opencode-rehydra-core 0.14.0

package/dist/core/anonymizer.d.ts

@@ -0,0 +1,309 @@
+/**
+ * Core Anonymizer Module
+ * Shared implementation for both browser and Node.js entry points
+ */
+import { AnonymizationResult, AnonymizationPolicy, AnonymizationMode, SemanticConfig, SecretsConfig, PIIType, TagFormat } from "../types/index.js";
+import { RecognizerRegistry } from "../recognizers/index.js";
+import { type INERModel, type OrtSessionOptions } from "../ner/index.js";
+import { type NERModelMode, type DownloadProgressCallback } from "../ner/model-manager.js";
+import { type RawPIIMap } from "../pipeline/tagger.js";
+import { type KeyProvider } from "../crypto/index.js";
+export type { RawPIIMap } from "../pipeline/tagger.js";
+import type { PIIStorageProvider, AnonymizerSession } from "../storage/types.js";
+export type { PIIStorageProvider, AnonymizerSession };
+/**
+ * NER configuration options
+ */
+export interface NERConfig {
+    /**
+     * NER model mode:
+     * - 'standard': Full-size multilingual model (~1.1 GB)
+     * - 'quantized': Smaller quantized model (~280 MB)
+     * - 'disabled': No NER, regex-only detection
+     * - 'custom': Use custom model paths
+     */
+    mode: NERModelMode;
+    /**
+     * Custom model path (required when mode is 'custom')
+     */
+    modelPath?: string;
+    /**
+     * Custom vocab path (required when mode is 'custom')
+     */
+    vocabPath?: string;
+    /**
+     * Whether to auto-download model if not present
+     * @default true
+     */
+    autoDownload?: boolean;
+    /**
+     * Callback for download progress
+     */
+    onDownloadProgress?: DownloadProgressCallback;
+    /**
+     * Callback for status messages
+     */
+    onStatus?: (status: string) => void;
+    /**
+     * Confidence thresholds per PII type (0.0 - 1.0)
+     * Overrides default thresholds for specified types
+     * @example { PERSON: 0.8, ORG: 0.7 }
+     */
+    thresholds?: Partial<Record<PIIType, number>>;
+    /**
+     * ONNX session options for performance tuning
+     * Allows customizing execution providers, threading, and graph optimizations
+     * @example { executionProviders: ['CoreMLExecutionProvider', 'CPUExecutionProvider'] }
+     */
+    sessionOptions?: OrtSessionOptions;
+    /**
+     * Inference backend: 'local' (default) or 'inference-server'
+     * - 'local': Local ONNX Runtime inference (CPU)
+     * - 'inference-server': Remote GPU inference via HTTP (enterprise deployment)
+     * @default 'local'
+     */
+    backend?: "local" | "inference-server";
+    /**
+     * Inference server URL (required when backend is 'inference-server')
+     * @example 'http://localhost:8080'
+     */
+    inferenceServerUrl?: string;
+    /**
+     * Inference server request timeout in milliseconds (default: 30000)
+     * Only used when backend is 'inference-server'
+     */
+    inferenceServerTimeout?: number;
+    /**
+     * Enable case-insensitive fallback for detecting lowercase names.
+     * Runs a second NER pass on title-cased text and merges new detections.
+     * Doubles NER inference time but catches names like "tom" that the
+     * case-sensitive model would otherwise miss.
+     * @default false
+     */
+    caseFallback?: boolean;
+    /**
+     * Confidence penalty multiplier for case-fallback detections (0.0 - 1.0).
+     * Applied as: confidence * caseFallbackPenalty
+     * @default 0.85
+     */
+    caseFallbackPenalty?: number;
+    /**
+     * Number of content tokens shared by adjacent windows when a long input is
+     * split into multiple inference windows. Larger overlap improves recall for
+     * entities near window boundaries at the cost of more windows.
+     * @default min(64, floor((maxLength - 2) / 4))
+     */
+    windowOverlapTokens?: number;
+    /**
+     * Maximum number of inference windows per input. Inputs that tokenize to
+     * more windows are scanned only up to this bound (onWindowCap fires once per
+     * capped inference pass — so twice for one input when caseFallback runs a
+     * second pass). Guards against unbounded inference cost on very large
+     * opaque payloads (e.g. multi-megabyte base64 attachments).
