tokenfold 0.4.0 → 0.5.0

package/README.md

@@ -1,19 +1,48 @@
-# tokenfold for Node.js
+# tokenfold for Node.js
 
-Zero-runtime-dependency TypeScript bindings for the tokenfold Rust CLI.
+Zero-runtime-dependency TypeScript bindings for the Tokenfold Rust CLI.
 
 ```sh
 npm install tokenfold
 ```
 
 ```ts
-import { compress } from "tokenfold";
+import { compress, decode, inspect } from "tokenfold";
 
-const { payload, report } = await compress(input, {
+const receipt = await inspect(input, { format: "json", preset: "balanced" });
+const { payload, report, text } = await compress(input, {
   format: "json",
-  mode: "balanced",
+  preset: "balanced",
 });
 ```
 
+Payloads are `Uint8Array`; the `text` convenience getter decodes UTF-8 strictly.
+`inspect` returns only the side-effect-free receipt.
+
+Recoverable pruning is explicit and generic-JSON-only:
+
+```ts
+import { compress, retrieve } from "tokenfold";
+
+const result = await compress(feed, {
+  format: "json",
+  targetTokens: 2_000,
+  pruning: {
+    keepRatio: 0.35,
+    preservePaths: ["meta"],
+    retrievalStore: ".tokenfold/retrieve",
+  },
+});
+const marker = JSON.parse(result.text).items.find((item) => item.$tf_ref);
+const original = await retrieve(marker, { retrievalStore: ".tokenfold/retrieve" });
+```
+
+Explicit TOON output is verified before emission and restored with `decode`:
+
+```ts
+const encoded = await compress(input, { format: "json", encoding: "toon" });
+const jsonBytes = await decode(encoded.payload, { from: "toon" });
+```
+
 Requires Node.js 22 or newer. The matching native CLI is installed through an
 optional platform package; set `TOKENFOLD_BINARY_PATH` to use a custom binary.

