the first pdf. See tex.stackexchange.com/questions/245801/…

[Rainb]

pdftk is a hell to install

[KCD]

Different pages size, want to "shrink to fit"? Try another tool included in poppler: pdftocairo -pdf -paper A4 in.pdf out.pdf ... and make backups as people point out you can override inputs

[tejasvi88]

Or you can install it anyway. Total size of dependencies is < 100 kb.

[tejasvi88]

pdftools is a wrapper for PyPDF2. See this answer.

[pts]

Ghostscript does many lossy transformations (e.g. page rotation, image quality degradation and dropping some metadata) during PDF-to-PDF conversion. If you want to have a peace of mind of having all PDF features intact, then don't use Ghostscript.

[Rainb]

@pts sure, or you can just use custom flags, so that ghostscript doesn't use lossy compression. superuser.com/questions/360216/…

[caram]

If pages in the even.pdf file are reversed (typical when you scan on a non-double-sided scanner), you will want to use this instead: qpdf --collate --empty --pages odd.pdf even.pdf z-1 -- merged.pdf

[pts]

@Rainb: The PDF output of Ghostscript is lossy even if it is configured to use lossless image compression, because Ghostscript ignores some PDF features (e.g. interactive).

[Ethan Brown]

I did not find that pdfunite created unreasonably large files. My two input files were both about 500k and the resulting output file was about 1 MB as one would expect.

[Antonin GAVREL]

Nice script. You can arrange the order by prefixing each page with "A_", "B_" etc or simply add 'z' if one document is at page 1 and you want it to be at the last page, assuming your files are named with letters and digits, and not starting with 'z'.

[Charles D Pantoga]

can be installed with homebrew +1 havent tried yet

[xZero]

I was fast to use it and not read the comments, so it overwrote the last file. Silly me run "pdfunite *.pdf" and then wondered where's the output file until I realized... Luckily I had a backup; It's always a good idea to have a backup.

[Tarick Welling]

This is quite the undervalued answer. The qpdf tool is great

[Andrea Vacondio]

This is no longer open source

[Doberon]

it's true, qpdf is multi platform, portable and possible used at scripts

[Doberon]

for extract qpdf "in.pdf" --pages . 1 -- "out.pdf"

[Ingo Mi]

What is the command for papersize A4? Thanks for your answer.

[Admin]

Qpdf is high in quality and actively maintained, which makes it a much better option than pdftk.

[Admin]

Pdftk is old, buggy, and not well maintained. It uses a fork of the iText library that is buggy. It has poor error handling and can fail on pdf 1.5 input files. Qpdf is a higher-quality alternative for most of its functionality.

[tripleee]

"Likely to be installed" depends entirely on your Linux distro. It's probably true e.g. for Ubuntu and very probably not true for Alpine or Gentoo.

[tripleee]

@bjd2385 An unquoted $@ is always an error; you want "$@". Without the quotes, it will break on file names which contain whitespace or shell metacharacters.

[tripleee]

The package name probably refers to a Debian package.

[rickhg12hs]

So far, qpdf seems to generate the best output where other viewing tools find no errors/issues.

[Rainb]

Ghostscript tends to fail if with some pdfs, what a shame

[chillitom]

In the case of an error about encrypted documents try the following first: ls *.pdf | xargs -I '{}' qpdf --decrypt {} --replace-input

[Gabriel Staples]

pdfunite broke my output PDF by deleting entire pages of fields I had filled in in this PDF here. I ran pdfunite in1.pdf in2.pdf out.pdf. I don't know why it did that, but it must be some issue with the metadata which stores the filled-in text fields inside the fillable PDF, or something. Using pdftk, as described here, however, worked perfectly: pdftk in1.pdf in2.pdf cat output out.pdf.

[Gabriel Staples]

I then ran my tool: pdf2searchablepdf out.pdf for good measure, to ensure the PDF metadata fillable fields get turned into an image and then OCRed back into a PDF, in order to remove the metadata fields to ensure I don't have problems printing it at Staples, which also was unable to print properly some of the metadata fields in the original PDF.

[Gabriel Staples]

pdftk worked when pdfunite did not. pdfunite erroneously removed metadata fields from a fillable PDF, as I described here. pdftk did not--it performed correctly.

[Timo]

I had two files on Mac Monterey, each 1 M resulting in output of 15 M. I then used --object-streams=generate and the output size was normal.

[Pavol Travnik]

I wonder why it is so low. This is exactly what I was looking for. Thank you.

[Doberon]

QPDF is great. Use it this way to extract pages 1-10 from input.pdf and save it as output.pdf: qpdf input.pdf --pages . 1-10 -- output.pdf