+     * @default undefined (unbounded — full coverage)
+     */
+    maxWindowsPerInput?: number;
+    /**
+     * Called when an input exceeds maxWindowsPerInput and its tail is left
+     * unscanned by NER. Callers should log/meter this — it is a coverage
+     * reduction, not an error. When absent, a console warning is emitted.
+     * `inputLength` is the full input character count.
+     */
+    onWindowCap?: (info: {
+        scannedWindows: number;
+        inputLength: number;
+    }) => void;
+}
+/**
+ * Anonymizer configuration
+ */
+export interface AnonymizerConfig {
+    /** Recognizer registry (uses default if not provided) */
+    registry?: RecognizerRegistry;
+    /**
+     * Anonymization mode:
+     * - 'pseudonymize': Reversible anonymization with encrypted PII map (default)
+     * - 'anonymize': Irreversible anonymization without PII map
+     * @default 'pseudonymize'
+     */
+    mode?: AnonymizationMode;
+    /**
+     * NER configuration
+     * @default { mode: 'disabled' }
+     */
+    ner?: NERConfig;
+    /**
+     * Semantic enrichment configuration
+     * Enables MT-friendly PII tags with gender/scope attributes
+     * @default { enabled: false }
+     */
+    semantic?: SemanticConfig;
+    /**
+     * Secrets/credentials detection configuration
+     * Detects API keys, private keys, JWTs, connection strings, etc.
+     * @default { enabled: false }
+     */
+    secrets?: SecretsConfig;
+    /** Key provider for encryption (generates random key if not provided) */
+    keyProvider?: KeyProvider;
+    /**
+     * PII storage provider for automatic session-based persistence
+     * When provided, enables the session() method for automatic PII map storage
+     */
+    piiStorageProvider?: PIIStorageProvider;
+    /** Default policy — merged with SDK defaults, so partial overrides are safe */
+    defaultPolicy?: Partial<AnonymizationPolicy>;
+    /** Model version string */
+    modelVersion?: string;
+    /** Policy version string */
+    policyVersion?: string;
+    /**
+     * Optional callback for validation warnings emitted during anonymization.
+     * When not provided, warnings are silently ignored.
+     * Previously these were logged via console.warn, which pollutes stdout
+     * in TUI-based editors (e.g. OpenCode).
+     */
+    onValidationWarning?: (warnings: Array<{
+        code: string;
+        message: string;
+    }>) => void;
+    /**
+     * Tag format configuration for PII placeholder tags.
+     * Controls the delimiters and keyword used in tags.
+     * @default { open: "<", close: "/>", keyword: "PII" } (XML-style)
+     * @example { open: "[[", close: "]]" } for bracket-style tags
+     */
+    tagFormat?: TagFormat;
+}
+/**
+ * Session implementation factory type
+ */
