@alete-ai/gate-ingest 0.1.0

package/dist/index.browser.d.ts

@@ -0,0 +1,106 @@
+import * as _mdream_js from '@mdream/js';
+
+interface RedactorOptions {
+    redactPii?: boolean;
+    redactFinancials?: boolean;
+    redactCredentials?: boolean;
+    redactInfrastructure?: boolean;
+    redactMedical?: boolean;
+    customPlaceholders?: Record<string, string>;
+}
+/**
+ * Sovereign wrapper for the PII Shield engine.
+ * Adopts a "Narrative-First" strategy: Preserves names, dates, and locations
+ * to maintain story coherence while redacting "Toxic Identifiers" (SSNs, Credit Cards, Secrets).
+ */
+declare class Redactor {
+    private engine;
+    constructor(options?: RedactorOptions);
+    /**
+     * Redacts sensitive information from the given text.
+     */
+    redact(text: string): Promise<string>;
+    /**
+     * Checks if the text contains any sensitive information without modifying it.
+     */
+    hasSensitiveInfo(text: string): Promise<boolean>;
+    /**
+     * Performs both detection and redaction in a single pass.
+     */
+    process(text: string): Promise<{
+        redacted: string;
+        hasSensitiveInfo: boolean;
+    }>;
+}
+
+/**
+ * Custom mdream plugin to retain structural artifacts.
+ * This preserves the "structural footprint" of sensitive portals
+ * by injecting intermediate markers with attribute context.
+ */
+declare const structuralPlugin: _mdream_js.TransformPlugin;
+
+/**
+ * Maps structural Markdown/HTML artifacts to explicit alphanumeric tokens.
+ * This prevents Apple's NLTokenizer from stripping critical punctuation.
+ * We use camelCase because NLTokenizer splits snake_case (STRUCT_FORM_START -> STRUCT, FORM, START).
+ */
+declare function mapToTokens(text: string): string;
+
+declare enum GateLabel {
+    SENSITIVE_PORTAL = "sensitive_portal",
+    DIGESTIBLE_ARTICLE = "digestible_article",
+    NOISE = "noise",
+    UNKNOWN = "unknown"
+}
+interface IngestionResult {
+    /**
+     * Alphanumeric tokenized text for Apple MaxEnt classifier.
+     * High structural fidelity, low natural language noise.
+     */
+    structural: string;
+    /**
+     * Clean, readable Markdown for LLM analysis.
+     * Low structural noise, high semantic fidelity.
+     */
+    semantic: string;
+    /**
+     * Whether the semantic content contains sensitive PII that was redacted.
+     */
+    hasSensitiveInfo?: boolean;
+    /**
+     * Extracted metadata from the HTML (title, description, author, etc.)
+     */
+    metadata?: Record<string, string>;
+    /**
+     * Whether the semantic content was truncated due to the token cap.
+     */
+    isTruncated?: boolean;
+}
+interface IngestionOptions {
+    /**
+     * Redaction configuration. If true, uses default settings.
+     */
+    redact?: RedactorOptions | boolean;
+    /**
+     * Override the semantic token cap for this specific call.
+     */
+    semanticTokenCap?: number;
+}
+interface GlobalConfig {
+    /**
+     * The default token cap for semantic returns. Defaults to 15,000.
+     */
+    defaultSemanticTokenCap?: number;
+}
+/**
+ * Initializes the Alete Gate ingestion substrate with global configuration.
+ */
+declare function initialize(config?: GlobalConfig): void;
+/**
+ * The unified ingestion pipeline for Alete Gate.
+ * Converts raw HTML into both structural tokens and semantic Markdown.
+ */
+declare function processHtml(html: string, options?: IngestionOptions): Promise<IngestionResult>;
+
+export { GateLabel, type GlobalConfig, type IngestionOptions, type IngestionResult, Redactor, type RedactorOptions, initialize, mapToTokens, structuralPlugin as plugin, processHtml };