@ontos-ai/knowhere-sdk 0.2.0 → 0.2.1

package/dist/index.d.mts

@@ -216,7 +216,7 @@ interface ParseParams {
     /** Generate table summaries */
     summaryTable?: boolean;
     /** Generate text summaries */
-    summaryText?: boolean;
+    summaryTxt?: boolean;
     /** Custom data identifier */
     dataId?: string;
     /** Additional fragment description */
@@ -282,6 +282,30 @@ interface Statistics {
 interface FileIndex {
     [chunkId: string]: string;
 }
+/**
+ * Processing cost details emitted by manifest v2
+ */
+interface ProcessingCost {
+    microDollars?: number;
+    credits?: number;
+}
+/**
+ * Processing timing details emitted by manifest v2
+ */
+interface ProcessingTiming {
+    startedAt?: Date;
+    completedAt?: Date;
+    durationMs?: number;
+}
+/**
+ * Processing metadata emitted by manifest v2
+ */
+interface ProcessingMetadata {
+    pageCount?: number;
+    billingStatus?: string;
+    cost?: ProcessingCost;
+    timing?: ProcessingTiming;
+}
 /**
  * Manifest containing metadata about the parse result
  */
@@ -295,11 +319,29 @@ interface Manifest {
     /** Original source file name */
     sourceFileName: string;
     /** Processing completion date */
-    processingDate: Date;
+    /** Processing completion date (optional: only present if emitted by the worker) */
+    processingDate?: Date;
+    /** Worker-side processing metadata emitted by manifest v2 */
+    processing?: ProcessingMetadata;
     /** Statistics */
     statistics: Statistics;
-    /** File index */
-    files: FileIndex;
+    /** Legacy file index from earlier ZIP manifests */
+    files?: FileIndex;
+}
+/**
+ * Chunk relationship entry (metadata.connect_to per schema v2.1)
+ */
+interface ConnectTo {
+    /** Target chunk_id */
+    target: string;
+    /** Relationship type */
+    relation: 'embeds' | 'related';
+    /** Placeholder ref in content, e.g. '[images/a.png]' (embeds only) */
+    ref?: string;
+    /** Semantic similarity score (related only) */
+    score?: number;
+    /** Shared keywords (related only) */
+    keywords?: string[];
 }
 /**
  * Base chunk properties
@@ -313,6 +355,17 @@ interface BaseChunk {
     content: string;
     /** Relative path in ZIP */
     path: string;
+    /** Page numbers spanned by this chunk when provided by the backend */
+    pageNums?: number[];
+}
+/**
+ * Minimal chunk representation emitted in chunks_slim.json
+ */
+interface SlimChunk {
+    type: 'text' | 'image' | 'table';
+    path: string;
+    content: string;
+    summary?: string;
 }
 /**
  * Text chunk
@@ -321,13 +374,18 @@ interface TextChunk extends BaseChunk {
     type: 'text';
     /** Content length */
     length: number;
-    /** Tokens or token count, depending on backend payload */
-    tokens?: number | string[];
+    /** Extracted tokens from the current backend payload */
+    tokens?: string[];
     /** Extracted keywords */
     keywords?: string[];
     /** Generated summary */
     summary?: string;
-    /** Related chunk IDs */
+    /** Chunk relationships (schema v2.1: metadata.connect_to) */
+    connectTo?: ConnectTo[];
+    /**
+     * @deprecated Use connectTo instead. Retained for backward compatibility.
+     * Previously populated from metadata.relationships which is no longer emitted by the API.
