@google-cloud/vertexai - npm Package Compare versions

Comparing version 1.7.0 to 1.8.0

.release-please-manifest.json

 {
-  ".": "1.7.0"
+  ".": "1.8.0"
 }

build/src/functions/generate_content.js

@@ -40,2 +40,3 @@ "use strict";
         systemInstruction: request.systemInstruction,
+        cachedContent: request.cachedContent,
         generationConfig: (_a = request.generationConfig) !== null && _a !== void 0 ? _a : generationConfig,
@@ -81,2 +82,3 @@ safetySettings: (_b = request.safetySettings) !== null && _b !== void 0 ? _b : safetySettings,
         systemInstruction: request.systemInstruction,
+        cachedContent: request.cachedContent,
         generationConfig: (_a = request.generationConfig) !== null && _a !== void 0 ? _a : generationConfig,
@@ -83,0 +85,0 @@ safetySettings: (_b = request.safetySettings) !== null && _b !== void 0 ? _b : safetySettings,

build/src/functions/pre_fetch_processing.js

@@ -63,3 +63,5 @@ "use strict";
 function getApiVersion(request) {
-    return hasVertexRagStore(request) ? 'v1beta1' : 'v1';
+    return hasVertexRagStore(request) || hasCachedContent(request)
+        ? 'v1beta1'
+        : 'v1';
 }
@@ -80,2 +82,5 @@ exports.getApiVersion = getApiVersion;
 exports.hasVertexRagStore = hasVertexRagStore;
+function hasCachedContent(request) {
+    return !!request.cachedContent;
+}
 function hasVertexAISearch(request) {
@@ -82,0 +87,0 @@ var _a;

build/src/models/chat_session.js

@@ -20,3 +20,3 @@ "use strict";
 exports.ChatSessionPreview = exports.ChatSession = void 0;
-const util_1 = require("./util");
+const util_1 = require("../functions/util");
 const generate_content_1 = require("../functions/generate_content");
@@ -23,0 +23,0 @@ const errors_1 = require("../types/errors");

build/src/models/generative_models.d.ts

@@ -17,3 +17,3 @@ /**
  */
-import { CountTokensRequest, CountTokensResponse, GenerateContentRequest, GenerateContentResult, GetGenerativeModelParams, StartChatParams, StreamGenerateContentResult } from '../types/content';
+import { CachedContent, Content, CountTokensRequest, CountTokensResponse, GenerateContentRequest, GenerateContentResult, GetGenerativeModelParams, StartChatParams, StreamGenerateContentResult } from '../types/content';
 import { ChatSession, ChatSessionPreview } from './chat_session';
@@ -160,2 +160,3 @@ /**
     private readonly apiEndpoint?;
+    private readonly cachedContent?;
     /**
@@ -259,2 +260,5 @@ * @constructor
     startChat(request?: StartChatParams): ChatSessionPreview;
+    getModelName(): string;
+    getCachedContent(): CachedContent | undefined;
+    getSystemInstruction(): Content | undefined;
 }

