remove length check for account id & move resource extraction into arn parser

Author: blairhyy-amazon

aws-distro-opentelemetry-node-autoinstrumentation/src/aws-metric-attribute-generator.ts

@@ -399,7 +399,9 @@ export class AwsMetricAttributeGenerator implements MetricAttributeGenerator {
       } else if (AwsSpanProcessingUtil.isKeyPresent(span, AWS_ATTRIBUTE_KEYS.AWS_DYNAMODB_TABLE_ARN)) {
         remoteResourceType = NORMALIZED_DYNAMO_DB_SERVICE_NAME + '::Table';
         remoteResourceIdentifier = AwsMetricAttributeGenerator.escapeDelimiters(
-          this.extractDynamoDbTableNameFromArn(span.attributes[AWS_ATTRIBUTE_KEYS.AWS_DYNAMODB_TABLE_ARN])
+          RegionalResourceArnParser.extractDynamoDbTableNameFromArn(
+            span.attributes[AWS_ATTRIBUTE_KEYS.AWS_DYNAMODB_TABLE_ARN]
+          )
         );
       } else if (AwsSpanProcessingUtil.isKeyPresent(span, AWS_ATTRIBUTE_KEYS.AWS_KINESIS_STREAM_NAME)) {
         remoteResourceType = NORMALIZED_KINESIS_SERVICE_NAME + '::Stream';
@@ -409,7 +411,9 @@ export class AwsMetricAttributeGenerator implements MetricAttributeGenerator {
       } else if (AwsSpanProcessingUtil.isKeyPresent(span, AWS_ATTRIBUTE_KEYS.AWS_KINESIS_STREAM_ARN)) {
         remoteResourceType = NORMALIZED_KINESIS_SERVICE_NAME + '::Stream';
         remoteResourceIdentifier = AwsMetricAttributeGenerator.escapeDelimiters(
-          this.extractKinesisStreamNameFromArn(span.attributes[AWS_ATTRIBUTE_KEYS.AWS_KINESIS_STREAM_ARN])
+          RegionalResourceArnParser.extractKinesisStreamNameFromArn(
+            span.attributes[AWS_ATTRIBUTE_KEYS.AWS_KINESIS_STREAM_ARN]
+          )
         );
       } else if (AwsSpanProcessingUtil.isKeyPresent(span, AWS_ATTRIBUTE_KEYS.AWS_S3_BUCKET)) {
         remoteResourceType = NORMALIZED_S3_SERVICE_NAME + '::Bucket';
@@ -421,31 +425,31 @@ export class AwsMetricAttributeGenerator implements MetricAttributeGenerator {
 
         remoteResourceType = NORMALIZED_SNS_SERVICE_NAME + '::Topic';
         remoteResourceIdentifier = AwsMetricAttributeGenerator.escapeDelimiters(
-          this.extractResourceNameFromArn(snsArn)
+          RegionalResourceArnParser.extractResourceNameFromArn(snsArn)
         );
         cloudFormationIdentifier = AwsMetricAttributeGenerator.escapeDelimiters(snsArn);
       } else if (AwsSpanProcessingUtil.isKeyPresent(span, AWS_ATTRIBUTE_KEYS.AWS_SECRETSMANAGER_SECRET_ARN)) {
         const secretsArn = span.attributes[AWS_ATTRIBUTE_KEYS.AWS_SECRETSMANAGER_SECRET_ARN];
 
         remoteResourceType = NORMALIZED_SECRETSMANAGER_SERVICE_NAME + '::Secret';
         remoteResourceIdentifier = AwsMetricAttributeGenerator.escapeDelimiters(
-          this.extractResourceNameFromArn(secretsArn)
+          RegionalResourceArnParser.extractResourceNameFromArn(secretsArn)
         );
         cloudFormationIdentifier = AwsMetricAttributeGenerator.escapeDelimiters(secretsArn);
       } else if (AwsSpanProcessingUtil.isKeyPresent(span, AWS_ATTRIBUTE_KEYS.AWS_STEPFUNCTIONS_STATEMACHINE_ARN)) {
         const stateMachineArn = span.attributes[AWS_ATTRIBUTE_KEYS.AWS_STEPFUNCTIONS_STATEMACHINE_ARN];
 