+     */
     relationships?: string[];
 }
 /**
@@ -378,10 +436,18 @@ interface ParseResult {
     manifest: Manifest;
     /** All chunks */
     chunks: Chunk[];
+    /** Minimal chunk projection from chunks_slim.json (if available) */
+    chunksSlim?: SlimChunk[];
     /** Full document as Markdown (if available) */
     fullMarkdown?: string;
     /** Document hierarchy (if available) */
     hierarchy?: unknown;
+    /** Table-of-contents hierarchy hints (if available) */
+    tocHierarchies?: unknown;
+    /** Knowledge-base CSV export (if available) */
+    kbCsv?: string;
+    /** Pre-rendered hierarchy HTML view (if available) */
+    hierarchyViewHtml?: string;
     /** Raw ZIP buffer */
     rawZip: Buffer;
     /** Text chunks only */

package/dist/index.d.ts

@@ -216,7 +216,7 @@ interface ParseParams {
     /** Generate table summaries */
     summaryTable?: boolean;
     /** Generate text summaries */
-    summaryText?: boolean;
+    summaryTxt?: boolean;
     /** Custom data identifier */
     dataId?: string;
     /** Additional fragment description */
@@ -282,6 +282,30 @@ interface Statistics {
 interface FileIndex {
     [chunkId: string]: string;
 }
+/**
+ * Processing cost details emitted by manifest v2
+ */
+interface ProcessingCost {
+    microDollars?: number;
+    credits?: number;
+}
+/**
+ * Processing timing details emitted by manifest v2
+ */
+interface ProcessingTiming {
+    startedAt?: Date;
+    completedAt?: Date;
+    durationMs?: number;
+}
+/**
+ * Processing metadata emitted by manifest v2
+ */
+interface ProcessingMetadata {
+    pageCount?: number;
+    billingStatus?: string;
+    cost?: ProcessingCost;
+    timing?: ProcessingTiming;
+}
 /**
  * Manifest containing metadata about the parse result
  */
@@ -295,11 +319,29 @@ interface Manifest {
     /** Original source file name */
     sourceFileName: string;
     /** Processing completion date */
-    processingDate: Date;
+    /** Processing completion date (optional: only present if emitted by the worker) */
+    processingDate?: Date;
+    /** Worker-side processing metadata emitted by manifest v2 */
+    processing?: ProcessingMetadata;
     /** Statistics */
     statistics: Statistics;
-    /** File index */
-    files: FileIndex;
+    /** Legacy file index from earlier ZIP manifests */
+    files?: FileIndex;
+}
+/**
+ * Chunk relationship entry (metadata.connect_to per schema v2.1)
+ */
+interface ConnectTo {
+    /** Target chunk_id */
+    target: string;
+    /** Relationship type */
+    relation: 'embeds' | 'related';
+    /** Placeholder ref in content, e.g. '[images/a.png]' (embeds only) */
+    ref?: string;
+    /** Semantic similarity score (related only) */
+    score?: number;
+    /** Shared keywords (related only) */
+    keywords?: string[];
 }
 /**
  * Base chunk properties
@@ -313,6 +355,17 @@ interface BaseChunk {
     content: string;
     /** Relative path in ZIP */
     path: string;
+    /** Page numbers spanned by this chunk when provided by the backend */
+    pageNums?: number[];
+}
+/**
+ * Minimal chunk representation emitted in chunks_slim.json
+ */
+interface SlimChunk {
+    type: 'text' | 'image' | 'table';
+    path: string;
+    content: string;
+    summary?: string;
 }
 /**
  * Text chunk
@@ -321,13 +374,18 @@ interface TextChunk extends BaseChunk {
     type: 'text';
     /** Content length */
     length: number;
-    /** Tokens or token count, depending on backend payload */
-    tokens?: number | string[];
+    /** Extracted tokens from the current backend payload */
+    tokens?: string[];
     /** Extracted keywords */
     keywords?: string[];
     /** Generated summary */
     summary?: string;
-    /** Related chunk IDs */
+    /** Chunk relationships (schema v2.1: metadata.connect_to) */
+    connectTo?: ConnectTo[];
+    /**
+     * @deprecated Use connectTo instead. Retained for backward compatibility.
+     * Previously populated from metadata.relationships which is no longer emitted by the API.
