@bevel-software/platform-core-backend 0.32.0 → 0.33.0

package/dist/modules/workspace/file-readers/extract-in-worker.js

@@ -0,0 +1,140 @@
+import { Worker } from 'node:worker_threads';
+import { exchange, MAX_TIMER_MS, sizingOption, WorkerPool } from '../worker-pool.js';
+import { extractionTimeoutNotice, } from './document-extractor.contract.js';
+import { logger } from '../../../shared/logging.js';
+const log = logger('documents');
+/**
+ * Extraction time one document may spend in its worker: the platform's rule,
+ * not a deployment's (decided 2026-10-19). A longer wait is too unresponsive
+ * for a read or a search; the hour a timeout is remembered for is what keeps a
+ * hanging file cheap (see `DocExtractService`).
+ */
+export const EXTRACTION_BUDGET_MS = 10_000;
+/** What a document answers when its worker failed outside the parser: never the worker's own error. */
+export const EXTRACTION_WORKER_FAILED = 'could not be read: its extraction failed on the server; try again later';
+/**
+ * The one place a document's extractor runs (see {@link IDocumentExtractor}):
+ * a pooled worker thread — the extraction worker kind of the shared
+ * {@link WorkerPool}, which `grep`'s regex matcher is the other kind of —
+ * under a budget per file.
+ *
+ * The bytes are COPIED once on the request thread and the copy transferred,
+ * so the caller's buffer (already hashed for the cache key, and possibly a
+ * slice of Node's shared pool) is never detached. Past the budget the worker is
+ * terminated, which stops a parser mid-loop, and the file answers the timeout
+ * notice. A worker that answers with a dangling rejection, raises, or dies is
+ * never reused; the pool replaces it. A rejection the worker cannot pin on
+ * this job — it had served another before (see `serveExtractions`) — runs the
+ * job again in a fresh worker, so one file's failure is never another's.
+ */
+export class WorkerDocumentExtractor {
+    budgetMs;
+    pool;
+    nextJob = 1;
+    closing = false;
+    constructor(opts = {}) {
+        this.budgetMs = sizingOption('WorkerDocumentExtractor', 'budgetMs', opts.budgetMs ?? EXTRACTION_BUDGET_MS, 1, MAX_TIMER_MS);
+        const entry = opts.entry ?? defaultEntry();
+        this.pool = new WorkerPool(() => new Worker(entry.url, { execArgv: entry.execArgv }), {
+            // Three documents extracted at once each get a worker of their own.
+            poolMax: sizingOption('WorkerDocumentExtractor', 'poolMax', opts.poolMax ?? 4, 1),
+            maxWaiting: sizingOption('WorkerDocumentExtractor', 'maxWaiting', opts.maxWaiting ?? 32, 0),
+        }, () => new ExtractionBusyError());
+    }
+    async extract(extension, bytes) {
+        const first = await this.attempt(extension, bytes, this.budgetMs, false);
+        if (!('unattributed' in first))
+            return first;
+        // A rejection came while a worker that had served another job ran this one,
+        // and may be that job's: run it again where it is the first job, so what
+        // the file is answered (and remembered) as is its own. With the WHOLE
+        // budget again: the first run was cut short by another file's leftover,
+        // not by this file, and a run given only what was left would time out a
+        // sound document — and remember that timeout for an hour.
+        const again = await this.attempt(extension, bytes, this.budgetMs, true);
+        // In a fresh worker the worker never answers `unattributed`; a crash of the moment if it ever did —
+        // and never with that rejection's words, which may be another file's.
+        return 'unattributed' in again ? { ok: false, failure: 'crashed', message: EXTRACTION_WORKER_FAILED } : again;
+    }
+    async attempt(extension, bytes, budgetMs, fresh) {
+        let w;
+        try {
+            w = await (fresh ? this.pool.acquireFresh() : this.pool.acquire());
+        }
+        catch {
+            return busy();
+        }
+        const copy = new Uint8Array(bytes);
+        const job = { id: this.nextJob++, extension, bytes: copy };
+        const outcome = await exchange(w, job, budgetMs, [copy.buffer]);
+        if ('timeout' in outcome) {
+            this.pool.discard(w);
+            return { ok: false, failure: 'timeout', message: extractionTimeoutNotice(this.budgetMs) };
+        }
+        if ('error' in outcome) {
+            // Raised outside every guard the worker installs — its script failed to
+            // load, or it ran out of memory: a fault of the moment rather than of
+            // the bytes. The worker is ending, and the pool replaces it. What went
+            // wrong is the server's (a module path, a heap size): logged here, and
+            // the reader is told — and the hour remembers — the same words each time.
+            log.error('an extraction worker failed', { err: outcome.error });
+            return { ok: false, failure: 'crashed', message: EXTRACTION_WORKER_FAILED };
+        }
+        if ('exited' in outcome) {
+            if (this.closing)
+                return busy();
+            // Exit 0 is the worker ending itself over an earlier job's straggler that
+            // landed between jobs (see `serveExtractions`), after this job was posted
+            // but before it was taken up: it says nothing about this file. Run it
+            // again in a fresh worker, as for any rejection it cannot pin on this job.
+            if (outcome.exited === 0 && !fresh)
+                return { unattributed: true, message: EXTRACTION_WORKER_FAILED };
+            return {
+                ok: false,
+                failure: 'crashed',
+                message: `could not be read: its extraction stopped unexpectedly (exit code ${outcome.exited})`,
+            };
+        }
+        const { reply } = outcome;
+        if (reply.retire)
+            this.pool.discard(w);
+        else
+            this.pool.release(w);
+        if (reply.unattributed && !reply.ok)
+            return { unattributed: true, message: reply.message };
+        return reply.ok
+            ? { ok: true, summary: reply.summary, text: reply.text }
+            : { ok: false, failure: reply.transient ? 'crashed' : 'parser', message: reply.message };
+    }
+    close() {
+        this.closing = true;
+        return this.pool.close();
+    }
+}
+/** Every worker busy and the queue for one full. */
+class ExtractionBusyError extends Error {
+    constructor() {
+        super('every extraction worker is busy');
+        this.name = 'ExtractionBusyError';
+    }
+}
+function busy() {
+    return {
+        ok: false,
+        failure: 'busy',
+        message: 'could not be read: too many documents are being extracted at once; try again in a few seconds',
+    };
+}
+/**
+ * The document extractors' worker, beside this module. Run from the build it
+ * is plain JavaScript; run from source (the test suites) it is TypeScript,
+ * which a worker thread loads through `tsx`, as the suites themselves are.
+ */
+function defaultEntry() {
+    const fromSource = import.meta.url.endsWith('.ts');
+    return {
+        url: new URL(fromSource ? './extract.worker.ts' : './extract.worker.js', import.meta.url),
+        execArgv: fromSource ? ['--import', 'tsx'] : [],
+    };
+}
+//# sourceMappingURL=extract-in-worker.js.map