package/dist/index.d.ts

@@ -3,17 +3,23 @@ import { TokenFoldProcessError } from "./errors.js";
 import { run, type Input, type ProcessResult, type RunOptions } from "./process.js";
 export { binaryPath, run, TokenFoldProcessError };
 export type { Input, ProcessResult, RunOptions };
-export type CompressionMode = "conservative" | "balanced" | "aggressive";
-export type TaskScope = "all" | "general" | "code_review" | "change_summary" | "debugging" | "generation" | "api_overview" | "retrieval_qa" | "agent_history";
+export type Preset = "conservative" | "balanced" | "aggressive";
+export type OutputEncoding = "json" | "toon";
+export type InputFormat = "auto" | "openai" | "anthropic" | "json" | "text" | "command" | "diff";
+export type DecodeFormat = "auto" | "json" | "toon" | "text";
+export interface PruningPolicy {
+    keepRatio?: number;
+    preservePaths?: readonly string[];
+    retrievalStore?: string;
+    retrievalNamespace?: string;
+}
 export interface CompressionOptions {
-    format?: "auto" | "openai" | "anthropic" | "json" | "text" | "command" | "diff";
-    mode?: CompressionMode;
+    format?: InputFormat;
+    preset?: Preset;
     targetTokens?: number;
-    disable?: readonly string[];
-    taskScope?: TaskScope;
-    experimental?: boolean;
-    storeOriginals?: boolean;
-    retrieveNamespace?: string;
+    requireTarget?: boolean;
+    encoding?: OutputEncoding;
+    pruning?: PruningPolicy;
     configPath?: string;
     signal?: AbortSignal;
 }
@@ -22,19 +28,6 @@ export interface EstimatorInfo {
     model: string | null;
     is_exact: boolean;
 }
-export interface BudgetReport {
-    target_tokens: number | null;
-    protected_floor: number;
-    achieved_tokens: number;
-}
-export interface QualityReport {
-    eval_profile_id: string;
-    task_scope: string;
-    validated_ratio_band: string | null;
-    quality_retention: number;
-    contrastive_failure_rate: number;
-    gate_passed: boolean;
-}
 export interface Warning {
     code: string;
     severity: "info" | "warn" | "critical";
@@ -53,12 +46,39 @@ export interface TransformReport {
     skipped_reason: string | null;
     warnings: readonly Warning[];
 }
-export interface CacheReport {
-    boundary_kind: string | null;
-    protected_bytes: number;
-    prefix_byte_identical: boolean;
+export interface BudgetReport {
+    status: "not_requested" | "met" | "best_effort" | "unreachable";
+    target_tokens: number | null;
+    protected_floor: number;
+    achieved_tokens: number;
+}
+export interface EncodingReport {
+    codec: string;
+    version: string;
+    roundtrip_verified: boolean;
+    tokens_before: number;
+    tokens_after: number;
+    token_delta: number;
     warnings: readonly Warning[];
 }
+export interface PruningReport {
+    requested: boolean;
+    applied: boolean;
+    preview: boolean;
+    candidate_items: number;
+    retained_items: number;
+    pruned_items: number;
+    evidence_refs: number;
+    preserve_paths: readonly string[];
+}
+export interface QualityReport {
+    eval_profile_id: string;
+    task_scope: string;
+    validated_ratio_band: string | null;
+    quality_retention: number | null;
+    contrastive_failure_rate: number | null;
+    gate_passed: boolean;
+}
 export interface RetrievalReport {
     store_namespace: string;
     hash_algorithm: string;
@@ -67,64 +87,81 @@ export interface RetrievalReport {
     persisted_original_bytes: number;
     skipped_original_bytes: number;
 }
-export interface OutputSavingsReport {
-    profile: string;
-    estimated_output_tokens_saved: number | null;
-    measured_output_tokens_saved: number | null;
-    provenance: string;
-}
-export interface BypassReport {
