@veryfront/ext-document-kreuzberg 0.1.986 → 0.1.988

package/esm/extensions/ext-document-kreuzberg/src/index.d.ts

@@ -9,7 +9,17 @@
 import "../../../_dnt.polyfills.js";
 import type { ExtensionFactory } from "veryfront/extensions";
 import type { DocumentExtractor, KreuzbergExtractor } from "veryfront/extensions/compat";
+/** Maximum time to wait for fallback worker extraction before aborting. */
+export declare const EXTRACTION_TIMEOUT_MS = 120000;
+declare function extractInWorkerDeno(buffer: ArrayBuffer, mimeType: string): Promise<string>;
+export interface KreuzbergDocumentExtractorDeps {
+    isDenoRuntime?: boolean;
+    loadNativeKreuzberg?: () => Promise<KreuzbergExtractor>;
+    extractInWorkerDeno?: typeof extractInWorkerDeno;
+}
 export declare class KreuzbergDocumentExtractor implements DocumentExtractor {
+    private readonly deps;
+    constructor(deps?: KreuzbergDocumentExtractorDeps);
     importKreuzberg(): Promise<KreuzbergExtractor>;
     extractInWorker(buffer: ArrayBuffer, mimeType: string): Promise<string>;
 }

package/esm/extensions/ext-document-kreuzberg/src/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../../src/extensions/ext-document-kreuzberg/src/index.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AACH,OAAO,4BAA4B,CAAC;AAKpC,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,kCAAkC,CAAC;AACzE,OAAO,KAAK,EAAE,iBAAiB,EAAE,kBAAkB,EAAE,MAAM,yCAAyC,CAAC;AAsDrG,qBAAa,0BAA2B,YAAW,iBAAiB;IAClE,eAAe,IAAI,OAAO,CAAC,kBAAkB,CAAC;IAIxC,eAAe,CAAC,MAAM,EAAE,WAAW,EAAE,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC;CAW9E;AAED,QAAA,MAAM,oBAAoB,EAAE,gBAkB3B,CAAC;AAEF,eAAe,oBAAoB,CAAC"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../../src/extensions/ext-document-kreuzberg/src/index.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AACH,OAAO,4BAA4B,CAAC;AAKpC,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,kCAAkC,CAAC;AACzE,OAAO,KAAK,EAAE,iBAAiB,EAAE,kBAAkB,EAAE,MAAM,yCAAyC,CAAC;AAIrG,2EAA2E;AAC3E,eAAO,MAAM,qBAAqB,SAAU,CAAC;AAE7C,iBAAS,mBAAmB,CAC1B,MAAM,EAAE,WAAW,EACnB,QAAQ,EAAE,MAAM,GACf,OAAO,CAAC,MAAM,CAAC,CA0CjB;AAED,MAAM,WAAW,8BAA8B;IAC7C,aAAa,CAAC,EAAE,OAAO,CAAC;IACxB,mBAAmB,CAAC,EAAE,MAAM,OAAO,CAAC,kBAAkB,CAAC,CAAC;IACxD,mBAAmB,CAAC,EAAE,OAAO,mBAAmB,CAAC;CAClD;AAgBD,qBAAa,0BAA2B,YAAW,iBAAiB;IACtD,OAAO,CAAC,QAAQ,CAAC,IAAI;gBAAJ,IAAI,GAAE,8BAAmC;IAEtE,eAAe,IAAI,OAAO,CAAC,kBAAkB,CAAC;IAIxC,eAAe,CAAC,MAAM,EAAE,WAAW,EAAE,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC;CA2B9E;AAED,QAAA,MAAM,oBAAoB,EAAE,gBAkB3B,CAAC;AAEF,eAAe,oBAAoB,CAAC"}