build/src/models/generative_models.js

@@ -20,3 +20,3 @@ "use strict";
 exports.GenerativeModelPreview = exports.GenerativeModel = void 0;
-const util_1 = require("./util");
+const util_1 = require("../functions/util");
 const count_tokens_1 = require("../functions/count_tokens");
@@ -214,2 +214,3 @@ const generate_content_1 = require("../functions/generate_content");
         this.toolConfig = getGenerativeModelParams.toolConfig;
+        this.cachedContent = getGenerativeModelParams.cachedContent;
         this.requestOptions = (_a = getGenerativeModelParams.requestOptions) !== null && _a !== void 0 ? _a : {};
@@ -252,4 +253,8 @@ if (getGenerativeModelParams.systemInstruction) {
     async generateContent(request) {
+        var _a;
         request = formulateRequestToGenerateContentRequest(request);
-        const formulatedRequest = formulateSystemInstructionIntoGenerateContentRequest(request, this.systemInstruction);
+        const formulatedRequest = {
+            ...formulateSystemInstructionIntoGenerateContentRequest(request, this.systemInstruction),
+            cachedContent: (_a = this.cachedContent) === null || _a === void 0 ? void 0 : _a.name,
+        };
         return (0, generate_content_1.generateContent)(this.location, this.resourcePath, this.fetchToken(), formulatedRequest, this.apiEndpoint, this.generationConfig, this.safetySettings, this.tools, this.toolConfig, this.requestOptions);
@@ -282,4 +287,8 @@ }
     async generateContentStream(request) {
+        var _a;
         request = formulateRequestToGenerateContentRequest(request);
-        const formulatedRequest = formulateSystemInstructionIntoGenerateContentRequest(request, this.systemInstruction);
+        const formulatedRequest = {
+            ...formulateSystemInstructionIntoGenerateContentRequest(request, this.systemInstruction),
+            cachedContent: (_a = this.cachedContent) === null || _a === void 0 ? void 0 : _a.name,
+        };
         return (0, generate_content_1.generateContentStream)(this.location, this.resourcePath, this.fetchToken(), formulatedRequest, this.apiEndpoint, this.generationConfig, this.safetySettings, this.tools, this.toolConfig, this.requestOptions);
@@ -333,3 +342,3 @@ }
     startChat(request) {
-        var _a, _b, _c, _d;
+        var _a, _b, _c, _d, _e, _f, _g;
         const startChatRequest = {
@@ -343,2 +352,3 @@ project: this.project,
             systemInstruction: this.systemInstruction,
+            cachedContent: (_a = this.cachedContent) === null || _a === void 0 ? void 0 : _a.name,
         };
@@ -348,11 +358,22 @@ if (request) {
             startChatRequest.generationConfig =
-                (_a = request.generationConfig) !== null && _a !== void 0 ? _a : this.generationConfig;
+                (_b = request.generationConfig) !== null && _b !== void 0 ? _b : this.generationConfig;
             startChatRequest.safetySettings =
-                (_b = request.safetySettings) !== null && _b !== void 0 ? _b : this.safetySettings;
-            startChatRequest.tools = (_c = request.tools) !== null && _c !== void 0 ? _c : this.tools;
+                (_c = request.safetySettings) !== null && _c !== void 0 ? _c : this.safetySettings;
+            startChatRequest.tools = (_d = request.tools) !== null && _d !== void 0 ? _d : this.tools;
             startChatRequest.systemInstruction =
-                (_d = request.systemInstruction) !== null && _d !== void 0 ? _d : this.systemInstruction;
+                (_e = request.systemInstruction) !== null && _e !== void 0 ? _e : this.systemInstruction;
+            startChatRequest.cachedContent =
+                (_f = request.cachedContent) !== null && _f !== void 0 ? _f : (_g = this.cachedContent) === null || _g === void 0 ? void 0 : _g.name;
         }
         return new chat_session_1.ChatSessionPreview(startChatRequest, this.requestOptions);
     }
+    getModelName() {
+        return this.model;
+    }
+    getCachedContent() {
+        return this.cachedContent;
+    }
+    getSystemInstruction() {
+        return this.systemInstruction;
+    }
 }
@@ -359,0 +380,0 @@ exports.GenerativeModelPreview = GenerativeModelPreview;

