@getrift/rift 0.1.0-beta.2 → 0.1.0-beta.20

package/dist/src/capture/recover-quarantine.d.ts

@@ -0,0 +1,260 @@
+import { z } from "zod";
+/**
+ * Default byte budget per summarization chunk. Sits well below the
+ * 8 MB auto-capture threshold so the Codex CLI input never hits the
+ * same wall that caused the quarantine in the first place. 512 KB
+ * fits comfortably in stdin and leaves room for the prompt.
+ */
+export declare const DEFAULT_RECOVERY_CHUNK_BYTES: number;
+/** Minimum chunk size we allow — anything smaller risks turn-truncation. */
+export declare const MIN_RECOVERY_CHUNK_BYTES: number;
+export declare const PROVENANCE_TOPIC = "recovered_oversized_quarantine";
+declare const QuarantineRecordSchema: z.ZodObject<{
+    source: z.ZodLiteral<"codex_cli">;
+    session_id: z.ZodString;
+    session_path: z.ZodString;
+    size_bytes: z.ZodNumber;
+    threshold_bytes: z.ZodNumber;
+    fingerprint: z.ZodString;
+    reason: z.ZodString;
+    quarantined_at: z.ZodString;
+}, "strip", z.ZodTypeAny, {
+    source: "codex_cli";
+    size_bytes: number;
+    quarantined_at: string;
+    reason: string;
+    fingerprint: string;
+    session_id: string;
+    session_path: string;
+    threshold_bytes: number;
+}, {
+    source: "codex_cli";
+    size_bytes: number;
+    quarantined_at: string;
+    reason: string;
+    fingerprint: string;
+    session_id: string;
+    session_path: string;
+    threshold_bytes: number;
+}>;
+export type QuarantineRecord = z.infer<typeof QuarantineRecordSchema> & {
+    /** Path to the quarantine record file itself. */
+    record_path: string;
+};
+declare const RecoveredEntrySchema: z.ZodObject<{
+    recovered_at: z.ZodString;
+    fingerprint: z.ZodString;
+    idempotency_key: z.ZodString;
+    chunks: z.ZodNumber;
+    session_path: z.ZodString;
+    size_bytes: z.ZodNumber;
+}, "strip", z.ZodTypeAny, {
+    idempotency_key: string;
+    size_bytes: number;
+    fingerprint: string;
+    session_path: string;
+    recovered_at: string;
+    chunks: number;
+}, {
+    idempotency_key: string;
+    size_bytes: number;
+    fingerprint: string;
+    session_path: string;
+    recovered_at: string;
+    chunks: number;
+}>;
+export type RecoveredEntry = z.infer<typeof RecoveredEntrySchema>;
+declare const RecoveryStateSchema: z.ZodObject<{
+    recovered: z.ZodDefault<z.ZodRecord<z.ZodString, z.ZodObject<{
+        recovered_at: z.ZodString;
+        fingerprint: z.ZodString;
+        idempotency_key: z.ZodString;
+        chunks: z.ZodNumber;
+        session_path: z.ZodString;
+        size_bytes: z.ZodNumber;
+    }, "strip", z.ZodTypeAny, {
+        idempotency_key: string;
+        size_bytes: number;
+        fingerprint: string;
+        session_path: string;
+        recovered_at: string;
+        chunks: number;
+    }, {
+        idempotency_key: string;
+        size_bytes: number;
+        fingerprint: string;
+        session_path: string;
+        recovered_at: string;
+        chunks: number;
+    }>>>;
+}, "strip", z.ZodTypeAny, {
+    recovered: Record<string, {
+        idempotency_key: string;
+        size_bytes: number;
+        fingerprint: string;
+        session_path: string;
+        recovered_at: string;
+        chunks: number;
+    }>;
+}, {
+    recovered?: Record<string, {
+        idempotency_key: string;
+        size_bytes: number;
+        fingerprint: string;
+        session_path: string;
+        recovered_at: string;
+        chunks: number;
+    }> | undefined;
+}>;
+export type RecoveryState = z.infer<typeof RecoveryStateSchema>;
+export type SkipReason = "already_recovered" | "missing_source" | "empty_session" | "summarize_failed" | "save_failed";
+export interface RecoveryResult {
+    session_id: string;
+    fingerprint: string;
+    outcome: "recovered" | "skipped" | "failed";
+    reason?: SkipReason;
+    chunks?: number;
+    idempotency_key?: string;
+    error?: string;
+}
+export interface RecoveryReport {
+    total_records: number;
+    total_sessions: number;
+    recovered: number;
+    skipped: number;
+    missing_source: number;
+    failed: number;
+    results: RecoveryResult[];
+}
+export interface ChunkSummary {
+    summary: string;
+    topics: string[];
+}
+export interface ChunkSummarizer {
+    summarize(opts: {
+        chunk: string;
+        index: number;
+        total: number;
+        sessionId: string;
+    }): Promise<ChunkSummary>;
+}
+export interface RecoverQuarantineSaveInput {
+    sessionId: string;
+    fingerprint: string;
+    sessionPath: string;
+    sizeBytes: number;
+    thresholdBytes: number;
+    chunks: ChunkSummary[];
+    content: string;
+    summary: string;
+    topics: string[];
+    idempotencyKey: string;
+    /**
+     * Set only when this session was already recovered under a *different*
+     * fingerprint (it grew since the last recovery). The save must replace
+     * the prior recovered memory rather than sit beside it, or search returns
+     * both the stale and the current recovery. Maps to `/save`'s
+     * `replace_idempotency_key`.