package/esm/extensions/ext-document-kreuzberg/src/index.js

@@ -8,10 +8,10 @@
  */
 import "../../../_dnt.polyfills.js";
 import * as dntShim from "../../../_dnt.shims.js";
-import { loadKreuzberg } from "./kreuzberg.js";
+import { isMissingPackageError, loadKreuzberg, loadKreuzbergNative } from "./kreuzberg.js";
 import { isDeno } from "./runtime.js";
-/** Maximum time to wait for document text extraction before aborting. */
-const EXTRACTION_TIMEOUT_MS = 30_000;
+/** Maximum time to wait for fallback worker extraction before aborting. */
+export const EXTRACTION_TIMEOUT_MS = 120_000;
 function extractInWorkerDeno(buffer, mimeType) {
     return new Promise((resolve, reject) => {
         // The worker ships as raw TypeScript in the compiled binary and from source
@@ -46,20 +46,43 @@ function extractInWorkerDeno(buffer, mimeType) {
         worker.postMessage({ buffer, mimeType }, [buffer]);
     });
 }
+function isPdfMimeType(mimeType) {
+    return mimeType.toLowerCase().split(";")[0]?.trim() === "application/pdf";
+}
+async function extractWithNativeKreuzberg(buffer, mimeType, loadNative) {
+    const { extractBytes } = await loadNative();
+    const result = await extractBytes(new Uint8Array(buffer), mimeType);
+    return result.content;
+}
 export class KreuzbergDocumentExtractor {
+    deps;
+    constructor(deps = {}) {
+        this.deps = deps;
+    }
     importKreuzberg() {
         return loadKreuzberg();
     }
     async extractInWorker(buffer, mimeType) {
-        // Only a real Deno runtime gets the isolated Worker; Node/Bun extract
-        // in-process via @kreuzberg/node. See ./runtime.ts for why a bare `Deno`
-        // check is unreliable in the dnt npm build.
-        if (!isDeno) {
+        const isDenoRuntime = this.deps.isDenoRuntime ?? isDeno;
+        const extractWithWorker = this.deps.extractInWorkerDeno ?? extractInWorkerDeno;
+        // Node/Bun extract in-process via @kreuzberg/node. Deno keeps the isolated
+        // Worker fallback, but PDFs first try the native extractor because the WASM
+        // PDF path can hang on valid large manuals.
+        if (!isDenoRuntime) {
             const { extractBytes } = await loadKreuzberg();
             const result = await extractBytes(new Uint8Array(buffer), mimeType);
             return result.content;
         }
-        return extractInWorkerDeno(buffer, mimeType);
+        if (isPdfMimeType(mimeType)) {
+            try {
+                return await extractWithNativeKreuzberg(buffer, mimeType, this.deps.loadNativeKreuzberg ?? loadKreuzbergNative);
+            }
+            catch (error) {
+                if (!isMissingPackageError(error))
+                    throw error;
+            }
+        }
+        return extractWithWorker(buffer, mimeType);
     }
 }
 const extDocumentKreuzberg = () => {

package/esm/extensions/ext-document-kreuzberg/src/kreuzberg.d.ts

@@ -1,3 +1,5 @@
 import type { KreuzbergExtractor } from "veryfront/extensions/compat";
+export declare function loadKreuzbergNative(): Promise<KreuzbergExtractor>;
 export declare function loadKreuzberg(): Promise<KreuzbergExtractor>;
+export declare function isMissingPackageError(error: unknown): boolean;
 //# sourceMappingURL=kreuzberg.d.ts.map