package/dist/modules/workspace/file-readers/extract-in-worker.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"extract-in-worker.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-in-worker.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,EAAE,MAAM,qBAAqB,CAAC;AAC7C,OAAO,EAAE,QAAQ,EAAE,YAAY,EAAE,YAAY,EAAE,UAAU,EAAE,MAAM,mBAAmB,CAAC;AACrF,OAAO,EACL,uBAAuB,GAGxB,MAAM,kCAAkC,CAAC;AAE1C,OAAO,EAAE,MAAM,EAAE,MAAM,4BAA4B,CAAC;AAEpD,MAAM,GAAG,GAAG,MAAM,CAAC,WAAW,CAAC,CAAC;AAEhC;;;;;GAKG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAG,MAAM,CAAC;AAE3C,uGAAuG;AACvG,MAAM,CAAC,MAAM,wBAAwB,GAAG,yEAAyE,CAAC;AA4BlH;;;;;;;;;;;;;;GAcG;AACH,MAAM,OAAO,uBAAuB;IACjB,QAAQ,CAAS;IACjB,IAAI,CAAa;IAC1B,OAAO,GAAG,CAAC,CAAC;IACZ,OAAO,GAAG,KAAK,CAAC;IAExB,YAAY,OAAuC,EAAE;QACnD,IAAI,CAAC,QAAQ,GAAG,YAAY,CAAC,yBAAyB,EAAE,UAAU,EAAE,IAAI,CAAC,QAAQ,IAAI,oBAAoB,EAAE,CAAC,EAAE,YAAY,CAAC,CAAC;QAC5H,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,IAAI,YAAY,EAAE,CAAC;QAC3C,IAAI,CAAC,IAAI,GAAG,IAAI,UAAU,CACxB,GAAG,EAAE,CAAC,IAAI,MAAM,CAAC,KAAK,CAAC,GAAG,EAAE,EAAE,QAAQ,EAAE,KAAK,CAAC,QAAQ,EAAE,CAAC,EACzD;YACE,oEAAoE;YACpE,OAAO,EAAE,YAAY,CAAC,yBAAyB,EAAE,SAAS,EAAE,IAAI,CAAC,OAAO,IAAI,CAAC,EAAE,CAAC,CAAC;YACjF,UAAU,EAAE,YAAY,CAAC,yBAAyB,EAAE,YAAY,EAAE,IAAI,CAAC,UAAU,IAAI,EAAE,EAAE,CAAC,CAAC;SAC5F,EACD,GAAG,EAAE,CAAC,IAAI,mBAAmB,EAAE,CAChC,CAAC;IACJ,CAAC;IAED,KAAK,CAAC,OAAO,CAAC,SAAiB,EAAE,KAAa;QAC5C,MAAM,KAAK,GAAG,MAAM,IAAI,CAAC,OAAO,CAAC,SAAS,EAAE,KAAK,EAAE,IAAI,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC;QACzE,IAAI,CAAC,CAAC,cAAc,IAAI,KAAK,CAAC;YAAE,OAAO,KAAK,CAAC;QAC7C,4EAA4E;QAC5E,yEAAyE;QACzE,sEAAsE;QACtE,wEAAwE;QACxE,wEAAwE;QACxE,0DAA0D;QAC1D,MAAM,KAAK,GAAG,MAAM,IAAI,CAAC,OAAO,CAAC,SAAS,EAAE,KAAK,EAAE,IAAI,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC;QACxE,oGAAoG;QACpG,sEAAsE;QACtE,OAAO,cAAc,IAAI,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,SAAS,EAAE,OAAO,EAAE,wBAAwB,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC;IAChH,CAAC;IAEO,KAAK,CAAC,OAAO,CACnB,SAAiB,EACjB,KAAa,EACb,QAAgB,EAChB,KAAc;QAEd,IAAI,CAAS,CAAC;QACd,IAAI,CAAC;YACH,CAAC,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,YAAY,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC,CAAC;QACrE,CAAC;QAAC,MAAM,CAAC;YACP,OAAO,IAAI,EAAE,CAAC;QAChB,CAAC;QACD,MAAM,IAAI,GAAG,IAAI,UAAU,CAAC,KAAK,CAAC,CAAC;QACnC,MAAM,GAAG,GAAkB,EAAE,EAAE,EAAE,IAAI,CAAC,OAAO,EAAE,EAAE,SAAS,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC;QAC1E,MAAM,OAAO,GAAG,MAAM,QAAQ,CAAkB,CAAC,EAAE,GAAG,EAAE,QAAQ,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC;QACjF,IAAI,SAAS,IAAI,OAAO,EAAE,CAAC;YACzB,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;YACrB,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,SAAS,EAAE,OAAO,EAAE,uBAAuB,CAAC,IAAI,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC5F,CAAC;QACD,IAAI,OAAO,IAAI,OAAO,EAAE,CAAC;YACvB,wEAAwE;YACxE,sEAAsE;YACtE,uEAAuE;YACvE,uEAAuE;YACvE,0EAA0E;YAC1E,GAAG,CAAC,KAAK,CAAC,6BAA6B,EAAE,EAAE,GAAG,EAAE,OAAO,CAAC,KAAK,EAAE,CAAC,CAAC;YACjE,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,SAAS,EAAE,OAAO,EAAE,wBAAwB,EAAE,CAAC;QAC9E,CAAC;QACD,IAAI,QAAQ,IAAI,OAAO,EAAE,CAAC;YACxB,IAAI,IAAI,CAAC,OAAO;gBAAE,OAAO,IAAI,EAAE,CAAC;YAChC,0EAA0E;YAC1E,0EAA0E;YAC1E,sEAAsE;YACtE,2EAA2E;YAC3E,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC,IAAI,CAAC,KAAK;gBAAE,OAAO,EAAE,YAAY,EAAE,IAAI,EAAE,OAAO,EAAE,wBAAwB,EAAE,CAAC;YACrG,OAAO;gBACL,EAAE,EAAE,KAAK;gBACT,OAAO,EAAE,SAAS;gBAClB,OAAO,EAAE,qEAAqE,OAAO,CAAC,MAAM,GAAG;aAChG,CAAC;QACJ,CAAC;QACD,MAAM,EAAE,KAAK,EAAE,GAAG,OAAO,CAAC;QAC1B,IAAI,KAAK,CAAC,MAAM;YAAE,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;;YAClC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;QAC1B,IAAI,KAAK,CAAC,YAAY,IAAI,CAAC,KAAK,CAAC,EAAE;YAAE,OAAO,EAAE,YAAY,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,CAAC;QAC3F,OAAO,KAAK,CAAC,EAAE;YACb,CAAC,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,EAAE;YACxD,CAAC,CAAC,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,QAAQ,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,CAAC;IAC7F,CAAC;IAED,KAAK;QACH,IAAI,CAAC,OAAO,GAAG,IAAI,CAAC;QACpB,OAAO,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE,CAAC;IAC3B,CAAC;CACF;AAED,oDAAoD;AACpD,MAAM,mBAAoB,SAAQ,KAAK;IACrC;QACE,KAAK,CAAC,iCAAiC,CAAC,CAAC;QACzC,IAAI,CAAC,IAAI,GAAG,qBAAqB,CAAC;IACpC,CAAC;CACF;AAED,SAAS,IAAI;IACX,OAAO;QACL,EAAE,EAAE,KAAK;QACT,OAAO,EAAE,MAAM;QACf,OAAO,EAAE,+FAA+F;KACzG,CAAC;AACJ,CAAC;AAED;;;;GAIG;AACH,SAAS,YAAY;IACnB,MAAM,UAAU,GAAG,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;IACnD,OAAO;QACL,GAAG,EAAE,IAAI,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,qBAAqB,CAAC,CAAC,CAAC,qBAAqB,EAAE,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC;QACzF,QAAQ,EAAE,UAAU,CAAC,CAAC,CAAC,CAAC,UAAU,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE;KAChD,CAAC;AACJ,CAAC"}

package/dist/modules/workspace/file-readers/extract-pdf.d.ts

@@ -1,3 +1,5 @@
 import type { ExtractResult } from './doc-extract.types.js';
 export declare function extractPdf(bytes: Buffer): Promise<ExtractResult>;
+/** What a PDF read answers when pdf.js itself failed to load: the same words, whatever the cause. */
+export declare const READER_DID_NOT_LOAD = "could not be read: the PDF reader did not load on the server; try again later";
 //# sourceMappingURL=extract-pdf.d.ts.map

package/dist/modules/workspace/file-readers/extract-pdf.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"extract-pdf.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-pdf.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,wBAAwB,CAAC;AAwD5D,wBAAsB,UAAU,CAAC,KAAK,EAAE,MAAM,GAAG,OAAO,CAAC,aAAa,CAAC,CA+GtE"}
+{"version":3,"file":"extract-pdf.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-pdf.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,wBAAwB,CAAC;AA2D5D,wBAAsB,UAAU,CAAC,KAAK,EAAE,MAAM,GAAG,OAAO,CAAC,aAAa,CAAC,CAyHtE;AAED,qGAAqG;AACrG,eAAO,MAAM,mBAAmB,kFAAkF,CAAC"}

package/dist/modules/workspace/file-readers/extract-pdf.js

@@ -1,4 +1,6 @@
 import { MAX_DOC_PART_BYTES } from './ooxml-text.js';
+import { logger } from '../../../shared/logging.js';
+const log = logger('documents');
 /**
  * Extract a PDF's TEXT LAYER page by page with pdf.js (`pdfjs-dist`,
  * Mozilla's maintained renderer — chosen over the unmaintained thin wrappers
@@ -61,9 +63,20 @@ export async function extractPdf(bytes) {
     // The loading TASK is what gets destroyed at the end, not the document:
     // pdf.js 6 dropped `PDFDocumentProxy.destroy()`, and destroying the task
     // frees the document (and its pages) with it — the shape the viewer uses.
+    let getDocument;
+    try {
+        ({ getDocument } = await loadPdfJs());
+    }
+    catch (err) {
+        // The reader, not the file: a later read must try this file again. Why
+        // it did not load is the server's to know (a module path, a missing
+        // dependency) — logged here, never quoted to the reader.
+        log.error('the PDF reader did not load', { err });
+        return { ok: false, transient: true, message: READER_DID_NOT_LOAD };
+    }
     let task;
     try {
-        task = await openPdf(bytes);
+        task = openPdf(getDocument, bytes);
     }
     catch (err) {
         return { ok: false, message: `could not be parsed as a PDF (${err.message})` };
@@ -167,8 +180,10 @@ export async function extractPdf(bytes) {
         await task.destroy();
     }
 }
+/** What a PDF read answers when pdf.js itself failed to load: the same words, whatever the cause. */
+export const READER_DID_NOT_LOAD = 'could not be read: the PDF reader did not load on the server; try again later';
 let pdfjsPromise;
-async function openPdf(bytes) {
+function loadPdfJs() {
     pdfjsPromise ??= import('pdfjs-dist/legacy/build/pdf.mjs').catch((err) => {
         // A FAILED load must not be memoized: left in place, the rejected promise
         // would answer every later read and disable PDF extraction for the whole
@@ -176,7 +191,9 @@ async function openPdf(bytes) {
         pdfjsPromise = undefined;
         throw err;
     });
-    const { getDocument } = await pdfjsPromise;
+    return pdfjsPromise;
+}
+function openPdf(getDocument, bytes) {
     // Returns the loading task; the caller awaits `.promise` for the document
     // and destroys the task when done (see `extractPdf`).
     return getDocument({

