fix: crop bloated code and binary COMPASS-8902

src/schema-analyzer.ts

@@ -209,7 +209,6 @@ function getBSONType(value: any): SchemaBSONType {
   const bsonType = value?._bsontype
     ? value._bsontype
     : Object.prototype.toString.call(value).replace(/^\[object (\w+)\]$/, '$1');
-
   if (bsonType === 'Object') {
     // In the resulting schema we rename `Object` to `Document`.
     return 'Document';
@@ -324,10 +323,30 @@ function simplifiedSchema(fields: SchemaAnalysisFieldsMap): SimplifiedSchema {
   return finalizeDocumentFieldSchema(fields);
 }
 
-function cropStringAt10kCharacters(value: string) {
-  return value.charCodeAt(10000 - 1) === value.codePointAt(10000 - 1)
-    ? value.slice(0, 10000)
-    : value.slice(0, 10000 - 1);
+function cropString(value: string, limit: number) {
+  if (limit < 1) return '';
+  return value.charCodeAt(limit - 1) === value.codePointAt(10000 - 1)
+    ? value.slice(0, limit)
+    : value.slice(0, limit - 1);
+}
+
+function getCappedValue(bsonType: SchemaBSONType, value: BSONValue) {
+  if (bsonType === 'String') {
+    return cropString(value as string, 10000);
+  }
+  if (bsonType === 'Binary') {
+    value = value as Binary;
+    return value.buffer.length > 10000
+      ? new Binary(value.buffer.slice(0, 10000), value.sub_type)
+      : value;
+  }
+  if (bsonType === 'Code') {
+    value = value as Code;
+    return (value.code.length >= 10000)
+      ? new Code(cropString(value.code, 10000), value.scope)
+      : value;
+  }
+  return value;
 }
 
 function computeHasDuplicatesForType(type: SchemaAnalysisType, unique?: number) {
@@ -530,7 +549,7 @@ export class SchemaAnalyzer {
         }
 
         type.values.pushSome(
-          type.name === 'String' ? cropStringAt10kCharacters(value as string) : value
+          getCappedValue(type.bsonType, value)
         );
       }
     };

test/bloated.test.ts

@@ -0,0 +1,44 @@
+import assert from 'assert';
+import { Binary, Code } from 'bson';
+
+import type { PrimitiveSchemaType } from '../src/schema-analyzer';
+import getSchema from '../src';
+
+function generateRandomString(length: number) {
+  const chars = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789';
+  let result = '';
+  for (let i = 0; i < length; i++) {
+    result += chars.charAt(Math.floor(Math.random() * chars.length));
+  }
+  return result;
+}
+
+describe('bloated documents', function() {
+  it('really long string is cropped', async function() {
+    const documents = [{
+      str: generateRandomString(20000)
+    }];
+    const schema = await getSchema(documents);
+    const stringLength = ((schema.fields[0].types[0] as PrimitiveSchemaType).values[0] as string).length;
+    assert.ok(stringLength <= 10000);
+  });
+
+  it('really long code is cropped', async function() {
+    const documents = [{
+      code: new Code(generateRandomString(20000))
+    }];
+    const schema = await getSchema(documents);
+    const codeLength = ((schema.fields[0].types[0] as PrimitiveSchemaType).values[0] as Code).code.length;
+    assert.ok(codeLength <= 10000);
+  });
+
+  it('really long binary is cropped', async function() {
+    const documents = [{
+      binData: new Binary(Buffer.from(generateRandomString(20000)), 2)
+    }];
+    const schema = await getSchema(documents);
+    const binary = ((schema.fields[0].types[0] as PrimitiveSchemaType).values[0] as Binary);
+    assert.ok(binary.length() <= 10000);
+    assert.strictEqual(binary.sub_type, 2);
+  });
+});