decoder-bundle-minimization/work/decode_proof.sh

#!/usr/bin/env bash
# Decode-correctness proof for candidate paq8px decoder builds.
# Every candidate decodes the SAME real project archives; output must be byte-identical
# to the known input (sha256), and to the original binary's output.
set -uo pipefail
B6=/root/hutter67/swarm67/B6-bundle
t=$B6/test
res=$t/decode_proof.tsv
: >"$res"
printf 'binary\tarchive\tout_bytes\tout_sha256\tmatch_expected\trss_kb\twall_s\n' >>"$res"

declare -A BIN=(
  [orig]="$B6/work/paq8px.orig"
  [dyn_O3]="$B6/out/build/dyn_O3"
  [dyn_Os]="$B6/out/build/dyn_Os"
  [static_Os]="$B6/out/build/static_Os"
  [dyn_Os_hard]="$B6/out/build/dyn_Os_hard"
)

check() { # $1 binname $2 binary $3 archive $4 expected file
  local name=$1 b=$2 arc=$3 exp=$4
  [[ -x $b ]] || { printf '%s\t%s\t-\t-\tBINARY_MISSING\t-\t-\n' "$name" "$(basename "$arc")" >>"$res"; return; }
  local out=$t/out_${name}_$(basename "$arc")
  local tt=$t/time_${name}_$(basename "$arc").time
  /usr/bin/time -v -o "$tt" timeout 900 "$b" -d "$arc" "$out" >"$t/${name}_$(basename "$arc").log" 2>&1
  local rc=$?
  local bytes=- sha=- match=NO rss=- wall=-
  [[ -f $out ]] && { bytes=$(stat -c%s "$out"); sha=$(sha256sum "$out" | cut -d' ' -f1); }
  local esha; esha=$(sha256sum "$exp" | cut -d' ' -f1)
  [[ $sha == "$esha" ]] && match=YES
  [[ $rc -ne 0 ]] && match="rc=$rc"
  rss=$(awk '/Maximum resident set size/ {print $NF}' "$tt" 2>/dev/null)
  wall=$(awk '/Elapsed \(wall clock\)/ {print $NF}' "$tt" 2>/dev/null)
  printf '%s\t%s\t%s\t%s\t%s\t%s\t%s\n' "$name" "$(basename "$arc")" "$bytes" "$sha" "$match" "${rss:--}" "${wall:--}" >>"$res"
}

for arc in "$t/level1/slice32k.paq8px216" "/root/hutter67/swarm67/P1-paq8px-cal/runs/small/small.paq8px216"; do
  case $arc in *slice32k*) exp=$t/level1/slice32k.bin ;; *) exp=/root/hutter67/swarm67/P1-paq8px-cal/runs/small/input.txt ;; esac
  for name in orig dyn_O3 dyn_Os static_Os dyn_Os_hard; do
    check "$name" "${BIN[$name]}" "$arc" "$exp"
  done
done
column -t -s$'\t' "$res"