+export type SessionFactory = (anonymizer: Anonymizer, sessionId: string, storage: PIIStorageProvider, keyProvider: KeyProvider) => AnonymizerSession;
+/**
+ * Anonymizer instance
+ * Main class for performing PII anonymization
+ */
+export declare class Anonymizer {
+    private registry;
+    private nerModel;
+    private nerConfig;
+    private semanticConfig;
+    private secretsConfig;
+    private mode;
+    private keyProvider;
+    private piiStorageProvider;
+    private defaultPolicy;
+    private modelVersion;
+    private policyVersion;
+    private initialized;
+    private semanticDataReady;
+    private sessionFactory;
+    private onValidationWarning;
+    private tagFormat;
+    constructor(config?: AnonymizerConfig, sessionFactory?: SessionFactory);
+    /**
+     * Returns the resolved tag format configuration
+     */
+    get resolvedTagFormat(): TagFormat;
+    /**
+     * Initializes the anonymizer
+     * Downloads NER model and semantic data if needed and loads them
+     */
+    initialize(): Promise<void>;
+    /**
+     * Anonymizes text, replacing PII with placeholder tags
+     * @param text - Input text to anonymize
+     * @param locale - Optional locale hint (e.g., 'de-DE', 'en-US')
+     * @param policy - Optional policy override
+     * @param existingPiiMap - Optional existing PII map for session-level ID reuse
+     * @returns Anonymization result with anonymized text and encrypted PII map
+     */
+    anonymize(text: string, locale?: string, policy?: Partial<AnonymizationPolicy>, existingPiiMap?: RawPIIMap): Promise<AnonymizationResult>;
+    /**
+     * Disposes of resources
+     */
+    dispose(): Promise<void>;
+    /**
+     * Gets the recognizer registry
+     */
+    getRegistry(): RecognizerRegistry;
+    /**
+     * Gets the NER model
+     */
+    getNERModel(): INERModel | null;
+    /**
+     * Whether the anonymizer is initialized
+     */
+    get isInitialized(): boolean;
+    /**
+     * Creates a session-bound interface for automatic PII map storage
+     *
+     * @param sessionId - Unique identifier for this session/conversation
+     * @returns AnonymizerSession that auto-saves on anonymize and auto-loads on rehydrate
+     * @throws Error if piiStorageProvider was not configured
+     *
+     * @example
+     * ```typescript
+     * const session = anonymizer.session('chat-123');
+     *
+     * // Anonymize - auto-saves to storage
+     * const result = await session.anonymize('Hello John Smith!');
+     *
+     * // Rehydrate - auto-loads and decrypts
+     * const original = await session.rehydrate(translatedText);
+     * ```
+     */
+    session(sessionId: string): AnonymizerSession;
+}
+/**
+ * Creates an anonymizer with the specified configuration
+ *
+ * @example
+ * ```typescript
+ * // Regex-only (no NER)
+ * const anonymizer = createAnonymizer();
+ *
+ * // With NER (auto-downloads model on first use)
