#!/bin/bash
# Usage: 1colpdf *.pdf,  where the pdf files are proceeding files in 2 columns
# Creates 1col-*.pdf file, 1-column files for ebook readers

# Needs: pdflatex pdftotext pdfcrop

for SOURCE in $@
do

# This parts outputs 2n pages, one per left col, one per right col
pdflatex -interaction batchmode -jobname="tmp1" \
"\documentclass[twoside]{article}" \
"\usepackage{pdfpages}" \
"\begin{document}" \
"\includepdf[pages=-,doublepages,offset={4.25in 0in}]{$SOURCE}" \
"\end{document}"

# This parts outputs the text in the PDF file up to the word Abstract
# It also bolds the first line it reads, hoping it is the title of the paper
pdftotext -eol unix -l 1 -layout $SOURCE - |
{ read x
echo "{\\bf $x}\\\\[\\medskip]"
while read x
do
   if [[ "foo$x" == foo*Abstract* ]]
     then break
   fi
   echo "$x\\\\"
done } > tmp2

# This parts merges the 2n pages + 1 page for title and author
pdflatex -interaction batchmode -jobname="tmp3" \
"\documentclass{article}" \
"\setlength{\textwidth}{3.5truein}" \
"\pdfpagewidth=3.5truein" \
"\usepackage{pdfpages}" \
"\pagestyle{empty}" \
"\begin{document}" \
"\begin{flushleft}\large" \
`cat tmp2` \
"\end{flushleft}" \
"\newpage" \
"\includepdf[pages=-,noautoscale]{tmp1}" \
"\end{document}"

# This part shrinks the margins to fit the reduced file
# It's very slow!
pdfcrop --margins '10 10 10 10' tmp3.pdf 1col-$SOURCE

# cleanup
rm tmp1.* tmp2 tmp3.*

done