         remoteResourceType = NORMALIZED_STEPFUNCTIONS_SERVICE_NAME + '::StateMachine';
         remoteResourceIdentifier = AwsMetricAttributeGenerator.escapeDelimiters(
-          this.extractResourceNameFromArn(stateMachineArn)
+          RegionalResourceArnParser.extractResourceNameFromArn(stateMachineArn)
         );
         cloudFormationIdentifier = AwsMetricAttributeGenerator.escapeDelimiters(stateMachineArn);
       } else if (AwsSpanProcessingUtil.isKeyPresent(span, AWS_ATTRIBUTE_KEYS.AWS_STEPFUNCTIONS_ACTIVITY_ARN)) {
         const activityArn = span.attributes[AWS_ATTRIBUTE_KEYS.AWS_STEPFUNCTIONS_ACTIVITY_ARN];
 
         remoteResourceType = NORMALIZED_STEPFUNCTIONS_SERVICE_NAME + '::Activity';
         remoteResourceIdentifier = AwsMetricAttributeGenerator.escapeDelimiters(
-          this.extractResourceNameFromArn(activityArn)
+          RegionalResourceArnParser.extractResourceNameFromArn(activityArn)
         );
         cloudFormationIdentifier = AwsMetricAttributeGenerator.escapeDelimiters(activityArn);
       } else if (AwsSpanProcessingUtil.isKeyPresent(span, AWS_ATTRIBUTE_KEYS.AWS_LAMBDA_FUNCTION_NAME)) {
@@ -725,24 +729,6 @@ export class AwsMetricAttributeGenerator implements MetricAttributeGenerator {
     return rpcService === 'Lambda' && span.attributes[SEMATTRS_RPC_METHOD] === LAMBDA_INVOKE_OPERATION;
   }
 
-  private static extractDynamoDbTableNameFromArn(attribute: AttributeValue | undefined): string | undefined {
-    return this.extractResourceNameFromArn(attribute)?.replace('table/', '');
-  }
-
-  private static extractKinesisStreamNameFromArn(attribute: AttributeValue | undefined): string | undefined {
-    return this.extractResourceNameFromArn(attribute)?.replace('stream/', '');
-  }
-
-  // Extracts the name of the resource from an arn
-  private static extractResourceNameFromArn(attribute: AttributeValue | undefined): string | undefined {
-    if (typeof attribute === 'string' && attribute.startsWith('arn:aws:')) {
-      const split = attribute.split(':');
-      return split[split.length - 1];
-    }
-
-    return undefined;
-  }
-
   /** Span kind is needed for differentiating metrics in the EMF exporter */
   private static setSpanKindForService(span: ReadableSpan, attributes: Attributes): void {
     let spanKind: string = SpanKind[span.kind];

aws-distro-opentelemetry-node-autoinstrumentation/src/regional-resource-arn-parser.ts

@@ -19,6 +19,24 @@ export class RegionalResourceArnParser {
     return undefined;
   }
 
+  public static extractDynamoDbTableNameFromArn(arn: AttributeValue | undefined): string | undefined {
+    return this.extractResourceNameFromArn(arn)?.replace('table/', '');
+  }
+
+  public static extractKinesisStreamNameFromArn(arn: AttributeValue | undefined): string | undefined {
+    return this.extractResourceNameFromArn(arn)?.replace('stream/', '');
+  }
+
+  // Extracts the name of the resource from an arn
+  public static extractResourceNameFromArn(arn: AttributeValue | undefined): string | undefined {
+    if (typeof arn === 'string' && this.isArn(arn)) {
+      const split = arn.split(':');
+      return split[split.length - 1];
+    }
+
+    return undefined;
+  }
+
   public static isArn(arn: string): boolean {
     // Check if arn follows the format:
     // arn:partition:service:region:account-id:resource-type/resource-id or

aws-distro-opentelemetry-node-autoinstrumentation/src/sqs-url-parser.ts

@@ -11,7 +11,7 @@ const HTTPS_SCHEMA: string = 'https://';
 // eslint-disable-next-line @typescript-eslint/typedef
 const ALPHABET_REGEX = /^[a-zA-Z]+$/;
 
