@krak-stack/registry 0.1.9 → 0.1.11

package/dist/services/file-extraction/index.js

@@ -0,0 +1,119 @@
+// ../../src/services/file-extraction/index.ts
+import { extractBatch, ExtractInputKind, OutputFormat } from "@xberg-io/xberg";
+import { Context, Effect, Layer, Semaphore } from "effect";
+
+// ../../src/services/file-extraction/schema.ts
+import { Schema } from "effect";
+var FileExtractedTextSchema = Schema.Struct({
+  content: Schema.String,
+  contentByteSize: Schema.Int,
+  truncated: Schema.Boolean
+}).annotate({ identifier: "FileExtractedText" });
+
+class FileExtractionFailed extends Schema.TaggedErrorClass()("FileExtractionFailed", { message: Schema.String }) {
+}
+
+// ../../src/services/file-extraction/index.ts
+import { OutputFormat as OutputFormat2 } from "@xberg-io/xberg";
+var defaultFileExtractionOptions = {
+  maxInputBytes: 20 * 1024 * 1024,
+  maxOutputBytes: 512 * 1024,
+  maxConcurrentExtractions: 2,
+  timeoutSeconds: 30
+};
+var truncateUtf8 = (content, maxBytes) => {
+  const bytes = new TextEncoder().encode(content);
+  if (bytes.byteLength <= maxBytes) {
+    return { content, byteSize: bytes.byteLength };
+  }
+  let end = maxBytes;
+  while (end > 0 && (bytes[end] & 192) === 128)
+    end -= 1;
+  return {
+    content: new TextDecoder().decode(bytes.slice(0, end)),
+    byteSize: end
+  };
+};
+var limitText = (content, maxBytes) => {
+  const normalized = content.replace(/\r\n?/g, `
+`).trim();
+  const originalByteSize = new TextEncoder().encode(normalized).byteLength;
+  const bounded = truncateUtf8(normalized, maxBytes);
+  return {
+    content: bounded.content,
+    contentByteSize: bounded.byteSize,
+    truncated: bounded.byteSize !== originalByteSize
+  };
+};
+var make = (options) => Effect.gen(function* () {
+  const semaphore = yield* Semaphore.make(options.maxConcurrentExtractions);
+  const extract = Effect.fn("FileExtractionService.extract")(function* ({
+    bytes,
+    filename,
+    mimeType,
+    outputFormat
+  }) {
+    if (bytes.byteLength > options.maxInputBytes) {
+      return yield* new FileExtractionFailed({
+        message: "Document exceeds the extraction size limit"
+      });
+    }
+    const output = yield* semaphore.withPermit(Effect.tryPromise({
+      try: () => extractBatch([
+        {
+          kind: ExtractInputKind.Bytes,
+          bytes,
+          filename,
+          mimeType,
+          config: { timeoutSecs: options.timeoutSeconds }
+        }
+      ], {
+        outputFormat,
+        extractionTimeoutSecs: options.timeoutSeconds,
+        maxConcurrentExtractions: 1,
+        maxEmbeddedFileBytes: 10 * 1024 * 1024,
+        securityLimits: {
+          maxArchiveSize: 50 * 1024 * 1024,
+          maxCompressionRatio: 100,
+          maxFilesInArchive: 1000,
+          maxNestingDepth: 50,
+          maxEntityLength: 1024 * 1024,
+          maxContentSize: 2 * 1024 * 1024,
+          maxIterations: 1e6,
+          maxXmlDepth: 50,
+          maxTableCells: 1e5
+        },
+        useCache: false
+      }),
+      catch: (cause) => new FileExtractionFailed({
+        message: cause instanceof Error ? cause.message : "Document extraction failed"
+      })
+    }));
+    const result = output.results?.[0];
+    if (!result?.content) {
+      return yield* new FileExtractionFailed({
+        message: output.errors?.[0]?.message ?? "Document extraction returned no readable content"
+      });
+    }
+    return limitText(result.content, options.maxOutputBytes);
+  });
+  return {
+    extract,
+    markdown: Effect.fn("FileExtractionService.markdown")((input) => extract({ ...input, outputFormat: OutputFormat.Markdown })),
+    text: Effect.fn("FileExtractionService.text")((input) => extract({ ...input, outputFormat: OutputFormat.Plain })),
+    html: Effect.fn("FileExtractionService.html")((input) => extract({ ...input, outputFormat: OutputFormat.Html }))
+  };
+});
+
+class FileExtractionService extends Context.Service()("FileExtractionService", { make: make(defaultFileExtractionOptions) }) {
+  static layer = Layer.effect(this, this.make);
+  static layerWith = (options) => Layer.effect(this, make({ ...defaultFileExtractionOptions, ...options }));
+  static testLayer = (service) => Layer.succeed(this, service);
+}
+export {
+  defaultFileExtractionOptions,
+  OutputFormat2 as OutputFormat,
+  FileExtractionService,
+  FileExtractionFailed,
+  FileExtractedTextSchema
+};

