@mastra/rag 0.0.2-alpha.15 → 0.0.2-alpha.17

package/CHANGELOG.md

@@ -1,5 +1,20 @@
 # @mastra/rag
 
+## 0.0.2-alpha.17
+
+### Patch Changes
+
+- e1dd94a: update the api for embeddings
+- Updated dependencies [e1dd94a]
+  - @mastra/core@0.1.27-alpha.33
+
+## 0.0.2-alpha.16
+
+### Patch Changes
+
+- Updated dependencies [2712098]
+  - @mastra/core@0.1.27-alpha.32
+
 ## 0.0.2-alpha.15
 
 ### Patch Changes

package/dist/document/document.d.ts

@@ -1,7 +1,8 @@
-import { Document } from 'llamaindex';
+import { EmbeddingOptions } from '@mastra/core';
+import { Document as Chunk } from 'llamaindex';
 import { ChunkOptions, ChunkStrategy, ExtractParams } from './types';
 export declare class MastraDocument {
-    private docs;
+    private chunks;
     private type;
     constructor({ docs, type }: {
         docs: {
@@ -28,8 +29,9 @@ export declare class MastraDocument {
         strategy?: ChunkStrategy;
         options?: ChunkOptions;
         extract?: ExtractParams;
-    }): Promise<MastraDocument>;
-    getDocs(): Document[];
+    }): Promise<MastraDocument['chunks']>;
+    embed(chunk: Chunk | string | string[] | Chunk[], options: Omit<EmbeddingOptions, 'value'>): Promise<import("ai").EmbedManyResult<string> | import("ai").EmbedResult<string>>;
+    getDocs(): Chunk[];
     getText(): string[];
     getMetadata(): Record<string, any>[];
 }

package/dist/rag.cjs.development.js

@@ -2,10 +2,10 @@
 
 Object.defineProperty(exports, '__esModule', { value: true });
 