package/dist/modules/workspace/file-readers/extract-pdf.js.map

@@ -1 +1 @@
-{"version":3,"file":"extract-pdf.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-pdf.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,kBAAkB,EAAE,MAAM,iBAAiB,CAAC;AAErD;;;;;;;;;;;GAWG;AACH;;;;;;;;GAQG;AACH,MAAM,kBAAkB,GAAG,EAAE,GAAG,IAAI,GAAG,IAAI,CAAC,CAAC,8BAA8B;AAE3E;;;;;;;;;GASG;AACH,MAAM,aAAa,GAAG,MAAM,CAAC;AAE7B;;;;;;;GAOG;AACH,MAAM,sBAAsB,GAAG,OAAO,CAAC;AAEvC,yDAAyD;AACzD,SAAS,UAAU;IACjB,OAAO;QACL,EAAE,EAAE,KAAK;QACT,OAAO,EAAE,6DAA6D,kBAAkB,0CAA0C;KACnI,CAAC;AACJ,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,UAAU,CAAC,KAAa;IAC5C,yEAAyE;IACzE,0EAA0E;IAC1E,0DAA0D;IAC1D,IAAI,KAAK,CAAC,MAAM,GAAG,kBAAkB,EAAE,CAAC;QACtC,OAAO;YACL,EAAE,EAAE,KAAK;YACT,OAAO,EAAE,gDAAgD,KAAK,CAAC,MAAM,qBAAqB,kBAAkB,iCAAiC;SAC9I,CAAC;IACJ,CAAC;IACD,wEAAwE;IACxE,yEAAyE;IACzE,0EAA0E;IAC1E,IAAI,IAAyC,CAAC;IAC9C,IAAI,CAAC;QACH,IAAI,GAAG,MAAM,OAAO,CAAC,KAAK,CAAC,CAAC;IAC9B,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,iCAAkC,GAAa,CAAC,OAAO,GAAG,EAAE,CAAC;IAC5F,CAAC;IACD,IAAI,GAAiC,CAAC;IACtC,IAAI,CAAC;QACH,GAAG,GAAG,MAAM,IAAI,CAAC,OAAO,CAAC;IAC3B,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,0EAA0E;QAC1E,yEAAyE;QACzE,MAAM,IAAI,CAAC,OAAO,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;QAC5C,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,iCAAkC,GAAa,CAAC,OAAO,GAAG,EAAE,CAAC;IAC5F,CAAC;IACD,IAAI,CAAC;QACH,IAAI,GAAG,CAAC,QAAQ,GAAG,aAAa,EAAE,CAAC;YACjC,OAAO;gBACL,EAAE,EAAE,KAAK;gBACT,OAAO,EAAE,gDAAgD,GAAG,CAAC,QAAQ,qBAAqB,aAAa,yBAAyB;aACjI,CAAC;QACJ,CAAC;QACD,MAAM,KAAK,GAAa,EAAE,CAAC;QAC3B,IAAI,SAAS,GAAG,CAAC,CAAC;QAClB,IAAI,OAAO,GAAG,KAAK,CAAC;QACpB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,GAAG,CAAC,QAAQ,EAAE,CAAC,EAAE,EAAE,CAAC;YACvC,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC;YAC1B,yEAAyE;YACzE,0EAA0E;YAC1E,SAAS,IAAI,CAAC,CAAC,QAAQ,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC;YACrC,IAAI,SAAS,GAAG,kBAAkB;gBAAE,OAAO,UAAU,EAAE,CAAC;YACxD,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;YAClC,sEAAsE;YACtE,qEAAqE;YACrE,sEAAsE;YACtE,sEAAsE;YACtE,4DAA4D;YAC5D,MAAM,MAAM,GACV,IAAI,CAAC,iBAAiB,EACvB,CAAC,SAAS,EAAE,CAAC;YACd,IAAI,SAAS,GAAG,CAAC,CAAC;YAClB,IAAI,IAAI,GAAG,EAAE,CAAC;YACd,IAAI,KAAyB,CAAC;YAC9B,MAAM,KAAK,GAAG,GAAS,EAAE;gBACvB,IAAI,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC;oBACvB,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;oBACjB,OAAO,GAAG,IAAI,CAAC;oBACf,oEAAoE;oBACpE,SAAS,IAAI,CAAC,CAAC;gBACjB,CAAC;gBACD,IAAI,GAAG,EAAE,CAAC;YACZ,CAAC,CAAC;YACF,SAAS,CAAC;gBACR,MAAM,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;gBACnD,IAAI,IAAI;oBAAE,MAAM;gBAChB,SAAS,IAAI,KAAK,CAAC,KAAK,CAAC,MAAM,CAAC;gBAChC,IAAI,SAAS,GAAG,sBAAsB,EAAE,CAAC;oBACvC,MAAM,MAAM,CAAC,MAAM,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;oBAC7C,IAAI,CAAC,OAAO,EAAE,CAAC;oBACf,OAAO;wBACL,EAAE,EAAE,KAAK;wBACT,OAAO,EAAE,yCAAyC,CAAC,oBAAoB,sBAAsB,0CAA0C;qBACxI,CAAC;gBACJ,CAAC;gBACD,KAAK,MAAM,IAAI,IAAI,KAAK,CAAC,KAAK,EAAE,CAAC;oBAC/B,IAAI,CAAC,CAAC,KAAK,IAAI,IAAI,CAAC;wBAAE,SAAS,CAAC,gCAAgC;oBAChE,MAAM,CAAC,GAAG,IAAI,CAAC,SAAS,EAAE,CAAC,CAAC,CAAC,CAAC;oBAC9B,wEAAwE;oBACxE,IAAI,OAAO,CAAC,KAAK,QAAQ,EAAE,CAAC;wBAC1B,IAAI,KAAK,KAAK,SAAS,IAAI,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,KAAK,CAAC,GAAG,CAAC;4BAAE,KAAK,EAAE,CAAC;wBAC5D,KAAK,GAAG,CAAC,CAAC;oBACZ,CAAC;oBACD,IAAI,IAAI,CAAC,GAAG,KAAK,EAAE,EAAE,CAAC;wBACpB,iEAAiE;wBACjE,gEAAgE;wBAChE,gEAAgE;wBAChE,SAAS,IAAI,IAAI,CAAC,GAAG,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;wBACrD,IAAI,SAAS,GAAG,kBAAkB,EAAE,CAAC;4BACnC,MAAM,MAAM,CAAC,MAAM,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;4BAC7C,OAAO,UAAU,EAAE,CAAC;wBACtB,CAAC;wBACD,IAAI,IAAI,CAAC,IAAI,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC;oBAC9C,CAAC;oBACD,IAAI,IAAI,CAAC,MAAM;wBAAE,KAAK,EAAE,CAAC;gBAC3B,CAAC;YACH,CAAC;YACD,KAAK,EAAE,CAAC;YACR,IAAI,CAAC,OAAO,EAAE,CAAC;QACjB,CAAC;QACD,MAAM,KAAK,GAAG,GAAG,GAAG,CAAC,QAAQ,QAAQ,GAAG,CAAC,QAAQ,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC;QACrE,OAAO,OAAO;YACZ,CAAC,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,OAAO,EAAE,GAAG,KAAK,yCAAyC,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE;YAClG,CAAC,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,OAAO,EAAE,GAAG,KAAK,qCAAqC,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;IACnG,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,qCAAsC,GAAa,CAAC,OAAO,GAAG,EAAE,CAAC;IAChG,CAAC;YAAS,CAAC;QACT,MAAM,IAAI,CAAC,OAAO,EAAE,CAAC;IACvB,CAAC;AACH,CAAC;AAGD,IAAI,YAAwC,CAAC;AAE7C,KAAK,UAAU,OAAO,CAAC,KAAa;IAClC,YAAY,KAAK,MAAM,CAAC,iCAAiC,CAAC,CAAC,KAAK,CAAC,CAAC,GAAY,EAAE,EAAE;QAChF,0EAA0E;QAC1E,yEAAyE;QACzE,sDAAsD;QACtD,YAAY,GAAG,SAAS,CAAC;QACzB,MAAM,GAAG,CAAC;IACZ,CAAC,CAAC,CAAC;IACH,MAAM,EAAE,WAAW,EAAE,GAAG,MAAM,YAAY,CAAC;IAC3C,0EAA0E;IAC1E,sDAAsD;IACtD,OAAO,WAAW,CAAC;QACjB,wEAAwE;QACxE,wEAAwE;QACxE,IAAI,EAAE,IAAI,UAAU,CAAC,KAAK,CAAC;QAC3B,mEAAmE;QACnE,eAAe,EAAE,IAAI;QACrB,cAAc,EAAE,IAAI;KACrB,CAAC,CAAC;AACL,CAAC"}
+{"version":3,"file":"extract-pdf.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract-pdf.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,kBAAkB,EAAE,MAAM,iBAAiB,CAAC;AACrD,OAAO,EAAE,MAAM,EAAE,MAAM,4BAA4B,CAAC;AAEpD,MAAM,GAAG,GAAG,MAAM,CAAC,WAAW,CAAC,CAAC;AAEhC;;;;;;;;;;;GAWG;AACH;;;;;;;;GAQG;AACH,MAAM,kBAAkB,GAAG,EAAE,GAAG,IAAI,GAAG,IAAI,CAAC,CAAC,8BAA8B;AAE3E;;;;;;;;;GASG;AACH,MAAM,aAAa,GAAG,MAAM,CAAC;AAE7B;;;;;;;GAOG;AACH,MAAM,sBAAsB,GAAG,OAAO,CAAC;AAEvC,yDAAyD;AACzD,SAAS,UAAU;IACjB,OAAO;QACL,EAAE,EAAE,KAAK;QACT,OAAO,EAAE,6DAA6D,kBAAkB,0CAA0C;KACnI,CAAC;AACJ,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,UAAU,CAAC,KAAa;IAC5C,yEAAyE;IACzE,0EAA0E;IAC1E,0DAA0D;IAC1D,IAAI,KAAK,CAAC,MAAM,GAAG,kBAAkB,EAAE,CAAC;QACtC,OAAO;YACL,EAAE,EAAE,KAAK;YACT,OAAO,EAAE,gDAAgD,KAAK,CAAC,MAAM,qBAAqB,kBAAkB,iCAAiC;SAC9I,CAAC;IACJ,CAAC;IACD,wEAAwE;IACxE,yEAAyE;IACzE,0EAA0E;IAC1E,IAAI,WAAiC,CAAC;IACtC,IAAI,CAAC;QACH,CAAC,EAAE,WAAW,EAAE,GAAG,MAAM,SAAS,EAAE,CAAC,CAAC;IACxC,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,uEAAuE;QACvE,oEAAoE;QACpE,yDAAyD;QACzD,GAAG,CAAC,KAAK,CAAC,6BAA6B,EAAE,EAAE,GAAG,EAAE,CAAC,CAAC;QAClD,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,EAAE,OAAO,EAAE,mBAAmB,EAAE,CAAC;IACtE,CAAC;IACD,IAAI,IAAgC,CAAC;IACrC,IAAI,CAAC;QACH,IAAI,GAAG,OAAO,CAAC,WAAW,EAAE,KAAK,CAAC,CAAC;IACrC,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,iCAAkC,GAAa,CAAC,OAAO,GAAG,EAAE,CAAC;IAC5F,CAAC;IACD,IAAI,GAAiC,CAAC;IACtC,IAAI,CAAC;QACH,GAAG,GAAG,MAAM,IAAI,CAAC,OAAO,CAAC;IAC3B,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,0EAA0E;QAC1E,yEAAyE;QACzE,MAAM,IAAI,CAAC,OAAO,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;QAC5C,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,iCAAkC,GAAa,CAAC,OAAO,GAAG,EAAE,CAAC;IAC5F,CAAC;IACD,IAAI,CAAC;QACH,IAAI,GAAG,CAAC,QAAQ,GAAG,aAAa,EAAE,CAAC;YACjC,OAAO;gBACL,EAAE,EAAE,KAAK;gBACT,OAAO,EAAE,gDAAgD,GAAG,CAAC,QAAQ,qBAAqB,aAAa,yBAAyB;aACjI,CAAC;QACJ,CAAC;QACD,MAAM,KAAK,GAAa,EAAE,CAAC;QAC3B,IAAI,SAAS,GAAG,CAAC,CAAC;QAClB,IAAI,OAAO,GAAG,KAAK,CAAC;QACpB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,GAAG,CAAC,QAAQ,EAAE,CAAC,EAAE,EAAE,CAAC;YACvC,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC;YAC1B,yEAAyE;YACzE,0EAA0E;YAC1E,SAAS,IAAI,CAAC,CAAC,QAAQ,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC;YACrC,IAAI,SAAS,GAAG,kBAAkB;gBAAE,OAAO,UAAU,EAAE,CAAC;YACxD,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC;YAClC,sEAAsE;YACtE,qEAAqE;YACrE,sEAAsE;YACtE,sEAAsE;YACtE,4DAA4D;YAC5D,MAAM,MAAM,GACV,IAAI,CAAC,iBAAiB,EACvB,CAAC,SAAS,EAAE,CAAC;YACd,IAAI,SAAS,GAAG,CAAC,CAAC;YAClB,IAAI,IAAI,GAAG,EAAE,CAAC;YACd,IAAI,KAAyB,CAAC;YAC9B,MAAM,KAAK,GAAG,GAAS,EAAE;gBACvB,IAAI,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC;oBACvB,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;oBACjB,OAAO,GAAG,IAAI,CAAC;oBACf,oEAAoE;oBACpE,SAAS,IAAI,CAAC,CAAC;gBACjB,CAAC;gBACD,IAAI,GAAG,EAAE,CAAC;YACZ,CAAC,CAAC;YACF,SAAS,CAAC;gBACR,MAAM,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;gBACnD,IAAI,IAAI;oBAAE,MAAM;gBAChB,SAAS,IAAI,KAAK,CAAC,KAAK,CAAC,MAAM,CAAC;gBAChC,IAAI,SAAS,GAAG,sBAAsB,EAAE,CAAC;oBACvC,MAAM,MAAM,CAAC,MAAM,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;oBAC7C,IAAI,CAAC,OAAO,EAAE,CAAC;oBACf,OAAO;wBACL,EAAE,EAAE,KAAK;wBACT,OAAO,EAAE,yCAAyC,CAAC,oBAAoB,sBAAsB,0CAA0C;qBACxI,CAAC;gBACJ,CAAC;gBACD,KAAK,MAAM,IAAI,IAAI,KAAK,CAAC,KAAK,EAAE,CAAC;oBAC/B,IAAI,CAAC,CAAC,KAAK,IAAI,IAAI,CAAC;wBAAE,SAAS,CAAC,gCAAgC;oBAChE,MAAM,CAAC,GAAG,IAAI,CAAC,SAAS,EAAE,CAAC,CAAC,CAAC,CAAC;oBAC9B,wEAAwE;oBACxE,IAAI,OAAO,CAAC,KAAK,QAAQ,EAAE,CAAC;wBAC1B,IAAI,KAAK,KAAK,SAAS,IAAI,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,KAAK,CAAC,GAAG,CAAC;4BAAE,KAAK,EAAE,CAAC;wBAC5D,KAAK,GAAG,CAAC,CAAC;oBACZ,CAAC;oBACD,IAAI,IAAI,CAAC,GAAG,KAAK,EAAE,EAAE,CAAC;wBACpB,iEAAiE;wBACjE,gEAAgE;wBAChE,gEAAgE;wBAChE,SAAS,IAAI,IAAI,CAAC,GAAG,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;wBACrD,IAAI,SAAS,GAAG,kBAAkB,EAAE,CAAC;4BACnC,MAAM,MAAM,CAAC,MAAM,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;4BAC7C,OAAO,UAAU,EAAE,CAAC;wBACtB,CAAC;wBACD,IAAI,IAAI,CAAC,IAAI,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC;oBAC9C,CAAC;oBACD,IAAI,IAAI,CAAC,MAAM;wBAAE,KAAK,EAAE,CAAC;gBAC3B,CAAC;YACH,CAAC;YACD,KAAK,EAAE,CAAC;YACR,IAAI,CAAC,OAAO,EAAE,CAAC;QACjB,CAAC;QACD,MAAM,KAAK,GAAG,GAAG,GAAG,CAAC,QAAQ,QAAQ,GAAG,CAAC,QAAQ,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC;QACrE,OAAO,OAAO;YACZ,CAAC,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,OAAO,EAAE,GAAG,KAAK,yCAAyC,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE;YAClG,CAAC,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,OAAO,EAAE,GAAG,KAAK,qCAAqC,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;IACnG,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,qCAAsC,GAAa,CAAC,OAAO,GAAG,EAAE,CAAC;IAChG,CAAC;YAAS,CAAC;QACT,MAAM,IAAI,CAAC,OAAO,EAAE,CAAC;IACvB,CAAC;AACH,CAAC;AAED,qGAAqG;AACrG,MAAM,CAAC,MAAM,mBAAmB,GAAG,+EAA+E,CAAC;AAGnH,IAAI,YAAwC,CAAC;AAE7C,SAAS,SAAS;IAChB,YAAY,KAAK,MAAM,CAAC,iCAAiC,CAAC,CAAC,KAAK,CAAC,CAAC,GAAY,EAAE,EAAE;QAChF,0EAA0E;QAC1E,yEAAyE;QACzE,sDAAsD;QACtD,YAAY,GAAG,SAAS,CAAC;QACzB,MAAM,GAAG,CAAC;IACZ,CAAC,CAAC,CAAC;IACH,OAAO,YAAY,CAAC;AACtB,CAAC;AAED,SAAS,OAAO,CAAC,WAAiC,EAAE,KAAa;IAC/D,0EAA0E;IAC1E,sDAAsD;IACtD,OAAO,WAAW,CAAC;QACjB,wEAAwE;QACxE,wEAAwE;QACxE,IAAI,EAAE,IAAI,UAAU,CAAC,KAAK,CAAC;QAC3B,mEAAmE;QACnE,eAAe,EAAE,IAAI;QACrB,cAAc,EAAE,IAAI;KACrB,CAAC,CAAC;AACL,CAAC"}

