#!/usr/bin/env bash
# Baseline (main worktree) and branch runs of `quber extract --review` over the seeded random sample.
# A side whose tables.json already exists is skipped, so main baselines from the first pass are kept.
set -u
S=/tmp/claude-1000/-home-mande-repo-quber/bc2f817a-5c6c-4515-9621-2302b82f7f68/scratchpad
OUT=/home/mande/repo/quber/data/que370-review/sample
MAIN=/home/mande/repo/quber-main-baseline
BRANCH=/home/mande/repo/quber
STATUS=$OUT/status.tsv
python3 -c "import json; [print(o['key'], o['folder'], o['pages']) for o in json.load(open('$OUT/sample10.json'))]" | while read key folder pages; do
  pdf=$S/hosted-pdfs/$key.pdf
  for side in main branch; do
    repo=$MAIN; [ $side = branch ] && repo=$BRANCH
    d=$OUT/$folder-$key/$side; mkdir -p $d
    if ls $d/*.tables.json >/dev/null 2>&1; then continue; fi
    t0=$(date +%s)
    (cd $repo && uv run quber extract $pdf --review -o $d > $d/extract.log 2>&1); rc=$?
    n=$(python3 -c "import json,glob; f=glob.glob('$d/*.tables.json'); print(len(json.load(open(f[0]))) if f else -1)")
    printf "%s\t%s\t%s\t%s\t%s\t%ss\n" "$folder" "$key" "$side" "rc=$rc" "tables=$n" "$(( $(date +%s) - t0 ))" >> $STATUS
  done
done
echo DONE2 >> $STATUS