#!/usr/bin/env bash
# Prospect reply watcher: polls sean@seanmattsson.com for inbound mail that may be
# replies to the prospecting engine's sent campaigns. Wakes a worker only when a
# candidate appears; stays silent otherwise (and during quiet hours 10pm-6am ET).
set -euo pipefail
source "$HATCH_HOOK_RUNTIME"

STATE_DIR="$HOME/hooks/state/prospect-reply-watch"
SEEN="$STATE_DIR/seen_ids"
PENDING="$STATE_DIR/pending.tsv"
mkdir -p "$STATE_DIR"
touch "$SEEN"
: > "$PENDING"

DRY=0
if [ "${HATCH_HOOK_DRY_RUN:-0}" = "1" ]; then
  DRY=1
fi

# Quiet hours: 10pm-6am ET. Stay silent WITHOUT marking seen, so the morning
# poll catches overnight replies and reports them after quiet hours.
HOUR=$(TZ=America/New_York date +%H)
if [ "$HOUR" -ge 22 ] || [ "$HOUR" -lt 6 ]; then
  silent "quiet hours" "{\"hour\":\"$HOUR\"}"
fi

ACCT="d48c9e2643f643aeb3c4dacf347245b8"
SENTLOG="$HOME/workspace/goals/sales-weapon-ae-copilot/hidden_files/prospect-queues/sent-log.json"

RAW=$(hatch_gws_cli gmail --account "$ACCT" +triage \
  --query 'newer_than:1d -from:sean@seanmattsson.com -from:smattsson@stratixsystems.com' \
  --max 30 --format json 2>/dev/null) || {
  log "gmail fetch failed" '{}'
  silent "fetch failed" '{}'
}

# Booking notifications come FROM Sean's own address (Google Calendar appointment
# scheduling), so the main query excludes them. Fetch them separately.
BOOKING=$(hatch_gws_cli gmail --account "$ACCT" +triage \
  --query 'subject:"Appointment booked" newer_than:1d' \
  --max 10 --format json 2>/dev/null) || BOOKING='{"messages":[]}'

ALL_MSGS=$(printf '%s\n%s' "$RAW" "$BOOKING" | jq -s '[.[].messages[]?] | unique_by(.id)')

mapfile -t WATCH_EMAILS < <(jq -r '.[].email' "$SENTLOG" 2>/dev/null | sort -u)
mapfile -t WATCH_DOMAINS < <(jq -r '.[].email | split("@")[1]' "$SENTLOG" 2>/dev/null | sort -u)

FOUND=0
while IFS=$'\t' read -r ID FROM SUBJ DATE; do
  [ -z "${ID:-}" ] && continue
  SENDER=$(printf '%s' "$FROM" | grep -oE '[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Za-z]{2,}' | tail -n 1 || true)
  [ -z "${SENDER:-}" ] && continue
  SENDER_LC=$(printf '%s' "$SENDER" | tr '[:upper:]' '[:lower:]')
  DOMAIN=${SENDER_LC#*@}
  MATCH=0
  for e in ${WATCH_EMAILS[@]+"${WATCH_EMAILS[@]}"}; do
    EL=$(printf '%s' "$e" | tr '[:upper:]' '[:lower:]')
    if [ "$SENDER_LC" = "$EL" ]; then MATCH=1; break; fi
  done
  if [ "$MATCH" = "0" ]; then
    for d in ${WATCH_DOMAINS[@]+"${WATCH_DOMAINS[@]}"}; do
      if [ "$DOMAIN" = "$d" ]; then MATCH=1; break; fi
    done
  fi
  # Delivery bounces are always candidates: the worker identifies the failed recipient.
  # Google Calendar booking notifications ("Appointment booked") are always
  # candidates: the worker logs the meeting and sends the confirmation + form.
  case "$SENDER_LC" in
    *mailer-daemon*|*postmaster*) MATCH=1 ;;
  esac
  case "$SUBJ" in
    *"Appointment booked"*) MATCH=1 ;;
  esac
  [ "$MATCH" = "0" ] && continue;
  if grep -qxF "$ID" "$SEEN" 2>/dev/null; then continue; fi
  printf '%s\t%s\t%s\t%s\n' "$ID" "$FROM" "$SUBJ" "$DATE" >> "$PENDING"
  if [ "$DRY" = "0" ]; then
    printf '%s\n' "$ID" >> "$SEEN"
  fi
  FOUND=$((FOUND + 1))
done < <(printf '%s' "$ALL_MSGS" | jq -r '.[]? | [.id, .from, .subject, .date] | @tsv')

if [ "$FOUND" -eq 0 ]; then
  silent "no campaign replies" '{}'
fi

PAYLOAD=$(jq -R -s '
  split("\n") | map(select(length > 0) | split("\t") | {id: .[0], from: .[1], subject: .[2], date: .[3]})
  | {candidates: .}
' "$PENDING")

log "campaign reply candidates" "$PAYLOAD"
wake "campaign reply detected" "$PAYLOAD"