Add sizeBytes to document metadata (#517)

* Add sizeBytes to document metadata

* use literal size and simpler multi-byte char for test

Author: Zee2413

.gitignore

@@ -15,3 +15,4 @@ tools/*
 /tmp-tst
 /jscpd
 /tmp
+/.idea/

src/document/Document.ts

@@ -2,6 +2,7 @@ import { TextDocument, Position, Range, DocumentUri } from 'vscode-languageserve
 import { TopLevelSection } from '../context/CloudFormationEnums';
 import { DefaultSettings } from '../settings/Settings';
 import { LoggerFactory } from '../telemetry/LoggerFactory';
+import { byteSize } from '../utils/String';
 import { DocumentMetadata } from './DocumentProtocol';
 import { detectDocumentType, uriToPath } from './DocumentUtils';
 import { parseValidYaml } from './YamlParser';
@@ -170,6 +171,7 @@ export class Document {
             languageId: this.languageId,
             version: this.version,
             lineCount: this.textDocument.lineCount,
+            sizeBytes: byteSize(this.textDocument.getText()),
         };
     }
 

src/document/DocumentManager.ts

@@ -7,7 +7,6 @@ import { ScopedTelemetry } from '../telemetry/ScopedTelemetry';
 import { Telemetry } from '../telemetry/TelemetryDecorator';
 import { Closeable } from '../utils/Closeable';
 import { Delayer } from '../utils/Delayer';
-import { byteSize } from '../utils/String';
 import { CloudFormationFileType, Document, DocumentType } from './Document';
 import { DocumentMetadata } from './DocumentProtocol';
 
@@ -170,7 +169,7 @@ export class DocumentManager implements SettingsConfigurable, Closeable {
     private emitDocSizeMetrics() {
         for (const doc of this.documentMap.values()) {
             if (doc.isTemplate()) {
-                this.telemetry.histogram('documents.template.size.bytes', byteSize(doc.contents()), { unit: 'By' });
+                this.telemetry.histogram('documents.template.size.bytes', doc.metadata().sizeBytes, { unit: 'By' });
             }
         }
     }

src/document/DocumentProtocol.ts

@@ -10,6 +10,7 @@ export type DocumentMetadata = {
     languageId: string;
     version: number;
     lineCount: number;
+    sizeBytes: number;
 };
 
 export type DocumentPreview = {

tst/unit/document/Document.test.ts

@@ -459,4 +459,23 @@ describe('Document', () => {
             expect(doc.getTemplateSizeCategory()).toBe('xlarge');
         });
     });
+
+    describe('metadata', () => {
+        it('should include sizeBytes matching content byte length', () => {
+            const content = 'Resources:\n  Bucket:\n    Type: AWS::S3::Bucket';
+            const textDocument = TextDocument.create('file:///test.yaml', 'yaml', 1, content);
+            const doc = new Document(textDocument);
+
+            expect(doc.metadata().sizeBytes).toBe(46);
+        });
+
+        it('should compute sizeBytes using byte length not character count', () => {
+            // 'é' is 1 character but 2 bytes in UTF-8
+            const content = 'é'.repeat(10);
+            const textDocument = TextDocument.create('file:///unicode.yaml', 'yaml', 1, content);
+            const doc = new Document(textDocument);
+
+            expect(doc.metadata().sizeBytes).toBe(20);
+        });
+    });
 });