-    reason: string;
-    source: string;
-}
-export interface CommandReport {
-    command_family: string | null;
-    child_exit_code: number | null;
-    duration_ms: number;
-    raw_output_bytes: number;
-    stdout_bytes: number;
-    stderr_bytes: number;
-    stderr_mode: string;
-    stderr_truncated: boolean;
-    compressed_output_bytes: number;
-    filter_pack_id: string | null;
-    filter_version: string | null;
-    never_worse_applied: boolean;
+export interface PipelineStageReport {
+    id: string;
+    version: string | null;
+    input_bytes: number | null;
+    output_bytes: number | null;
+    saved_bytes: number | null;
+    input_tokens: number | null;
+    output_tokens: number | null;
+    saved_tokens: number | null;
+    estimator: EstimatorInfo | null;
+    status: string;
+    duration_ms: number | null;
     bypass_reason: string | null;
+    provenance: string;
+    recoverability: string;
+    evidence_ref: string | null;
 }
-export interface LedgerReport {
-    recorded: boolean;
-    scope: string | null;
-    project_hash: string | null;
-    record_id: string | null;
+export interface PipelineReport {
+    raw_input_bytes: number | null;
+    raw_input_tokens: number | null;
+    final_output_bytes: number;
+    final_output_tokens: number;
+    total_saved_tokens: number | null;
+    raw_capture: string;
+    upstream_recoverability: string;
+    stages: readonly PipelineStageReport[];
 }
-export interface CompressionReport {
+export interface CompressionReceipt {
     schema_version: string;
+    status: "compressed" | "passthrough";
     original_tokens: number;
     compressed_tokens: number;
     saved_tokens: number;
     savings_ratio: number;
     savings_pct: number;
     estimator: EstimatorInfo;
-    status: "compressed" | "passthrough" | "best_effort" | "unreachable_target";
-    mode: string;
+    preset: Preset;
     format: string;
+    output_encoding: string;
     task_scope: string;
     request_id: string | null;
+    pipeline: PipelineReport | null;
     quality: QualityReport | null;
     budget: BudgetReport | null;
-    cache: CacheReport | null;
+    encoding: EncodingReport | null;
+    pruning: PruningReport | null;
     retrieval: RetrievalReport | null;
-    output_savings: OutputSavingsReport | null;
-    bypass: BypassReport | null;
-    command: CommandReport | null;
-    ledger: LedgerReport | null;
     transforms: readonly TransformReport[];
     warnings: readonly Warning[];
+    cache: unknown;
+    output_savings: unknown;
+    bypass: unknown;
+    command: unknown;
+    ledger: unknown;
 }
+export type CompressionReport = CompressionReceipt;
 export interface CompressionResult {
     payload: Uint8Array;
-    report: CompressionReport;
+    readonly text: string;
+    report: CompressionReceipt;
+}
+export declare class BudgetUnmetError extends Error {
+    readonly receipt: CompressionReceipt;
+    constructor(receipt: CompressionReceipt);
 }
 export declare function compress(input: Input, options?: CompressionOptions): Promise<CompressionResult>;
-export declare function inspect(input: Input, options?: CompressionOptions): Promise<CompressionResult>;
+export declare function inspect(input: Input, options?: CompressionOptions): Promise<CompressionReceipt>;
+export declare function decode(input: Input, options?: {
+    from?: DecodeFormat;
+    signal?: AbortSignal;
+}): Promise<Uint8Array>;
+export interface RetrieveOptions {
+    retrievalStore?: string;
+    namespace?: string;
+    configPath?: string;
+    signal?: AbortSignal;
+}
+export declare function retrieve(reference: string | Record<string, unknown>, options?: RetrieveOptions): Promise<Uint8Array>;

