run_fuzz.sh

   1 #!/bin/sh
   2
   3 set -eu
   4
   5 # Download a random, recent paper from arXiv and try to run llf on it.
   6
   7 # Note the most common problems are unmatched {, which LLF rejects,
   8 # or an encoding other than UTF-8. These are not considered errors
   9 # in LLF.
  10 tmpdir=$(mktemp -d)
  11 root=$(dirname $(readlink -f "$0") )
  12 cd "${tmpdir}"
  13
  14 discipline=$(dd if=/dev/urandom count=1 bs=1 2>/dev/null | od -A none -d)
  15 discipline=$(( ${discipline} % 20 ))
  16 case ${discipline} in
  17 0) discipline="astro-ph"; ;;
  18 1) discipline="cond-mat"; ;;
  19 2) discipline="cs"; ;;
  20 3) discipline="econ"; ;;
  21 4) discipline="eess"; ;;
  22 5) discipline="gr-qc"; ;;
  23 6) discipline="hep-ex"; ;;
  24 7) discipline="hep-lat"; ;;
  25 8) discipline="hep-ph"; ;;
  26 9) discipline="hep-th"; ;;
  27 10) discipline="math"; ;;
  28 11) discipline="math-ph"; ;;
  29 12) discipline="nlin"; ;;
  30 13) discipline="nucl-ex"; ;;
  31 14) discipline="nucl-th"; ;;
  32 15) discipline="physics"; ;;
  33 16) discipline="q-bio"; ;;
  34 17) discipline="q-fin"; ;;
  35 18) discipline="quant-ph"; ;;
  36 19) discipline="stat"; ;;
  37 esac
  38
  39 absurl=$(curl -s "http://export.arxiv.org/rss/${discipline}" | grep -Eo '[^"<>]+arxiv.org/abs/[^"<>]+' | sort -R | head -n 1)
  40 srcurl=$(printf %s\\n "${absurl}" | sed -re 's/\/abs\//\/e-print\//')
  41 srcname=$(printf %s\\n "${srcurl}" | sed -re 's/.*\///')
  42 curl -Ls "${srcurl}" -o "${srcname}"
  43 mime=$(file -bi "${srcname}"  | sed -re 's/;.*//')
  44
  45 case ${mime} in
  46 application/x-gzip)
  47         tar -xf "${srcname}" 2>/dev/null >/dev/null || gunzip -c "${srcname}" > "${srcname}.raw"
  48         ;;
  49 application/gzip)
  50         tar -xf "${srcname}" 2>/dev/null >/dev/null || gunzip -c "${srcname}" > "${srcname}.raw"
  51         ;;
  52 text/x-tex) ;;
  53 application/pdf)
  54         printf 'No TeX source for %s\n' "${absurl}"
  55         printf 'tmpdir = %s\n' "${tmpdir}"
  56         exit 1
  57         ;;
  58 *)
  59         printf 'Unknown mimetype %s at %s\n' "${mime}" ${tmpdir}
  60         printf 'tmpdir = %s\n' "${tmpdir}"
  61         exit 1
  62         ;;
  63 esac
  64
  65 oldifs="${IFS}"
  66 IFS="
  67 "
  68 err=0
  69 for f in $(ls); do
  70         mime=$(file -bi "${f}"  | sed -re 's/;.*//')
  71         [ "${mime}" = "text/x-tex" ] || continue
  72         case "${f}" in
  73         *sty) continue ;;
  74         *bst) continue ;;
  75         esac
  76
  77         "${root}/llf.lua" -c "${root}/llfrc.lua" < "${f}" 2>/dev/null > /dev/null && continue
  78         err=1
  79         printf 'LLF could not handle %s\n' "${f}"
  80 done
  81
  82 if [ "${err}" -eq 0 ]; then
  83         printf 'All okay for %s\n' "${absurl}"
  84         rm -rf "${tmpdir}"
  85
  86         exit 0
  87 else
  88         printf 'There were some errors for %s\n' "${absurl}"
  89         printf 'tmpdir = %s\n' "${tmpdir}"
  90 fi