+     */
+    replaceIdempotencyKey?: string;
+}
+export interface RecoverQuarantineDeps {
+    dataDir: string;
+    /** Override quarantine directory (defaults to `<dataDir>/quarantine`). */
+    quarantineDir?: string;
+    /**
+     * Chunk byte budget. Clamped to at least `MIN_RECOVERY_CHUNK_BYTES`.
+     * Defaults to `DEFAULT_RECOVERY_CHUNK_BYTES`.
+     */
+    chunkBytes?: number;
+    summarizer: ChunkSummarizer;
+    /**
+     * Save the recovered artifact through the daemon. Implementations
+     * should be idempotent w.r.t. `idempotencyKey` so a partial failure
+     * mid-batch doesn't double-insert when the recovery state write
+     * happened after a successful save (it doesn't here — state is
+     * persisted on success only — but the contract still holds).
+     */
+    saveFn: (input: RecoverQuarantineSaveInput) => Promise<void>;
+    /**
+     * Dry-run: discover, parse, and chunk only. The Codex CLI summarizer
+     * is NOT invoked (each chunk call costs real time/quota on real data),
+     * and saving + state persistence are skipped. Use this to plan a real
+     * run cheaply; the report's `chunks` count reflects what a real run
+     * would summarize.
+     */
+    dryRun?: boolean;
+    /** Optional clock for deterministic tests. */
+    now?: () => Date;
+    /**
+     * Optional per-session gate. When it returns true for a grouped record,
+     * that session is skipped entirely — not attempted, not summarized, not
+     * added to the report. The background auto-repair caller uses this to
+     * exclude sessions whose bounded retry budget is exhausted, so a
+     * permanently-failing session cannot be retried forever. The manual
+     * `rift doctor repair capture-large-sessions` path passes no gate, so its
+     * behavior (including regrown-session replacement) is unchanged.
+     */
+    skipSession?: (record: QuarantineRecord) => boolean;
+}
+/**
+ * Read all Codex-CLI quarantine records under the configured directory.
+ * Malformed JSON / schema mismatches are skipped silently with a stderr
+ * note so one corrupt record can never block the rest of the batch.
+ */
+export declare function loadQuarantineRecords(deps: Pick<RecoverQuarantineDeps, "dataDir" | "quarantineDir">): QuarantineRecord[];
+/**
+ * Collapse multiple quarantine records that point at the same session
+ * (the same jsonl that hit the threshold at different sizes as it grew)
+ * down to one: the latest `quarantined_at`. The collapsed record still
+ * carries the recovery key, so reruns can mark every related record as
+ * already-recovered for that session.
+ */
+export declare function groupBySession(records: QuarantineRecord[]): QuarantineRecord[];
+export declare function loadRecoveryState(deps: Pick<RecoverQuarantineDeps, "dataDir">): RecoveryState;
+export declare function saveRecoveryState(deps: Pick<RecoverQuarantineDeps, "dataDir">, state: RecoveryState): Promise<void>;
+/**
+ * Count distinct oversized Codex sessions that are still parked — i.e. whose
+ * LATEST quarantine fingerprint has not yet been recovered. This is the
+ * honest "state of the world" the status surface should report, NOT the
+ * per-run `quarantined` count from the most recent capture run: that count
+ * flickers on/off as a stable oversized file dedups on one run and
+ * re-quarantines on the next, which teaches users to distrust the warning.
+ *
+ * Fingerprint-aware by design (mirrors the skip logic in
+ * `runRecoverQuarantine`): a session recovered at an earlier size is NOT
+ * counted as covered once the file grows to a new fingerprint, because the
+ * old recovery doesn't include the new turns. Counting by bare session id
+ * would under-report exactly those grown-since-recovery sessions.
