etc/mirrors.d/50-eprint-bib: Collect IACR ePrint bibliography.

[mirror-admin] / etc / mirrors.d / 50-eprint-bib

#! /bin/bash

set -e

## Initial setup.
mkdir -p $MIRRORS/eprint-bib; cd $MIRRORS/eprint-bib
eprint=http://eprint.iacr.org
rm -f list tags

## Fetch the master index.
curl -sS -o list -z '1 hour ago' $eprint/complete/
[ -r list ] || exit 0

## Pick out the article tags.
sed -n '/^<a href="\/\([0-9]\+\/[0-9]\+\)">.*$/ s::\1:p' list >tags

## For each tag, make sure we have the article information.
while read tag; do
  dir=${tag%/*}
  mkdir -p $dir
  file=$dir/eprint-${tag/\//-}.bib
  [ -r $file ] || {
    curl -sS -o $file.in "$eprint/cgi-bin/cite.pl?entry=$tag"
    sed -n "
    /^<PRE>/,/<\/PRE>/ {
      /^<PRE>/ d
      s:</PRE>.*::
      s%note =.*\$%url = {http://eprint.iacr.org/$tag},%
      p
    }" $file.in >$file.sed
    mv $file.sed $file
    rm -f $file.in
    echo $tag
  }
done <tags

## Build the main bibliography.
{
  cat header
  find . -name 'eprint-*.bib' -print | sort | xargs cat
} >eprint.bib.new
mv eprint.bib.new eprint.bib