config

type DiagnosticsConfig struct {
	// Enabled turns on diagnostic-log publishing.  When true (the default) the
	// worker's slog records are mirrored to sqi-server in addition to stderr.
	// Env: SQI_DIAGNOSTICS_ENABLED
	Enabled bool `yaml:"enabled"`
}

type DiscoveryConfig struct {
	// EnableMDNS enables mDNS browsing for "_sqi._tcp" services on the local
	// network. When true and NATS.URL is empty, the worker uses the first
	// discovered server address.
	// Env: SQI_WORKER_DISCOVERY_ENABLE_MDNS
	EnableMDNS bool `yaml:"enable_mdns"`

	// MDNSTimeout is the maximum duration to wait for mDNS discovery results
	// before falling back to an error.
	// Env: SQI_WORKER_DISCOVERY_MDNS_TIMEOUT
	MDNSTimeout time.Duration `yaml:"mdns_timeout"`
}

type FlagOverrides struct {
	// LogLevel overrides Log.Level when non-empty.
	LogLevel string
	// LogFormat overrides Log.Format when non-empty.
	LogFormat string
	// DryRun, when true, makes the start command print resolved config and exit.
	DryRun bool
	// NATSInsecureSkipVerify overrides NATS.InsecureSkipVerify when true.
	// Corresponds to --nats-insecure-skip-verify on the start subcommand.
	NATSInsecureSkipVerify bool
}

type LogConfig struct {
	// Level is the minimum log level: debug, info, warn, error.
	// Env: SQI_WORKER_LOG_LEVEL
	Level string `yaml:"level"`

	// Format is the log output format: json or text.
	// Env: SQI_WORKER_LOG_FORMAT
	Format string `yaml:"format"`
}

type LogStreamerConfig struct {
	// MaxLinesPerChunk is the maximum number of output lines batched into a
	// single NATS message.  A flush is triggered immediately when the buffer
	// reaches this count.  Increase for verbose processes; decrease for more
	// granular live log updates.
	// Default: 50
	// Env: SQI_WORKER_LOG_STREAMER_MAX_LINES_PER_CHUNK
	MaxLinesPerChunk int `yaml:"max_lines_per_chunk"`

	// MaxBytesPerChunk is the maximum total byte count of line content in a
	// single NATS message.  A flush is triggered when the accumulated bytes
	// reach this limit after adding a line.
	// Default: 16384 (16 KB)
	// Env: SQI_WORKER_LOG_STREAMER_MAX_BYTES_PER_CHUNK
	MaxBytesPerChunk int `yaml:"max_bytes_per_chunk"`

	// FlushInterval is the maximum time a line may sit in the buffer before
	// being flushed by the background goroutine regardless of chunk thresholds.
	// Smaller values make the web UI log viewer feel more live at the cost of
	// more frequent small NATS publishes.
	// Default: 500ms
	// Env: SQI_WORKER_LOG_STREAMER_FLUSH_INTERVAL
	FlushInterval time.Duration `yaml:"flush_interval"`
}

type MetricsConfig struct {
	// Addr is the TCP address the local HTTP server listens on.
	// Env: SQI_WORKER_METRICS_ADDR
	Addr string `yaml:"addr"`

	// EnablePprof exposes Go runtime profiling endpoints at /debug/pprof/.
	// Env: SQI_WORKER_METRICS_ENABLE_PPROF
	EnablePprof bool `yaml:"enable_pprof"`
}

type NATSConfig struct {
	// URL is the NATS server URL, e.g. "nats://localhost:4222".
	// Env: SQI_WORKER_NATS_URL
	URL string `yaml:"url"`

	// TLSCertFile is the path to the client TLS certificate (PEM).
	// Env: SQI_WORKER_NATS_TLS_CERT_FILE
	TLSCertFile string `yaml:"tls_cert_file"`

	// TLSKeyFile is the path to the client TLS key (PEM).
	// Env: SQI_WORKER_NATS_TLS_KEY_FILE
	TLSKeyFile string `yaml:"tls_key_file"`

	// TLSCAFile is the path to the CA certificate used to verify the NATS
	// server certificate (PEM).
	// Env: SQI_WORKER_NATS_TLS_CA_FILE
	TLSCAFile string `yaml:"tls_ca_file"`

	// InsecureSkipVerify disables TLS certificate verification. For
	// development environments only.
	// Env: SQI_WORKER_NATS_INSECURE_SKIP_VERIFY
	InsecureSkipVerify bool `yaml:"insecure_skip_verify"`

	// MaxReconnectAttempts is the maximum number of reconnection attempts
	// before giving up. -1 means retry indefinitely.
	// Env: SQI_WORKER_NATS_MAX_RECONNECT_ATTEMPTS
	MaxReconnectAttempts int `yaml:"max_reconnect_attempts"`

	// ReconnectWait is the base wait duration between reconnection attempts.
	// Actual wait uses exponential backoff with jitter.
	// Env: SQI_WORKER_NATS_RECONNECT_WAIT
	ReconnectWait time.Duration `yaml:"reconnect_wait"`
}

type StagingConfig struct {
	// ScratchDir is the base directory for per-attempt staged copies.
	ScratchDir string `yaml:"scratch_dir"`
	// SyncCommand is the command template invoked per path, with {src}, {dest},
	// and optional {object_type} placeholders, e.g. "rsync -a {src} {dest}".
	SyncCommand string `yaml:"sync_command"`

