Add complete WCX update workflow

2026-07-17 23:04:57 +02:00
parent 86c672de1e
commit 4bea8077ec
1 changed files with 152 additions and 0 deletions
--- a/scripts/update_wcx.sh
+++ b/scripts/update_wcx.sh
@ -0,0 +1,152 @@
+#!/usr/bin/env bash
+
+#
+# Run the complete WCX index update:
+#
+#   1. Download the current site index.
+#   2. Import site metadata into SQLite.
+#   3. OCR-process newly discovered movies.
+#
+# Usage:
+#   scripts/update_wcx.sh
+#   scripts/update_wcx.sh --skip-ocr
+#
+# GOOGLE_VISION_API_KEY must be set unless --skip-ocr is used.
+#
+
+set -Eeuo pipefail
+
+ROOT_DIR="/storage/disk1/WCX"
+
+SYNC_SCRIPT="${ROOT_DIR}/wcx_sync.py"
+IMPORT_SCRIPT="${ROOT_DIR}/scripts/import_site.py"
+OCR_BATCH_SCRIPT="${ROOT_DIR}/scripts/process_pending_ocr.py"
+
+DATABASE_FILE="${ROOT_DIR}/database/wcx.db"
+INDEX_FILE="${ROOT_DIR}/import/wcx_site_index.json"
+
+LOCK_FILE="${ROOT_DIR}/database/update_wcx.lock"
+
+SKIP_OCR=false
+
+
+log() {
+    printf '[%s] %s\n' \
+        "$(date '+%Y-%m-%d %H:%M:%S')" \
+        "$*"
+}
+
+
+fail() {
+    log "ERROR: $*" >&2
+    exit 1
+}
+
+
+on_error() {
+    local exit_code=$?
+    local line_number=$1
+
+    log "Update failed on line ${line_number} with exit code ${exit_code}." >&2
+    exit "${exit_code}"
+}
+
+
+show_usage() {
+    cat <<'EOF'
+Usage:
+    update_wcx.sh
+    update_wcx.sh --skip-ocr
+
+Options:
+    --skip-ocr  Run site synchronization and database import without OCR.
+    --help      Show this help.
+EOF
+}
+
+
+while [[ $# -gt 0 ]]; do
+    case "$1" in
+        --skip-ocr)
+            SKIP_OCR=true
+            shift
+            ;;
+        --help|-h)
+            show_usage
+            exit 0
+            ;;
+        *)
+            fail "Unknown argument: $1"
+            ;;
+    esac
+done
+
+
+trap 'on_error "$LINENO"' ERR
+
+
+[[ -f "${SYNC_SCRIPT}" ]] ||
+    fail "Sync script not found: ${SYNC_SCRIPT}"
+
+[[ -x "${IMPORT_SCRIPT}" ]] ||
+	    fail "Import script is missing or not executable: ${IMPORT_SCRIPT}"
+
+[[ -x "${OCR_BATCH_SCRIPT}" ]] ||
+    fail "OCR batch script is missing or not executable: ${OCR_BATCH_SCRIPT}"
+
+[[ -f "${DATABASE_FILE}" ]] ||
+    fail "Database not found: ${DATABASE_FILE}"
+
+
+if [[ "${SKIP_OCR}" == false ]] &&
+   [[ -z "${GOOGLE_VISION_API_KEY:-}" ]]; then
+    fail "GOOGLE_VISION_API_KEY is not set. Use --skip-ocr to omit OCR."
+fi
+
+
+# Prevent two update processes from running simultaneously.
+exec 9>"${LOCK_FILE}"
+
+if ! flock -n 9; then
+    fail "Another WCX update is already running."
+fi
+
+
+cd "${ROOT_DIR}"
+
+log "Starting WCX update."
+log "Database: ${DATABASE_FILE}"
+
+
+log "Step 1/3: Downloading current WCX site index."
+
+python3 "${SYNC_SCRIPT}" --out "${INDEX_FILE}"
+
+[[ -s "${INDEX_FILE}" ]] ||
+    fail "Site index was not created or is empty: ${INDEX_FILE}"
+
+log "Site index downloaded successfully."
+
+
+log "Step 2/3: Importing site metadata."
+
+"${IMPORT_SCRIPT}" \
+    --database "${DATABASE_FILE}" \
+    --input "${INDEX_FILE}"
+
+log "Site metadata imported successfully."
+
+
+if [[ "${SKIP_OCR}" == true ]]; then
+    log "Step 3/3: OCR skipped."
+else
+    log "Step 3/3: Processing pending OCR records."
+
+    "${OCR_BATCH_SCRIPT}" \
+        --database "${DATABASE_FILE}"
+
+    log "OCR processing finished."
+fi
+
+
+log "WCX update completed successfully."