package/dist/modules/workspace/file-readers/extract.worker.d.ts

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=extract.worker.d.ts.map

package/dist/modules/workspace/file-readers/extract.worker.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"extract.worker.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract.worker.ts"],"names":[],"mappings":""}

package/dist/modules/workspace/file-readers/extract.worker.js

@@ -0,0 +1,5 @@
+import { DOCUMENT_EXTRACTORS } from './document-extractors.js';
+import { serveExtractions } from './extraction-worker.js';
+/** The extraction worker's entry: the document extractors, served one job at a time (see `extract-in-worker.ts`). */
+serveExtractions(DOCUMENT_EXTRACTORS);
+//# sourceMappingURL=extract.worker.js.map

package/dist/modules/workspace/file-readers/extract.worker.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"extract.worker.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extract.worker.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,mBAAmB,EAAE,MAAM,0BAA0B,CAAC;AAC/D,OAAO,EAAE,gBAAgB,EAAE,MAAM,wBAAwB,CAAC;AAE1D,qHAAqH;AACrH,gBAAgB,CAAC,mBAAmB,CAAC,CAAC"}

package/dist/modules/workspace/file-readers/extraction-cache.d.ts

@@ -8,13 +8,26 @@ import type { ExtractedDoc } from './doc-extract.types.js';
  * when the file is untracked or dirty. One code path, no fallback branch.
  */
 export declare function gitBlobSha(bytes: Buffer): string;
+/**
+ * What the cache holds for a key: the extraction, or the notice a failed one
+ * answered with. A failure with `until` (epoch ms) is remembered only until
+ * then — a lookup past it is a miss; one without is remembered like a success.
+ */
+export type CachedExtraction = ({
+    failed?: undefined;
+} & ExtractedDoc) | {
+    failed: true;
+    message: string;
+    until?: number;
+};
 /**
  * On-disk cache of document extractions. The KEY is the caller's business —
  * `DocExtractService` passes git blob sha + extension (content hash, so a
  * re-upload of identical bytes, or the same document on another branch or
  * path, hits the same entry; extension, so identical bytes under another
  * FORMAT never return the wrong extractor's output). One JSON file per entry
- * (`<key>.json` holding the `{ summary, text }`), under a dedicated root
+ * (`<key>.json` holding the `{ summary, text }`, or a failure's
+ * `{ failed, message, until? }`), under a dedicated root
  * that — like the spill
  * store's — sits BESIDE the workspaces root, never inside a workspace and
  * never committed.
@@ -31,8 +44,11 @@ export declare class DocExtractionCache {
     private readonly root;
     private readonly maxTotalBytes;
     constructor(root: string, maxTotalBytes?: number);
-    /** The cached extraction for `sha`, or undefined on miss/corrupt entry. */
-    get(sha: string): Promise<ExtractedDoc | undefined>;
+    /**
+     * The cached entry for `sha`, or undefined on miss/corrupt entry — and for a
+     * failure remembered only until a moment `now` (epoch ms) has passed.
+     */
+    get(sha: string, now: number): Promise<CachedExtraction | undefined>;
     /**
      * Bytes written since the last full scan. The scan costs a `readdir` plus a
      * `stat` per entry, and running it on EVERY cold extraction made a cache that
@@ -44,8 +60,8 @@ export declare class DocExtractionCache {
     private totalAtLastScan;
     /** The prune in flight, so concurrent puts share ONE scan instead of racing their own. */
     private pruning;
-    /** Store an extraction under `sha`; prunes towards the size bound. Never throws. */
-    put(sha: string, doc: ExtractedDoc): Promise<void>;
+    /** Store an extraction (or a failure) under `sha`; prunes towards the size bound. Never throws. */
+    put(sha: string, entry: CachedExtraction): Promise<void>;
     /**
      * The file backing `key`, always INSIDE the cache root.
      *

