ingest

type GateOptions struct {
	// Force skips gates 1–6 (the source is written as kept); classify and
	// link still run.
	Force bool
	// Refetch performs the one fresh stage-4 scrape of a URL source (nil for
	// other sources).
	Refetch func(ctx context.Context) (*firecrawl.ScrapeResult, error)
	// RequestedURL is the URL the fetch was asked for; FinalURL and
	// StatusCode are what the fetcher reported.
	RequestedURL string
	FinalURL     string
	StatusCode   int
	// SiteName is the site name the fetcher reported (Firecrawl
	// ogSiteName), for the "title equals the site name" quality rule.
	SiteName string
	// PlatformID is the `youtube:<id>` / `instagram:<shortcode>` identity.
	PlatformID string
	// Prefetch is the stage-2 verdict of a bulk item (nil for single
	// sources and rescued items).
	Prefetch *gate.PrefetchDecision
}

type Options struct {
	VaultPath        string
	Topic            string
	SourceKind       models.SourceKind
	SourcePath       string
	SourceURL        string
	Title            string
	Markdown         string
	ExtraFrontmatter map[string]any
	ConvertOptions   map[string]any
	Registry         Registry
	ScrapedAt        time.Time
	// Batch names the ingest run (spec §4.4) and is written as ingest_batch.
	// Bulk runs pass the same value for every item. Empty omits the key.
	Batch string
	// Query is what produced the item (search query, channel URL, bookmark
	// label or file name) and is written as ingest_query. Empty omits the key.
	Query string
	// Session, when set, runs the decision-backed path for this one source:
	// gates (spec §7), owned-key writes, quarantine, then classify and link,
	// with one log.md run entry. Without it Ingest writes the source as
	// before (codebase ingest never uses a session).
	Session *session.Session
	// Gate carries the per-source gate inputs of the decision-backed path.
	Gate GateOptions
}

type Registry interface {
	Convert(ctx context.Context, input models.ConvertInput) (*models.ConvertResult, error)
}

type Result struct {
	models.IngestResult
	Triage       string   `json:"triage,omitempty"`
	TriageReason string   `json:"triage_reason,omitempty"`
	Quality      string   `json:"quality,omitempty"`
	DuplicateOf  string   `json:"duplicate_of,omitempty"`
	Shadow       []string `json:"shadow,omitempty"`
	Supersedes   []string `json:"supersedes,omitempty"`
	Refetched    bool     `json:"refetched,omitempty"`
	ReviewItem   string   `json:"review_item,omitempty"`
	Undecided    []string `json:"undecided,omitempty"`
	// Excluded reports that the source matches decisions.exclude: it was
	// written without any decision call and is left out of classify and
	// link; Note says so.
	Excluded bool   `json:"excluded,omitempty"`
	Note     string `json:"note,omitempty"`
}

type Run struct {
	// contains filtered or unexported fields
}

type RunOptions struct {
	// Command names the ingest subcommand ("url", "channel", ...).
	Command string
	// Batch and Query are the run provenance (ingest_batch, ingest_query);
	// per-source Options.Batch/Query override them.
	Batch string
	Query string
	// Force skips gates 1–6 for every source of the run.
	Force bool
}

type Summary struct {
	Tally gate.Tally `json:"tally"`
	// Modes names the gate mode per gate.
	Modes []string `json:"modes"`
	// Classify and Link are the reports of the in-run classification and
	// linking of the written sources (nil when nothing was written).
	Classify *classify.Report `json:"classify,omitempty"`
	Link     *link.Report     `json:"link,omitempty"`
	CostUSD  float64          `json:"cost_usd"`
	// Folders are the source counts per raw/ folder after the run, set when
	// the run quarantined a source (spec §7.1: the agent updates prose
	// counts from them).
	Folders []scope.FolderCount `json:"folders,omitempty"`
}