build/src/types/content.d.ts

@@ -61,2 +61,7 @@ /**
     systemInstruction?: string | Content;
+    /**
+     * Optional. The name of the cached content used as context to serve the prediction.
+     * This is the name of a `CachedContent` and not the cache object itself.
+     */
+    cachedContent?: string;
 }
@@ -132,2 +137,8 @@ /**
     model: string;
+    /**
+     * Optional. The cached content used as context to serve the prediction.
+     * Note: only used in explicit caching, where users can have control over caching
+     * (e.g. what content to cache) and enjoy guaranteed cost savings.
+     */
+    cachedContent?: CachedContent;
 }
@@ -957,2 +968,7 @@ /**
     systemInstruction?: string | Content;
+    /**
+     * Optional. The name of the cached content used as context to serve the prediction.
+     * This is the name of a `CachedContent` and not the cache object itself.
+     */
+    cachedContent?: string;
 }
@@ -996,1 +1012,70 @@ /**
 }
+/**
+ * A resource used in LLM queries for users to explicitly specify
+ * what to cache and how to cache.
+ */
+export interface CachedContent {
+    /**
+     * Immutable. Identifier. The server-generated resource name of the cached content.
+     * Format: projects/{project}/locations/{location}/cachedContents/{cached_content}
+     */
+    name?: string;
+    /** Optional. Immutable. The user-generated meaningful display name of the cached content. */
+    displayName?: string;
+    /**
+     * Immutable. The name of the publisher model to use for cached content.
+     * Format: projects/{project}/locations/{location}/publishers/{publisher}/models/{model}
+     */
+    model?: string;
+    /** Developer set system instruction. Currently, text only. */
+    systemInstruction?: Content | string;
+    /** Optional. Input only. Immutable. The content to cache. */
+    contents?: Content[];
+    /** Optional. Input only. Immutable. A list of `Tools` the model may use to generate the next response. */
+    tools?: Tool[];
+    /** Optional. Input only. Immutable. Tool config. This config is shared for all tools. */
+    toolConfig?: ToolConfig;
+    /**
+     * Output only. Creatation time of the cache entry.
+     * Format: google-datetime. See {@link https://cloud.google.com/docs/discovery/type-format}
+     */
+    createTime?: string;
+    /**
+     * Output only. When the cache entry was last updated in UTC time.
+     * Format: google-datetime. See {@link https://cloud.google.com/docs/discovery/type-format}
+     */
+    updateTime?: string;
+    /** Output only. Metadata on the usage of the cached content. */
+    usageMetadata?: CachedContentUsageMetadata;
+    /**
+     * Timestamp of when this resource is considered expired.
+     * This is *always* provided on output, regardless of what was sent on input.
+     */
+    expireTime?: string;
+    /**
+     * Input only. The TTL seconds for this resource. The expiration time
+     * is computed: now + TTL.
+     * Format: google-duration. See {@link https://cloud.google.com/docs/discovery/type-format}
+     */
+    ttl?: string;
+}
+/** Metadata on the usage of the cached content. */
+export interface CachedContentUsageMetadata {
+    /** Total number of tokens that the cached content consumes. */
+    totalTokenCount?: number;
+    /** Number of text characters. */
+    textCount?: number;
+    /** Number of images. */
+    imageCount?: number;
+    /** Duration of video in seconds. */
+    videoDurationSeconds?: number;
+    /** Duration of audio in seconds. */
+    audioDurationSeconds?: number;
+}
+/** Response with a list of CachedContents. */
+export interface ListCachedContentsResponse {
+    /** List of cached contents. */
+    cachedContents?: CachedContent[];
+    /** A token, which can be sent as `page_token` to retrieve the next page. If this field is omitted, there are no subsequent pages. */
+    nextPageToken?: string;
+}

build/src/util/constants.d.ts

@@ -23,3 +23,3 @@ /**
 export declare const SYSTEM_ROLE = "system";
-export declare const USER_AGENT = "model-builder/1.7.0 grpc-node/1.7.0";
+export declare const USER_AGENT = "model-builder/1.8.0 grpc-node/1.8.0";
 export declare const CREDENTIAL_ERROR_MESSAGE = "\nUnable to authenticate your request        \nDepending on your run time environment, you can get authentication by        \n- if in local instance or cloud shell: `!gcloud auth login`        \n- if in Colab:        \n    -`from google.colab import auth`        \n    -`auth.authenticate_user()`        \n- if in service account or other: please follow guidance in https://cloud.google.com/docs/authentication";