-export interface ParsedSqsUrl { 
+export interface ParsedSqsUrl {
   queueName: string;
   accountId: string;
   region?: string;
@@ -22,7 +22,7 @@ export class SqsUrlParser {
    * Best-effort logic to extract queue name from an HTTP url. This method should only be used with
    * a string that is, with reasonably high confidence, an SQS queue URL. Handles new/legacy/some
    * custom URLs. Essentially, we require that the URL should have exactly three parts, delimited by
-   * /'s (excluding schema), the second part should be a 12-digit account id, and the third part
+   * /'s (excluding schema), the second part should be a account id consisting of digits, and the third part
    * should be a valid queue name, per SQS naming conventions.
    */
   public static getQueueName(url: AttributeValue | undefined): string | undefined {

aws-distro-opentelemetry-node-autoinstrumentation/src/utils.ts

@@ -23,7 +23,7 @@ export const checkDigits = (str: string): boolean => {
 };
 
 export const isAccountId = (input: string): boolean => {
-  if (input == null || input.length !== 12) {
+  if (input == null) {
     return false;
   }
 

aws-distro-opentelemetry-node-autoinstrumentation/test/aws-metric-attribute-generator.test.ts

@@ -1145,11 +1145,7 @@ describe('AwsMetricAttributeGeneratorTest', () => {
       AWS_ATTRIBUTE_KEYS.AWS_STEPFUNCTIONS_STATEMACHINE_ARN,
       'arn:aws:states:us-east-1:invalid_account_id:stateMachine:testStateMachine'
     );
-    validateRemoteResourceAttributes(
-      'AWS::StepFunctions::StateMachine',
-      'testStateMachine',
-      'arn:aws:states:us-east-1:invalid_account_id:stateMachine:testStateMachine'
-    );
+    validateRemoteResourceAttributes(undefined, undefined);
     mockAttribute(AWS_ATTRIBUTE_KEYS.AWS_STEPFUNCTIONS_STATEMACHINE_ARN, undefined);
 
     // Arn with invalid region

aws-distro-opentelemetry-node-autoinstrumentation/test/regional-resource-arn-parser.test.ts

@@ -13,8 +13,8 @@ describe('RegionalResourceArnParserTest', () => {
     validateAccountId('::::::', undefined);
     validateAccountId('not:an:arn:string', undefined);
     validateAccountId('arn:aws:ec2:us-west-2:123456', undefined);
-    validateAccountId('arn:aws:ec2:us-west-2:1234567xxxxx', undefined);
     validateAccountId('arn:aws:ec2:us-west-2:123456789012', undefined);
+    validateAccountId('arn:aws:ec2:us-west-2:1234567xxxxx:table/test_table', undefined);
     validateAccountId('arn:aws:dynamodb:us-west-2:123456789012:table/test_table', '123456789012');
     validateAccountId('arn:aws:acm:us-east-1:123456789012:certificate:abc-123', '123456789012');
   });
@@ -34,6 +34,52 @@ describe('RegionalResourceArnParserTest', () => {
   });
 });
 
