feat(construct): fixed issues in image summarization (#483)

* feat(construct): fixed issues in image summarization

* chore: self mutation

Signed-off-by: github-actions <[email protected]>

---------

Signed-off-by: github-actions <[email protected]>
Co-authored-by: Dinesh Sajwan <[email protected]>
Co-authored-by: github-actions <[email protected]>

Author: 3 people

apidocs/classes/SummarizationAppsyncStepfn.md

@@ -22,6 +22,8 @@
 - [eventBridgeBus](SummarizationAppsyncStepfn.md#eventbridgebus)
 - [fieldLogLevel](SummarizationAppsyncStepfn.md#fieldloglevel)
 - [graphqlApi](SummarizationAppsyncStepfn.md#graphqlapi)
+- [graphqlApiId](SummarizationAppsyncStepfn.md#graphqlapiid)
+- [graphqlUrl](SummarizationAppsyncStepfn.md#graphqlurl)
 - [inputAssetBucket](SummarizationAppsyncStepfn.md#inputassetbucket)
 - [inputValidationLambdaFunction](SummarizationAppsyncStepfn.md#inputvalidationlambdafunction)
 - [lambdaTracing](SummarizationAppsyncStepfn.md#lambdatracing)
@@ -145,6 +147,22 @@ Returns an instance of appsync.CfnGraphQLApi for summary created by the construc
 
 ___
 
+### graphqlApiId
+
+• `Readonly` **graphqlApiId**: `string`
+
+Graphql Api Id value
+
+___
+
+### graphqlUrl
+
+• `Readonly` **graphqlUrl**: `string`
+
+Graphql Url value
+
+___
+
 ### inputAssetBucket
 
 • `Readonly` **inputAssetBucket**: `IBucket`

lambda/aws-summarization-appsync-stepfn/document_reader/helper.py

@@ -11,9 +11,11 @@
 # and limitations under the License.
 #
 from typing import Dict
+from aiohttp import ClientError
 import boto3
 from PyPDF2 import PdfReader
-
+import tempfile
+import os
 from aws_lambda_powertools import Logger, Tracer
 from s3inmemoryloader import S3FileLoaderInMemory
 
@@ -23,6 +25,8 @@
 s3 = boto3.resource('s3')
 rekognition_client=boto3.client('rekognition')
 
+s3_client = boto3.client('s3')
+
 @tracer.capture_method
 def read_file_from_s3(bucket, key):
     logger.info(f"Fetching file from S3: bucket: {bucket}, key: {key}")
@@ -42,10 +46,10 @@ def check_file_exists(bucket,key):
         return True
     except s3_client.exceptions.ClientError as exp:
         if exp.response['Error']['Code'] == '404':
-            logger.exception('Object doesn\'t exist')
+            logger.info('Object doesn\'t already exist in tranformed bucket, processing it now.. ')
             return False 
         else:
-            logger.exception('An error occured')
+            logger.exception('Object doesn\'t already exist in tranformed bucket, processing it now..')
             return False
 
 
@@ -73,11 +77,15 @@ def get_file_transformation(transformed_asset_bucket,transformed_file_name,
                 response['name'] = transformed_file_name
                 response['summary']=''
             else:
-                with open(original_file_name, "rb") as img_file:
+                input_file= download_file(input_asset_bucket,original_file_name)
+                with open(input_file, "rb") as img_file:
                     image_bytes = {"Bytes": img_file.read()}
                 if(moderate_image(image_bytes) is False):
                      logger.info("Upload image to processed assets bucket")
-                     s3.Bucket(transformed_asset_bucket).put_object(Key=original_file_name, Body=image_bytes)
+                     s3_client.copy_object(Bucket=transformed_asset_bucket, 
+                                           Key=original_file_name, 
+                                           CopySource=input_asset_bucket+'/'+original_file_name
+                                            )
                      response['status'] = 'File transformed'
                      response['name'] = original_file_name
                      response['summary']=''
@@ -86,6 +94,19 @@ def get_file_transformation(transformed_asset_bucket,transformed_file_name,
             logger.info("File already exists,skip transformation.")           
     return response
 