build/src/util/constants.js

@@ -27,3 +27,3 @@ "use strict";
 const USER_AGENT_PRODUCT = 'model-builder';
-const CLIENT_LIBRARY_VERSION = '1.7.0'; // x-release-please-version
+const CLIENT_LIBRARY_VERSION = '1.8.0'; // x-release-please-version
 const CLIENT_LIBRARY_LANGUAGE = `grpc-node/${CLIENT_LIBRARY_VERSION}`;
@@ -30,0 +30,0 @@ exports.USER_AGENT = `${USER_AGENT_PRODUCT}/${CLIENT_LIBRARY_VERSION} ${CLIENT_LIBRARY_LANGUAGE}`;

build/src/vertex_ai.d.ts

@@ -19,3 +19,4 @@ /**
 import { GenerativeModelPreview, GenerativeModel } from './models';
-import { ModelParams, RequestOptions, VertexInit } from './types/content';
+import { CachedContent, ModelParams, RequestOptions, VertexInit } from './types/content';
+import * as Resources from './resources';
 /**
@@ -100,2 +101,4 @@ * The `VertexAI` class is the base class for authenticating to Vertex AI.
     private readonly apiEndpoint?;
+    private readonly apiClient;
+    readonly cachedContents: Resources.CachedContents;
     /**
@@ -123,3 +126,4 @@ * @constructor
     getGenerativeModel(modelParams: ModelParams, requestOptions?: RequestOptions): GenerativeModelPreview;
+    getGenerativeModelFromCachedContent(cachedContent: CachedContent, modelParams?: Partial<ModelParams>, requestOptions?: RequestOptions): GenerativeModelPreview;
 }
 export {};

build/src/vertex_ai.js

@@ -24,2 +24,4 @@ "use strict";
 const errors_1 = require("./types/errors");
+const Resources = require("./resources");
+const cached_contents_1 = require("./resources/cached_contents");
 /**
@@ -142,2 +144,4 @@ * The `VertexAI` class is the base class for authenticating to Vertex AI.
         this.apiEndpoint = apiEndpoint;
+        this.apiClient = new Resources.ApiClient(this.project, this.location, 'v1beta1', this.googleAuth);
+        this.cachedContents = new Resources.CachedContents(this.apiClient);
     }
@@ -165,3 +169,63 @@ /**
     }
+    getGenerativeModelFromCachedContent(cachedContent, modelParams, requestOptions) {
+        if (!cachedContent.name) {
+            throw new errors_1.ClientError('Cached content must contain a `name` field.');
+        }
+        if (!cachedContent.model) {
+            throw new errors_1.ClientError('Cached content must contain a `model` field.');
+        }
+        validateCachedContentModel(cachedContent.model);
+        /**
+         * Not checking tools and toolConfig for now as it would require a deep
+         * equality comparison and isn't likely to be a common case.
+         */
+        const disallowedDuplicates = ['model', 'systemInstruction'];
+        for (const key of disallowedDuplicates) {
+            if ((modelParams === null || modelParams === void 0 ? void 0 : modelParams[key]) &&
+                cachedContent[key] &&
+                (modelParams === null || modelParams === void 0 ? void 0 : modelParams[key]) !== cachedContent[key]) {
+                if (key === 'model') {
+                    const modelParamsComp = parseModelName(modelParams[key]);
+                    const cachedContentComp = parseModelName(cachedContent[key]);
+                    if (modelParamsComp === cachedContentComp) {
+                        continue;
+                    }
+                }
+                throw new errors_1.ClientError(`Different value for "${key}" specified in modelParams` +
+                    ` (${modelParams[key]}) and cachedContent (${cachedContent[key]})`);
+            }
+        }
+        cachedContent.name = (0, cached_contents_1.inferFullResourceName)(this.project, this.location, cachedContent.name);
+        const modelParamsFromCache = {
+            model: cachedContent.model,
+            project: this.project,
+            location: this.location,
+            googleAuth: this.googleAuth,
+            apiEndpoint: this.apiEndpoint,
+            safetySettings: modelParams === null || modelParams === void 0 ? void 0 : modelParams.safetySettings,
+            generationConfig: modelParams === null || modelParams === void 0 ? void 0 : modelParams.generationConfig,
+            tools: cachedContent.tools,
+            toolConfig: cachedContent.toolConfig,
+            requestOptions: requestOptions,
+            systemInstruction: cachedContent.systemInstruction,
+            cachedContent,
+        };
+        return new models_1.GenerativeModelPreview(modelParamsFromCache);
+    }
 }
+function validateCachedContentModel(modelName) {
+    if (modelName.startsWith('models/') ||
+        (modelName.startsWith('projects/') &&
+            modelName.includes('/publishers/google/models/')) ||
+        !modelName.includes('/')) {
+        return;
+    }
+    throw new errors_1.ClientError(`Cached content model name must start with "models/" or match "projects/.*/publishers/google/models/.*" or is a model name listed at https://cloud.google.com/vertex-ai/generative-ai/docs/learn/model-versions. Received: ${modelName}`);
+}
+function parseModelName(modelName) {
+    if (!modelName.includes('/')) {
+        return modelName;
+    }
+    return modelName.split('/').pop();
+}
 function validateGoogleAuthOptions(project, googleAuthOptions) {
@@ -168,0 +232,0 @@ let opts;

