tas_councils/run_all.sh

#!/usr/bin/env bash
set -euo pipefail

echo "Starting run at $(date -u +"%Y-%m-%d %H:%M:%S")"

echo "Running schema migrations…"
ruby /app/lib/migrate.rb

ONLY_LIST="${ONLY:-}"     # e.g. "meandervalley" or "kentish,break_oday"
SKIP_LIST="${SKIP:-}"     # e.g. "hobartcity,latrobe"
DEBUG_FLAG="${DEBUG:-}"   # pass through to scrapers
DRY_FLAG="${DRY_RUN:-}"   # pass through to scrapers

shopt -s nullglob
SCRIPTS=(/app/scrapers/*.rb)

should_run() {
  local name="$1"

  if [[ -n "$ONLY_LIST" ]]; then
    IFS=',' read -ra arr <<< "$ONLY_LIST"
    local pick
    for pick in "${arr[@]}"; do
      pick="${pick// /}"
      if [[ "$name" == "$pick" ]]; then
        echo "1"
        return
      fi
    done
    echo "0"
    return
  fi

  if [[ -n "$SKIP_LIST" ]]; then
    IFS=',' read -ra arr <<< "$SKIP_LIST"
    local skip
    for skip in "${arr[@]}"; do
      skip="${skip// /}"
      if [[ "$name" == "$skip" ]]; then
        echo "0"
        return
      fi
    done
  fi

  echo "1"
}

count=0
total_saved=0
total_warns=0

# Each entry: "name|saved|warns|status"
SUMMARY=()

for f in "${SCRIPTS[@]}"; do
  name="$(basename "$f" .rb)"
  table="da_${name}"

  if [[ "$(should_run "$name")" != "1" ]]; then
    continue
  fi

  echo ""
  echo "Running ${name} -> table ${table}"

  tmpfile=$(mktemp /tmp/scraper_XXXXXX.log)

  # Run scraper; merge stderr into stdout so tee captures both.
  # Disable pipefail temporarily so a non-zero ruby exit doesn't abort the loop.
  set +e
  TABLE_NAME="$table" DEBUG="$DEBUG_FLAG" DRY_RUN="$DRY_FLAG" ruby "$f" 2>&1 | tee "$tmpfile"
  ruby_exit=${PIPESTATUS[0]}
  set -e

  if [[ $ruby_exit -ne 0 ]]; then
    echo "  [run_all] scraper exited with code ${ruby_exit}"
  fi

  # --- Parse summary fields from captured output ---

  # "Saved N item(s)" or "saved N" — case-insensitive, last occurrence wins.
  # Fallback: count "Upserted" lines (every scraper prints one per DB write).
  saved=$(grep -oiE 'saved [0-9]+' "$tmpfile" | tail -1 | grep -oE '[0-9]+' || true)
  if [[ -z "$saved" || "$saved" == "0" ]]; then
    upsert_count=$(grep -cE '^(Upserted|  Upserted)' "$tmpfile" 2>/dev/null || true)
    [[ "${upsert_count:-0}" -gt 0 ]] && saved="$upsert_count"
  fi
  saved="${saved:-0}"

  # Count WARN lines (from Log.warn)
  warns=$(grep -c '^\s*WARN' "$tmpfile" || true)
  warns="${warns:-0}"

  # Determine status
  if [[ $ruby_exit -ne 0 ]]; then
    status="ERROR"
  elif grep -qiE 'cloudflare|blocked by|challenge page' "$tmpfile" 2>/dev/null; then
    status="blocked"
  elif [[ $warns -gt 0 ]]; then
    status="warn"
  else
    status="ok"
  fi

  rm -f "$tmpfile"

  SUMMARY+=("${name}|${saved}|${warns}|${status}")
  total_saved=$((total_saved + saved))
  total_warns=$((total_warns + warns))
  count=$((count+1))
done

finish_time=$(date -u +"%Y-%m-%d %H:%M:%S")

# ---------------------------------------------------------------------------
# Summary table
# ---------------------------------------------------------------------------
echo ""
echo "========================================================================"
printf "  SCRAPE SUMMARY — finished %s UTC\n" "$finish_time"
echo "========================================================================"
printf "  %-32s  %6s  %5s  %s\n" "Council" "Saved" "Warns" "Status"
echo "  ------------------------------------------------------------------------"

for entry in "${SUMMARY[@]}"; do
  IFS='|' read -r n s w st <<< "$entry"
  # Colour-code status when output is a terminal; plain text otherwise
  if [[ -t 1 ]]; then
    case "$st" in
      ok)      colour="\033[0;32m" ;;   # green
      warn)    colour="\033[0;33m" ;;   # yellow
      blocked) colour="\033[0;33m" ;;   # yellow
      ERROR)   colour="\033[0;31m" ;;   # red
      *)       colour="" ;;
    esac
    reset="\033[0m"
    printf "  %-32s  %6s  %5s  ${colour}%s${reset}\n" "$n" "$s" "$w" "$st"
  else
    printf "  %-32s  %6s  %5s  %s\n" "$n" "$s" "$w" "$st"
  fi
done

echo "  ------------------------------------------------------------------------"
printf "  %-32s  %6s  %5s\n" "TOTAL (${count} scrapers)" "$total_saved" "$total_warns"
echo "========================================================================"