package/dist/services/file-extraction/schema.d.ts

@@ -0,0 +1,13 @@
+import { Schema } from "effect";
+export declare const FileExtractedTextSchema: Schema.Struct<{
+    readonly content: Schema.String;
+    readonly contentByteSize: Schema.Int;
+    readonly truncated: Schema.Boolean;
+}>;
+export type FileExtractedText = typeof FileExtractedTextSchema.Type;
+declare const FileExtractionFailed_base: Schema.Class<FileExtractionFailed, Schema.TaggedStruct<"FileExtractionFailed", {
+    readonly message: Schema.String;
+}>, import("effect/Cause").YieldableError>;
+export declare class FileExtractionFailed extends FileExtractionFailed_base {
+}
+export {};

package/dist/services/file-extraction/schema.js

@@ -0,0 +1,14 @@
+// ../../src/services/file-extraction/schema.ts
+import { Schema } from "effect";
+var FileExtractedTextSchema = Schema.Struct({
+  content: Schema.String,
+  contentByteSize: Schema.Int,
+  truncated: Schema.Boolean
+}).annotate({ identifier: "FileExtractedText" });
+
+class FileExtractionFailed extends Schema.TaggedErrorClass()("FileExtractionFailed", { message: Schema.String }) {
+}
+export {
+  FileExtractionFailed,
+  FileExtractedTextSchema
+};

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@krak-stack/registry",
-  "version": "0.1.9",
+  "version": "0.1.11",
   "description": "Tree-shakable KrakStack components and Effect services.",
   "license": "MIT",
   "repository": {
@@ -172,6 +172,14 @@
       "types": "./dist/services/s3/schema.d.ts",
       "import": "./dist/services/s3/schema.js"
     },
+    "./service-file-extraction": {
+      "types": "./dist/services/file-extraction/index.d.ts",
+      "import": "./dist/services/file-extraction/index.js"
+    },
+    "./service-file-extraction/schema": {
+      "types": "./dist/services/file-extraction/schema.d.ts",
+      "import": "./dist/services/file-extraction/schema.js"
+    },
     "./tailwind.css": "./tailwind.css",
     "./package.json": "./package.json"
   },
@@ -198,8 +206,9 @@
     "@streamdown/code": "^1.1.1",
     "@tanstack/react-form": "^1.33.0",
     "@tanstack/react-router": "^1.170.15",
-    "@tanstack/react-table": "^8.21.3",
+    "@tanstack/react-table": "^9.1.2",
     "@tanstack/react-virtual": "^3.14.5",
+    "@xberg-io/xberg": "1.0.14",
     "class-variance-authority": "^0.7.1",
     "clsx": "^2.1.1",
     "cmdk": "^1.1.1",
@@ -261,6 +270,9 @@
     "@tanstack/react-virtual": {
       "optional": true
     },
+    "@xberg-io/xberg": {
+      "optional": true
+    },
     "shiki": {
       "optional": true
     },