Initial import of transcript pipeline
This commit is contained in:
maddin
Executable
+294
@@ -0,0 +1,294 @@
|
||||
#!/usr/bin/env bash
|
||||
set -euo pipefail
|
||||
|
||||
SOURCE_PATH="${BASH_SOURCE[0]}"
|
||||
while [[ -L "$SOURCE_PATH" ]]; do
|
||||
SOURCE_DIR="$(cd "$(dirname "$SOURCE_PATH")" && pwd)"
|
||||
SOURCE_PATH="$(readlink "$SOURCE_PATH")"
|
||||
[[ "$SOURCE_PATH" != /* ]] && SOURCE_PATH="$SOURCE_DIR/$SOURCE_PATH"
|
||||
done
|
||||
SCRIPT_DIR="$(cd "$(dirname "$SOURCE_PATH")" && pwd)"
|
||||
PROJECT_DIR="${PROJECT_DIR:-$(cd "$SCRIPT_DIR/.." && pwd)}"
|
||||
PIPELINE_SCRIPT="$PROJECT_DIR/transcript_pipeline.py"
|
||||
PIPELINE_LOG="$PROJECT_DIR/pipeline.log"
|
||||
STATE_DB="$PROJECT_DIR/pipeline_state.sqlite3"
|
||||
PLATFORM="$(uname -s)"
|
||||
LABEL="com.maddin.whisper-transcript-pipeline"
|
||||
PYTHON_BIN="${PYTHON_BIN:-$(command -v python3 || true)}"
|
||||
|
||||
if [[ -z "$PYTHON_BIN" ]]; then
|
||||
echo "python3 not found in PATH" >&2
|
||||
exit 1
|
||||
fi
|
||||
|
||||
macos_is_loaded() {
|
||||
local domain
|
||||
domain="gui/$(id -u)"
|
||||
launchctl print "$domain/$LABEL" >/dev/null 2>&1
|
||||
}
|
||||
|
||||
linux_service_name() {
|
||||
echo "$LABEL.service"
|
||||
}
|
||||
|
||||
linux_unit_path() {
|
||||
echo "$HOME/.config/systemd/user/$(linux_service_name)"
|
||||
}
|
||||
|
||||
show_pending_uploads() {
|
||||
if [[ -f "$STATE_DB" ]] && command -v sqlite3 >/dev/null 2>&1; then
|
||||
local pending
|
||||
pending="$(sqlite3 "$STATE_DB" "select basename || '|' || coalesce(remote_audio_status,'') || '|' || coalesce(remote_audio_last_error,'') from source_state where coalesce(remote_audio_status,'') != '' and remote_audio_status != 'uploaded';" || true)"
|
||||
if [[ -n "$pending" ]]; then
|
||||
echo "pending uploads:"
|
||||
echo "$pending"
|
||||
fi
|
||||
fi
|
||||
}
|
||||
|
||||
macos_start() {
|
||||
exec "$PROJECT_DIR/install_launch_agent.sh"
|
||||
}
|
||||
|
||||
macos_stop() {
|
||||
if [[ -x "$PROJECT_DIR/uninstall_launch_agent.sh" ]]; then
|
||||
"$PROJECT_DIR/uninstall_launch_agent.sh" --keep-files
|
||||
fi
|
||||
pkill -f "$PIPELINE_SCRIPT watch" >/dev/null 2>&1 || true
|
||||
echo "stopped: $LABEL"
|
||||
}
|
||||
|
||||
macos_status() {
|
||||
local pids
|
||||
if macos_is_loaded; then
|
||||
echo "launchd: loaded"
|
||||
launchctl list | grep "$LABEL" || true
|
||||
else
|
||||
echo "launchd: not loaded"
|
||||
fi
|
||||
|
||||
pids="$(pgrep -af "$PIPELINE_SCRIPT watch" || true)"
|
||||
if [[ -n "$pids" ]]; then
|
||||
echo "process:"
|
||||
echo "$pids"
|
||||
else
|
||||
echo "process: not running"
|
||||
fi
|
||||
|
||||
if [[ -f "$PIPELINE_LOG" ]]; then
|
||||
echo "pipeline log:"
|
||||
tail -n 5 "$PIPELINE_LOG"
|
||||
fi
|
||||
show_pending_uploads
|
||||
}
|
||||
|
||||
linux_start() {
|
||||
exec "$PROJECT_DIR/deploy/linux/install_systemd_service.sh"
|