package/dist/modules/workspace/file-readers/extraction-cache.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"extraction-cache.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extraction-cache.ts"],"names":[],"mappings":"AAGA,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,wBAAwB,CAAC;AAE3D;;;;;;;GAOG;AACH,wBAAgB,UAAU,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAEhD;AAKD;;;;;;;;;;;;;;;;;;GAkBG;AACH,qBAAa,kBAAkB;IAE3B,OAAO,CAAC,QAAQ,CAAC,IAAI;IACrB,OAAO,CAAC,QAAQ,CAAC,aAAa;gBADb,IAAI,EAAE,MAAM,EACZ,aAAa,GAAE,MAAgC;IAGlE,2EAA2E;IACrE,GAAG,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAAC,YAAY,GAAG,SAAS,CAAC;IAgBzD;;;;;OAKG;IACH,OAAO,CAAC,gBAAgB,CAAK;IAE7B,kHAAkH;IAClH,OAAO,CAAC,eAAe,CAAqB;IAE5C,0FAA0F;IAC1F,OAAO,CAAC,OAAO,CAA4B;IAE3C,oFAAoF;IAC9E,GAAG,CAAC,GAAG,EAAE,MAAM,EAAE,GAAG,EAAE,YAAY,GAAG,OAAO,CAAC,IAAI,CAAC;IAoBxD;;;;;;;OAOG;IACH,OAAO,CAAC,SAAS;IAKjB,uEAAuE;YACzD,KAAK;CA8BpB"}
+{"version":3,"file":"extraction-cache.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extraction-cache.ts"],"names":[],"mappings":"AAGA,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,wBAAwB,CAAC;AAE3D;;;;;;;GAOG;AACH,wBAAgB,UAAU,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAEhD;AAED;;;;GAIG;AACH,MAAM,MAAM,gBAAgB,GACxB,CAAC;IAAE,MAAM,CAAC,EAAE,SAAS,CAAA;CAAE,GAAG,YAAY,CAAC,GACvC;IAAE,MAAM,EAAE,IAAI,CAAC;IAAC,OAAO,EAAE,MAAM,CAAC;IAAC,KAAK,CAAC,EAAE,MAAM,CAAA;CAAE,CAAC;AAKtD;;;;;;;;;;;;;;;;;;;GAmBG;AACH,qBAAa,kBAAkB;IAE3B,OAAO,CAAC,QAAQ,CAAC,IAAI;IACrB,OAAO,CAAC,QAAQ,CAAC,aAAa;gBADb,IAAI,EAAE,MAAM,EACZ,aAAa,GAAE,MAAgC;IAGlE;;;OAGG;IACG,GAAG,CAAC,GAAG,EAAE,MAAM,EAAE,GAAG,EAAE,MAAM,GAAG,OAAO,CAAC,gBAAgB,GAAG,SAAS,CAAC;IAoB1E;;;;;OAKG;IACH,OAAO,CAAC,gBAAgB,CAAK;IAE7B,kHAAkH;IAClH,OAAO,CAAC,eAAe,CAAqB;IAE5C,0FAA0F;IAC1F,OAAO,CAAC,OAAO,CAA4B;IAE3C,mGAAmG;IAC7F,GAAG,CAAC,GAAG,EAAE,MAAM,EAAE,KAAK,EAAE,gBAAgB,GAAG,OAAO,CAAC,IAAI,CAAC;IAwB9D;;;;;;;OAOG;IACH,OAAO,CAAC,SAAS;IAKjB,uEAAuE;YACzD,KAAK;CA8BpB"}

package/dist/modules/workspace/file-readers/extraction-cache.js

@@ -20,7 +20,8 @@ const DEFAULT_MAX_TOTAL_BYTES = 512 * 1024 * 1024; // 512MB
  * re-upload of identical bytes, or the same document on another branch or
  * path, hits the same entry; extension, so identical bytes under another
  * FORMAT never return the wrong extractor's output). One JSON file per entry
- * (`<key>.json` holding the `{ summary, text }`), under a dedicated root
+ * (`<key>.json` holding the `{ summary, text }`, or a failure's
+ * `{ failed, message, until? }`), under a dedicated root
  * that — like the spill
  * store's — sits BESIDE the workspaces root, never inside a workspace and
  * never committed.
@@ -40,14 +41,25 @@ export class DocExtractionCache {
         this.root = root;
         this.maxTotalBytes = maxTotalBytes;
     }
-    /** The cached extraction for `sha`, or undefined on miss/corrupt entry. */
-    async get(sha) {
+    /**
+     * The cached entry for `sha`, or undefined on miss/corrupt entry — and for a
+     * failure remembered only until a moment `now` (epoch ms) has passed.
+     */
+    async get(sha, now) {
         try {
             const parsed = JSON.parse(await fs.readFile(this.entryPath(sha), 'utf8'));
-            if (typeof parsed === 'object' && parsed !== null &&
-                typeof parsed.summary === 'string' &&
-                typeof parsed.text === 'string') {
-                return { summary: parsed.summary, text: parsed.text };
+            if (typeof parsed !== 'object' || parsed === null)
+                return undefined;
+            const entry = parsed;
+            if (entry.failed === true && typeof entry.message === 'string') {
+                if (entry.until === undefined)
+                    return { failed: true, message: entry.message };
+                return typeof entry.until === 'number' && now < entry.until
+                    ? { failed: true, message: entry.message, until: entry.until }
+                    : undefined;
+            }
+            if (typeof entry.summary === 'string' && typeof entry.text === 'string') {
+                return { summary: entry.summary, text: entry.text };
             }
             return undefined;
         }
@@ -66,11 +78,13 @@ export class DocExtractionCache {
     totalAtLastScan;
     /** The prune in flight, so concurrent puts share ONE scan instead of racing their own. */
     pruning;
-    /** Store an extraction under `sha`; prunes towards the size bound. Never throws. */
-    async put(sha, doc) {
+    /** Store an extraction (or a failure) under `sha`; prunes towards the size bound. Never throws. */
+    async put(sha, entry) {
         try {
             await fs.mkdir(this.root, { recursive: true });
-            const payload = JSON.stringify({ summary: doc.summary, text: doc.text });
+            const payload = JSON.stringify(entry.failed
+                ? { failed: true, message: entry.message, ...(entry.until === undefined ? {} : { until: entry.until }) }
+                : { summary: entry.summary, text: entry.text });
             await fs.writeFile(this.entryPath(sha), payload, 'utf8');
             this.writtenSinceScan += Buffer.byteLength(payload);
             if (this.totalAtLastScan === undefined ||

