4 ms·

Wkhtmltopdf - Shell utility to convert html to PDF using webkit.

37 points by xd 16y ago

cstuder 16y ago
On a related note: I'm looking for a command line utility (*nix) to go the other way. Any recommendations for a tool which extracts text from a PDF?
- sho_hn 16y ago
  pdftotext and pdftohtml are shipped with poppler, the de-facto PDF implementation on Linux systems today.
  - hassy 16y ago
    pdftotext can produce garbled output, especially from two-column PDFs.* Depending on what you need, scripting Adobe Reader may be an option as it does a better job. * I did a sizeable data mining project a while ago which required converting a lot of scientific papers to plain text. pdftotext didn't work well on a lot of them, so I had a script sending click events to Adobe Reader running overnight.
    - silentbicycle 16y ago
      While still a very new project, pdfsplit (http://dmwit.com/pdfsplit/ http://dmwit.com/pdfsplit/) may be worth a try.
      - cstuder 16y ago
        Thanks for all the links, I'll check those out.
daeken 16y ago
This tool works impressively well. I use it to convert my CV to PDF, as I don't trust my CSS to work everywhere.
acabal 16y ago
This is a great tool, I've been using it for a long time. Only downside is that on Ubuntu Server certain graphics libraries aren't in the PPA so you have to download a static binary instead of using apt-get. But it's a minor quibble for a great program.
silentbicycle 16y ago
I tried porting it to OpenBSD and found building it from source to be surprisingly complicated. IIRC, it wanted me to install their own fork of qt, just to render PDFs. Pass. I went with htmldoc (http://www.htmldoc.org/ http://www.htmldoc.org/) instead. Perhaps the rendered PDFs aren't as pretty, but for my purposes, it works quite well.
billybob 16y ago
We are using this. Seems to do fine. My only complaint is that trying to pronounce the name makes me choke on my own uuvula.
alanh 16y ago
Very useful tool. Beware a tough time getting various fonts to work, though — and some characters (thin space anyone?) will just completely crash it.
- togasystems 16y ago
  I am using this for two different projects. The font issue can lead to a lot of head banging. Try messing around with your css to get it close to what you expected.
acangiano 16y ago
I use the excellent Pandoc for my conversion needs: http://johnmacfarlane.net/pandoc/ http://johnmacfarlane.net/pandoc/
sim0n 16y ago
We're using this on http://interstateapp.com http://interstateapp.com to allow users to download a PDF version of their roadmap. The script works very well!
thangalin 16y ago
Use wkhtmltoimage to create images of book-quality syntax-highlighted source code: http://superuser.com/questions/213217/convert-html-to-image http://superuser.com/questions/213217/convert-html-to-image Use gvim to pick from 100 different ready-made colour schemes: http://www.vi-improved.org/color_sampler_pack/ http://www.vi-improved.org/color_sampler_pack/ Shameless plug: I describe this technique in an appendix of my technical manual. http://www.whitemagicsoftware.com/books/indispensable/ http://www.whitemagicsoftware.com/books/indispensable/ For example: #!/bin/bash echo Converting $1 to $1.html... # Schemes: pyte vim $1 -c "set nobackup" ${2:+"-c set number"} \ -c ":colorscheme tango-morning" \ -c ":TOhtml" \ -c "wq" -c ":q" echo Converting $1.html to $1.png... wkhtmltoimage --transparent --minimum-font-size 80 \ --format png --quality 100 --width 4000 \ $1.html $1.png mv $1.png $1.orig.png # For scripts that are under 80 characters, make the generated image always # the same size (4000 pixels = width + (border x 2) = 3950 + 25 x 2). LENGTH=$(awk 'BEGIN {x=0} {if( length($0)>x ) x=length()} END {print x;}' $1) if [ $LENGTH -le 80 ]; then EXTENT=true fi echo Trimming $1.png... convert -format png $1.orig.png -background none -antialias -trim +repage \ -density 150x150 ${EXTENT:+-extent 3950x} \ -bordercolor none -border 25 \ $1.png echo Removing temporary files... rm -f $1.orig.png rm -f $1.html
jbaker 16y ago
Cool. Can you get an image out of it ?
- xd 16y ago
  There is a utility on the same project page called: wkhtmltoimage
asymptotic 16y ago
This is fantastic, thank you so much for posting this. I felt like pointing out another headless Webkit project, PhantomJS (http://www.phantomjs.org/ http://www.phantomjs.org/), which can also perform HTML to PDF conversion. edit: I've just run a comparison between wkhtmltopdf and PhantomJS and wkhtmltopdf is far superior. wkhtmltopdf produces correct output and bookmarks but has a slightly lower quality, where PhantomJS's output was not correct and without bookmarks but at a slightly higher quality. From my perspective correctness beats quality.
Jetlag 16y ago
I was briefly looking for exactly this, after our lawyer suggested manually printing out thousands of emails and scanning to PDF. Luckily my encouragement to use a PST file finally won out. Not a confidence inspiring moment.

