Crate gaze_types

Structs§

Candidate: Candidate PII span emitted by a recognizer before final conflict resolution.
CodecAuditRow: Metadata-only audit row emitted by a document codec.
CodecCapabilitySet: Orthogonal document codec capabilities delivered or advertised by a codec.
DetectContext: Context supplied to recognizers during detection.
Detection: A detected span and its class/source metadata.
DictionaryBundle: Value-only dictionary bundle shared with recognizers.
DictionaryEntry: Value-only dictionary entry; compiled automatons live outside gaze-types.
DictionaryStats: Dictionary metadata used for diagnostics and tests.
DocumentExtension: Signed document-context metadata carried inside a session snapshot envelope.
DocumentExtensionBuilder: Builder for DocumentExtension that requires signed integrity-binding fields.
EmittedTokenSpan: A replacement emitted by the pseudonymization pipeline.
LeakReport: A suspected missed PII span reported by a SafetyNet.
LeakReportStats: Aggregate leak report statistics.
LeakSuspect: Suspected leak reported by an observer-only safety net.
LocaleChain: Ordered locale fallback chain.
Manifest: Set of emitted token spans for one clean text segment.
RedactionEntry: One row of redaction metadata emitted to a RedactionLogger.
RulepackDict: Dictionary declared by a rulepack.
SafetyNetContext: Context passed to a privacy safety net.

Action: Disposition applied to a detected PII span.
CleanDocument: The pseudonymized output from Pipeline::redact.
ConflictTier: Conflict resolution tier that selected or rejected a candidate.
DictionaryLoadError: Error raised when constructing invalid dictionary entries.
DictionarySource: Source of a dictionary entry.
DocumentExtensionError: Errors returned while building a DocumentExtension.
DocumentKind: Source document kind for metadata-only audit logging.
ExtractionDensityPolicy: Per-codec declaration for text extraction density checks.
LeakKind: The category of a suspected missed PII span.
LeakReportTelemetry: Bytes-free telemetry emitted by safety-net orchestration.
LocaleError: Locale parsing error.
LocaleTag: Locale tag recognized by policy and recognizers.
OpenAiPrivateLabel: Closed set of upstream OpenAI Privacy Filter labels accepted by Gaze.
PiiClass: The category of a detected PII span.
RawDocument: The input document submitted for pseudonymization.
RedactionLogError: Closed error set for redaction log sinks.
SafetyNetError: Exhaustive, closed error set for safety-net execution.
SafetyNetPiiClass: Closed safety-net PII vocabulary before mapping into PiiClass.
TextOrigin: Provenance of text extracted from a document or transcript source.
Value: Minimal structured value representation that avoids a serde_json dependency.

Detector: Shared detector contract for text-only PII detection.
Recognizer: Shared recognizer contract for locale-aware PII candidates.
RedactionLogger: Trait for audit sinks that receive redaction metadata.
SafetyNet: Observer-only post-clean check (Pass 3 in the detection pipeline).