+var core = require('@mastra/core');
 var llamaindex = require('llamaindex');
 var jsdom = require('jsdom');
 var tiktoken = require('@dqbd/tiktoken');
-var core = require('@mastra/core');
 var pg = require('pg');
 var pinecone = require('@pinecone-database/pinecone');
 var astraDbTs = require('@datastax/astra-db-ts');
@@ -1602,9 +1602,9 @@ var MastraDocument = /*#__PURE__*/function () {
   function MastraDocument(_ref) {
     var docs = _ref.docs,
       type = _ref.type;
-    this.docs = void 0;
+    this.chunks = void 0;
     this.type = void 0;
-    this.docs = docs.map(function (d) {
+    this.chunks = docs.map(function (d) {
       return new llamaindex.Document({
         text: d.text,
         metadata: d.metadata
@@ -1638,11 +1638,11 @@ var MastraDocument = /*#__PURE__*/function () {
             });
             _context.next = 9;
             return pipeline.run({
-              documents: this.docs
+              documents: this.chunks
             });
           case 9:
             nodes = _context.sent;
-            this.docs = this.docs.map(function (doc, i) {
+            this.chunks = this.chunks.map(function (doc, i) {
               var _nodes$i;
               return new llamaindex.Document({
                 text: doc.text,
@@ -1777,8 +1777,8 @@ var MastraDocument = /*#__PURE__*/function () {
               break;
             }
             _rt = RecursiveCharacterTransformer.fromLanguage(options.language, options);
-            _textSplit = _rt.transformDocuments(this.docs);
-            this.docs = _textSplit;
+            _textSplit = _rt.transformDocuments(this.chunks);
+            this.chunks = _textSplit;
             return _context3.abrupt("return");
           case 5:
             rt = new RecursiveCharacterTransformer({
@@ -1786,8 +1786,8 @@ var MastraDocument = /*#__PURE__*/function () {
               isSeparatorRegex: options == null ? void 0 : options.isSeparatorRegex,
               options: options
             });
-            textSplit = rt.transformDocuments(this.docs);
-            this.docs = textSplit;
+            textSplit = rt.transformDocuments(this.chunks);
+            this.chunks = textSplit;
           case 8:
           case "end":
             return _context3.stop();
@@ -1810,8 +1810,8 @@ var MastraDocument = /*#__PURE__*/function () {
               isSeparatorRegex: options == null ? void 0 : options.isSeparatorRegex,
               options: options
             });
-            textSplit = rt.transformDocuments(this.docs);
-            this.docs = textSplit;
+            textSplit = rt.transformDocuments(this.chunks);
+            this.chunks = textSplit;
           case 3:
           case "end":
             return _context4.stop();
@@ -1835,8 +1835,8 @@ var MastraDocument = /*#__PURE__*/function () {
               break;
             }
             rt = new HTMLHeaderTransformer(options.headers, options == null ? void 0 : options.returnEachLine);
-            textSplit = rt.transformDocuments(this.docs);
-            this.docs = textSplit;
+            textSplit = rt.transformDocuments(this.chunks);
+            this.chunks = textSplit;
             return _context5.abrupt("return");
           case 5:
             if (!(options != null && (_options$sections = options.sections) != null && _options$sections.length)) {
@@ -1844,8 +1844,8 @@ var MastraDocument = /*#__PURE__*/function () {
               break;
             }
             _rt2 = new HTMLSectionTransformer(options.sections);
-            _textSplit2 = _rt2.transformDocuments(this.docs);
-            this.docs = _textSplit2;
+            _textSplit2 = _rt2.transformDocuments(this.chunks);
+            this.chunks = _textSplit2;
             return _context5.abrupt("return");
           case 10:
             throw new Error('HTML chunking requires either headers or sections to be specified');
@@ -1877,11 +1877,11 @@ var MastraDocument = /*#__PURE__*/function () {
               minChunkSize: options == null ? void 0 : options.minChunkSize
             });
             textSplit = rt.transformDocuments({
-              documents: this.docs,
+              documents: this.chunks,
               ensureAscii: options == null ? void 0 : options.ensureAscii,
               convertLists: options == null ? void 0 : options.convertLists
             });
-            this.docs = textSplit;
+            this.chunks = textSplit;
           case 5:
           case "end":
             return _context6.stop();
@@ -1900,8 +1900,8 @@ var MastraDocument = /*#__PURE__*/function () {
         while (1) switch (_context7.prev = _context7.next) {
           case 0:
             rt = new LatexTransformer(options);
-            textSplit = rt.transformDocuments(this.docs);
-            this.docs = textSplit;
+            textSplit = rt.transformDocuments(this.chunks);
+            this.chunks = textSplit;
           case 3:
           case "end":
             return _context7.stop();
@@ -1924,8 +1924,8 @@ var MastraDocument = /*#__PURE__*/function () {
               encodingName: options == null ? void 0 : options.encodingName,
               modelName: options == null ? void 0 : options.modelName
             });
-            textSplit = rt.transformDocuments(this.docs);
-            this.docs = textSplit;
+            textSplit = rt.transformDocuments(this.chunks);
+            this.chunks = textSplit;
             rt.dispose();
           case 4:
           case "end":
@@ -1949,13 +1949,13 @@ var MastraDocument = /*#__PURE__*/function () {
               break;
             }
             _rt3 = new MarkdownHeaderTransformer(options.headers, options == null ? void 0 : options.returnEachLine, options == null ? void 0 : options.stripHeaders);
-            _textSplit3 = _rt3.transformDocuments(this.docs);
-            this.docs = _textSplit3;
+            _textSplit3 = _rt3.transformDocuments(this.chunks);
+            this.chunks = _textSplit3;
             return _context9.abrupt("return");
           case 5:
             rt = new MarkdownTransformer(options);
-            textSplit = rt.transformDocuments(this.docs);
-            this.docs = textSplit;
+            textSplit = rt.transformDocuments(this.chunks);
+            this.chunks = textSplit;
           case 8:
           case "end":
             return _context9.stop();
@@ -1985,7 +1985,7 @@ var MastraDocument = /*#__PURE__*/function () {
             _context10.next = 6;
             return this.extract(params.extract);
           case 6:
-            return _context10.abrupt("return", this);
+            return _context10.abrupt("return", this.chunks);
           case 7:
           case "end":
             return _context10.stop();
@@ -1997,16 +1997,45 @@ var MastraDocument = /*#__PURE__*/function () {
     }
     return chunk;
   }();
+  _proto.embed = /*#__PURE__*/function () {
+    var _embed2 = /*#__PURE__*/_asyncToGenerator(/*#__PURE__*/_regeneratorRuntime().mark(function _callee11(chunk, options) {
+      var value;
+      return _regeneratorRuntime().wrap(function _callee11$(_context11) {
+        while (1) switch (_context11.prev = _context11.next) {
+          case 0:
+            if (Array.isArray(chunk)) {
+              value = chunk.map(function (chunk) {
+                return typeof chunk === 'string' ? chunk : chunk.getText();
+              });
+            } else if (chunk instanceof llamaindex.Document) {
+              value = chunk.getText();
+            } else {
+              value = chunk;
+            }
+            return _context11.abrupt("return", core.embed(_extends({}, options, {
+              value: value
+            })));
+          case 2:
+          case "end":
+            return _context11.stop();
+        }
+      }, _callee11);
+    }));
+    function embed(_x12, _x13) {
+      return _embed2.apply(this, arguments);
+    }
+    return embed;
+  }();
   _proto.getDocs = function getDocs() {
-    return this.docs;
+    return this.chunks;
   };
   _proto.getText = function getText() {
-    return this.docs.map(function (doc) {
+    return this.chunks.map(function (doc) {
       return doc.text;
     });
   };
   _proto.getMetadata = function getMetadata() {
-    return this.docs.map(function (doc) {
+    return this.chunks.map(function (doc) {
       return doc.metadata;
     });
   };