+def download_file(bucket,key )-> str:
+        try: 
+            file_path = os.path.join(tempfile.gettempdir(), os.path.basename(key))
+            s3_client.download_file(bucket, key,file_path)
+            logger.info(f"file downloaded {file_path}")
+            return file_path
+        except ClientError as client_err:
+            logger.error(f"Couldn\'t download file {key}/{file_path} from {bucket}: {client_err.response['Error']['Message']}")
+        
+        except Exception as exp:
+            logger.error(f"Couldn\'t download file {key}/{file_path} from {bucket}: {exp}")
+
+
 def moderate_image(image_bytes)-> str:
         isToxicImage = False
         try: 

lambda/aws-summarization-appsync-stepfn/summary_generator/lambda.py

@@ -48,9 +48,9 @@
 
 transformed_bucket_name = os.environ["ASSET_BUCKET_NAME"]
 chain_type = os.environ["SUMMARY_LLM_CHAIN_TYPE"]
-
 bedrock_client = boto3.client('bedrock-runtime')
 
+
 @logger.inject_lambda_context(log_event=True)
 @tracer.capture_lambda_handler
 @metrics.log_metrics(capture_cold_start_metric=True)
@@ -142,9 +142,9 @@ def generate_summary(_summary_llm,inputFile,language)-> str:
 def generate_summary_for_image(_summary_llm,base64_images,language)-> str:
 
     logger.info(f" generating image description...") 
-    prompt ="Describe this image."
-    
-    system_message = "Respond only in {language}."
+    prompt =f"Think step by step and then describe this image in {language} language."
+     
+    system_message = f"Respond only in {language}."
     human_message=[
 
             {
@@ -161,8 +161,7 @@ def generate_summary_for_image(_summary_llm,base64_images,language)-> str:
 
             }
        ]
-    #ai_message = "Here is my response after thinking step by step ."
-    ai_message = "Here is my response in 50 words." 
+    ai_message = "Here is my response after thinking step by step ." 
     prompt = ChatPromptTemplate.from_messages(
     [
         SystemMessage(content=system_message),
@@ -172,4 +171,4 @@ def generate_summary_for_image(_summary_llm,base64_images,language)-> str:
     )
 
     chain = prompt | _summary_llm | StrOutputParser()
-    chain.invoke({"question": "Describe the image"})
+    chain.invoke({"question": prompt})

src/patterns/gen-ai/aws-summarization-appsync-stepfn/index.ts

@@ -11,7 +11,7 @@
  *  and limitations under the License.
  */
 import * as path from 'path';
-import { Duration, Aws, Stack } from 'aws-cdk-lib';
+import { Duration, Aws } from 'aws-cdk-lib';
 import * as appsync from 'aws-cdk-lib/aws-appsync';
 import * as cognito from 'aws-cdk-lib/aws-cognito';
 import * as ec2 from 'aws-cdk-lib/aws-ec2';
@@ -31,7 +31,7 @@ import { ConstructName } from '../../../common/base-class/construct-name-enum';
 import * as eventBridge from '../../../common/helpers/eventbridge-helper';
 import { buildDockerLambdaFunction } from '../../../common/helpers/lambda-builder-helper';
 import * as s3BucketHelper from '../../../common/helpers/s3-bucket-helper';
-import { generatePhysicalName, lambdaMemorySizeLimiter } from '../../../common/helpers/utils';
+import { lambdaMemorySizeLimiter, generatePhysicalNameV2 } from '../../../common/helpers/utils';
 import { DockerLambdaCustomProps } from '../../../common/props/DockerLambdaCustomProps';
 
 export interface SummarizationAppsyncStepfnProps {
@@ -197,6 +197,16 @@ export class SummarizationAppsyncStepfn extends BaseClass {
    */
   public readonly graphqlApi: appsync.IGraphqlApi ;
 