+ *
+ * Sessions whose source jsonl no longer exists are excluded: recovery skips
+ * them as `missing_source` (it has nothing to read), so counting them would
+ * surface a "run recovery" advisory that can never clear — recovery cannot
+ * recover a file that's gone.
+ */
+export declare function listParkedOversizedSessions(deps: Pick<RecoverQuarantineDeps, "dataDir" | "quarantineDir">): QuarantineRecord[];
+export declare function countParkedOversizedSessions(deps: Pick<RecoverQuarantineDeps, "dataDir" | "quarantineDir">): number;
+/**
+ * Split parser-emitted Codex content into byte-budgeted chunks at turn
+ * boundaries. The parser format is `User: …\n\nAssistant: …\n\n…`, so
+ * we split on `\n\n` immediately preceding a `User:` or `Assistant:`
+ * marker. A single turn larger than `maxBytes` becomes its own chunk
+ * (oversized but preserved — better than dropping signal).
+ */
+export declare function chunkByTurns(content: string, maxBytes: number): string[];
+export declare function runRecoverQuarantine(deps: RecoverQuarantineDeps): Promise<RecoveryReport>;
+export declare class CodexCliChunkSummarizer implements ChunkSummarizer {
+    private readonly options;
+    constructor(options?: {
+        cwd?: string;
+        timeoutMs?: number;
+    });
+    summarize(opts: {
+        chunk: string;
+        index: number;
+        total: number;
+        sessionId: string;
+    }): Promise<ChunkSummary>;
+}
+export {};
+//# sourceMappingURL=recover-quarantine.d.ts.map

package/dist/src/capture/recover-quarantine.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"recover-quarantine.d.ts","sourceRoot":"","sources":["../../../src/capture/recover-quarantine.ts"],"names":[],"mappings":"AA8BA,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAUxB;;;;;GAKG;AACH,eAAO,MAAM,4BAA4B,QAAa,CAAC;AAEvD,4EAA4E;AAC5E,eAAO,MAAM,wBAAwB,QAAY,CAAC;AAElD,eAAO,MAAM,gBAAgB,mCAAmC,CAAC;AAEjE,QAAA,MAAM,sBAAsB;;;;;;;;;;;;;;;;;;;;;;;;;;;EAS1B,CAAC;AAiBH,MAAM,MAAM,gBAAgB,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,sBAAsB,CAAC,GAAG;IACtE,iDAAiD;IACjD,WAAW,EAAE,MAAM,CAAC;CACrB,CAAC;AAEF,QAAA,MAAM,oBAAoB;;;;;;;;;;;;;;;;;;;;;EAOxB,CAAC;AAEH,MAAM,MAAM,cAAc,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,oBAAoB,CAAC,CAAC;AAElE,QAAA,MAAM,mBAAmB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;EAEvB,CAAC;AAEH,MAAM,MAAM,aAAa,GAAG,CAAC,CAAC,KAAK,CAAC,OAAO,mBAAmB,CAAC,CAAC;AAEhE,MAAM,MAAM,UAAU,GAClB,mBAAmB,GACnB,gBAAgB,GAChB,eAAe,GACf,kBAAkB,GAClB,aAAa,CAAC;AAElB,MAAM,WAAW,cAAc;IAC7B,UAAU,EAAE,MAAM,CAAC;IACnB,WAAW,EAAE,MAAM,CAAC;IACpB,OAAO,EAAE,WAAW,GAAG,SAAS,GAAG,QAAQ,CAAC;IAC5C,MAAM,CAAC,EAAE,UAAU,CAAC;IACpB,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED,MAAM,WAAW,cAAc;IAC7B,aAAa,EAAE,MAAM,CAAC;IACtB,cAAc,EAAE,MAAM,CAAC;IACvB,SAAS,EAAE,MAAM,CAAC;IAClB,OAAO,EAAE,MAAM,CAAC;IAChB,cAAc,EAAE,MAAM,CAAC;IACvB,MAAM,EAAE,MAAM,CAAC;IACf,OAAO,EAAE,cAAc,EAAE,CAAC;CAC3B;AAED,MAAM,WAAW,YAAY;IAC3B,OAAO,EAAE,MAAM,CAAC;IAChB,MAAM,EAAE,MAAM,EAAE,CAAC;CAClB;AAED,MAAM,WAAW,eAAe;IAC9B,SAAS,CAAC,IAAI,EAAE;QACd,KAAK,EAAE,MAAM,CAAC;QACd,KAAK,EAAE,MAAM,CAAC;QACd,KAAK,EAAE,MAAM,CAAC;QACd,SAAS,EAAE,MAAM,CAAC;KACnB,GAAG,OAAO,CAAC,YAAY,CAAC,CAAC;CAC3B;AAED,MAAM,WAAW,0BAA0B;IACzC,SAAS,EAAE,MAAM,CAAC;IAClB,WAAW,EAAE,MAAM,CAAC;IACpB,WAAW,EAAE,MAAM,CAAC;IACpB,SAAS,EAAE,MAAM,CAAC;IAClB,cAAc,EAAE,MAAM,CAAC;IACvB,MAAM,EAAE,YAAY,EAAE,CAAC;IACvB,OAAO,EAAE,MAAM,CAAC;IAChB,OAAO,EAAE,MAAM,CAAC;IAChB,MAAM,EAAE,MAAM,EAAE,CAAC;IACjB,cAAc,EAAE,MAAM,CAAC;IACvB;;;;;;OAMG;IACH,qBAAqB,CAAC,EAAE,MAAM,CAAC;CAChC;AAED,MAAM,WAAW,qBAAqB;IACpC,OAAO,EAAE,MAAM,CAAC;IAChB,0EAA0E;IAC1E,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB;;;OAGG;IACH,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,UAAU,EAAE,eAAe,CAAC;IAC5B;;;;;;OAMG;IACH,MAAM,EAAE,CAAC,KAAK,EAAE,0BAA0B,KAAK,OAAO,CAAC,IAAI,CAAC,CAAC;IAC7D;;;;;;OAMG;IACH,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB,8CAA8C;IAC9C,GAAG,CAAC,EAAE,MAAM,IAAI,CAAC;IACjB;;;;;;;;OAQG;IACH,WAAW,CAAC,EAAE,CAAC,MAAM,EAAE,gBAAgB,KAAK,OAAO,CAAC;CACrD;AAWD;;;;GAIG;AACH,wBAAgB,qBAAqB,CACnC,IAAI,EAAE,IAAI,CAAC,qBAAqB,EAAE,SAAS,GAAG,eAAe,CAAC,GAC7D,gBAAgB,EAAE,CAgCpB;AAED;;;;;;GAMG;AACH,wBAAgB,cAAc,CAAC,OAAO,EAAE,gBAAgB,EAAE,GAAG,gBAAgB,EAAE,CAY9E;AAED,wBAAgB,iBAAiB,CAC/B,IAAI,EAAE,IAAI,CAAC,qBAAqB,EAAE,SAAS,CAAC,GAC3C,aAAa,CAef;AAED,wBAAsB,iBAAiB,CACrC,IAAI,EAAE,IAAI,CAAC,qBAAqB,EAAE,SAAS,CAAC,EAC5C,KAAK,EAAE,aAAa,GACnB,OAAO,CAAC,IAAI,CAAC,CAGf;AAED;;;;;;;;;;;;;;;;;;GAkBG;AACH,wBAAgB,2BAA2B,CACzC,IAAI,EAAE,IAAI,CAAC,qBAAqB,EAAE,SAAS,GAAG,eAAe,CAAC,GAC7D,gBAAgB,EAAE,CAapB;AAED,wBAAgB,4BAA4B,CAC1C,IAAI,EAAE,IAAI,CAAC,qBAAqB,EAAE,SAAS,GAAG,eAAe,CAAC,GAC7D,MAAM,CAER;AAED;;;;;;GAMG;AACH,wBAAgB,YAAY,CAAC,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,MAAM,EAAE,CAwCxE;AAqDD,wBAAsB,oBAAoB,CACxC,IAAI,EAAE,qBAAqB,GAC1B,OAAO,CAAC,cAAc,CAAC,CA4MzB;AAqBD,qBAAa,uBAAwB,YAAW,eAAe;IAE3D,OAAO,CAAC,QAAQ,CAAC,OAAO;gBAAP,OAAO,GAAE;QAAE,GAAG,CAAC,EAAE,MAAM,CAAC;QAAC,SAAS,CAAC,EAAE,MAAM,CAAA;KAAO;IAG/D,SAAS,CAAC,IAAI,EAAE;QACpB,KAAK,EAAE,MAAM,CAAC;QACd,KAAK,EAAE,MAAM,CAAC;QACd,KAAK,EAAE,MAAM,CAAC;QACd,SAAS,EAAE,MAAM,CAAC;KACnB,GAAG,OAAO,CAAC,YAAY,CAAC;CAsB1B"}