+ * const anonymizer = createAnonymizer({
+ *   ner: { mode: 'quantized' }
+ * });
+ *
+ * // With NER and progress callback
+ * const anonymizer = createAnonymizer({
+ *   ner: {
+ *     mode: 'standard',
+ *     onStatus: (status) => console.log(status),
+ *     onDownloadProgress: (p) => console.log(`${p.file}: ${p.percent}%`)
+ *   }
+ * });
+ * ```
+ */
+export declare function createAnonymizer(config?: AnonymizerConfig, sessionFactory?: SessionFactory): Anonymizer;
+/**
+ * Convenience function for one-off anonymization
+ * Creates a temporary anonymizer with default settings (regex-only)
+ */
+export declare function anonymize(text: string, locale?: string, policy?: Partial<AnonymizationPolicy>): Promise<AnonymizationResult>;
+/**
+ * Quick regex-only anonymization (no NER, faster)
+ */
+export declare function anonymizeRegexOnly(text: string, policy?: Partial<AnonymizationPolicy>): Promise<AnonymizationResult>;
+/**
+ * Full anonymization with NER
+ * Auto-downloads the quantized model on first use
+ *
+ * @example
+ * ```typescript
+ * const result = await anonymizeWithNER(
+ *   'Contact John Smith at john@example.com',
+ *   {
+ *     mode: 'quantized',
+ *     onStatus: console.log
+ *   }
+ * );
+ * ```
+ */
+export declare function anonymizeWithNER(text: string, nerConfig: Omit<NERConfig, "mode"> & {
+    mode?: "standard" | "quantized";
+}, policy?: Partial<AnonymizationPolicy>): Promise<AnonymizationResult>;
+//# sourceMappingURL=anonymizer.d.ts.map

package/dist/core/anonymizer.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"anonymizer.d.ts","sourceRoot":"","sources":["../../src/core/anonymizer.ts"],"names":[],"mappings":"AAAA;;;GAGG;AAEH,OAAO,EACL,mBAAmB,EACnB,mBAAmB,EAEnB,iBAAiB,EAEjB,cAAc,EACd,aAAa,EAEb,OAAO,EACP,SAAS,EAIV,MAAM,mBAAmB,CAAC;AAE3B,OAAO,EAEL,kBAAkB,EAEnB,MAAM,yBAAyB,CAAC;AAGjC,OAAO,EACL,KAAK,SAAS,EAId,KAAK,iBAAiB,EACvB,MAAM,iBAAiB,CAAC;AAEzB,OAAO,EACL,KAAK,YAAY,EAGjB,KAAK,wBAAwB,EAC9B,MAAM,yBAAyB,CAAC;AAIjC,OAAO,EAGL,KAAK,SAAS,EACf,MAAM,uBAAuB,CAAC;AAY/B,OAAO,EAGL,KAAK,WAAW,EACjB,MAAM,oBAAoB,CAAC;AAI5B,YAAY,EAAE,SAAS,EAAE,MAAM,uBAAuB,CAAC;AAGvD,OAAO,KAAK,EACV,kBAAkB,EAClB,iBAAiB,EAClB,MAAM,qBAAqB,CAAC;AAG7B,YAAY,EAAE,kBAAkB,EAAE,iBAAiB,EAAE,CAAC;AAsCtD;;GAEG;AACH,MAAM,WAAW,SAAS;IACxB;;;;;;OAMG;IACH,IAAI,EAAE,YAAY,CAAC;IAEnB;;OAEG;IACH,SAAS,CAAC,EAAE,MAAM,CAAC;IAEnB;;OAEG;IACH,SAAS,CAAC,EAAE,MAAM,CAAC;IAEnB;;;OAGG;IACH,YAAY,CAAC,EAAE,OAAO,CAAC;IAEvB;;OAEG;IACH,kBAAkB,CAAC,EAAE,wBAAwB,CAAC;IAE9C;;OAEG;IACH,QAAQ,CAAC,EAAE,CAAC,MAAM,EAAE,MAAM,KAAK,IAAI,CAAC;IAEpC;;;;OAIG;IACH,UAAU,CAAC,EAAE,OAAO,CAAC,MAAM,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC,CAAC;IAE9C;;;;OAIG;IACH,cAAc,CAAC,EAAE,iBAAiB,CAAC;IAEnC;;;;;OAKG;IACH,OAAO,CAAC,EAAE,OAAO,GAAG,kBAAkB,CAAC;IAEvC;;;OAGG;IACH,kBAAkB,CAAC,EAAE,MAAM,CAAC;IAE5B;;;OAGG;IACH,sBAAsB,CAAC,EAAE