CHANGELOG.md

 # Changelog
 
+## [1.8.0](https://github.com/googleapis/nodejs-vertexai/compare/v1.7.0...v1.8.0) (2024-09-18)
+
+
+### Features
+
+* Add CachedContent resource to Vertex AI client library. ([8c8963e](https://github.com/googleapis/nodejs-vertexai/commit/8c8963e5c62bf491d5a47f7c5a0db64fafaea0cd))
+* Implement cached_content with generateContent methods ([c604b8c](https://github.com/googleapis/nodejs-vertexai/commit/c604b8caf4138537b38bdf9f57e8086d55216981))
+
 ## [1.7.0](https://github.com/googleapis/nodejs-vertexai/compare/v1.6.0...v1.7.0) (2024-08-30)
@@ -4,0 +12,0 @@

package.json

 {
   "name": "@google-cloud/vertexai",
   "description": "Vertex Generative AI client for Node.js",
-  "version": "1.7.0",
+  "version": "1.8.0",
   "license": "Apache-2.0",
@@ -6,0 +6,0 @@ "author": "Google LLC",

src/functions/generate_content.ts

@@ -75,2 +75,3 @@ /**
     systemInstruction: request.systemInstruction,
+    cachedContent: request.cachedContent,
     generationConfig: request.generationConfig ?? generationConfig,
@@ -130,2 +131,3 @@ safetySettings: request.safetySettings ?? safetySettings,
     systemInstruction: request.systemInstruction,
+    cachedContent: request.cachedContent,
     generationConfig: request.generationConfig ?? generationConfig,
@@ -132,0 +134,0 @@ safetySettings: request.safetySettings ?? safetySettings,

src/functions/pre_fetch_processing.ts

@@ -83,3 +83,5 @@ /**
 ): 'v1' | 'v1beta1' {
-  return hasVertexRagStore(request) ? 'v1beta1' : 'v1';
+  return hasVertexRagStore(request) || hasCachedContent(request)
+    ? 'v1beta1'
+    : 'v1';
 }
@@ -98,2 +100,6 @@
 