package/dist/modules/workspace/file-readers/extraction-cache.js.map

@@ -1 +1 @@
-{"version":3,"file":"extraction-cache.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extraction-cache.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AACzC,OAAO,EAAE,QAAQ,IAAI,EAAE,EAAE,MAAM,SAAS,CAAC;AACzC,OAAO,IAAI,MAAM,WAAW,CAAC;AAG7B;;;;;;;GAOG;AACH,MAAM,UAAU,UAAU,CAAC,KAAa;IACtC,OAAO,UAAU,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC,QAAQ,KAAK,CAAC,MAAM,IAAI,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;AACzF,CAAC;AAED,2DAA2D;AAC3D,MAAM,uBAAuB,GAAG,GAAG,GAAG,IAAI,GAAG,IAAI,CAAC,CAAC,QAAQ;AAE3D;;;;;;;;;;;;;;;;;;GAkBG;AACH,MAAM,OAAO,kBAAkB;IAEV;IACA;IAFnB,YACmB,IAAY,EACZ,gBAAwB,uBAAuB;QAD/C,SAAI,GAAJ,IAAI,CAAQ;QACZ,kBAAa,GAAb,aAAa,CAAkC;IAC/D,CAAC;IAEJ,2EAA2E;IAC3E,KAAK,CAAC,GAAG,CAAC,GAAW;QACnB,IAAI,CAAC;YACH,MAAM,MAAM,GAAY,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,QAAQ,CAAC,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,EAAE,MAAM,CAAC,CAAC,CAAC;YACnF,IACE,OAAO,MAAM,KAAK,QAAQ,IAAI,MAAM,KAAK,IAAI;gBAC7C,OAAQ,MAAuB,CAAC,OAAO,KAAK,QAAQ;gBACpD,OAAQ,MAAuB,CAAC,IAAI,KAAK,QAAQ,EACjD,CAAC;gBACD,OAAO,EAAE,OAAO,EAAG,MAAuB,CAAC,OAAO,EAAE,IAAI,EAAG,MAAuB,CAAC,IAAI,EAAE,CAAC;YAC5F,CAAC;YACD,OAAO,SAAS,CAAC;QACnB,CAAC;QAAC,MAAM,CAAC;YACP,OAAO,SAAS,CAAC,CAAC,oDAAoD;QACxE,CAAC;IACH,CAAC;IAED;;;;;OAKG;IACK,gBAAgB,GAAG,CAAC,CAAC;IAE7B,kHAAkH;IAC1G,eAAe,CAAqB;IAE5C,0FAA0F;IAClF,OAAO,CAA4B;IAE3C,oFAAoF;IACpF,KAAK,CAAC,GAAG,CAAC,GAAW,EAAE,GAAiB;QACtC,IAAI,CAAC;YACH,MAAM,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC;YAC/C,MAAM,OAAO,GAAG,IAAI,CAAC,SAAS,CAAC,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,EAAE,CAAC,CAAC;YACzE,MAAM,EAAE,CAAC,SAAS,CAAC,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC;YACzD,IAAI,CAAC,gBAAgB,IAAI,MAAM,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC;YACpD,IACE,IAAI,CAAC,eAAe,KAAK,SAAS;gBAClC,IAAI,CAAC,eAAe,GAAG,IAAI,CAAC,gBAAgB,GAAG,IAAI,CAAC,aAAa,EACjE,CAAC;gBACD,IAAI,CAAC,OAAO,KAAK,IAAI,CAAC,KAAK,EAAE,CAAC,OAAO,CAAC,GAAG,EAAE;oBACzC,IAAI,CAAC,OAAO,GAAG,SAAS,CAAC;gBAC3B,CAAC,CAAC,CAAC;gBACH,MAAM,IAAI,CAAC,OAAO,CAAC;YACrB,CAAC;QACH,CAAC;QAAC,MAAM,CAAC;YACP,2EAA2E;QAC7E,CAAC;IACH,CAAC;IAED;;;;;;;OAOG;IACK,SAAS,CAAC,GAAW;QAC3B,MAAM,IAAI,GAAG,GAAG,CAAC,OAAO,CAAC,kBAAkB,EAAE,GAAG,CAAC,CAAC;QAClD,OAAO,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,GAAG,IAAI,OAAO,CAAC,CAAC;IAC9C,CAAC;IAED,uEAAuE;IAC/D,KAAK,CAAC,KAAK;QACjB,yEAAyE;QACzE,sEAAsE;QACtE,yEAAyE;QACzE,wEAAwE;QACxE,iEAAiE;QACjE,MAAM,OAAO,GAAG,IAAI,CAAC,gBAAgB,CAAC;QACtC,MAAM,OAAO,GAAsD,EAAE,CAAC;QACtE,IAAI,KAAK,GAAG,CAAC,CAAC;QACd,KAAK,MAAM,IAAI,IAAI,MAAM,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAC/C,IAAI,CAAC;gBACH,MAAM,EAAE,GAAG,MAAM,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC,CAAC;gBACrD,IAAI,CAAC,EAAE,CAAC,MAAM,EAAE;oBAAE,SAAS;gBAC3B,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,EAAE,IAAI,EAAE,EAAE,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,OAAO,EAAE,CAAC,CAAC;gBAClF,KAAK,IAAI,EAAE,CAAC,IAAI,CAAC;YACnB,CAAC;YAAC,MAAM,CAAC;gBACP,6BAA6B;YAC/B,CAAC;QACH,CAAC;QACD,IAAI,KAAK,GAAG,IAAI,CAAC,aAAa,EAAE,CAAC;YAC/B,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC;YAC1C,KAAK,MAAM,CAAC,IAAI,OAAO,EAAE,CAAC;gBACxB,IAAI,KAAK,IAAI,IAAI,CAAC,aAAa;oBAAE,MAAM;gBACvC,MAAM,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAC;gBAClC,KAAK,IAAI,CAAC,CAAC,IAAI,CAAC;YAClB,CAAC;QACH,CAAC;QACD,IAAI,CAAC,eAAe,GAAG,KAAK,CAAC;QAC7B,IAAI,CAAC,gBAAgB,IAAI,OAAO,CAAC;IACnC,CAAC;CACF"}
+{"version":3,"file":"extraction-cache.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extraction-cache.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AACzC,OAAO,EAAE,QAAQ,IAAI,EAAE,EAAE,MAAM,SAAS,CAAC;AACzC,OAAO,IAAI,MAAM,WAAW,CAAC;AAG7B;;;;;;;GAOG;AACH,MAAM,UAAU,UAAU,CAAC,KAAa;IACtC,OAAO,UAAU,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC,QAAQ,KAAK,CAAC,MAAM,IAAI,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;AACzF,CAAC;AAWD,2DAA2D;AAC3D,MAAM,uBAAuB,GAAG,GAAG,GAAG,IAAI,GAAG,IAAI,CAAC,CAAC,QAAQ;AAE3D;;;;;;;;;;;;;;;;;;;GAmBG;AACH,MAAM,OAAO,kBAAkB;IAEV;IACA;IAFnB,YACmB,IAAY,EACZ,gBAAwB,uBAAuB;QAD/C,SAAI,GAAJ,IAAI,CAAQ;QACZ,kBAAa,GAAb,aAAa,CAAkC;IAC/D,CAAC;IAEJ;;;OAGG;IACH,KAAK,CAAC,GAAG,CAAC,GAAW,EAAE,GAAW;QAChC,IAAI,CAAC;YACH,MAAM,MAAM,GAAY,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,QAAQ,CAAC,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,EAAE,MAAM,CAAC,CAAC,CAAC;YACnF,IAAI,OAAO,MAAM,KAAK,QAAQ,IAAI,MAAM,KAAK,IAAI;gBAAE,OAAO,SAAS,CAAC;YACpE,MAAM,KAAK,GAAG,MAAiC,CAAC;YAChD,IAAI,KAAK,CAAC,MAAM,KAAK,IAAI,IAAI,OAAO,KAAK,CAAC,OAAO,KAAK,QAAQ,EAAE,CAAC;gBAC/D,IAAI,KAAK,CAAC,KAAK,KAAK,SAAS;oBAAE,OAAO,EAAE,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,CAAC;gBAC/E,OAAO,OAAO,KAAK,CAAC,KAAK,KAAK,QAAQ,IAAI,GAAG,GAAG,KAAK,CAAC,KAAK;oBACzD,CAAC,CAAC,EAAE,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,KAAK,EAAE,KAAK,CAAC,KAAK,EAAE;oBAC9D,CAAC,CAAC,SAAS,CAAC;YAChB,CAAC;YACD,IAAI,OAAO,KAAK,CAAC,OAAO,KAAK,QAAQ,IAAI,OAAO,KAAK,CAAC,IAAI,KAAK,QAAQ,EAAE,CAAC;gBACxE,OAAO,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,EAAE,CAAC;YACtD,CAAC;YACD,OAAO,SAAS,CAAC;QACnB,CAAC;QAAC,MAAM,CAAC;YACP,OAAO,SAAS,CAAC,CAAC,oDAAoD;QACxE,CAAC;IACH,CAAC;IAED;;;;;OAKG;IACK,gBAAgB,GAAG,CAAC,CAAC;IAE7B,kHAAkH;IAC1G,eAAe,CAAqB;IAE5C,0FAA0F;IAClF,OAAO,CAA4B;IAE3C,mGAAmG;IACnG,KAAK,CAAC,GAAG,CAAC,GAAW,EAAE,KAAuB;QAC5C,IAAI,CAAC;YACH,MAAM,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC;YAC/C,MAAM,OAAO,GAAG,IAAI,CAAC,SAAS,CAC5B,KAAK,CAAC,MAAM;gBACV,CAAC,CAAC,EAAE,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,GAAG,CAAC,KAAK,CAAC,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,KAAK,CAAC,KAAK,EAAE,CAAC,EAAE;gBACxG,CAAC,CAAC,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,EAAE,CACjD,CAAC;YACF,MAAM,EAAE,CAAC,SAAS,CAAC,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC;YACzD,IAAI,CAAC,gBAAgB,IAAI,MAAM,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC;YACpD,IACE,IAAI,CAAC,eAAe,KAAK,SAAS;gBAClC,IAAI,CAAC,eAAe,GAAG,IAAI,CAAC,gBAAgB,GAAG,IAAI,CAAC,aAAa,EACjE,CAAC;gBACD,IAAI,CAAC,OAAO,KAAK,IAAI,CAAC,KAAK,EAAE,CAAC,OAAO,CAAC,GAAG,EAAE;oBACzC,IAAI,CAAC,OAAO,GAAG,SAAS,CAAC;gBAC3B,CAAC,CAAC,CAAC;gBACH,MAAM,IAAI,CAAC,OAAO,CAAC;YACrB,CAAC;QACH,CAAC;QAAC,MAAM,CAAC;YACP,2EAA2E;QAC7E,CAAC;IACH,CAAC;IAED;;;;;;;OAOG;IACK,SAAS,CAAC,GAAW;QAC3B,MAAM,IAAI,GAAG,GAAG,CAAC,OAAO,CAAC,kBAAkB,EAAE,GAAG,CAAC,CAAC;QAClD,OAAO,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,GAAG,IAAI,OAAO,CAAC,CAAC;IAC9C,CAAC;IAED,uEAAuE;IAC/D,KAAK,CAAC,KAAK;QACjB,yEAAyE;QACzE,sEAAsE;QACtE,yEAAyE;QACzE,wEAAwE;QACxE,iEAAiE;QACjE,MAAM,OAAO,GAAG,IAAI,CAAC,gBAAgB,CAAC;QACtC,MAAM,OAAO,GAAsD,EAAE,CAAC;QACtE,IAAI,KAAK,GAAG,CAAC,CAAC;QACd,KAAK,MAAM,IAAI,IAAI,MAAM,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAC/C,IAAI,CAAC;gBACH,MAAM,EAAE,GAAG,MAAM,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC,CAAC;gBACrD,IAAI,CAAC,EAAE,CAAC,MAAM,EAAE;oBAAE,SAAS;gBAC3B,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,EAAE,IAAI,EAAE,EAAE,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,OAAO,EAAE,CAAC,CAAC;gBAClF,KAAK,IAAI,EAAE,CAAC,IAAI,CAAC;YACnB,CAAC;YAAC,MAAM,CAAC;gBACP,6BAA6B;YAC/B,CAAC;QACH,CAAC;QACD,IAAI,KAAK,GAAG,IAAI,CAAC,aAAa,EAAE,CAAC;YAC/B,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC;YAC1C,KAAK,MAAM,CAAC,IAAI,OAAO,EAAE,CAAC;gBACxB,IAAI,KAAK,IAAI,IAAI,CAAC,aAAa;oBAAE,MAAM;gBACvC,MAAM,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAC;gBAClC,KAAK,IAAI,CAAC,CAAC,IAAI,CAAC;YAClB,CAAC;QACH,CAAC;QACD,IAAI,CAAC,eAAe,GAAG,KAAK,CAAC;QAC7B,IAAI,CAAC,gBAAgB,IAAI,OAAO,CAAC;IACnC,CAAC;CACF"}

package/dist/modules/workspace/file-readers/extraction-worker.d.ts

@@ -0,0 +1,42 @@
+import type { ExtractFn, ExtractResult } from './doc-extract.types.js';
+/** One job, host → worker: the bytes (transferred, not copied) and the format to read them as. */
+export interface ExtractionJob {
+    id: number;
+    extension: string;
+    bytes: Uint8Array;
+}
+/**
+ * One answer, worker → host. `retire` says the worker must not run another
+ * job: a parser left a promise rejected behind it, so its state is no longer
+ * something a next job should inherit. `unattributed` says the rejection came
+ * while a worker that had served an earlier job was running this one, so it
+ * may be that earlier job's: the answer says nothing about this job's file,
+ * and the host runs the job again in a fresh worker.
+ */
+export type ExtractionReply = {
+    id: number;
+    retire?: true;
+    unattributed?: true;
+} & ExtractResult;
+/**
+ * The worker side of document extraction: answer each job with what its
+ * format's extractor returns, ONE job at a time.
+ *
+ * A parser is contracted to answer for bad content rather than throw, but it
+ * is third-party code: one that throws is answered as the job's failure, and
+ * one that rejects a promise nobody awaits — pdf.js did, on a damaged
+ * cross-reference table, and took a whole server down with it (2026-10-15) —
+ * is caught HERE, in the worker's own process events, and answered as the
+ * job's failure with that rejection's message. Nothing a parser does reaches
+ * the server's process: the worker's events are the worker's.
+ *
+ * A rejection is pinned on a job only when that job is the first this worker
+ * ran: no other job's parser has run here, so the rejection is the job's own.
+ * One arriving while a LATER job runs may be a straggler from an earlier one —
+ * the host hands a worker its next job the moment it answers — so it is
+ * answered `unattributed`, and the host runs the job again where it is first.
+ * One arriving between jobs has nothing left to fail; the worker ends itself
+ * instead of carrying it into the next job.
+ */
+export declare function serveExtractions(extractors: Readonly<Record<string, ExtractFn>>): void;
+//# sourceMappingURL=extraction-worker.d.ts.map