package/esm/extensions/ext-document-kreuzberg/src/kreuzberg.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"kreuzberg.d.ts","sourceRoot":"","sources":["../../../../src/extensions/ext-document-kreuzberg/src/kreuzberg.ts"],"names":[],"mappings":"AAaA,OAAO,KAAK,EAAE,kBAAkB,EAAE,MAAM,yCAAyC,CAAC;AAwBlF,wBAAsB,aAAa,IAAI,OAAO,CAAC,kBAAkB,CAAC,CA2BjE"}
+{"version":3,"file":"kreuzberg.d.ts","sourceRoot":"","sources":["../../../../src/extensions/ext-document-kreuzberg/src/kreuzberg.ts"],"names":[],"mappings":"AAaA,OAAO,KAAK,EAAE,kBAAkB,EAAE,MAAM,yCAAyC,CAAC;AAWlF,wBAAsB,mBAAmB,IAAI,OAAO,CAAC,kBAAkB,CAAC,CAWvE;AAED,wBAAsB,aAAa,IAAI,OAAO,CAAC,kBAAkB,CAAC,CA2BjE;AAeD,wBAAgB,qBAAqB,CAAC,KAAK,EAAE,OAAO,GAAG,OAAO,CAqB7D"}

package/esm/extensions/ext-document-kreuzberg/src/kreuzberg.js

@@ -10,23 +10,22 @@
  */
 import * as dntShim from "../../../_dnt.shims.js";
 import { isDeno } from "./runtime.js";
-// deno-lint-ignore no-explicit-any
-async function loadKreuzbergNode() {
+export async function loadKreuzbergNative() {
     try {
         return await import("@kreuzberg/node");
     }
     catch (error) {
         if (!isMissingPackageError(error))
             throw error;
-        throw new Error('Document extraction on Node requires the optional package "@kreuzberg/node". ' +
-            "Install @kreuzberg/node@^4.4.2 or disable document extraction.");
+        throw new Error('Native document extraction requires the optional package "@kreuzberg/node". ' +
+            "Install @kreuzberg/node@^4.4.2 or disable document extraction.", { cause: error });
     }
 }
 export async function loadKreuzberg() {
     // Node/Bun load the native @kreuzberg/node; only a real Deno runtime uses the
     // WASM build. See ./runtime.ts for why a bare `Deno` check is unreliable here.
     if (!isDeno) {
-        return loadKreuzbergNode();
+        return loadKreuzbergNative();
     }
     const mod = await importKreuzbergWasm();
     const mainModule = typeof dntShim.Deno.mainModule === "string"
@@ -55,13 +54,23 @@ async function importKreuzbergWasm() {
         if (!isMissingPackageError(error))
             throw error;
         throw new Error('Document extraction on Deno requires the optional package "@kreuzberg/wasm". ' +
-            "Install @kreuzberg/wasm@4.5.2 or disable document extraction.");
+            "Install @kreuzberg/wasm@4.5.2 or disable document extraction.", { cause: error });
     }
 }
-function isMissingPackageError(error) {
-    const message = error instanceof Error ? error.message : String(error);
-    return message.includes("Cannot find package") ||
-        message.includes("Cannot find module") ||
-        message.includes("ERR_MODULE_NOT_FOUND") ||
-        message.includes("Module not found");
+export function isMissingPackageError(error) {
+    const seen = new Set();
+    let current = error;
+    while (current !== undefined && current !== null && !seen.has(current)) {
+        seen.add(current);
+        const message = current instanceof Error ? current.message : String(current);
+        if (message.includes("Cannot find package") ||
+            message.includes("Cannot find module") ||
+            message.includes("Cannot find native binding") ||
+            message.includes("ERR_MODULE_NOT_FOUND") ||
+            message.includes("Module not found")) {
+            return true;
+        }
+        current = current instanceof Error ? current.cause : undefined;
+    }
+    return false;
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@veryfront/ext-document-kreuzberg",
-  "version": "0.1.986",
+  "version": "0.1.988",
   "description": "Veryfront first-party extension package for ext-document-kreuzberg",
   "keywords": [
     "veryfront",
@@ -53,7 +53,7 @@
     "@kreuzberg/wasm": "4.5.2"
   },
   "peerDependencies": {
-    "veryfront": "^0.1.986"
+    "veryfront": "^0.1.988"
   },
   "type": "module",
   "types": "./esm/extensions/ext-document-kreuzberg/src/index.d.ts",
@@ -63,4 +63,4 @@
     "NOTICE",
     "README.md"
   ]
-}
+}