+function hasCachedContent(request: GenerateContentRequest): boolean {
+  return !!request.cachedContent;
+}
+
 export function hasVertexAISearch(request: GenerateContentRequest): boolean {
@@ -100,0 +106,0 @@ for (const tool of request?.tools ?? []) {

src/models/chat_session.ts

@@ -21,3 +21,3 @@ /**
 
-import {formulateSystemInstructionIntoContent} from './util';
+import {formulateSystemInstructionIntoContent} from '../functions/util';
 import {
@@ -24,0 +24,0 @@ generateContent,

src/models/generative_models.ts

@@ -21,3 +21,3 @@ /**
 
-import {formulateSystemInstructionIntoContent} from './util';
+import {formulateSystemInstructionIntoContent} from '../functions/util';
 import {countTokens} from '../functions/count_tokens';
@@ -29,2 +29,3 @@ import {
 import {
+  CachedContent,
   Content,
@@ -300,2 +301,3 @@ CountTokensRequest,
   private readonly apiEndpoint?: string;
+  private readonly cachedContent?: CachedContent;
 
@@ -316,2 +318,3 @@ /**
     this.toolConfig = getGenerativeModelParams.toolConfig;
+    this.cachedContent = getGenerativeModelParams.cachedContent;
     this.requestOptions = getGenerativeModelParams.requestOptions ?? {};
@@ -365,7 +368,9 @@ if (getGenerativeModelParams.systemInstruction) {
     request = formulateRequestToGenerateContentRequest(request);
-    const formulatedRequest =
-      formulateSystemInstructionIntoGenerateContentRequest(
+    const formulatedRequest = {
+      ...formulateSystemInstructionIntoGenerateContentRequest(
         request,
         this.systemInstruction
-      );
+      ),
+      cachedContent: this.cachedContent?.name,
+    };
     return generateContent(
@@ -413,7 +418,9 @@ this.location,
     request = formulateRequestToGenerateContentRequest(request);
-    const formulatedRequest =
-      formulateSystemInstructionIntoGenerateContentRequest(
+    const formulatedRequest = {
+      ...formulateSystemInstructionIntoGenerateContentRequest(
         request,
         this.systemInstruction
-      );
+      ),
+      cachedContent: this.cachedContent?.name,
+    };
     return generateContentStream(
@@ -495,2 +502,3 @@ this.location,
       systemInstruction: this.systemInstruction,
+      cachedContent: this.cachedContent?.name,
     };
@@ -507,5 +515,19 @@
         request.systemInstruction ?? this.systemInstruction;
+      startChatRequest.cachedContent =
+        request.cachedContent ?? this.cachedContent?.name;
     }
     return new ChatSessionPreview(startChatRequest, this.requestOptions);
   }
+
+  getModelName(): string {
+    return this.model;
+  }
+
+  getCachedContent(): CachedContent | undefined {
+    return this.cachedContent;
+  }
+
+  getSystemInstruction(): Content | undefined {
+    return this.systemInstruction;
+  }
 }
@@ -512,0 +534,0 @@