37 points by xd 16y ago

On a related note: I'm looking for a command line utility (*nix) to go the other way. Any recommendations for a tool which extracts text from a PDF?

sho_hn 16y ago
pdftotext and pdftohtml are shipped with poppler, the de-facto PDF implementation on Linux systems today.
- hassy 16y ago
  pdftotext can produce garbled output, especially from two-column PDFs.* Depending on what you need, scripting Adobe Reader may be an option as it does a better job. * I did a sizeable data mining project a while ago which required converting a lot of scientific papers to plain text. pdftotext didn't work well on a lot of them, so I had a script sending click events to Adobe Reader running overnight.
  - silentbicycle 16y ago
    While still a very new project, pdfsplit (http://dmwit.com/pdfsplit/ http://dmwit.com/pdfsplit/) may be worth a try.
    - cstuder 16y ago
      Thanks for all the links, I'll check those out.

daeken 16y ago

This tool works impressively well. I use it to convert my CV to PDF, as I don't trust my CSS to work everywhere.

acabal 16y ago

This is a great tool, I've been using it for a long time. Only downside is that on Ubuntu Server certain graphics libraries aren't in the PPA so you have to download a static binary instead of using apt-get. But it's a minor quibble for a great program.

silentbicycle 16y ago

I tried porting it to OpenBSD and found building it from source to be surprisingly complicated. IIRC, it wanted me to install their own fork of qt, just to render PDFs. Pass. I went with htmldoc (http://www.htmldoc.org/ http://www.htmldoc.org/) instead. Perhaps the rendered PDFs aren't as pretty, but for my purposes, it works quite well.

billybob 16y ago

We are using this. Seems to do fine. My only complaint is that trying to pronounce the name makes me choke on my own uuvula.

alanh 16y ago

Very useful tool. Beware a tough time getting various fonts to work, though — and some characters (thin space anyone?) will just completely crash it.

togasystems 16y ago
I am using this for two different projects. The font issue can lead to a lot of head banging. Try messing around with your css to get it close to what you expected.

acangiano 16y ago

I use the excellent Pandoc for my conversion needs: http://johnmacfarlane.net/pandoc/ http://johnmacfarlane.net/pandoc/

sim0n 16y ago

We're using this on http://interstateapp.com http://interstateapp.com to allow users to download a PDF version of their roadmap. The script works very well!

thangalin 16y ago

Use wkhtmltoimage to create images of book-quality syntax-highlighted source code: http://superuser.com/questions/213217/convert-html-to-image http://superuser.com/questions/213217/convert-html-to-image Use gvim to pick from 100 different ready-made colour schemes: http://www.vi-improved.org/color_sampler_pack/ http://www.vi-improved.org/color_sampler_pack/ Shameless plug: I describe this technique in an appendix of my technical manual. http://www.whitemagicsoftware.com/books/indispensable/ http://www.whitemagicsoftware.com/books/indispensable/ For example: #!/bin/bash echo Converting $1 to $1.html... # Schemes: pyte vim $1 -c "set nobackup" ${2:+"-c set number"} \ -c ":colorscheme tango-morning" \ -c ":TOhtml" \ -c "wq" -c ":q" echo Converting $1.html to $1.png... wkhtmltoimage --transparent --minimum-font-size 80 \ --format png --quality 100 --width 4000 \ $1.html $1.png mv $1.png $1.orig.png # For scripts that are under 80 characters, make the generated image always # the same size (4000 pixels = width + (border x 2) = 3950 + 25 x 2). LENGTH=$(awk 'BEGIN {x=0} {if( length($0)>x ) x=length()} END {print x;}' $1) if [ $LENGTH -le 80 ]; then EXTENT=true fi echo Trimming $1.png... convert -format png $1.orig.png -background none -antialias -trim +repage \ -density 150x150 ${EXTENT:+-extent 3950x} \ -bordercolor none -border 25 \ $1.png echo Removing temporary files... rm -f $1.orig.png rm -f $1.html

jbaker 16y ago

Cool. Can you get an image out of it ?

xd 16y ago
There is a utility on the same project page called: wkhtmltoimage

asymptotic 16y ago

This is fantastic, thank you so much for posting this. I felt like pointing out another headless Webkit project, PhantomJS (http://www.phantomjs.org/ http://www.phantomjs.org/), which can also perform HTML to PDF conversion. edit: I've just run a comparison between wkhtmltopdf and PhantomJS and wkhtmltopdf is far superior. wkhtmltopdf produces correct output and bookmarks but has a slightly lower quality, where PhantomJS's output was not correct and without bookmarks but at a slightly higher quality. From my perspective correctness beats quality.

Jetlag 16y ago

I was briefly looking for exactly this, after our lawyer suggested manually printing out thousands of emails and scanning to PDF. Luckily my encouragement to use a PST file finally won out. Not a confidence inspiring moment.