@kreuzberg/node 4.6.1 → 4.7.0

package/dist/index.mjs

@@ -506,6 +506,7 @@ function convertChunk(rawChunk) {
   if (!rawChunk || typeof rawChunk !== "object") {
     return {
       content: "",
+      chunkType: null,
       metadata: {
         byteStart: 0,
         byteEnd: 0,
@@ -519,26 +520,17 @@ function convertChunk(rawChunk) {
   const chunk = rawChunk;
   const metadata = chunk["metadata"] ?? {};
   return {
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     content: chunk["content"] ?? "",
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
+    chunkType: chunk["chunk_type"] ?? chunk["chunkType"] ?? null,
     embedding: chunk["embedding"] ?? null,
     metadata: {
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
       byteStart: metadata["byte_start"] ?? metadata["charStart"] ?? 0,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
       byteEnd: metadata["byte_end"] ?? metadata["charEnd"] ?? 0,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
       tokenCount: metadata["token_count"] ?? metadata["tokenCount"] ?? null,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
       chunkIndex: metadata["chunk_index"] ?? metadata["chunkIndex"] ?? 0,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
       totalChunks: metadata["total_chunks"] ?? metadata["totalChunks"] ?? 0,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
       firstPage: metadata["first_page"] ?? metadata["firstPage"] ?? null,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
       lastPage: metadata["last_page"] ?? metadata["lastPage"] ?? null,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
       headingContext: (() => {
         const hc = metadata["heading_context"] ?? metadata["headingContext"];
         if (!hc) return null;
@@ -548,9 +540,7 @@ function convertChunk(rawChunk) {
           headings: headings.map((h) => {
             const heading = h;
             return {
-              // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
               level: heading["level"] ?? 0,
-              // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
               text: heading["text"] ?? ""
             };
           })
@@ -571,22 +561,14 @@ function convertElement(rawElement) {
   const element = rawElement;
   const elementMetadata = element["metadata"] ?? {};
   return {
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     elementId: element["element_id"] ?? element["elementId"] ?? "",
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     elementType: element["element_type"] ?? element["elementType"] ?? "narrative_text",
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     text: element["text"] ?? "",
     metadata: {
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
       pageNumber: elementMetadata["page_number"] ?? elementMetadata["pageNumber"] ?? null,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
       filename: elementMetadata["filename"] ?? null,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
       coordinates: elementMetadata["coordinates"] ? elementMetadata["coordinates"] : null,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
       elementIndex: elementMetadata["element_index"] ?? elementMetadata["elementIndex"] ?? null,
-      // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
       additional: elementMetadata["additional"] ?? {}
     }
   };
@@ -609,27 +591,16 @@ function convertImage(rawImage) {
   }
   const image = rawImage;
   return {
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     data: ensureUint8Array(image["data"]),
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     format: image["format"] ?? "unknown",
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     imageIndex: image["imageIndex"] ?? 0,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     pageNumber: image["pageNumber"] ?? null,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     width: image["width"] ?? null,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     height: image["height"] ?? null,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     colorspace: image["colorspace"] ?? null,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     bitsPerComponent: image["bitsPerComponent"] ?? null,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     isMask: image["isMask"] ?? false,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     description: image["description"] ?? null,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     ocrResult: image["ocrResult"] ? convertResult(image["ocrResult"]) : null
   };
 }
@@ -644,15 +615,10 @@ function convertPageContent(rawPage) {
   }
   const page = rawPage;
   return {
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     pageNumber: page["pageNumber"] ?? 0,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     content: page["content"] ?? "",
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     tables: Array.isArray(page["tables"]) ? page["tables"] : [],
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     images: Array.isArray(page["images"]) ? page["images"].map((image) => convertImage(image)) : [],
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     isBlank: page["isBlank"] ?? null
   };
 }
@@ -675,20 +641,15 @@ function convertResult(rawResult) {
   const metadata = result["metadata"];
   const metadataValue = typeof metadata === "string" ? parseMetadata(metadata) : metadata ?? {};
   const returnObj = {
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     content: result["content"] ?? "",
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     mimeType: result["mimeType"] ?? "application/octet-stream",
     metadata: metadataValue,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     tables: Array.isArray(result["tables"]) ? result["tables"] : [],
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     detectedLanguages: Array.isArray(result["detectedLanguages"]) ? result["detectedLanguages"] : null,
     chunks: null,
     images: null,
     elements: null,
     pages: null,
-    // biome-ignore lint/complexity/useLiteralKeys: required for strict TypeScript noPropertyAccessFromIndexSignature
     document: result["document"] ?? null
   };
   const chunksData = result["chunks"];
@@ -760,6 +721,42 @@ async function batchExtractBytes(dataList, mimeTypes, config = null) {
   return rawResults.map(convertResult);
 }
 
+// typescript/extraction/render.ts
+function renderPdfPageSync(filePath, pageIndex, options) {
+  return getBinding().renderPdfPageSync(filePath, pageIndex, options?.dpi ?? null);
+}
+async function renderPdfPage(filePath, pageIndex, options) {
+  return getBinding().renderPdfPage(filePath, pageIndex, options?.dpi ?? null);
+}
+function iteratePdfPagesSync(filePath, options) {
+  return getBinding().iteratePdfPagesSync(filePath, options?.dpi ?? null);
+}
+async function iteratePdfPages(filePath, options) {
+  return getBinding().iteratePdfPages(filePath, options?.dpi ?? null);
+}
+function pdfPageCount(filePath) {
+  return getBinding().pdfPageCount(filePath);
+}
+var PdfPageIterator = class {
+  inner;
+  constructor(filePath, options) {
+    const Ctor = getBinding().JsPdfPageIterator;
+    this.inner = new Ctor(filePath, options?.dpi ?? null);
+  }
+  /** Advance and return the next page, or null when exhausted. */
+  next() {
+    return this.inner.next();
+  }
+  /** Total number of pages in the PDF. */
+  pageCount() {
+    return this.inner.pageCount();
+  }
+  /** Free native resources. Safe to call multiple times. */
+  close() {
+    this.inner.close();
+  }
+};
+
 // typescript/extraction/single.ts
 import { readFileSync } from "fs";
 function extractFileSync(filePath, mimeTypeOrConfig, maybeConfig) {
@@ -1158,7 +1155,7 @@ function getEmbeddingPreset(name) {
 }
 
 // typescript/index.ts
-var __version__ = "4.6.1";
+var __version__ = "4.7.0";
 export {
   CacheError,
   ErrorCode,
@@ -1168,6 +1165,7 @@ export {
   MissingDependencyError,
   OcrError,
   ParsingError,
+  PdfPageIterator,
   PluginError,
   ValidationError,
   __resetBindingForTests,
@@ -1199,6 +1197,8 @@ export {
   getLastErrorCode,
   getLastPanicContext,
   getWorkerPoolStats,
+  iteratePdfPages,
+  iteratePdfPagesSync,
   listDocumentExtractors,
   listEmbeddingPresets,
   listOcrBackends,
@@ -1206,9 +1206,12 @@ export {
   listValidators,
   loadConfigFile,
   loadConfigFromPath,
+  pdfPageCount,
   registerOcrBackend,
   registerPostProcessor,
   registerValidator,
+  renderPdfPage,
+  renderPdfPageSync,
   unregisterDocumentExtractor,
   unregisterOcrBackend,
   unregisterPostProcessor,