src/types/content.ts

@@ -65,2 +65,8 @@ /**
   systemInstruction?: string | Content;
+
+  /**
+   * Optional. The name of the cached content used as context to serve the prediction.
+   * This is the name of a `CachedContent` and not the cache object itself.
+   */
+  cachedContent?: string;
 }
@@ -140,2 +146,9 @@
   model: string;
+
+  /**
+   * Optional. The cached content used as context to serve the prediction.
+   * Note: only used in explicit caching, where users can have control over caching
+   * (e.g. what content to cache) and enjoy guaranteed cost savings.
+   */
+  cachedContent?: CachedContent;
 }
@@ -1026,2 +1039,7 @@
   systemInstruction?: string | Content;
+  /**
+   * Optional. The name of the cached content used as context to serve the prediction.
+   * This is the name of a `CachedContent` and not the cache object itself.
+   */
+  cachedContent?: string;
 }
@@ -1067,1 +1085,88 @@
 }
+
+/**
+ * A resource used in LLM queries for users to explicitly specify
+ * what to cache and how to cache.
+ */
+export interface CachedContent {
+  /**
+   * Immutable. Identifier. The server-generated resource name of the cached content.
+   * Format: projects/{project}/locations/{location}/cachedContents/{cached_content}
+   */
+  name?: string;
+
+  /** Optional. Immutable. The user-generated meaningful display name of the cached content. */
+  displayName?: string;
+
+  /**
+   * Immutable. The name of the publisher model to use for cached content.
+   * Format: projects/{project}/locations/{location}/publishers/{publisher}/models/{model}
+   */
+  model?: string;
+
+  /** Developer set system instruction. Currently, text only. */
+  systemInstruction?: Content | string;
+
+  /** Optional. Input only. Immutable. The content to cache. */
+  contents?: Content[];
+
+  /** Optional. Input only. Immutable. A list of `Tools` the model may use to generate the next response. */
+  tools?: Tool[];
+
+  /** Optional. Input only. Immutable. Tool config. This config is shared for all tools. */
+  toolConfig?: ToolConfig;
+
+  /**
+   * Output only. Creatation time of the cache entry.
+   * Format: google-datetime. See {@link https://cloud.google.com/docs/discovery/type-format}
+   */
+  createTime?: string;
+
+  /**
+   * Output only. When the cache entry was last updated in UTC time.
+   * Format: google-datetime. See {@link https://cloud.google.com/docs/discovery/type-format}
+   */
+  updateTime?: string;
+
+  /** Output only. Metadata on the usage of the cached content. */
+  usageMetadata?: CachedContentUsageMetadata;
+
+  /**
+   * Timestamp of when this resource is considered expired.
+   * This is *always* provided on output, regardless of what was sent on input.
+   */
+  expireTime?: string;
+
+  /**
+   * Input only. The TTL seconds for this resource. The expiration time
+   * is computed: now + TTL.
+   * Format: google-duration. See {@link https://cloud.google.com/docs/discovery/type-format}
+   */
+  ttl?: string;
+}
+
+/** Metadata on the usage of the cached content. */
+export interface CachedContentUsageMetadata {
+  /** Total number of tokens that the cached content consumes. */
+  totalTokenCount?: number;
+
+  /** Number of text characters. */
+  textCount?: number;
+
+  /** Number of images. */
+  imageCount?: number;
+
+  /** Duration of video in seconds. */
+  videoDurationSeconds?: number;
+
+  /** Duration of audio in seconds. */
+  audioDurationSeconds?: number;
+}
+
+/** Response with a list of CachedContents. */
+export interface ListCachedContentsResponse {
+  /** List of cached contents. */
+  cachedContents?: CachedContent[];
+  /** A token, which can be sent as `page_token` to retrieve the next page. If this field is omitted, there are no subsequent pages. */
+  nextPageToken?: string;
+}

src/util/constants.ts

@@ -24,3 +24,3 @@ /**
 const USER_AGENT_PRODUCT = 'model-builder';
-const CLIENT_LIBRARY_VERSION = '1.7.0'; // x-release-please-version
+const CLIENT_LIBRARY_VERSION = '1.8.0'; // x-release-please-version
 const CLIENT_LIBRARY_LANGUAGE = `grpc-node/${CLIENT_LIBRARY_VERSION}`;
@@ -27,0 +27,0 @@ export const USER_AGENT = `${USER_AGENT_PRODUCT}/${CLIENT_LIBRARY_VERSION} ${CLIENT_LIBRARY_LANGUAGE}`;

src/vertex_ai.ts

@@ -23,2 +23,3 @@ /**
 import {
+  CachedContent,
   GetGenerativeModelParams,
@@ -29,3 +30,9 @@ ModelParams,
 } from './types/content';
-import {GoogleAuthError, IllegalArgumentError} from './types/errors';
+import {
+  GoogleAuthError,
+  IllegalArgumentError,
+  ClientError,
+} from './types/errors';
+import * as Resources from './resources';
+import {inferFullResourceName} from './resources/cached_contents';
 