package/dist/modules/workspace/file-readers/extraction-worker.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"extraction-worker.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extraction-worker.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,SAAS,EAAE,aAAa,EAAE,MAAM,wBAAwB,CAAC;AAEvE,kGAAkG;AAClG,MAAM,WAAW,aAAa;IAC5B,EAAE,EAAE,MAAM,CAAC;IACX,SAAS,EAAE,MAAM,CAAC;IAClB,KAAK,EAAE,UAAU,CAAC;CACnB;AAED;;;;;;;GAOG;AACH,MAAM,MAAM,eAAe,GAAG;IAAE,EAAE,EAAE,MAAM,CAAC;IAAC,MAAM,CAAC,EAAE,IAAI,CAAC;IAAC,YAAY,CAAC,EAAE,IAAI,CAAA;CAAE,GAAG,aAAa,CAAC;AAEjG;;;;;;;;;;;;;;;;;;;GAmBG;AACH,wBAAgB,gBAAgB,CAAC,UAAU,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,SAAS,CAAC,CAAC,GAAG,IAAI,CAwCtF"}

package/dist/modules/workspace/file-readers/extraction-worker.js

@@ -0,0 +1,69 @@
+import { parentPort } from 'node:worker_threads';
+/**
+ * The worker side of document extraction: answer each job with what its
+ * format's extractor returns, ONE job at a time.
+ *
+ * A parser is contracted to answer for bad content rather than throw, but it
+ * is third-party code: one that throws is answered as the job's failure, and
+ * one that rejects a promise nobody awaits — pdf.js did, on a damaged
+ * cross-reference table, and took a whole server down with it (2026-10-15) —
+ * is caught HERE, in the worker's own process events, and answered as the
+ * job's failure with that rejection's message. Nothing a parser does reaches
+ * the server's process: the worker's events are the worker's.
+ *
+ * A rejection is pinned on a job only when that job is the first this worker
+ * ran: no other job's parser has run here, so the rejection is the job's own.
+ * One arriving while a LATER job runs may be a straggler from an earlier one —
+ * the host hands a worker its next job the moment it answers — so it is
+ * answered `unattributed`, and the host runs the job again where it is first.
+ * One arriving between jobs has nothing left to fail; the worker ends itself
+ * instead of carrying it into the next job.
+ */
+export function serveExtractions(extractors) {
+    const port = parentPort;
+    if (!port)
+        throw new Error('serveExtractions runs inside a worker thread.');
+    let current;
+    let served = 0;
+    const answer = (reply) => {
+        if (current !== reply.id)
+            return; // already answered — by a dangling rejection, say
+        current = undefined;
+        port.postMessage(reply);
+    };
+    const failCurrent = (reason) => {
+        // A straggler from a job already answered: this worker's parser state is
+        // no longer to be trusted, and left running it could pin the rejection on
+        // the NEXT job's file. It ends itself; the pool replaces it. Exit code 0 is
+        // this signal: the host may already have posted the next job, and runs that
+        // job again in a fresh worker rather than answering it as crashed.
+        if (current === undefined)
+            process.exit(0);
+        const ownJob = served === 1;
+        answer({ id: current, ok: false, message: couldNotBeExtracted(reason), retire: true, ...(ownJob ? {} : { unattributed: true }) });
+    };
+    process.on('unhandledRejection', failCurrent);
+    process.on('uncaughtException', failCurrent);
+    port.on('message', (job) => {
+        current = job.id;
+        served++;
+        void (async () => {
+            let result;
+            try {
+                const extract = extractors[job.extension];
+                result = extract
+                    ? await extract(Buffer.from(job.bytes.buffer, job.bytes.byteOffset, job.bytes.byteLength))
+                    : { ok: false, message: `could not be extracted (no extractor reads "${job.extension}" files)` };
+            }
+            catch (err) {
+                result = { ok: false, message: couldNotBeExtracted(err) };
+            }
+            answer({ id: job.id, ...result });
+        })();
+    });
+}
+/** The notice a parser's own error earns: the same words a thrown extractor always earned in `read_file`. */
+function couldNotBeExtracted(reason) {
+    return `could not be extracted (${reason instanceof Error ? reason.message : String(reason)})`;
+}
+//# sourceMappingURL=extraction-worker.js.map

package/dist/modules/workspace/file-readers/extraction-worker.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"extraction-worker.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/extraction-worker.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAoBjD;;;;;;;;;;;;;;;;;;;GAmBG;AACH,MAAM,UAAU,gBAAgB,CAAC,UAA+C;IAC9E,MAAM,IAAI,GAAG,UAAU,CAAC;IACxB,IAAI,CAAC,IAAI;QAAE,MAAM,IAAI,KAAK,CAAC,+CAA+C,CAAC,CAAC;IAC5E,IAAI,OAA2B,CAAC;IAChC,IAAI,MAAM,GAAG,CAAC,CAAC;IAEf,MAAM,MAAM,GAAG,CAAC,KAAsB,EAAQ,EAAE;QAC9C,IAAI,OAAO,KAAK,KAAK,CAAC,EAAE;YAAE,OAAO,CAAC,kDAAkD;QACpF,OAAO,GAAG,SAAS,CAAC;QACpB,IAAI,CAAC,WAAW,CAAC,KAAK,CAAC,CAAC;IAC1B,CAAC,CAAC;IACF,MAAM,WAAW,GAAG,CAAC,MAAe,EAAQ,EAAE;QAC5C,yEAAyE;QACzE,0EAA0E;QAC1E,4EAA4E;QAC5E,4EAA4E;QAC5E,mEAAmE;QACnE,IAAI,OAAO,KAAK,SAAS;YAAE,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;QAC3C,MAAM,MAAM,GAAG,MAAM,KAAK,CAAC,CAAC;QAC5B,MAAM,CAAC,EAAE,EAAE,EAAE,OAAO,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,mBAAmB,CAAC,MAAM,CAAC,EAAE,MAAM,EAAE,IAAI,EAAE,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,YAAY,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC,CAAC;IACpI,CAAC,CAAC;IACF,OAAO,CAAC,EAAE,CAAC,oBAAoB,EAAE,WAAW,CAAC,CAAC;IAC9C,OAAO,CAAC,EAAE,CAAC,mBAAmB,EAAE,WAAW,CAAC,CAAC;IAE7C,IAAI,CAAC,EAAE,CAAC,SAAS,EAAE,CAAC,GAAkB,EAAE,EAAE;QACxC,OAAO,GAAG,GAAG,CAAC,EAAE,CAAC;QACjB,MAAM,EAAE,CAAC;QACT,KAAK,CAAC,KAAK,IAAI,EAAE;YACf,IAAI,MAAqB,CAAC;YAC1B,IAAI,CAAC;gBACH,MAAM,OAAO,GAAG,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;gBAC1C,MAAM,GAAG,OAAO;oBACd,CAAC,CAAC,MAAM,OAAO,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,MAAM,EAAE,GAAG,CAAC,KAAK,CAAC,UAAU,EAAE,GAAG,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC;oBAC1F,CAAC,CAAC,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,+CAA+C,GAAG,CAAC,SAAS,UAAU,EAAE,CAAC;YACrG,CAAC;YAAC,OAAO,GAAG,EAAE,CAAC;gBACb,MAAM,GAAG,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,mBAAmB,CAAC,GAAG,CAAC,EAAE,CAAC;YAC5D,CAAC;YACD,MAAM,CAAC,EAAE,EAAE,EAAE,GAAG,CAAC,EAAE,EAAE,GAAG,MAAM,EAAE,CAAC,CAAC;QACpC,CAAC,CAAC,EAAE,CAAC;IACP,CAAC,CAAC,CAAC;AACL,CAAC;AAED,6GAA6G;AAC7G,SAAS,mBAAmB,CAAC,MAAe;IAC1C,OAAO,2BAA2B,MAAM,YAAY,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,GAAG,CAAC;AACjG,CAAC"}

package/dist/modules/workspace/file-readers/file-reader.d.ts

@@ -39,17 +39,26 @@ export declare const oneLine: typeof sanitizedPath;
  * legacy .doc/.ppt/.xls count too), `image`, `archive`, or any other `binary`.
  */
 export type FileKind = 'text' | 'document' | 'image' | 'archive' | 'binary';
+/** Where a read came from: the workspace whose file it is. A document reader reports a failed extraction there. */
+export interface ReadOrigin {
+    workspaceId: string;
+}
 /** A per-format file reader. Register implementations in `createFileReaderRegistry`. */
 export interface FileReader {
     /** The extensions this reader owns — lowercase, with the dot. Empty for the default (fallback) reader. */
     readonly extensions: readonly string[];
-    /** Read `bytes` (read at `path`) into what the read tools return. Must not throw for bad file content. */
-    read(bytes: Buffer, path: string): Promise<ReadResult>;
+    /**
+     * Read `bytes` (read at `path`) into what the read tools return. Must not
+     * throw for bad file content. `origin`, when the caller has one, is the
+     * workspace the read happened in.
+     */
+    read(bytes: Buffer, path: string, origin?: ReadOrigin): Promise<ReadResult>;
     /**
      * Text `grep` may search, or null when there is nothing (cheaply) searchable.
      * Default readers: the decoded text content (TextReader; null on NUL bytes) /
-     * the CACHED extraction (DocumentReader — cold extraction is grep's call,
-     * under its per-walk budget) / absent = never greppable (ImageReader).
+     * absent = never greppable this way (ImageReader; DocumentReader, whose
+     * remembered extraction grep asks for through `cachedRead` — cold
+     * extraction is grep's call, under its per-walk budget).
      */
     greppableText?(bytes: Buffer, path: string): Promise<string | null>;
     /** May the agent TEXT-editing tools (write_file/write_files/edit_file) touch this file? */

