`poll-and-download.sh`

67 lines
#!/bin/bash
# Detached poll-until-key-then-download job for USPTO TRTYRAP.
# Polls the ODP key-reveal page via openclaw; when a key (or Generate button)
# appears, captures/generates the key -> .apikey (chmod 600), then runs the
# downloader. Logs everything to poll.log. Caps at ~2h so it can't run forever.
set -uo pipefail
cd "$(dirname "$0")"
LOG=poll.log
KEY_URL="https://data.uspto.gov/apikey/key-reveal"
MAX=60          # iterations
SLEEP=120       # seconds between polls
log(){ echo "[$(date '+%H:%M:%S')] $*" >> "$LOG"; }

extract_state(){
  openclaw browser open "$KEY_URL" >/dev/null 2>&1; sleep 5
  local tid
  tid=$(openclaw browser tabs 2>/dev/null | grep -iB1 "Manage API\|apikey\|key-reveal\|key-warning" | grep "id:" | head -1 | awk '{print $2}')
  openclaw browser evaluate ${tid:+--target-id "$tid"} --fn "() => {
    const t=document.body.innerText||'';
    const cand=[...document.querySelectorAll('input,code,pre,span,td')].map(e=>(e.value||e.textContent||'').trim()).find(v=>/^[A-Za-z0-9_-]{28,60}\$/.test(v));
    const gen=[...document.querySelectorAll('a,button')].find(b=>/generate|create.*key|new api key/i.test(b.textContent));
    let state='BLOCKED';
    if(cand) state='KEY_PRESENT';
    else if(gen) state='CAN_GENERATE';
    else if(/not currently verified with ID\.me/i.test(t)) state='NOT_VERIFIED';
    else if(/sign in|registration requirement/i.test(t)) state='SIGNED_OUT';
    return JSON.stringify({state, key:cand||''});
  }" 2>/dev/null | tail -1
}

click_generate(){
  local tid
  tid=$(openclaw browser tabs 2>/dev/null | grep -iB1 "Manage API\|apikey\|key-reveal" | grep "id:" | head -1 | awk '{print $2}')
  openclaw browser evaluate ${tid:+--target-id "$tid"} --fn "() => { const b=[...document.querySelectorAll('a,button')].find(x=>/generate|create.*key|new api key/i.test(x.textContent)); if(b){b.click();return 'gen-clicked'} return 'no-gen' }" >/dev/null 2>&1
  sleep 6
}

log "=== poll-and-download started (pid $$) ==="
for i in $(seq 1 $MAX); do
  raw=$(extract_state)
  parsed=$(printf '%s' "$raw" | python3 parse_state.py)
  state=${parsed%%$'\t'*}; key=${parsed#*$'\t'}
  log "poll $i/$MAX -> $state"
  if [ "$state" = "CAN_GENERATE" ]; then
    log "clicking Generate API key"; click_generate; raw=$(extract_state)
    parsed=$(printf '%s' "$raw" | python3 parse_state.py)
    state=${parsed%%$'\t'*}; key=${parsed#*$'\t'}
    log "post-generate -> $state"
  fi
  if [ "$state" = "KEY_PRESENT" ]; then
    if [ -n "$key" ]; then
      printf '%s' "$key" > .apikey; chmod 600 .apikey
      log "KEY captured (…${key: -4}); starting download"
      break
    fi
  fi
  sleep $SLEEP
done

if [ -s .apikey ]; then
  log "=== running download_backfill.py ==="
  ./.venv/bin/python download_backfill.py >> "$LOG" 2>&1
  log "=== download finished (exit $?) ==="
else
  log "=== gave up: no key after $MAX polls (still not signed-in/verified) ==="
fi