@@ -154,2 +161,5 @@ /**
 
+  private readonly apiClient: Resources.ApiClient;
+  readonly cachedContents: Resources.CachedContents;
+
   /**
@@ -180,2 +190,10 @@ * @constructor
     this.apiEndpoint = apiEndpoint;
+
+    this.apiClient = new Resources.ApiClient(
+      this.project,
+      this.location,
+      'v1beta1',
+      this.googleAuth
+    );
+    this.cachedContents = new Resources.CachedContents(this.apiClient);
   }
@@ -207,4 +225,86 @@
   }
+
+  getGenerativeModelFromCachedContent(
+    cachedContent: CachedContent,
+    modelParams?: Partial<ModelParams>,
+    requestOptions?: RequestOptions
+  ) {
+    if (!cachedContent.name) {
+      throw new ClientError('Cached content must contain a `name` field.');
+    }
+    if (!cachedContent.model) {
+      throw new ClientError('Cached content must contain a `model` field.');
+    }
+    validateCachedContentModel(cachedContent.model);
+    /**
+     * Not checking tools and toolConfig for now as it would require a deep
+     * equality comparison and isn't likely to be a common case.
+     */
+    const disallowedDuplicates: Array<keyof ModelParams & keyof CachedContent> =
+      ['model', 'systemInstruction'];
+
+    for (const key of disallowedDuplicates) {
+      if (
+        modelParams?.[key] &&
+        cachedContent[key] &&
+        modelParams?.[key] !== cachedContent[key]
+      ) {
+        if (key === 'model') {
+          const modelParamsComp = parseModelName(modelParams[key]!);
+          const cachedContentComp = parseModelName(cachedContent[key]!);
+          if (modelParamsComp === cachedContentComp) {
+            continue;
+          }
+        }
+        throw new ClientError(
+          `Different value for "${key}" specified in modelParams` +
+            ` (${modelParams[key]}) and cachedContent (${cachedContent[key]})`
+        );
+      }
+    }
+
+    cachedContent.name = inferFullResourceName(
+      this.project,
+      this.location,
+      cachedContent.name
+    );
+    const modelParamsFromCache: GetGenerativeModelParams = {
+      model: cachedContent.model,
+      project: this.project,
+      location: this.location,
+      googleAuth: this.googleAuth,
+      apiEndpoint: this.apiEndpoint,
+      safetySettings: modelParams?.safetySettings,
+      generationConfig: modelParams?.generationConfig,
+      tools: cachedContent.tools,
+      toolConfig: cachedContent.toolConfig,
+      requestOptions: requestOptions,
+      systemInstruction: cachedContent.systemInstruction,
+      cachedContent,
+    };
+    return new GenerativeModelPreview(modelParamsFromCache);
+  }
 }
 
+function validateCachedContentModel(modelName: string) {
+  if (
+    modelName.startsWith('models/') ||
+    (modelName.startsWith('projects/') &&
+      modelName.includes('/publishers/google/models/')) ||
+    !modelName.includes('/')
+  ) {
+    return;
+  }
+  throw new ClientError(
+    `Cached content model name must start with "models/" or match "projects/.*/publishers/google/models/.*" or is a model name listed at https://cloud.google.com/vertex-ai/generative-ai/docs/learn/model-versions. Received: ${modelName}`
+  );
+}
+
+function parseModelName(modelName: string): string {
+  if (!modelName.includes('/')) {
+    return modelName;
+  }
+  return modelName.split('/').pop()!;
+}
+
 function validateGoogleAuthOptions(
@@ -211,0 +311,0 @@ project?: string,

New alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Fixed alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Improved metrics

Total package byte prevSize

1342268

increased by12.2%

Number of package files

197

increased by13.87%

Lines of code

25186

increased by13.02%

Worsened metrics

Number of medium supply chain risk alerts

2

increased by100%

No dependency changes