+  /**
+   * Graphql Api Id value
+   */
+  public readonly graphqlApiId: string ;
+
+  /**
+   *  Graphql Url value
+   */
+  public readonly graphqlUrl: string ;
+
   /**
    * Returns the instance of ec2.IVpc used by the construct
    */
@@ -313,8 +323,10 @@ export class SummarizationAppsyncStepfn extends BaseClass {
       this.inputAssetBucket = new s3.Bucket(this,
         'inputAssetsSummaryBucket'+this.stage, props.bucketInputsAssetsProps);
     } else {
-      const bucketName= 'input-assets-summary-bucket'+this.stage+'-'+Aws.ACCOUNT_ID;
-      this.inputAssetBucket = new s3.Bucket(this, 'inputAssetsSummaryBucket'+this.stage,
+      const bucketName= generatePhysicalNameV2(this,
+        'input-assets-bucket'+this.stage,
+        { maxLength: 63, lower: true });
+      this.inputAssetBucket = new s3.Bucket(this, bucketName,
         {
           blockPublicAccess: s3.BlockPublicAccess.BLOCK_ALL,
           encryption: s3.BucketEncryption.S3_MANAGED,
@@ -340,9 +352,11 @@ export class SummarizationAppsyncStepfn extends BaseClass {
       this.processedAssetBucket = new s3.Bucket(this,
         'processedAssetsSummaryBucket'+this.stage, props.bucketProcessedAssetsProps);
     } else {
-      const bucketName= 'processed-assets-summary-bucket'+this.stage+'-'+Aws.ACCOUNT_ID;
+      const bucketName= generatePhysicalNameV2(this,
+        'processed-assets-bucket'+this.stage,
+        { maxLength: 63, lower: true });
 
-      this.processedAssetBucket = new s3.Bucket(this, 'processedAssetsSummaryBucket'+this.stage,
+      this.processedAssetBucket = new s3.Bucket(this, bucketName,
         {
           blockPublicAccess: s3.BlockPublicAccess.BLOCK_ALL,
           encryption: s3.BucketEncryption.S3_MANAGED,
@@ -376,10 +390,12 @@ export class SummarizationAppsyncStepfn extends BaseClass {
       ],
     };
 
-    const apiName = props.summaryApiName || 'summaryApi';
+    const apiName = props.summaryApiName || generatePhysicalNameV2(this,
+      'summaryApi'+this.stage,
+      { maxLength: 32, lower: true });
 
     // graphql api for summary. client invoke this api with given schema and cognito user pool auth.
-    const summarizationGraphqlApi = new appsync.GraphqlApi(this, 'summarizationGraphqlApi'+this.stage,
+    const summarizationGraphqlApi = new appsync.GraphqlApi(this, apiName,
       {
         name: apiName+this.stage,
         logConfig: {
@@ -393,6 +409,8 @@ export class SummarizationAppsyncStepfn extends BaseClass {
         xrayEnabled: this.enablexray,
       });
     this.graphqlApi= summarizationGraphqlApi;
+    this.graphqlApiId = summarizationGraphqlApi.apiId;
+    this.graphqlUrl= summarizationGraphqlApi.graphqlUrl;
 
     // If the user provides a mergedApi endpoint, the lambda
     // functions will use this endpoint to send their status updates
@@ -473,7 +491,9 @@ export class SummarizationAppsyncStepfn extends BaseClass {
 
     const construct_input_validation_lambda_props = {
       code: lambda.DockerImageCode.fromImageAsset(path.join(__dirname, '../../../../lambda/aws-summarization-appsync-stepfn/input_validator')),
-      functionName: 'summary_input_validator'+this.stage,
+      functionName: generatePhysicalNameV2(this,
+        'summary_input_validator'+this.stage,
+        { maxLength: 63, lower: true }),
       description: 'Lambda function to validate input for summary api',
       vpc: this.vpc,
       tracing: this.lambdaTracing,
@@ -669,8 +689,11 @@ export class SummarizationAppsyncStepfn extends BaseClass {
       true,
     );
 