||||
}
|
||||
|
||||
linux_stop() {
|
||||
local service
|
||||
service="$(linux_service_name)"
|
||||
systemctl --user stop "$service" >/dev/null 2>&1 || true
|
||||
pkill -f "$PIPELINE_SCRIPT watch" >/dev/null 2>&1 || true
|
||||
echo "stopped: $service"
|
||||
}
|
||||
|
||||
linux_status() {
|
||||
local service pids
|
||||
service="$(linux_service_name)"
|
||||
if systemctl --user status "$service" >/dev/null 2>&1; then
|
||||
echo "systemd --user: loaded"
|
||||
else
|
||||
echo "systemd --user: not loaded"
|
||||
fi
|
||||
|
||||
pids="$(pgrep -af "$PIPELINE_SCRIPT watch" || true)"
|
||||
if [[ -n "$pids" ]]; then
|
||||
echo "process:"
|
||||
echo "$pids"
|
||||
else
|
||||
echo "process: not running"
|
||||
fi
|
||||
|
||||
if [[ -f "$PIPELINE_LOG" ]]; then
|
||||
echo "pipeline log:"
|
||||
tail -n 5 "$PIPELINE_LOG"
|
||||
fi
|
||||
show_pending_uploads
|
||||
}
|
||||
|
||||
cmd_start() {
|
||||
case "$PLATFORM" in
|
||||
Darwin) macos_start ;;
|
||||
Linux) linux_start ;;
|
||||
*)
|
||||
echo "unsupported platform: $PLATFORM" >&2
|
||||
exit 1
|
||||
;;
|
||||
esac
|
||||
}
|
||||
|
||||
cmd_stop() {
|
||||
case "$PLATFORM" in
|
||||
Darwin) macos_stop ;;
|
||||
Linux) linux_stop ;;
|
||||
*)
|
||||
echo "unsupported platform: $PLATFORM" >&2
|
||||
exit 1
|
||||
;;
|
||||
esac
|
||||
}
|
||||
|
||||
cmd_restart() {
|
||||
cmd_stop
|
||||
cmd_start
|
||||
}
|
||||
|
||||
cmd_install() {
|
||||
cmd_start
|
||||
}
|
||||
|
||||
cmd_uninstall() {
|
||||
case "$PLATFORM" in
|
||||
Darwin) exec "$PROJECT_DIR/uninstall_launch_agent.sh" ;;
|
||||
Linux) exec "$PROJECT_DIR/deploy/linux/uninstall_systemd_service.sh" ;;
|
||||
*)
|
||||
echo "unsupported platform: $PLATFORM" >&2
|
||||
exit 1
|
||||
;;
|
||||
esac
|
||||
}
|
||||
|
||||
cmd_status() {
|
||||
case "$PLATFORM" in
|
||||
Darwin) macos_status ;;
|
||||
Linux) linux_status ;;
|
||||
*)
|
||||
echo "unsupported platform: $PLATFORM" >&2
|
||||
exit 1
|
||||
;;
|
||||
esac
|
||||
}
|
||||
|
||||
cmd_logs() {
|
||||
local which_log="${1:-pipeline}"
|
||||
case "$PLATFORM" in
|
||||
Darwin)
|
||||
case "$which_log" in
|
||||
pipeline) tail -f "$PIPELINE_LOG" ;;
|
||||
stdout) tail -f "$PROJECT_DIR/launchd.stdout.log" ;;
|
||||
stderr) tail -f "$PROJECT_DIR/launchd.stderr.log" ;;
|
||||
all) tail -f "$PIPELINE_LOG" "$PROJECT_DIR/launchd.stdout.log" "$PROJECT_DIR/launchd.stderr.log" ;;
|
||||
*)
|
||||
echo "unknown log target: $which_log" >&2
|
||||
exit 1
|
||||
;;
|
||||
esac
|
||||
;;
|
||||
Linux)
|
||||
case "$which_log" in
|
||||
pipeline) tail -f "$PIPELINE_LOG" ;;
|
||||
service|journal) journalctl --user -u "$(linux_service_name)" -f ;;
|
||||
all) tail -f "$PIPELINE_LOG" ;;
|
||||
*)
|
||||
echo "unknown log target: $which_log" >&2
|
||||
exit 1
|
||||
;;
|
||||
esac
|
||||
;;
|
||||
*)
|
||||
echo "unsupported platform: $PLATFORM" >&2
|
||||
exit 1