	// Defaults enables the built-in copy + TEMP scratch when staging is
	// otherwise unconfigured. Default true; set false to make an unconfigured
	// stage_locally job fail hard instead.
	// Env: SQI_STAGING_DEFAULTS
	Defaults bool `yaml:"defaults"`
}

type ValidationError struct {
	Field   string
	Message string
}

type WorkerConfig struct {
	// NATS is the connection configuration for the remote NATS instance
	// embedded in sqi-server.
	NATS NATSConfig `yaml:"nats"`

	// Worker contains identity and runtime behavior settings.
	Worker WorkerSettings `yaml:"worker"`

	// Log controls structured logging output.
	Log LogConfig `yaml:"log"`

	// Metrics controls the local HTTP server used for health probes and
	// Prometheus metrics.
	Metrics MetricsConfig `yaml:"metrics"`

	// Discovery controls mDNS-based server auto-discovery.
	Discovery DiscoveryConfig `yaml:"discovery"`

	// LogStreamer controls the log-chunk publisher that batches task process
	// output and streams it to sqi-server via NATS.
	LogStreamer LogStreamerConfig `yaml:"log_streamer"`

	// Diagnostics controls publishing the worker's own slog output to
	// sqi-server for display in the web UI.
	Diagnostics DiagnosticsConfig `yaml:"diagnostics"`

	// Staging configures local input/output staging for the stage_locally path
	// delivery. Both fields must be set for staging to function.
	Staging StagingConfig `yaml:"staging"`

	// Capabilities configures software capability auto-detection (built-in DCC
	// detectors plus custom detectors and a disable list).
	Capabilities capabilities.CapabilitiesConfig `yaml:"capabilities"`
}

type WorkerSettings struct {
	// Name is a human-readable label for this worker shown in the web UI.
	// Defaults to the hostname.
	// Env: SQI_WORKER_NAME
	Name string `yaml:"name"`

	// FarmID is the farm this worker belongs to. When set, the worker only
	// receives tasks belonging to that farm. When empty (the default), the
	// worker is unaffiliated and accepts tasks from any farm — suitable for
	// single-farm or development setups.
	// Env: SQI_WORKER_FARM_ID
	FarmID string `yaml:"farm_id"`

	// DataDir is the directory used to persist the worker ID file and
	// session working directories.
	// Env: SQI_WORKER_DATA_DIR
	DataDir string `yaml:"data_dir"`

	// ComputeLocation is the named location (matching a storage location in
	// sqi-server) where this worker's filesystem lives. Used for resolved-mode
	// path mapping.
	// Env: SQI_WORKER_COMPUTE_LOCATION
	ComputeLocation string `yaml:"compute_location"`

	// CapabilityTags is a list of arbitrary capability tags merged with
	// auto-detected capabilities at registration time, e.g. ["maya-2025",
	// "gpu", "highram"].
	// Env: SQI_WORKER_CAPABILITY_TAGS (comma-separated)
	CapabilityTags []string `yaml:"capability_tags"`

	// HeartbeatInterval is how often the worker publishes a heartbeat to the
	// server. Should be shorter than the server's heartbeat sweep interval.
	// Env: SQI_WORKER_HEARTBEAT_INTERVAL
	HeartbeatInterval time.Duration `yaml:"heartbeat_interval"`

	// ShutdownGracePeriod is the maximum time the worker waits for in-flight
	// tasks to finish before force-killing them on shutdown.
	// Env: SQI_WORKER_SHUTDOWN_GRACE_PERIOD
	ShutdownGracePeriod time.Duration `yaml:"shutdown_grace_period"`

	// AllowRoot permits the worker to run as the root user on Linux/macOS.
	// Disabled by default because executing render processes as root is a
	// security risk (see docs/worker-configuration.md, "worker.allow_root").
	// Env: SQI_WORKER_ALLOW_ROOT
	AllowRoot bool `yaml:"allow_root"`

	// KeepFailedSessions retains a session's working directory after the
	// session ends in failure (task cancellation, non-zero exit code, or
	// environment setup error). Useful for post-mortem inspection of partial
	// outputs and environment state. Disabled by default to avoid filling the
	// data directory on busy workers.
	// Env: SQI_WORKER_KEEP_FAILED_SESSIONS
	KeepFailedSessions bool `yaml:"keep_failed_sessions"`

	// QueueIDs restricts this worker to serving the listed queue IDs. An empty
	// list means the worker accepts assignments from all queues via a wildcard
	// JetStream consumer.  Set this when running a heterogeneous farm where
	// some workers specialise in a subset of queues.
	// Env: SQI_WORKER_QUEUE_IDS (comma-separated)
	QueueIDs []string `yaml:"queue_ids"`

	// PullIdleBackoff is the wait duration between pull attempts when the
	// work queue is empty. Prevents tight polling on idle queues.
	// Resets to zero immediately when a task is received.
	// Default: 2s
	// Env: SQI_WORKER_PULL_IDLE_BACKOFF
	PullIdleBackoff time.Duration `yaml:"pull_idle_backoff"`

	// PullNackDelay is the redelivery delay applied to an assignment when
	// pre-execution validation fails. The delay gives other workers
	// a chance to pick it up before NATS redelivers to this one.
	// Default: 5s
	// Env: SQI_WORKER_PULL_NACK_DELAY
	PullNackDelay time.Duration `yaml:"pull_nack_delay"`
}