10

A friend of mine recently completed his doctoral thesis. In order to save on toner, he printed the pages with figures on a color printer, but printed everything else on a black and white printer. In order to do this, he went through the document and printed out each page with a figure, by hand.

While he was careful, it seems extremely error prone, and I'd like to know if there is a way to generate a document containing only the pages with figures on them while retaining the formatting and pagination of the full document?

rcollyer
  • 587

2 Answers2

8

It might be easiest to do this sort of thing with the pdfpages package. It's not automatic, but you wouldn't have to print each page individually, and you could generate the list of pages from the list of figures.

Here's an example:

\documentclass{article}
\usepackage{pdfpages}
\begin{document}
\includepdf[pages={3-6, 17, 28, 29-31}]{yourdocument}
\end{document}

One thing to bear in mind when using this method is that the page numbers correspond to the number of the physical pages in yourdocument.pdf, not the page numbers in your document. This means that if you have n \frontmatter pages (which in a thesis, you would) then all of the relevant page numbers will be off by n. The easiest way around this is to simply delete all the frontmatter pages from the pdf document in order to generate the extracted page document, so that the first physical page corresponds to page 1 of the thesis.

Automating

Theoretically, it would be possible to automate the process slightly, although for something that is being done once or twice it is probably not worth it. For those who like the challenge, here are some ideas (based on the discussion in the comments).

  1. Modify the \addtocontents code for figures to also write out a list of page numbers for subsequent use in the pdfpages document. This seems the most work for little gain in actual automation.
  2. Use your favourite scripting language to to extract the lof entries from the .aux file, and parse out the page numbers. The script could then generate the entire pdfpages document. This would probably be the best route if you really needed to automate the process.
  3. You're only doing this once or twice; do 2 above by hand and spend the rest of your time enjoying a beer.
Alan Munn
  • 218,180
  • I hadn't considered pdfpages, but it does make it relatively simple to do, especially if I work from the table of figures. – rcollyer Apr 18 '11 at 19:14
  • @Alan Munn Could one modify \contentsline and pick up the page number from the .lof then pass it on onto a list etc. to automate it? – yannisl Apr 18 '11 at 19:28
  • @rcollyer I'll be to see anyone comes up with a more automatic solution, but since this is something you only do once in a while, doing it automatically may not be worth the effort. – Alan Munn Apr 18 '11 at 19:29
  • @Yiannis, I was thinking along those lines myself. – rcollyer Apr 18 '11 at 19:33
  • @Alan, I've been told the same thing, but I'm lazy, and if an automated solution existed ... – rcollyer Apr 18 '11 at 19:34
  • @Yiannis probably. The main issue would be to avoid duplicate page numbers (i.e. from two or more figures on the same page.) Although this reminds me of "A programmer is someone spends 8 hours writing a program to solve a problem that would take 10 minutes to do by hand." The tradeoff really is how many more times you are likely to do the 10 minutes of work by hand. Of course if automating it only takes an hour... – Alan Munn Apr 18 '11 at 19:35
  • @Alan Munn @rcollyer ...You right about the ten minutes and especially if one was printing a thesis to submit! It would probably take the best part of a weekend to do it properly, duplicates shouldn't be that hard to remove as one would capture only the page number (i.e no parsing or sorting of any strings). – yannisl Apr 18 '11 at 19:49
  • @Yiannis it might be simplest to just work off of the .aux file itself: grab all the lines with {lof} in them and then parse out the page number and construct the whole extraction document in your favourite scripting language. But cutting and pasting from the actual LoF and a couple of regex replaces in a text editor would probably take only a few minutes. – Alan Munn Apr 18 '11 at 19:50
  • @Alan Munn Yes that would be the simplest and quickest. – yannisl Apr 18 '11 at 19:54
  • @Alan, as you're the only one to submit an answer, and it is more than reasonable, I will gladly accept it. But, would you summarize the comment discussion to round out your answer for the lazy among us, first? – rcollyer Apr 19 '11 at 15:24
  • @rcollyer Summary done! – Alan Munn Apr 20 '11 at 02:39
3

I gave a shot at this and came up with a solution. Probably not the best that can be made, but it works. The solution is based on @Alan Munn's answer.

It's a shell script that should work on all Unix systems with pdfLaTeX and pdfpages installed. You use it by giving it the file name of the TeX file (no .tex ending!) and the offset in pages (caused by front matter) as the second argument:

./only-figures.sh document 1

First we load the .aux and prepare to store it in a variable.

STR=$(cat $1.aux |\

We only care about the lines with figures.

 grep figure |\

And of these lines we only care about the number referring to unique pages.

 sed 's/.*}}{\([1-9]*\)}}/\1/' | uniq |\

We then add the offset to the page numbers.

 awk '{print $0+'$2'}' |\

And finally pack them all up in a comma-separated list.

 sed ':a;N;$!ba;s/\n/,/g')

A template file is used for the new temporary document.

cat template.tex | sed s/@@@/$STR/ > only-figures-tmp.tex

The template looks as following:

\documentclass{article}
\usepackage{pdfpages}
\begin{document}
\includepdf[pages={@@@}]{document}
\end{document}

We call pdfLaTeX on the temporary file, move the new PDF and cleans up.

pdflatex only-figures-tmp
mv -f only-figures-tmp.pdf $1-only-figures.pdf
rm only-figures-tmp.*

The whole thing can also be found in this Gist

Henrik Hansen
  • 2,927