+     */
     relationships?: string[];
 }
 /**
@@ -378,10 +436,18 @@ interface ParseResult {
     manifest: Manifest;
     /** All chunks */
     chunks: Chunk[];
+    /** Minimal chunk projection from chunks_slim.json (if available) */
+    chunksSlim?: SlimChunk[];
     /** Full document as Markdown (if available) */
     fullMarkdown?: string;
     /** Document hierarchy (if available) */
     hierarchy?: unknown;
+    /** Table-of-contents hierarchy hints (if available) */
+    tocHierarchies?: unknown;
+    /** Knowledge-base CSV export (if available) */
+    kbCsv?: string;
+    /** Pre-rendered hierarchy HTML view (if available) */
+    hierarchyViewHtml?: string;
     /** Raw ZIP buffer */
     rawZip: Buffer;
     /** Text chunks only */

package/dist/index.js

@@ -839,11 +839,39 @@ async function parseResult(httpClient, resultUrl, options) {
     const hierarchyContent = await hierarchyFile.async("string");
     hierarchy = JSON.parse(hierarchyContent);
   }
+  let chunksSlim;
+  const chunksSlimFile = zip.file("chunks_slim.json");
+  if (chunksSlimFile) {
+    const chunksSlimContent = await chunksSlimFile.async("string");
+    let chunksSlimData = JSON.parse(chunksSlimContent);
+    chunksSlimData = keysToCamel(chunksSlimData);
+    chunksSlim = extractSlimChunks(chunksSlimData);
+  }
+  let tocHierarchies;
+  const tocHierarchiesFile = zip.file("toc_hierarchies.json");
+  if (tocHierarchiesFile) {
+    const tocHierarchiesContent = await tocHierarchiesFile.async("string");
+    tocHierarchies = keysToCamel(JSON.parse(tocHierarchiesContent));
+  }
+  let kbCsv;
+  const kbCsvFile = zip.file("kb.csv");
+  if (kbCsvFile) {
+    kbCsv = await kbCsvFile.async("string");
+  }
+  let hierarchyViewHtml;
+  const hierarchyViewFile = zip.file("hierarchy_view.html");
+  if (hierarchyViewFile) {
+    hierarchyViewHtml = await hierarchyViewFile.async("string");
+  }
   const result = {
     manifest,
     chunks,
+    chunksSlim,
     fullMarkdown,
     hierarchy,
+    tocHierarchies,
+    kbCsv,
+    hierarchyViewHtml,
     rawZip: zipBuffer,
     get textChunks() {
       return chunks.filter((c) => c.type === "text");
@@ -867,12 +895,30 @@ async function parseResult(httpClient, resultUrl, options) {
       await import_fs2.promises.mkdir(directory, { recursive: true });
       await import_fs2.promises.writeFile((0, import_path.join)(directory, "manifest.json"), JSON.stringify(manifest, null, 2));
       await import_fs2.promises.writeFile((0, import_path.join)(directory, "chunks.json"), JSON.stringify(chunks, null, 2));
+      if (chunksSlim) {
+        await import_fs2.promises.writeFile(
+          (0, import_path.join)(directory, "chunks_slim.json"),
+          JSON.stringify({ chunks: chunksSlim }, null, 2)
+        );
+      }
       if (fullMarkdown) {
         await import_fs2.promises.writeFile((0, import_path.join)(directory, "full.md"), fullMarkdown);
       }
       if (hierarchy) {
         await import_fs2.promises.writeFile((0, import_path.join)(directory, "hierarchy.json"), JSON.stringify(hierarchy, null, 2));
       }
+      if (tocHierarchies) {
+        await import_fs2.promises.writeFile(
+          (0, import_path.join)(directory, "toc_hierarchies.json"),
+          JSON.stringify(tocHierarchies, null, 2)
+        );
+      }
+      if (kbCsv) {
+        await import_fs2.promises.writeFile((0, import_path.join)(directory, "kb.csv"), kbCsv);
+      }
+      if (hierarchyViewHtml) {
+        await import_fs2.promises.writeFile((0, import_path.join)(directory, "hierarchy_view.html"), hierarchyViewHtml);
+      }
       for (const imageChunk of this.imageChunks) {
         await imageChunk.save(directory);
       }
@@ -894,6 +940,15 @@ function extractChunks(payload) {
   }
   return [];
 }
+function extractSlimChunks(payload) {
+  if (Array.isArray(payload)) {
+    return payload;
+  }
+  if (Array.isArray(payload.chunks)) {
+    return payload.chunks;
+  }
+  return [];
+}
 function getChunkMetadata(chunkData) {
   if (!chunkData.metadata) {
     return {};
@@ -904,18 +959,38 @@ function getChunkFilePath(chunkData) {
   const metadata = getChunkMetadata(chunkData);
   return chunkData.filePath ?? metadata.filePath ?? chunkData.path;
