Jobber/scripts/jobber-pipeline-telegram.sh

#!/usr/bin/env bash
# Run Jobber pipeline, wait until it finishes, send summary + job links to Telegram.
# Secrets: copy scripts/jobber-cron.env.example to /root/.jobber-cron.env (chmod 600).
set -euo pipefail

ENV_FILE="${JOBBER_CRON_ENV:-/root/.jobber-cron.env}"
if [[ ! -f "$ENV_FILE" ]]; then
  echo "Missing env file: $ENV_FILE (set JOBBER_CRON_ENV or create the default path)" >&2
  exit 1
fi
# shellcheck source=/dev/null
source "$ENV_FILE"

: "${TELEGRAM_BOT_TOKEN:?Set TELEGRAM_BOT_TOKEN in $ENV_FILE}"
: "${TELEGRAM_CHAT_ID:?Set TELEGRAM_CHAT_ID in $ENV_FILE}"

BASE="${JOBOPS_URL:-http://127.0.0.1:3005}"
MAX_JOBS="${JOB_TELEGRAM_MAX_JOBS:-25}"
AUTH=()
if [[ -n "${BASIC_AUTH_USER:-}" && -n "${BASIC_AUTH_PASSWORD:-}" ]]; then
  AUTH=(-u "${BASIC_AUTH_USER}:${BASIC_AUTH_PASSWORD}")
fi

tg_html_escape() {
  printf '%s' "$1" | sed -e 's/&/\&amp;/g' -e 's/</\&lt;/g' -e 's/>/\&gt;/g'
}

tg_href_escape() {
  printf '%s' "$1" | sed -e 's/&/\&amp;/g' -e 's/"/\&quot;/g'
}

send_tg_html() {
  local msg="$1"
  curl -sS -X POST "https://api.telegram.org/bot${TELEGRAM_BOT_TOKEN}/sendMessage" \
    -H "Content-Type: application/json" \
    -d "$(jq -n \
      --arg c "$TELEGRAM_CHAT_ID" \
      --arg t "$msg" \
      '{chat_id: $c, text: $t, parse_mode: "HTML", disable_web_page_preview: true}')" >/dev/null
}

fetch_status() {
  curl -sS "${AUTH[@]}" "${BASE}/api/pipeline/status"
}

fetch_jobs_list() {
  curl -sS "${AUTH[@]}" "${BASE}/api/jobs?view=list"
}

build_job_lines_html() {
  local jobs_json="$1"
  local started="$2"
  local completed="$3"
  local max_n="$4"

  echo "$jobs_json" | jq -c --arg s "$started" --arg e "$completed" --argjson max "$max_n" '
    def pickurl:
      if (.jobUrl // "") != "" then .jobUrl
      elif (.applicationLink // "") != "" then .applicationLink
      else "" end;
    def pickrows($all):
      ($all | map(select($s != "" and $e != "" and (.discoveredAt >= $s) and (.discoveredAt <= $e)))) as $win |
      if ($win | length) > 0 then $win
      else ($all | map(select($s != "" and (.discoveredAt >= $s))))
      end;
    (.data.jobs // []) as $all |
    if ($all | length) == 0 then
      {total: 0, lines: []}
    else
      (pickrows($all) | sort_by(.discoveredAt) | reverse) as $sorted |
      ($sorted | length) as $total |
      ($sorted | .[0:max]) as $slice |
      {
        total: $total,
        lines: [
          $slice[] |
          {
            title: (.title // "Untitled"),
            url: pickurl,
            employer: (.employer // "")
          }
        ]
      }
    end
  '
}