,MAAM,CAAC;IAEhC;;;;;;OAMG;IACH,YAAY,CAAC,EAAE,OAAO,CAAC;IAEvB;;;;OAIG;IACH,mBAAmB,CAAC,EAAE,MAAM,CAAC;IAE7B;;;;;OAKG;IACH,mBAAmB,CAAC,EAAE,MAAM,CAAC;IAE7B;;;;;;;OAOG;IACH,kBAAkB,CAAC,EAAE,MAAM,CAAC;IAE5B;;;;;OAKG;IACH,WAAW,CAAC,EAAE,CAAC,IAAI,EAAE;QAAE,cAAc,EAAE,MAAM,CAAC;QAAC,WAAW,EAAE,MAAM,CAAA;KAAE,KAAK,IAAI,CAAC;CAC/E;AAED;;GAEG;AACH,MAAM,WAAW,gBAAgB;IAC/B,yDAAyD;IACzD,QAAQ,CAAC,EAAE,kBAAkB,CAAC;IAE9B;;;;;OAKG;IACH,IAAI,CAAC,EAAE,iBAAiB,CAAC;IAEzB;;;OAGG;IACH,GAAG,CAAC,EAAE,SAAS,CAAC;IAEhB;;;;OAIG;IACH,QAAQ,CAAC,EAAE,cAAc,CAAC;IAE1B;;;;OAIG;IACH,OAAO,CAAC,EAAE,aAAa,CAAC;IAExB,yEAAyE;IACzE,WAAW,CAAC,EAAE,WAAW,CAAC;IAE1B;;;OAGG;IACH,kBAAkB,CAAC,EAAE,kBAAkB,CAAC;IAExC,+EAA+E;IAC/E,aAAa,CAAC,EAAE,OAAO,CAAC,mBAAmB,CAAC,CAAC;IAE7C,2BAA2B;IAC3B,YAAY,CAAC,EAAE,MAAM,CAAC;IAEtB,4BAA4B;IAC5B,aAAa,CAAC,EAAE,MAAM,CAAC;IAEvB;;;;;OAKG;IACH,mBAAmB,CAAC,EAAE,CAAC,QAAQ,EAAE,KAAK,CAAC;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,OAAO,EAAE,MAAM,CAAA;KAAE,CAAC,KAAK,IAAI,CAAC;IAEnF;;;;;OAKG;IACH,SAAS,CAAC,EAAE,SAAS,CAAC;CACvB;AAED;;GAEG;AACH,MAAM,MAAM,cAAc,GAAG,CAC3B,UAAU,EAAE,UAAU,EACtB,SAAS,EAAE,MAAM,EACjB,OAAO,EAAE,kBAAkB,EAC3B,WAAW,EAAE,WAAW,KACrB,iBAAiB,CAAC;AAEvB;;;GAGG;AACH,qBAAa,UAAU;IACrB,OAAO,CAAC,QAAQ,CAAqB;IACrC,OAAO,CAAC,QAAQ,CAA0B;IAC1C,OAAO,CAAC,SAAS,CAAY;IAC7B,OAAO,CAAC,cAAc,CAAiB;IACvC,OAAO,CAAC,aAAa,CAAgB;IACrC,OAAO,CAAC,IAAI,CAAoB;IAChC,OAAO,CAAC,WAAW,CAAqB;IACxC,OAAO,CAAC,kBAAkB,CAA4B;IACtD,OAAO,CAAC,aAAa,CAAsB;IAC3C,OAAO,CAAC,YAAY,CAAS;IAC7B,OAAO,CAAC,aAAa,CAAS;IAC9B,OAAO,CAAC,WAAW,CAAS;IAC5B,OAAO,CAAC,iBAAiB,CAAS;IAClC,OAAO,CAAC,cAAc,CAAwB;IAC9C,OAAO,CAAC,mBAAmB,CAAwE;IACnG,OAAO,CAAC,SAAS,CAAY;gBAEjB,MAAM,GAAE,gBAAqB,EAAE,cAAc,CAAC,EAAE,cAAc;IA4E1E;;OAEG;IACH,IAAI,iBAAiB,IAAI,SAAS,CAEjC;IAED;;;OAGG;IACG,UAAU,IAAI,OAAO,CAAC,IAAI,CAAC;IA6JjC;;;;;;;OAOG;IACG,SAAS,CACb,IAAI,EAAE,MAAM,EACZ,MAAM,CAAC,EAAE,MAAM,EACf,MAAM,CAAC,EAAE,OAAO,CAAC,mBAAmB,CAAC,EACrC,cAAc,CAAC,EAAE,SAAS,GACzB,OAAO,CAAC,mBAAmB,CAAC;IA4I/B;;OAEG;IACG,OAAO,IAAI,OAAO,CAAC,IAAI,CAAC;IAO9B;;OAEG;IACH,WAAW,IAAI,kBAAkB;IAIjC;;OAEG;IACH,WAAW,IAAI,SAAS,GAAG,IAAI;IAI/B;;OAEG;IACH,IAAI,aAAa,IAAI,OAAO,CAE3B;IAED;;;;;;;;;;;;;;;;;OAiBG;IACH,OAAO,CAAC,SAAS,EAAE,MAAM,GAAG,iBAAiB;CAmD9C;AAED;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,wBAAgB,gBAAgB,CAC9B,MAAM,CAAC,EAAE,gBAAgB,EACzB,cAAc,CAAC,EAAE,cAAc,GAC9B,UAAU,CAEZ;AAED;;;GAGG;AACH,wBAAsB,SAAS,CAC7B,IAAI,EAAE,MAAM,EACZ,MAAM,CAAC,EAAE,MAAM,EACf,MAAM,CAAC,EAAE,OAAO,CAAC,mBAAmB,CAAC,GACpC,OAAO,CAAC,mBAAmB,CAAC,CAS9B;AAED;;GAEG;AACH,wBAAsB,kBAAkB,CACtC,IAAI,EAAE,MAAM,EACZ,MAAM,CAAC,EAAE,OAAO,CAAC,mBAAmB,CAAC,GACpC,OAAO,CAAC,mBAAmB,CAAC,CAQ9B;AAED;;;;;;;;;;;;;;GAcG;AACH,wBAAsB,gBAAgB,CACpC,IAAI,EAAE,MAAM,EACZ,SAAS,EAAE,IAAI,CAAC,SAAS,EAAE,MAAM,CAAC,GAAG;IAAE,IAAI,CAAC,EAAE,UAAU,GAAG,WAAW,CAAA;CAAE,EACxE,MAAM,CAAC,EAAE,OAAO,CAAC,mBAAmB,CAAC,GACpC,OAAO,CAAC,mBAAmB,CAAC,CAe9B"}