 }
+function normalizePageNums(pageNums) {
+  if (!Array.isArray(pageNums)) {
+    return void 0;
+  }
+  const normalized = pageNums.filter((pageNum) => typeof pageNum === "number");
+  return normalized.length > 0 ? normalized : void 0;
+}
+function normalizeTokens(tokens) {
+  if (!Array.isArray(tokens)) {
+    return void 0;
+  }
+  if (!tokens.every((token) => typeof token === "string")) {
+    return void 0;
+  }
+  return tokens;
+}
 function normalizeTextChunk(chunkData) {
   const metadata = getChunkMetadata(chunkData);
+  const connectTo = metadata.connectTo ?? chunkData.connectTo;
+  const relationships = metadata.relationships ?? chunkData.relationships;
   return {
     chunkId: chunkData.chunkId ?? "",
     type: "text",
     content: chunkData.content ?? "",
     path: chunkData.path ?? "",
+    pageNums: normalizePageNums(metadata.pageNums ?? chunkData.pageNums),
     length: metadata.length ?? chunkData.length ?? 0,
-    tokens: metadata.tokens ?? chunkData.tokens,
+    tokens: normalizeTokens(metadata.tokens ?? chunkData.tokens),
     keywords: metadata.keywords ?? chunkData.keywords,
     summary: metadata.summary ?? chunkData.summary,
-    relationships: metadata.relationships ?? chunkData.relationships
+    ...connectTo !== void 0 && { connectTo },
+    ...relationships !== void 0 && { relationships }
   };
 }
 async function processChunk(zip, chunkData) {
@@ -939,6 +1014,7 @@ async function processChunk(zip, chunkData) {
       type: "image",
       content: chunkData.content ?? "",
       path: chunkData.path ?? "",
+      pageNums: normalizePageNums(metadata.pageNums ?? chunkData.pageNums),
       length: metadata.length ?? chunkData.length ?? 0,
       filePath,
       summary: metadata.summary ?? chunkData.summary,
@@ -973,6 +1049,7 @@ async function processChunk(zip, chunkData) {
       type: "table",
       content: chunkData.content ?? "",
       path: chunkData.path ?? "",
+      pageNums: normalizePageNums(metadata.pageNums ?? chunkData.pageNums),
       length: metadata.length ?? chunkData.length ?? 0,
       filePath,
       tableType: metadata.tableType ?? chunkData.tableType,
@@ -1181,7 +1258,7 @@ var Knowhere = class {
       smartTitleParse: params.smartTitleParse,
       summaryImage: params.summaryImage,
       summaryTable: params.summaryTable,
-      summaryTxt: params.summaryText,
+      summaryTxt: params.summaryTxt,
       addFragDesc: params.addFragDesc,
       kbDir: params.kbDir
     };

package/dist/index.mjs

@@ -781,11 +781,39 @@ async function parseResult(httpClient, resultUrl, options) {
     const hierarchyContent = await hierarchyFile.async("string");
     hierarchy = JSON.parse(hierarchyContent);
   }
+  let chunksSlim;
+  const chunksSlimFile = zip.file("chunks_slim.json");
+  if (chunksSlimFile) {
+    const chunksSlimContent = await chunksSlimFile.async("string");
+    let chunksSlimData = JSON.parse(chunksSlimContent);
+    chunksSlimData = keysToCamel(chunksSlimData);
+    chunksSlim = extractSlimChunks(chunksSlimData);
+  }
+  let tocHierarchies;
+  const tocHierarchiesFile = zip.file("toc_hierarchies.json");
+  if (tocHierarchiesFile) {
+    const tocHierarchiesContent = await tocHierarchiesFile.async("string");
+    tocHierarchies = keysToCamel(JSON.parse(tocHierarchiesContent));
+  }
+  let kbCsv;
+  const kbCsvFile = zip.file("kb.csv");
+  if (kbCsvFile) {
+    kbCsv = await kbCsvFile.async("string");
+  }
+  let hierarchyViewHtml;
+  const hierarchyViewFile = zip.file("hierarchy_view.html");
+  if (hierarchyViewFile) {
+    hierarchyViewHtml = await hierarchyViewFile.async("string");
+  }
   const result = {
     manifest,
     chunks,
+    chunksSlim,
     fullMarkdown,
     hierarchy,
+    tocHierarchies,
+    kbCsv,
+    hierarchyViewHtml,
     rawZip: zipBuffer,
     get textChunks() {
       return chunks.filter((c) => c.type === "text");
@@ -809,12 +837,30 @@ async function parseResult(httpClient, resultUrl, options) {
       await fs2.mkdir(directory, { recursive: true });
       await fs2.writeFile(join(directory, "manifest.json"), JSON.stringify(manifest, null, 2));
       await fs2.writeFile(join(directory, "chunks.json"), JSON.stringify(chunks, null, 2));
