Struct IngestArgs 

pub struct IngestArgs {

    pub dir: PathBuf,
    pub type: MemoryType,
    pub pattern: String,
    pub recursive: bool,
    pub enable_ner: bool,
    pub auto_describe: bool,
    pub no_auto_describe: bool,
    pub skip_extraction: bool,
    pub fail_fast: bool,
    pub dry_run: bool,
    pub max_files: usize,
    pub namespace: Option<String>,
    pub db: Option<String>,
    pub format: JsonOutputFormat,
    pub json: bool,
    pub ingest_parallelism: Option<usize>,
    pub low_memory: bool,
    pub max_rss_mb: u64,
    pub llm_parallelism: u64,
    pub max_name_length: usize,
    pub name_prefix: Option<String>,
    pub mode: IngestMode,
    pub max_cost_usd: Option<f64>,
    pub wait_job_singleton: Option<u64>,
    pub force_job_singleton: bool,
    pub enrich_after: bool,
    pub force_merge: bool,
}

Ingest args.

Fields

dir: PathBuf

Directory containing files to ingest.

type: MemoryType

Memory type stored in memories.type for every ingested file. Defaults to document.

pattern: String

Glob pattern matched against file basenames (default: *.md). Supports *.<ext>, <prefix>*, and exact filename match.

recursive: bool

Recurse into subdirectories.

enable_ner: bool

Enable NER.

auto_describe: bool

GAP-E2E-011: generates a heuristic description from the first meaningful line of the body, instead of “ingested from <path>”. When --no-auto-describe is passed, keeps the legacy behaviour.

no_auto_describe: bool

No auto describe.

skip_extraction: bool

Deprecated: NER is now disabled by default. Kept for backwards compatibility.

fail_fast: bool

Stop on first per-file error instead of continuing with the next file.

dry_run: bool

Preview file-to-name mapping without loading model or persisting.

max_files: usize

Maximum number of files to ingest (safety cap to prevent runaway ingestion).

namespace: Option<String>

Namespace for the ingested memories.

db: Option<String>

Database path. Falls back to XDG db.path, then graphrag.sqlite under the XDG data directory.

format: JsonOutputFormat

Output format.

json: bool

Emit machine-readable JSON on stdout.

ingest_parallelism: Option<usize>

Number of files to extract+embed in parallel; default = max(1, cpus/2).min(4).

low_memory: bool

Force single-threaded ingest to reduce RSS pressure.

Equivalent to --ingest-parallelism 1, takes precedence over any explicit value. Recommended for environments with <4 GB available RAM or container/cgroup constraints. Trade-off: 3-4x longer wall time. Also available via XDG config set ingest.low_memory 1, which the flag overrides. No environment variable supplies this value.

max_rss_mb: u64

Maximum process RSS in MiB; abort if exceeded during embedding.

llm_parallelism: u64

G42/S3 (v1.0.79): maximum simultaneous LLM embedding subprocesses PER FILE. Multiplies with –ingest-parallelism (files staged concurrently), hence the conservative default of 2. The effective value is further bounded by CPU count and available RAM.

max_name_length: usize

Maximum character length for derived memory names from file basenames.

Overrides the compile-time DERIVED_NAME_MAX_LEN constant (default 60). Shorter values leave more headroom for collision suffix resolution.

name_prefix: Option<String>

v1.1.1 (P12): kebab-case prefix prepended to every derived memory name, AFTER the basename is normalized. Namespaces a corpus inside a shared database (e.g. --name-prefix projx- yields projx-<derived>). The derived part’s budget shrinks so the final name always respects the 80-char name cap. Only supported with --mode none.

mode: IngestMode

Extraction mode: none (body-only, default).

max_cost_usd: Option<f64>

Maximum cumulative cost in USD before aborting.

wait_job_singleton: Option<u64>

G30: poll for the job singleton every second for up to N seconds when another invocation holds the lock. Default: 0 (fail fast).

force_job_singleton: bool

G30: force acquisition of the singleton lock by removing a stale lock file from a previously crashed invocation.

enrich_after: bool

v1.0.93 (GAP-OR-INGEST): run enrich --operation memory-bindings after all files are embedded, using the active --llm-backend.

force_merge: bool

GAP-SG-54: update existing memories instead of skipping them. Without this flag a file whose derived name already exists is reported skipped; with it the existing memory’s body, embedding and chunks are refreshed (the remember --force-merge update path applied per file).

Trait Implementations

impl Args for IngestArgs

fn group_id() -> Option<Id>

Report the ArgGroup::id for this set of arguments

fn augment_args<'b>(__clap_app: Command) -> Command

Append to Command so it can instantiate Self via FromArgMatches::from_arg_matches_mut

fn augment_args_for_update<'b>(__clap_app: Command) -> Command

Append to Command so it can instantiate self via FromArgMatches::update_from_arg_matches_mut

impl FromArgMatches for IngestArgs

fn from_arg_matches(__clap_arg_matches: &ArgMatches) -> Result<Self, Error>

Instantiate Self from ArgMatches, parsing the arguments as needed.

fn from_arg_matches_mut( __clap_arg_matches: &mut ArgMatches, ) -> Result<Self, Error>

Instantiate Self from ArgMatches, parsing the arguments as needed.

fn update_from_arg_matches( &mut self, __clap_arg_matches: &ArgMatches, ) -> Result<(), Error>

Assign values from ArgMatches to self.

fn update_from_arg_matches_mut( &mut self, __clap_arg_matches: &mut ArgMatches, ) -> Result<(), Error>

Assign values from ArgMatches to self.