#!/usr/bin/env bash
# Run the set-of-mark table extractor across the Visa + TMUS corpus to validate
# the title-anchored declash seam. Writes review artifacts too (--review adds no
# LLM cost). Smallest / most stacked-table-relevant docs first; heavy
# 10-Q/factbook PDFs last.
set -u
# Extraction artifacts (.tables.json / .review.html / .boxed.pdf) land in the
# standard output/ dir; only run logs + the summary live in a sidecar.
LOGDIR=output/.sweep_logs
mkdir -p "$LOGDIR"
SUMMARY="$LOGDIR/summary.tsv"
: > "$SUMMARY"

docs=(
  # Visa operational performance data — the stacked period tables
  s3://qubera-docs/visa/Q1FY26-Visa-Operational-Performance-Data.pdf
  s3://qubera-docs/visa/Q1FY25-Visa-Operational-Performance-Data-FINAL.pdf
  s3://qubera-docs/visa/Q2FY25-Visa-Operational-Performance-Data-FINAL.pdf
  s3://qubera-docs/visa/Q2FY26-Visa-Operational-Performance-Data.pdf
  s3://qubera-docs/visa/Q4FY25-Visa-Operational-Performance-Data.pdf
  # Visa earnings / presentations / 8-K
  s3://qubera-docs/visa/Q1-2026-Earnings-Release_vF.pdf
  s3://qubera-docs/visa/Q2-2026-Earnings-Release_vF.pdf
  s3://qubera-docs/visa/VISA_991_Q126.pdf
  s3://qubera-docs/visa/VISA_991_Q425.pdf
  s3://qubera-docs/visa/Visa-Inc-First-Quarter-2026-Financial-Results-Presentation.pdf
  s3://qubera-docs/visa/Visa-Inc-Fourth-Quarter-2025-Financial-Results-Presentation.pdf
  s3://qubera-docs/visa/Visa-Inc-Second-Quarter-2026-Financial-Results-Presentation.pdf
  # Heavy multi-page docs last
  s3://qubera-docs/visa/Q1-2026-Investor-Factbook-vFinal.pdf
  s3://qubera-docs/tmus/TMUS_992_Q324.pdf
  s3://qubera-docs/tmus/TMUS_992_Q224.pdf
  s3://qubera-docs/tmus/TMUS_992_Q323.pdf
  s3://qubera-docs/tmus/TMUS_992_Q423.pdf
)

for d in "${docs[@]}"; do
  name=$(basename "$d")
  start=$(date +%s)
  if uv run quber table "$d" --review > "$LOGDIR/$name.log" 2>&1; then
    status=OK
  else
    status=FAIL
  fi
  end=$(date +%s)
  tables=$(grep -oE 'Tables: [0-9]+' "$LOGDIR/$name.log" | tail -1 | grep -oE '[0-9]+$')
  printf '%s\t%s\t%ss\ttables=%s\n' "$status" "$name" "$((end-start))" "${tables:-?}" >> "$SUMMARY"
done
echo "DONE" >> "$SUMMARY"