+      if (chunksSlim) {
+        await fs2.writeFile(
+          join(directory, "chunks_slim.json"),
+          JSON.stringify({ chunks: chunksSlim }, null, 2)
+        );
+      }
       if (fullMarkdown) {
         await fs2.writeFile(join(directory, "full.md"), fullMarkdown);
       }
       if (hierarchy) {
         await fs2.writeFile(join(directory, "hierarchy.json"), JSON.stringify(hierarchy, null, 2));
       }
+      if (tocHierarchies) {
+        await fs2.writeFile(
+          join(directory, "toc_hierarchies.json"),
+          JSON.stringify(tocHierarchies, null, 2)
+        );
+      }
+      if (kbCsv) {
+        await fs2.writeFile(join(directory, "kb.csv"), kbCsv);
+      }
+      if (hierarchyViewHtml) {
+        await fs2.writeFile(join(directory, "hierarchy_view.html"), hierarchyViewHtml);
+      }
       for (const imageChunk of this.imageChunks) {
         await imageChunk.save(directory);
       }
@@ -836,6 +882,15 @@ function extractChunks(payload) {
   }
   return [];
 }
+function extractSlimChunks(payload) {
+  if (Array.isArray(payload)) {
+    return payload;
+  }
+  if (Array.isArray(payload.chunks)) {
+    return payload.chunks;
+  }
+  return [];
+}
 function getChunkMetadata(chunkData) {
   if (!chunkData.metadata) {
     return {};
@@ -846,18 +901,38 @@ function getChunkFilePath(chunkData) {
   const metadata = getChunkMetadata(chunkData);
   return chunkData.filePath ?? metadata.filePath ?? chunkData.path;
 }
+function normalizePageNums(pageNums) {
+  if (!Array.isArray(pageNums)) {
+    return void 0;
+  }
+  const normalized = pageNums.filter((pageNum) => typeof pageNum === "number");
+  return normalized.length > 0 ? normalized : void 0;
+}
+function normalizeTokens(tokens) {
+  if (!Array.isArray(tokens)) {
+    return void 0;
+  }
+  if (!tokens.every((token) => typeof token === "string")) {
+    return void 0;
+  }
+  return tokens;
+}
 function normalizeTextChunk(chunkData) {
   const metadata = getChunkMetadata(chunkData);
+  const connectTo = metadata.connectTo ?? chunkData.connectTo;
+  const relationships = metadata.relationships ?? chunkData.relationships;
   return {
     chunkId: chunkData.chunkId ?? "",
     type: "text",
     content: chunkData.content ?? "",
     path: chunkData.path ?? "",
+    pageNums: normalizePageNums(metadata.pageNums ?? chunkData.pageNums),
     length: metadata.length ?? chunkData.length ?? 0,
-    tokens: metadata.tokens ?? chunkData.tokens,
+    tokens: normalizeTokens(metadata.tokens ?? chunkData.tokens),
     keywords: metadata.keywords ?? chunkData.keywords,
     summary: metadata.summary ?? chunkData.summary,
-    relationships: metadata.relationships ?? chunkData.relationships
+    ...connectTo !== void 0 && { connectTo },
+    ...relationships !== void 0 && { relationships }
   };
 }
 async function processChunk(zip, chunkData) {
@@ -881,6 +956,7 @@ async function processChunk(zip, chunkData) {
       type: "image",
       content: chunkData.content ?? "",
       path: chunkData.path ?? "",
+      pageNums: normalizePageNums(metadata.pageNums ?? chunkData.pageNums),
       length: metadata.length ?? chunkData.length ?? 0,
       filePath,
       summary: metadata.summary ?? chunkData.summary,
@@ -915,6 +991,7 @@ async function processChunk(zip, chunkData) {
       type: "table",
       content: chunkData.content ?? "",
       path: chunkData.path ?? "",
+      pageNums: normalizePageNums(metadata.pageNums ?? chunkData.pageNums),
       length: metadata.length ?? chunkData.length ?? 0,
       filePath,
       tableType: metadata.tableType ?? chunkData.tableType,
@@ -1123,7 +1200,7 @@ var Knowhere = class {
       smartTitleParse: params.smartTitleParse,
       summaryImage: params.summaryImage,
       summaryTable: params.summaryTable,
-      summaryTxt: params.summaryText,
+      summaryTxt: params.summaryTxt,
       addFragDesc: params.addFragDesc,
       kbDir: params.kbDir
     };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ontos-ai/knowhere-sdk",
-  "version": "0.2.0",
+  "version": "0.2.1",
   "description": "Official Node.js SDK for Knowhere document parsing API",
   "keywords": [
     "knowhere",