|
||||
;;
|
||||
esac
|
||||
}
|
||||
|
||||
cmd_scan() {
|
||||
exec "$PYTHON_BIN" "$PIPELINE_SCRIPT" scan "$@"
|
||||
}
|
||||
|
||||
cmd_cleanup() {
|
||||
exec "$PYTHON_BIN" "$PIPELINE_SCRIPT" cleanup
|
||||
}
|
||||
|
||||
cmd_memos_sync() {
|
||||
exec "$PYTHON_BIN" "$PIPELINE_SCRIPT" memos-sync
|
||||
}
|
||||
|
||||
cmd_memos_auth() {
|
||||
exec "$PYTHON_BIN" "$PIPELINE_SCRIPT" memos-auth
|
||||
}
|
||||
|
||||
cmd_retry_uploads() {
|
||||
exec "$PYTHON_BIN" "$PIPELINE_SCRIPT" retry-uploads
|
||||
}
|
||||
|
||||
cmd_migrate_archive() {
|
||||
exec "$PYTHON_BIN" "$PIPELINE_SCRIPT" migrate-archive
|
||||
}
|
||||
|
||||
cmd_reprocess() {
|
||||
if [[ $# -lt 1 ]]; then
|
||||
echo "usage: transcript reprocess <basename>" >&2
|
||||
exit 1
|
||||
fi
|
||||
exec "$PYTHON_BIN" "$PIPELINE_SCRIPT" reprocess "$1"
|
||||
}
|
||||
|
||||
cmd_bundle() {
|
||||
exec "$PROJECT_DIR/scripts/create_migration_bundle.sh"
|
||||
}
|
||||
|
||||
cmd_help() {
|
||||
cat <<'EOF'
|
||||
Usage: transcript <command> [args]
|
||||
|
||||
Commands:
|
||||
install Install and start the background watcher
|
||||
start Start or install the background watcher
|
||||
stop Stop the background watcher
|
||||
restart Restart the background watcher
|
||||
uninstall Uninstall the background watcher service
|
||||
status Show service/process status and recent pipeline log lines
|
||||
logs [pipeline|stdout|stderr|service|journal|all]
|
||||
Follow log output
|
||||
scan Process currently available transcript/audio pairs once
|
||||
cleanup Delete archived originals older than the retention period
|
||||
memos-sync Export published notes for Quartz and rebuild memos.maddin.app
|
||||
memos-auth Generate the Basic Auth htpasswd file from .env
|
||||
retry-uploads Retry remote audio uploads from the local archive
|
||||
migrate-archive Retroactively migrate archived sources to the new schema
|
||||
reprocess <basename> Reprocess one basename from the watch folder
|
||||
bundle Build a portable migration tarball in dist/
|
||||
help Show this help
|
||||
EOF
|
||||
}
|
||||
|
||||
command="${1:-help}"
|
||||
shift || true
|
||||
|
||||
case "$command" in
|
||||
install) cmd_install ;;
|
||||
start) cmd_start ;;
|
||||
stop) cmd_stop ;;
|
||||
restart) cmd_restart ;;
|
||||
uninstall) cmd_uninstall ;;
|
||||
status) cmd_status ;;
|
||||
logs) cmd_logs "$@" ;;
|
||||
scan) cmd_scan "$@" ;;
|
||||
cleanup) cmd_cleanup ;;
|
||||
memos-sync) cmd_memos_sync ;;
|
||||
memos-auth) cmd_memos_auth ;;
|
||||
retry-uploads) cmd_retry_uploads ;;
|
||||
migrate-archive) cmd_migrate_archive ;;
|
||||
reprocess) cmd_reprocess "$@" ;;
|
||||
bundle) cmd_bundle ;;
|
||||
help|-h|--help) cmd_help ;;
|
||||
*)
|
||||
echo "unknown command: $command" >&2
|
||||
echo >&2
|
||||
cmd_help >&2
|
||||
exit 1
|
||||
;;
|
||||
esac
|
||||
Reference in New Issue
Block a user