#!/usr/bin/env bash
set -euo pipefail
# Import the existing live homepage of a site into Microweber as a "Site
# Homepage" page so it's editable in /mw/admin/page from day one.
#
# Idempotent: if a page with url='site-homepage' already exists, its content is
# overwritten only if --force is passed; otherwise the script logs and exits 0.
#
# Usage:
# bash import_homepage_into_mw.sh \
# --site-root /srv/www/seethruit \
# --homepage-url 'https://seethruit.net/' \
# [--mw-at-root] # also sets is_home=1
# [--force] # overwrite existing 'site-homepage' content
#
# bash import_homepage_into_mw.sh \
# --site-root /srv/www/seethruit \
# --from-file /srv/www/seethruit/public/index-classic.php \
# [--mw-at-root] [--force]
#
# --from-file is the canonical-source mode: it reads a local PHP/HTML file
# (e.g. the original static index-classic.php) and extracts ...
# the same way the URL fetch path does. This is the right mode after MW has
# already taken over /, since curl would otherwise return MW-themed markup.
#
# The script reads MySQL credentials from /private/mw/.env so each
# install hits the right database without us re-passing them on the CLI.
SITE_ROOT=""
HOMEPAGE_URL=""
FROM_FILE=""
MW_AT_ROOT=0
FORCE=0
TITLE_OVERRIDE=""
while (($#)); do
case "$1" in
--site-root) SITE_ROOT="${2:-}"; shift 2 ;;
--homepage-url) HOMEPAGE_URL="${2:-}"; shift 2 ;;
--from-file) FROM_FILE="${2:-}"; shift 2 ;;
--title) TITLE_OVERRIDE="${2:-}"; shift 2 ;;
--mw-at-root) MW_AT_ROOT=1; shift ;;
--force) FORCE=1; shift ;;
*) echo "Unknown arg: $1" >&2; exit 1 ;;
esac
done
if [[ -z "$SITE_ROOT" || ! -d "$SITE_ROOT/private/mw" ]]; then
echo "Error: --site-root must point to a site with private/mw/ deployed" >&2
exit 1
fi
if [[ -z "$HOMEPAGE_URL" && -z "$FROM_FILE" ]]; then
echo "Error: one of --homepage-url or --from-file is required" >&2
exit 1
fi
if [[ -n "$HOMEPAGE_URL" && -n "$FROM_FILE" ]]; then
echo "Error: --homepage-url and --from-file are mutually exclusive" >&2
exit 1
fi
if [[ -n "$FROM_FILE" && ! -f "$FROM_FILE" ]]; then
echo "Error: --from-file path '$FROM_FILE' is not a regular file" >&2
exit 1
fi
ENV_FILE="$SITE_ROOT/private/mw/.env"
[[ -f "$ENV_FILE" ]] || { echo "Error: $ENV_FILE missing" >&2; exit 1; }
read_env() {
grep -E "^${1}=" "$ENV_FILE" | head -n1 | cut -d= -f2- | sed -e 's/^"//' -e 's/"$//'
}
DB_HOST="$(read_env DB_HOST)"
DB_PORT="$(read_env DB_PORT)"
DB_NAME="$(read_env DB_DATABASE)"
DB_USER="$(read_env DB_USERNAME)"
DB_PASS="$(read_env DB_PASSWORD)"
: "${DB_HOST:=127.0.0.1}"
: "${DB_PORT:=3306}"
if [[ -z "$DB_NAME" || -z "$DB_USER" ]]; then
echo "Error: DB_DATABASE/DB_USERNAME missing from $ENV_FILE" >&2
exit 1
fi
# Source the homepage HTML (live URL or local file).
TMP_HTML="$(mktemp)"
trap 'rm -f "$TMP_HTML" "$TMP_BODY" "$TMP_SQL"' EXIT
TMP_BODY="$(mktemp)"
TMP_SQL="$(mktemp)"
if [[ -n "$HOMEPAGE_URL" ]]; then
echo "Fetching $HOMEPAGE_URL ..."
if ! curl -fsSL --max-time 30 -o "$TMP_HTML" "$HOMEPAGE_URL"; then
echo "Error: curl failed for $HOMEPAGE_URL" >&2
exit 1
fi
else
echo "Reading $FROM_FILE ..."
cp -f "$FROM_FILE" "$TMP_HTML"
fi
BYTES="$(wc -c < "$TMP_HTML" | tr -d ' ')"
if (( BYTES < 200 )); then
echo "Error: source HTML is suspiciously small ($BYTES bytes)" >&2
exit 1
fi
# Extract the ... body if present, else fall back to everything
# between .... Strip #is", "", $out);
$out = preg_replace("#]*>.*?#is", "", $out);
$out = preg_replace("#]*>.*?#is", "", $out);
$out = preg_replace("#]*>.*?#is", "", $out);
$out = preg_replace("#]*>.*?#is", "", $out);
$out = preg_replace("##s", "", $out);
$out = trim($out);
if ($out === "") { fwrite(STDERR, "warning: extracted body is empty\n"); }
file_put_contents($argv[2], $out);
' "$TMP_HTML" "$TMP_BODY"
BODY_BYTES="$(wc -c < "$TMP_BODY" | tr -d ' ')"
echo "Extracted body: $BODY_BYTES bytes"
if (( BODY_BYTES < 50 )); then
echo "Error: extracted body is too small to be useful" >&2
exit 1
fi
# Default the Site Homepage title to a friendly per-site label
if [[ -n "$HOMEPAGE_URL" ]]; then
HOST_LABEL="$(echo "$HOMEPAGE_URL" | awk -F/ '{print $3}')"
else
HOST_LABEL="$(basename "$SITE_ROOT")"
fi
TITLE="${TITLE_OVERRIDE:-Site Homepage}"
# UPSERT via mysql. We base64-encode the body so we don't need to escape
# arbitrary HTML for shell + SQL.
BODY_B64="$(base64 -w0 < "$TMP_BODY")"
IS_HOME=0
if [[ "$MW_AT_ROOT" == "1" ]]; then IS_HOME=1; fi
cat > "$TMP_SQL" < 'site-homepage';"
fi
echo "Imported homepage content into MW at $ENV_FILE -> /mw/admin/page (url='site-homepage')."