append_lines_from_json() {
  local sel="$1"
  local -n _out="$2"
  local item line title url emp
  while IFS= read -r item; do
    [[ -z "$item" || "$item" == "null" ]] && continue
    title="$(echo "$item" | jq -r '.title // "Untitled"')"
    url="$(echo "$item" | jq -r '.url // ""')"
    emp="$(echo "$item" | jq -r '.employer // ""')"
    if [[ -n "$url" && "$url" != "null" ]]; then
      line="$(tg_html_escape "$emp") — <a href=\"$(tg_href_escape "$url")\">$(tg_html_escape "$title")</a>"
    else
      line="$(tg_html_escape "$emp") — $(tg_html_escape "$title") <i>(no URL)</i>"
    fi
    _out+=$'\n'"${line}"
  done < <(echo "$sel" | jq -c '.lines[]? // empty')
}

body="$(fetch_status)"
if ! echo "$body" | jq -e '.ok == true' >/dev/null 2>&1; then
  send_tg_html "Jobber: /api/pipeline/status failed (before run). Check container."
  exit 1
fi

if echo "$body" | jq -e '.data.isRunning == true' >/dev/null 2>&1; then
  send_tg_html "Jobber: pipeline already running; skipping scheduled run."
  exit 0
fi

resp="$(curl -sS "${AUTH[@]}" -X POST "${BASE}/api/pipeline/run" \
  -H "Content-Type: application/json" -d '{}')"
if ! echo "$resp" | jq -e '.ok == true' >/dev/null 2>&1; then
  _fail_json="$(echo "$resp" | jq -c . 2>/dev/null || echo "$resp")"
  send_tg_html "Jobber: POST /api/pipeline/run failed: $(tg_html_escape "$_fail_json")"
  exit 1
fi

was_running=0
for _ in $(seq 1 720); do
  sleep 30
  body="$(fetch_status)"
  if ! echo "$body" | jq -e '.ok == true' >/dev/null 2>&1; then
    send_tg_html "Jobber: status check failed mid-run."
    exit 1
  fi
  running="$(echo "$body" | jq -r '.data.isRunning')"
  if [[ "$running" == "true" ]]; then
    was_running=1
  elif [[ "$was_running" -eq 1 ]]; then
    lr="$(echo "$body" | jq '.data.lastRun')"
    st="$(echo "$lr" | jq -r '.status // "unknown"')"
    disc="$(echo "$lr" | jq -r '.jobsDiscovered // 0')"
    proc="$(echo "$lr" | jq -r '.jobsProcessed // 0')"
    err="$(echo "$lr" | jq -r '.errorMessage // empty')"
    started="$(echo "$lr" | jq -r '.startedAt // ""')"
    completed="$(echo "$lr" | jq -r '.completedAt // ""')"

    msg="<b>Jobber</b> pipeline: <b>$(tg_html_escape "$st")</b>"
    msg+=$'\n'"Discovered: ${disc}, processed: ${proc}."
    [[ -n "$err" ]] && msg+=$'\n'"<b>Error:</b> $(tg_html_escape "$err")"

    jobs_resp="$(fetch_jobs_list)"
    if echo "$jobs_resp" | jq -e '.ok == true' >/dev/null 2>&1; then
      sel="$(build_job_lines_html "$jobs_resp" "$started" "$completed" "$MAX_JOBS")" ||
        sel='{"total":0,"lines":[]}'
      total="$(echo "$sel" | jq -r '.total // 0')"
      shown="$(echo "$sel" | jq -r '.lines | length')"
      if [[ "$total" -gt 0 ]]; then
        msg+=$'\n\n'"<b>Jobs in this run</b> (showing ${shown} of ${total}):"
        append_lines_from_json "$sel" msg
        rest=$((total - shown))
        if [[ "$rest" -gt 0 ]]; then
          msg+=$'\n\n'"<i>…and ${rest} more not shown.</i>"
        fi
      else
        msg+=$'\n\n'"<i>No job rows matched this run window (time filter). Open the app for the full list.</i>"
      fi
    else
      msg+=$'\n\n'"<i>Could not load GET /api/jobs for links.</i>"
    fi

    send_tg_html "$msg"
    exit 0
  fi
done

send_tg_html "Jobber: timed out waiting for pipeline (6h). Check server."
exit 1