+it('testExtractDynamoDbTableNameFromArn', () => {
+  validateDynamoDbTableName(undefined, undefined);
+  validateDynamoDbTableName('', undefined);
+  validateDynamoDbTableName(' ', undefined);
+  validateDynamoDbTableName(':', undefined);
+  validateDynamoDbTableName('::::::', undefined);
+  validateDynamoDbTableName('not:an:arn:string', undefined);
+  validateDynamoDbTableName('arn:aws:dynamodb:us-west-2:123456789012:table/test_table', 'test_table');
+  validateDynamoDbTableName('arn:aws:dynamodb:us-west-2:123456789012:table/my-table-name', 'my-table-name');
+});
+
+it('testExtractKinesisStreamNameFromArn', () => {
+  validateKinesisStreamName(undefined, undefined);
+  validateKinesisStreamName('', undefined);
+  validateKinesisStreamName(' ', undefined);
+  validateKinesisStreamName(':', undefined);
+  validateKinesisStreamName('::::::', undefined);
+  validateKinesisStreamName('not:an:arn:string', undefined);
+  validateKinesisStreamName('arn:aws:kinesis:us-west-2:123456789012:stream/test_stream', 'test_stream');
+  validateKinesisStreamName('arn:aws:kinesis:us-west-2:123456789012:stream/my-stream-name', 'my-stream-name');
+});
+
+it('testExtractResourceNameFromArn', () => {
+  validateResourceName(undefined, undefined);
+  validateResourceName('', undefined);
+  validateResourceName(' ', undefined);
+  validateResourceName(':', undefined);
+  validateResourceName('::::::', undefined);
+  validateResourceName('not:an:arn:string', undefined);
+  validateResourceName('arn:aws:dynamodb:us-west-2:123456789012:table/test_table', 'table/test_table');
+  validateResourceName('arn:aws:kinesis:us-west-2:123456789012:stream/test_stream', 'stream/test_stream');
+  validateResourceName('arn:aws:s3:us-west-2:123456789012:my-bucket', 'my-bucket');
+});
+
+function validateDynamoDbTableName(arn: string | undefined, expectedName: string | undefined): void {
+  expect(RegionalResourceArnParser.extractDynamoDbTableNameFromArn(arn)).toEqual(expectedName);
+}
+
+function validateKinesisStreamName(arn: string | undefined, expectedName: string | undefined): void {
+  expect(RegionalResourceArnParser.extractKinesisStreamNameFromArn(arn)).toEqual(expectedName);
+}
+
+function validateResourceName(arn: string | undefined, expectedName: string | undefined): void {
+  expect(RegionalResourceArnParser.extractResourceNameFromArn(arn)).toEqual(expectedName);
+}
+
 function validateAccountId(arn: string | undefined, expectedAccountId: string | undefined): void {
   expect(RegionalResourceArnParser.getAccountId(arn)).toEqual(expectedAccountId);
 }

aws-distro-opentelemetry-node-autoinstrumentation/test/sqs-url-parser.test.ts

@@ -47,7 +47,7 @@ describe('SqsUrlParserTest', () => {
     validateGetQueueName('invalidUrl', undefined);
     validateGetQueueName('https://www.amazon.com', undefined);
     validateGetQueueName('https://sqs.us-east-1.amazonaws.com/123412341234/.', undefined);
-    validateGetQueueName('https://sqs.us-east-1.amazonaws.com/12/Queue', undefined);
+    validateGetQueueName('https://sqs.us-east-1.amazonaws.com/12341234xxxx/.', undefined);
     validateGetQueueName('https://sqs.us-east-1.amazonaws.com/A/A', undefined);
     validateGetQueueName('https://sqs.us-east-1.amazonaws.com/123412341234/A/ThisShouldNotBeHere', undefined);
   });
@@ -63,7 +63,7 @@ describe('SqsUrlParserTest', () => {
     validateGetAccountId('/123412341234/as&df', undefined);
     validateGetAccountId('invalidUrl', undefined);
     validateGetAccountId('https://www.amazon.com', undefined);
-    validateGetAccountId('https://sqs.us-east-1.amazonaws.com/12341234/Queue', undefined);
+    validateGetAccountId('https://sqs.us-east-1.amazonaws.com/12341234/Queue', '12341234');
     validateGetAccountId('https://sqs.us-east-1.amazonaws.com/1234123412xx/Queue', undefined);
     validateGetAccountId('https://sqs.us-east-1.amazonaws.com/1234123412xx', undefined);
     validateGetAccountId('https://sqs.us-east-1.amazonaws.com/123412341234/Q_Namez-5', '123412341234');
@@ -80,7 +80,7 @@ describe('SqsUrlParserTest', () => {
     validateGetRegion('/123412341234/as&df', undefined);
     validateGetRegion('invalidUrl', undefined);
     validateGetRegion('https://www.amazon.com', undefined);
-    validateGetRegion('https://sqs.us-east-1.amazonaws.com/12341234/Queue', undefined);
+    validateGetRegion('https://sqs.us-east-1.amazonaws.com/12341234/Queue', 'us-east-1');
     validateGetRegion('https://sqs.us-east-1.amazonaws.com/1234123412xx/Queue', undefined);
     validateGetRegion('https://sqs.us-east-1.amazonaws.com/1234123412xx', undefined);
     validateGetRegion('https://sqs.us-east-1.amazonaws.com/123412341234/Q_Namez-5', 'us-east-1');