#!/bin/bash
PAGES_DIR="/data/hermes-pages"
PERSISTENT_FILE="$PAGES_DIR/.persistent"
LOG_FILE="/var/log/hermes-pages-cleanup.log"

# Delete HTML files older than 24 hours unless in persistent list
find "$PAGES_DIR" -maxdepth 1 -name '*.html' -mmin +1440 | while read -r f; do
    slug=$(basename "$f" .html)
    if grep -qxF "$slug" "$PERSISTENT_FILE" 2>/dev/null; then
        echo "[$(date)] SKIP persistent: $slug" >> "$LOG_FILE"
        continue
    fi
    rm -f "$f"
    rm -f "$PAGES_DIR/.meta/${slug}.json"
    echo "[$(date)] DELETED: $slug" >> "$LOG_FILE"
done

# Also clean orphaned .meta files
find "$PAGES_DIR/.meta" -name '*.json' | while read -r mf; do
    slug=$(basename "$mf" .json)
    if [ ! -f "$PAGES_DIR/${slug}.html" ]; then
        rm -f "$mf"
    fi
done