package/dist/modules/workspace/file-readers/file-reader.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"file-reader.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/file-reader.ts"],"names":[],"mappings":"AAWA,OAAO,EAAE,aAAa,EAAE,MAAM,8BAA8B,CAAC;AAE7D;;;;;;;GAOG;AACH,MAAM,MAAM,UAAU,GAClB;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,MAAM,CAAA;CAAE,GAC9B;IAAE,IAAI,EAAE,OAAO,CAAC;IAAC,IAAI,EAAE,MAAM,CAAC;IAAC,QAAQ,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,MAAM,CAAA;CAAE,GAC/D;IAAE,IAAI,EAAE,SAAS,CAAC;IAAC,OAAO,EAAE,MAAM,CAAA;CAAE,CAAC;AAEzC;;;;;;GAMG;AACH,eAAO,MAAM,WAAW,sBAAgB,CAAC;AAEzC;;;;GAIG;AACH,eAAO,MAAM,OAAO,sBAAc,CAAC;AAEnC;;;;GAIG;AACH,MAAM,MAAM,QAAQ,GAAG,MAAM,GAAG,UAAU,GAAG,OAAO,GAAG,SAAS,GAAG,QAAQ,CAAC;AAE5E,wFAAwF;AACxF,MAAM,WAAW,UAAU;IACzB,0GAA0G;IAC1G,QAAQ,CAAC,UAAU,EAAE,SAAS,MAAM,EAAE,CAAC;IACvC,0GAA0G;IAC1G,IAAI,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,OAAO,CAAC,UAAU,CAAC,CAAC;IACvD;;;;;OAKG;IACH,aAAa,CAAC,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,GAAG,IAAI,CAAC,CAAC;IACpE,2FAA2F;IAC3F,QAAQ,CAAC,YAAY,EAAE,OAAO,CAAC;IAC/B,6FAA6F;IAC7F,QAAQ,CAAC,QAAQ,EAAE,QAAQ,CAAC;IAC5B;;;;OAIG;IACH,OAAO,CAAC,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS,CAAC;IAC3C;;;;OAIG;IACH,WAAW,CAAC,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAAC;IACnC;;;;;;;OAOG;IACH,sBAAsB,CAAC,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CAAC;CACrE;AAED;;;;GAIG;AACH,qBAAa,kBAAkB;IAK3B,OAAO,CAAC,QAAQ,CAAC,QAAQ;IAJ3B,OAAO,CAAC,QAAQ,CAAC,WAAW,CAAiC;gBAG3D,OAAO,EAAE,SAAS,UAAU,EAAE,EACb,QAAQ,EAAE,UAAU;IAgBvC,gGAAgG;IAChG,eAAe,IAAI,MAAM,EAAE;IAI3B,6FAA6F;IAC7F,SAAS,CAAC,IAAI,EAAE,MAAM,GAAG,UAAU;CAGpC"}
+{"version":3,"file":"file-reader.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/file-reader.ts"],"names":[],"mappings":"AAWA,OAAO,EAAE,aAAa,EAAE,MAAM,8BAA8B,CAAC;AAE7D;;;;;;;GAOG;AACH,MAAM,MAAM,UAAU,GAClB;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,MAAM,CAAA;CAAE,GAC9B;IAAE,IAAI,EAAE,OAAO,CAAC;IAAC,IAAI,EAAE,MAAM,CAAC;IAAC,QAAQ,EAAE,MAAM,CAAC;IAAC,IAAI,EAAE,MAAM,CAAA;CAAE,GAC/D;IAAE,IAAI,EAAE,SAAS,CAAC;IAAC,OAAO,EAAE,MAAM,CAAA;CAAE,CAAC;AAEzC;;;;;;GAMG;AACH,eAAO,MAAM,WAAW,sBAAgB,CAAC;AAEzC;;;;GAIG;AACH,eAAO,MAAM,OAAO,sBAAc,CAAC;AAEnC;;;;GAIG;AACH,MAAM,MAAM,QAAQ,GAAG,MAAM,GAAG,UAAU,GAAG,OAAO,GAAG,SAAS,GAAG,QAAQ,CAAC;AAE5E,mHAAmH;AACnH,MAAM,WAAW,UAAU;IACzB,WAAW,EAAE,MAAM,CAAC;CACrB;AAED,wFAAwF;AACxF,MAAM,WAAW,UAAU;IACzB,0GAA0G;IAC1G,QAAQ,CAAC,UAAU,EAAE,SAAS,MAAM,EAAE,CAAC;IACvC;;;;OAIG;IACH,IAAI,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,CAAC,EAAE,UAAU,GAAG,OAAO,CAAC,UAAU,CAAC,CAAC;IAC5E;;;;;;OAMG;IACH,aAAa,CAAC,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,GAAG,IAAI,CAAC,CAAC;IACpE,2FAA2F;IAC3F,QAAQ,CAAC,YAAY,EAAE,OAAO,CAAC;IAC/B,6FAA6F;IAC7F,QAAQ,CAAC,QAAQ,EAAE,QAAQ,CAAC;IAC5B;;;;OAIG;IACH,OAAO,CAAC,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS,CAAC;IAC3C;;;;OAIG;IACH,WAAW,CAAC,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAAC;IACnC;;;;;;;OAOG;IACH,sBAAsB,CAAC,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CAAC;CACrE;AAED;;;;GAIG;AACH,qBAAa,kBAAkB;IAK3B,OAAO,CAAC,QAAQ,CAAC,QAAQ;IAJ3B,OAAO,CAAC,QAAQ,CAAC,WAAW,CAAiC;gBAG3D,OAAO,EAAE,SAAS,UAAU,EAAE,EACb,QAAQ,EAAE,UAAU;IAgBvC,gGAAgG;IAChG,eAAe,IAAI,MAAM,EAAE;IAI3B,6FAA6F;IAC7F,SAAS,CAAC,IAAI,EAAE,MAAM,GAAG,UAAU;CAGpC"}

package/dist/modules/workspace/file-readers/file-reader.js.map

@@ -1 +1 @@
-{"version":3,"file":"file-reader.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/file-reader.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AACH,OAAO,EAAE,aAAa,EAAE,MAAM,wBAAwB,CAAC;AACvD,OAAO,EAAE,aAAa,EAAE,MAAM,8BAA8B,CAAC;AAe7D;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG,aAAa,CAAC;AAEzC;;;;GAIG;AACH,MAAM,CAAC,MAAM,OAAO,GAAG,WAAW,CAAC;AAiDnC;;;;GAIG;AACH,MAAM,OAAO,kBAAkB;IAKV;IAJF,WAAW,GAAG,IAAI,GAAG,EAAsB,CAAC;IAE7D,YACE,OAA8B,EACb,QAAoB;QAApB,aAAQ,GAAR,QAAQ,CAAY;QAErC,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;YAC7B,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,UAAU,EAAE,CAAC;gBACpC,wEAAwE;gBACxE,yEAAyE;gBACzE,6DAA6D;gBAC7D,MAAM,GAAG,GAAG,GAAG,CAAC,WAAW,EAAE,CAAC;gBAC9B,mEAAmE;gBACnE,8DAA8D;gBAC9D,IAAI,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,GAAG,CAAC;oBAAE,MAAM,IAAI,KAAK,CAAC,yCAAyC,GAAG,GAAG,CAAC,CAAC;gBAChG,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC;YACpC,CAAC;QACH,CAAC;IACH,CAAC;IAED,gGAAgG;IAChG,eAAe;QACb,OAAO,CAAC,GAAG,IAAI,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC,CAAC;IACtC,CAAC;IAED,6FAA6F;IAC7F,SAAS,CAAC,IAAY;QACpB,OAAO,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC;IACpE,CAAC;CACF"}
+{"version":3,"file":"file-reader.js","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/file-reader.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AACH,OAAO,EAAE,aAAa,EAAE,MAAM,wBAAwB,CAAC;AACvD,OAAO,EAAE,aAAa,EAAE,MAAM,8BAA8B,CAAC;AAe7D;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG,aAAa,CAAC;AAEzC;;;;GAIG;AACH,MAAM,CAAC,MAAM,OAAO,GAAG,WAAW,CAAC;AA2DnC;;;;GAIG;AACH,MAAM,OAAO,kBAAkB;IAKV;IAJF,WAAW,GAAG,IAAI,GAAG,EAAsB,CAAC;IAE7D,YACE,OAA8B,EACb,QAAoB;QAApB,aAAQ,GAAR,QAAQ,CAAY;QAErC,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;YAC7B,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,UAAU,EAAE,CAAC;gBACpC,wEAAwE;gBACxE,yEAAyE;gBACzE,6DAA6D;gBAC7D,MAAM,GAAG,GAAG,GAAG,CAAC,WAAW,EAAE,CAAC;gBAC9B,mEAAmE;gBACnE,8DAA8D;gBAC9D,IAAI,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,GAAG,CAAC;oBAAE,MAAM,IAAI,KAAK,CAAC,yCAAyC,GAAG,GAAG,CAAC,CAAC;gBAChG,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC;YACpC,CAAC;QACH,CAAC;IACH,CAAC;IAED,gGAAgG;IAChG,eAAe;QACb,OAAO,CAAC,GAAG,IAAI,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC,CAAC;IACtC,CAAC;IAED,6FAA6F;IAC7F,SAAS,CAAC,IAAY;QACpB,OAAO,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC;IACpE,CAAC;CACF"}

package/dist/modules/workspace/file-readers/file-reader.registry.d.ts

@@ -5,8 +5,9 @@ import { FileReaderRegistry } from './file-reader.js';
  * which extension. Adding a format is one entry here (plus its reader/extract
  * function); read_file, grep and the write-refusal all follow automatically.
  *
- * `docExtract` is the shared content-hash extraction cache the document
- * readers wrap their pure extract functions with (see DocumentReader).
+ * `docExtract` is the shared content-hash extraction service every document
+ * reader extracts through (see DocumentReader); the extractors themselves run
+ * in its worker, never here.
  * Everything not claimed below falls back to the TextReader.
  */
 export declare function createFileReaderRegistry(docExtract: DocExtractService): FileReaderRegistry;

package/dist/modules/workspace/file-readers/file-reader.registry.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"file-reader.registry.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/file-reader.registry.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,iBAAiB,EAAE,MAAM,0BAA0B,CAAC;AAYlE,OAAO,EAAE,kBAAkB,EAAE,MAAM,kBAAkB,CAAC;AAItD;;;;;;;;GAQG;AACH,wBAAgB,wBAAwB,CAAC,UAAU,EAAE,iBAAiB,GAAG,kBAAkB,CA2B1F"}
+{"version":3,"file":"file-reader.registry.d.ts","sourceRoot":"","sources":["../../../../src/modules/workspace/file-readers/file-reader.registry.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,iBAAiB,EAAE,MAAM,0BAA0B,CAAC;AAIlE,OAAO,EAAE,kBAAkB,EAAE,MAAM,kBAAkB,CAAC;AAItD;;;;;;;;;GASG;AACH,wBAAgB,wBAAwB,CAAC,UAAU,EAAE,iBAAiB,GAAG,kBAAkB,CAoB1F"}