package/dist/index.js

@@ -2,68 +2,98 @@ import { binaryPath } from "./binary.js";
 import { TokenFoldProcessError } from "./errors.js";
 import { run } from "./process.js";
 export { binaryPath, run, TokenFoldProcessError };
+export class BudgetUnmetError extends Error {
+    receipt;
+    constructor(receipt) { super(`token budget unmet: achieved ${receipt.compressed_tokens} tokens`); this.name = "BudgetUnmetError"; this.receipt = receipt; }
+}
 function argumentsFor(command, options) {
-    const args = [command, "--json"];
+    const args = [command, "--receipt-format", "json"];
     if (options.format)
         args.push("--format", options.format);
-    if (options.mode)
-        args.push("--mode", options.mode);
+    if (options.preset)
+        args.push("--preset", options.preset);
     if (options.targetTokens !== undefined)
         args.push("--target-tokens", String(options.targetTokens));
-    if (options.disable?.length && command === "compress")
-        args.push("--disable", options.disable.join(","));
-    if (options.taskScope)
-        args.push("--task-scope", options.taskScope);
-    if (options.experimental)
-        args.push("--experimental");
-    if (options.storeOriginals && command === "compress")
-        args.push("--store-originals");
-    if (options.retrieveNamespace && command === "compress") {
-        args.push("--retrieve-namespace", options.retrieveNamespace);
+    if (options.requireTarget)
+        args.push("--require-target");
+    if (options.encoding)
+        args.push("--encoding", options.encoding);
+    if (options.pruning) {
+        args.push("--prune");
+        if (options.pruning.keepRatio !== undefined)
+            args.push("--keep-ratio", String(options.pruning.keepRatio));
+        for (const path of options.pruning.preservePaths ?? [])
+            args.push("--preserve", path);
+        if (options.pruning.retrievalStore)
+            args.push("--retrieval-store", options.pruning.retrievalStore);
+        if (options.pruning.retrievalNamespace)
+            args.push("--retrieval-namespace", options.pruning.retrievalNamespace);
     }
     if (options.configPath)
         args.push("--config", options.configPath);
     return args;
 }
-function parseReport(bytes, result) {
+function parseReceipt(bytes, result) {
+    const text = Buffer.from(bytes).toString("utf8");
+    const json = text.split("\ntokenfold:", 1)[0] ?? "";
     try {
-        return JSON.parse(Buffer.from(bytes).toString("utf8"));
+        return JSON.parse(json);
     }
     catch (cause) {
-        throw new TokenFoldProcessError("tokenfold returned an invalid JSON report", {
-            code: "invalid_report",
-            exitCode: result.exitCode,
-            signal: result.signal,
-            stderr: result.stderr,
-            cause,
-        });
+        throw new TokenFoldProcessError("tokenfold returned an invalid JSON receipt", { code: "invalid_report", exitCode: result.exitCode, signal: result.signal, stderr: result.stderr, cause });
     }
 }
-async function execute(command, input, options) {
-    const runOptions = {
-        stdin: input,
-        env: { TOKENFOLD_ANALYTICS_ENABLED: "false" },
-    };
-    if (options.signal)
-        runOptions.signal = options.signal;
-    const result = await run(argumentsFor(command, options), runOptions);
-    if (result.exitCode !== 0) {
-        throw new TokenFoldProcessError(`tokenfold exited with status ${result.exitCode ?? result.signal}`, {
-            code: "tokenfold_exit",
-            exitCode: result.exitCode,
-            signal: result.signal,
-            stderr: result.stderr,
-        });
-    }
-    const reportBytes = command === "compress" ? result.stderr : result.stdout;
-    return {
-        payload: command === "compress" ? result.stdout : Uint8Array.from(Buffer.from(input)),
-        report: parseReport(reportBytes, result),
-    };
+function optionsFor(input, signal) {
+    const options = { env: { TOKENFOLD_ANALYTICS_ENABLED: "false" } };
+    if (input !== undefined)
+        options.stdin = input;
+    if (signal)
+        options.signal = signal;
+    return options;
+}
+function withText(payload, report) {
+    return { payload, report, get text() { return new TextDecoder("utf-8", { fatal: true }).decode(payload); } };
+}
+export async function compress(input, options = {}) {
+    const result = await run(argumentsFor("compress", options), optionsFor(input, options.signal));
+    if (result.exitCode !== 0 && result.exitCode !== 7)
+        throwProcess(result);
+    const receipt = parseReceipt(result.stderr, result);
+    if (result.exitCode === 7)
+        throw new BudgetUnmetError(receipt);
+    return withText(result.stdout, receipt);
 }
-export function compress(input, options = {}) {
-    return execute("compress", input, options);
+export async function inspect(input, options = {}) {
+    const result = await run(argumentsFor("inspect", options), optionsFor(input, options.signal));
+    if (result.exitCode !== 0 && result.exitCode !== 7)
+        throwProcess(result);
+    const receipt = parseReceipt(result.stdout, result);
+    if (result.exitCode === 7)
+        throw new BudgetUnmetError(receipt);
+    return receipt;
+}
+export async function decode(input, options = {}) {
+    const args = ["decode"];
+    if (options.from)
+        args.push("--from", options.from);
+    const result = await run(args, optionsFor(input, options.signal));
+    if (result.exitCode !== 0)
+        throwProcess(result);
+    return result.stdout;
+}
+export async function retrieve(reference, options = {}) {
+    const args = ["retrieve", typeof reference === "string" ? reference : JSON.stringify(reference)];
+    if (options.retrievalStore)
+        args.push("--retrieval-store", options.retrievalStore);
+    if (options.namespace)
+        args.push("--retrieval-namespace", options.namespace);
+    if (options.configPath)
+        args.push("--config", options.configPath);
+    const result = await run(args, optionsFor(undefined, options.signal));
+    if (result.exitCode !== 0)
+        throwProcess(result);
+    return result.stdout;
 }
-export function inspect(input, options = {}) {
-    return execute("inspect", input, options);
+function throwProcess(result) {
+    throw new TokenFoldProcessError(`tokenfold exited with status ${result.exitCode ?? result.signal}`, { code: "tokenfold_exit", exitCode: result.exitCode, signal: result.signal, stderr: result.stderr });
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tokenfold",
-  "version": "0.4.0",
+  "version": "0.5.0",
   "description": "Token-aware compression for LLM payloads, backed by the tokenfold Rust CLI.",
   "license": "Apache-2.0",
   "repository": {
@@ -49,11 +49,11 @@
     "prepack": "npm run build"
   },
   "optionalDependencies": {
-    "@tokenfold/cli-darwin-x64": "0.4.0",
-    "@tokenfold/cli-darwin-arm64": "0.4.0",
-    "@tokenfold/cli-linux-x64": "0.4.0",
-    "@tokenfold/cli-linux-arm64": "0.4.0",
-    "@tokenfold/cli-win32-x64": "0.4.0"
+    "@tokenfold/cli-darwin-x64": "0.5.0",
+    "@tokenfold/cli-darwin-arm64": "0.5.0",
+    "@tokenfold/cli-linux-x64": "0.5.0",
+    "@tokenfold/cli-linux-arm64": "0.5.0",
+    "@tokenfold/cli-win32-x64": "0.5.0"
   },
   "devDependencies": {
     "@types/node": "^24.0.0",