+    const functionName = generatePhysicalNameV2(this,
+      'summary_generator'+this.stage,
+      { maxLength: 32, lower: true });
     const construct_generate_summary_lambda_props = {
-      functionName: 'summary_generator'+this.stage,
+      functionName: functionName,
       description: 'Lambda function to generate the summary',
       code: lambda.DockerImageCode.fromImageAsset(path.join(__dirname, '../../../../lambda/aws-summarization-appsync-stepfn/summary_generator')),
       vpc: this.vpc,
@@ -689,7 +712,7 @@ export class SummarizationAppsyncStepfn extends BaseClass {
 
     // Lambda function used to generate the summary in the step function
     const generateSummarylambda = buildDockerLambdaFunction(this,
-      'generateSummarylambda' + this.stage,
+      functionName,
       construct_generate_summary_lambda_props,
       props.customSummaryGeneratorDockerLambdaProps,
     );
@@ -729,7 +752,7 @@ export class SummarizationAppsyncStepfn extends BaseClass {
       resultPath: '$.validation_result',
     });
 
-    const documentReaderTask = new sfnTask.LambdaInvoke(this, 'Read document and check summary in cache ', {
+    const documentReaderTask = new sfnTask.LambdaInvoke(this, 'Read document.', {
       lambdaFunction: documentReaderLambda,
       resultPath: '$.document_result',
     });
@@ -741,7 +764,9 @@ export class SummarizationAppsyncStepfn extends BaseClass {
     });
 
     const dlq: sqs.Queue = new sqs.Queue(this, 'dlq', {
-      queueName: 'summarydlq'+this.stage,
+      queueName: generatePhysicalNameV2(this,
+        'summarydlq'+this.stage,
+        { maxLength: 32, lower: true }),
       retentionPeriod: Duration.days(7),
       enforceSSL: true,
     });
@@ -783,12 +808,9 @@ export class SummarizationAppsyncStepfn extends BaseClass {
     const maxLogGroupNameLength = 255;
     const logGroupPrefix = '/aws/vendedlogs/states/constructs/';
     const maxGeneratedNameLength = maxLogGroupNameLength - logGroupPrefix.length;
-    const nameParts: string[] = [
-      Stack.of(scope).stackName, // Name of the stack
-      scope.node.id, // Construct ID
-      'StateMachineLogSummarization', // Literal string for log group name portion
-    ];
-    const logGroupName = generatePhysicalName(logGroupPrefix, nameParts, maxGeneratedNameLength);
+
+    const logGroupName = generatePhysicalNameV2(this, logGroupPrefix,
+      { maxLength: maxGeneratedNameLength, lower: true });
     const summarizationLogGroup = new logs.LogGroup(this, 'summarizationLogGroup', {
       logGroupName: logGroupName,
     });

test/patterns/gen-ai/aws-summarization-appsync-stepfn/aws-summarization-appsync-stepfn.test.ts

@@ -126,9 +126,9 @@ describe('Summarization Appsync Stepfn construct', () => {
               'GraphQLUrl',
             ],
           },
-          INPUT_ASSET_BUCKET: { Ref: Match.stringLikeRegexp('testinputAssetsSummaryBucketdev') },
+          INPUT_ASSET_BUCKET: { Ref: Match.stringLikeRegexp('testinputassetsbucket') },
           IS_FILE_TRANSFORMED: 'false',
-          TRANSFORMED_ASSET_BUCKET: { Ref: Match.stringLikeRegexp('testprocessedAssetsSummaryBucket') },
+          TRANSFORMED_ASSET_BUCKET: { Ref: Match.stringLikeRegexp('testprocessedassetsbucket') },
         },
       },
     });
@@ -139,7 +139,7 @@ describe('Summarization Appsync Stepfn construct', () => {
         Variables: {
           ASSET_BUCKET_NAME: {
             Ref: Match.stringLikeRegexp
-            ('testprocessedAssetsSummaryBucket'),
+            ('testprocessedassetsbucket'),
           },
           GRAPHQL_URL: {
             'Fn::GetAtt': [