Remove dead bookmarks (404 / DNS fail / timeout) from a Chrome profile

chrome-deadlink-purge.sh

#!/usr/bin/env bash
# chrome-deadlink-purge.sh — remove dead bookmarks (404 / DNS fail / timeout) from a Chrome profile.
#
# Usage:
#   chrome-deadlink-purge.sh [PROFILE_DIR]
#     PROFILE_DIR defaults to: ~/Library/Application Support/Google/Chrome/Default  (macOS)
#
# Requirements: bash, jq, curl, xargs.
# Quit Chrome before running — Chrome rewrites the Bookmarks file on exit and will
# clobber your edits otherwise.
#
# What it does:
#   1. Backs up the Bookmarks file.
#   2. Extracts every URL.
#   3. Probes each URL in parallel (30-way) with a real-browser User-Agent.
#   4. Marks dead = HTTP 404 OR DNS fail (curl exit 6) OR timeout (curl exit 28).
#   5. Rewrites Bookmarks with dead URLs removed and the checksum field dropped
#      (Chrome recomputes it on next launch).
#   6. Deletes Bookmarks.bak so Chrome doesn't auto-restore the old file.

set -euo pipefail

PROFILE_DIR="${1:-$HOME/Library/Application Support/Google/Chrome/Default}"
BOOKMARKS="$PROFILE_DIR/Bookmarks"
WORK="$(mktemp -d -t chrome-deadlink-XXXXXX)"
PARALLEL="${PARALLEL:-30}"
CONNECT_TIMEOUT="${CONNECT_TIMEOUT:-8}"
MAX_TIME="${MAX_TIME:-15}"

[ -f "$BOOKMARKS" ] || { echo "no Bookmarks file at: $BOOKMARKS" >&2; exit 1; }
command -v jq >/dev/null || { echo "jq required" >&2; exit 1; }

if pgrep -x "Google Chrome" >/dev/null 2>&1; then
  echo "Chrome is running. Quit it first (File > Quit, or Cmd+Q). Aborting." >&2
  exit 1
fi

echo "==> workdir: $WORK"
echo "==> profile: $PROFILE_DIR"

STAMP="$(date +%Y%m%d-%H%M%S)"
BACKUP="$WORK/Bookmarks.backup.$STAMP"
cp "$BOOKMARKS" "$BACKUP"
echo "==> backup: $BACKUP"

# Probe one URL. Output: STATUS<TAB>URL.
cat > "$WORK/probe.sh" <<'PROBE'
#!/usr/bin/env bash
url="$1"
case "$url" in http://*|https://*) ;; *) printf 'SKIP\t%s\n' "$url"; exit 0 ;; esac
code=$(curl -sS -o /dev/null -L \
  --max-time "${MAX_TIME:-15}" --connect-timeout "${CONNECT_TIMEOUT:-8}" \
  -A "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36" \
  -w "%{http_code}" "$url" 2>/dev/null)
rc=$?
case $rc in
  0)  printf '%s\t%s\n' "$code" "$url" ;;
  6)  printf 'DNS\t%s\n' "$url" ;;
  28) printf 'TIMEOUT\t%s\n' "$url" ;;
  7)  printf 'CONNREFUSED\t%s\n' "$url" ;;
  *)  printf 'ERR%s\t%s\n' "$rc" "$url" ;;
esac
PROBE
chmod +x "$WORK/probe.sh"

jq -r '[.. | objects | select(.type=="url") | .url] | unique | .[]' "$BOOKMARKS" > "$WORK/urls.txt"
TOTAL=$(wc -l < "$WORK/urls.txt" | tr -d ' ')
echo "==> probing $TOTAL unique URLs ($PARALLEL parallel) ..."

export MAX_TIME CONNECT_TIMEOUT
# -S 65536 raises BSD xargs replstr buffer; -0 handles long lines safely.
tr '\n' '\0' < "$WORK/urls.txt" \
  | xargs -0 -n1 -P "$PARALLEL" -S 65536 -I{} "$WORK/probe.sh" "{}" \
  > "$WORK/results.tsv" 2> "$WORK/probe.err" || true

echo "==> status counts:"
awk -F'\t' '{print $1}' "$WORK/results.tsv" | sort | uniq -c | sort -rn | sed 's/^/    /'

awk -F'\t' '$1=="404" || $1=="DNS" || $1=="TIMEOUT" {print $2}' "$WORK/results.tsv" > "$WORK/dead.txt"
DEAD=$(wc -l < "$WORK/dead.txt" | tr -d ' ')
echo "==> dead URLs (404 + DNS + TIMEOUT): $DEAD"
[ "$DEAD" -eq 0 ] && { echo "nothing to remove."; exit 0; }

jq -R . "$WORK/dead.txt" | jq -s . > "$WORK/dead.json"

jq --slurpfile dead "$WORK/dead.json" '
  ($dead[0] | map({(.): true}) | add) as $deadset
  | walk(
      if (type == "object") and (.children | type == "array")
      then .children |= map(select((.type != "url") or (($deadset[.url] // false) | not)))
      else .
      end
    )
  | del(.checksum)
' "$BOOKMARKS" > "$WORK/Bookmarks.cleaned"

BEFORE=$(jq '[.. | objects | select(.type=="url") | .url] | length' "$BOOKMARKS")
AFTER=$(jq '[.. | objects | select(.type=="url") | .url] | length' "$WORK/Bookmarks.cleaned")
echo "==> bookmarks: $BEFORE -> $AFTER  (removed $((BEFORE-AFTER)))"

if pgrep -x "Google Chrome" >/dev/null 2>&1; then
  echo "Chrome restarted mid-run. Aborting before write." >&2
  exit 1
fi

cp "$WORK/Bookmarks.cleaned" "$BOOKMARKS"
chmod 600 "$BOOKMARKS"
rm -f "$PROFILE_DIR/Bookmarks.bak"
echo "==> deployed. Reopen Chrome to verify."
echo "==> artifacts kept in: $WORK"