Add Anonymizer class

Author: IvanGoncharov

src/index.ts

@@ -394,6 +394,7 @@ export type {
 
 // Utilities for operating on GraphQL type schema and parsed sources.
 export {
+  Anonymizer,
   // Produce the GraphQL query recommended for a full schema introspection.
   // Accepts optional IntrospectionOptions.
   getIntrospectionQuery,

src/utilities/Anonymizer.ts

@@ -0,0 +1,265 @@
+import { inspect } from '../jsutils/inspect';
+import { invariant } from '../jsutils/invariant';
+
+import type {
+  ASTNode,
+  DocumentNode,
+  FloatValueNode,
+  IntValueNode,
+  NameNode,
+  StringValueNode,
+} from '../language/ast';
+import { isNode } from '../language/ast';
+import { Kind } from '../language/kinds';
+import { parseValue } from '../language/parser';
+import type { ASTVisitor } from '../language/visitor';
+import { visit } from '../language/visitor';
+
+import { GraphQLSchema } from '../type/schema';
+import { specifiedScalarTypes } from '../type/scalars';
+
+import { TypeInfo, visitWithTypeInfo } from './TypeInfo';
+
+interface AnonymizerOptions {
+  hashSalt: string;
+  hashFunction: (value: ArrayBuffer) => Promise<ArrayBuffer>;
+  safeListedSchema?: GraphQLSchema | null;
+}
+
+export class Anonymizer {
+  private _valueMap: Map<string, string> = new Map();
+  private _safeListedSchema: GraphQLSchema | null;
+  private _hashFunction: (value: ArrayBuffer) => Promise<ArrayBuffer>;
+  private _hashSalt: string;
+
+  constructor(options: AnonymizerOptions) {
+    const {
+      hashSalt,
+      hashFunction,
+      safeListedSchema = new GraphQLSchema({ types: specifiedScalarTypes }),
+    } = options;
+
+    this._safeListedSchema = safeListedSchema;
+    this._hashSalt = hashSalt;
+    this._hashFunction = hashFunction;
+  }
+
+  get [Symbol.toStringTag]() {
+    return 'Anonymizer';
+  }
+
+  async anonymizeDocumentNode(
+    documentAST: DocumentNode,
+  ): Promise<DocumentNode> {
+    const nodesToAnonymize: Array<
+      NameNode | StringValueNode | IntValueNode | FloatValueNode
+    > = [];
+
+    const typeInfo =
+      this._safeListedSchema !== null
+        ? new TypeInfo(this._safeListedSchema)
+        : null;
+    const safeListedSchema = this._safeListedSchema;
+    const visitor: ASTVisitor = {
+      Name(node, key, parent) {
+        if (
+          safeListedSchema === null ||
+          typeInfo === null ||
+          !isSafeListedName(safeListedSchema, typeInfo, node, key, parent)
+        ) {
+          nodesToAnonymize.push(node);
+        }
+      },
+      StringValue(node) {
+        nodesToAnonymize.push(node);
+      },
+      IntValue(node) {
+        nodesToAnonymize.push(node);
+      },
+      FloatValue(node) {
+        nodesToAnonymize.push(node);
+      },
+    };
+
+    const typeInfoVisitor =
+      typeInfo !== null ? visitWithTypeInfo(typeInfo, visitor) : visitor;
+    visit(documentAST, typeInfoVisitor);
+
+    const anonymizedValues = await Promise.all(
+      nodesToAnonymize.map(({ value }) => this.anonymizeStringValue(value)),
+    );
+
+    const anonymizedMap = new Map<ASTNode, ASTNode>();
+    for (const [i, value] of anonymizedValues.entries()) {
+      const node = nodesToAnonymize[i];
+      anonymizedMap.set(node, { ...node, loc: undefined, value });
+    }
+
+    return visit(documentAST, {
+      enter(node) {
+        const anonymizedNode = anonymizedMap.get(node);
+        if (anonymizedNode !== undefined) {
+          return anonymizedNode;
+        }
+
+        return {
+          ...node,
+          // Remove `loc` on all nodes
+          loc: undefined,
+        };
+      },
+    });
+  }
+
+  async anonymizeValue(oldValue: unknown): Promise<unknown> {
+    switch (typeof oldValue) {
+      case 'undefined':
+      case 'boolean':
+        return oldValue;
+      case 'number':
+        return Number.isFinite(oldValue)
+          ? Number(await this.anonymizeStringValue(oldValue.toString()))
+          : oldValue;
+      case 'bigint':
+        return BigInt(await this.anonymizeStringValue(oldValue.toString()));
+      case 'string':
+        return this.anonymizeStringValue(oldValue);
+      case 'symbol':
+        throw new TypeError('Can not anonymize symbol:' + inspect(oldValue));
+      case 'function':
+        throw new TypeError('Can not anonymize function:' + inspect(oldValue));
+      case 'object':
+        if (oldValue === null) {
+          return oldValue;
+        }
+
+        if (Array.isArray(oldValue)) {
+          return Promise.all(oldValue.map((item) => this.anonymizeValue(item)));
+        }
+
+        if (isPlainObject(oldValue)) {
+          return Object.fromEntries(
+            await Promise.all(
+              Object.entries(oldValue).map(async ([key, value]) => [
+                await this.anonymizeStringValue(key),
+                await this.anonymizeValue(value),
+              ]),
+            ),
+          );
+        }
+        throw new TypeError('Can not anonymize object:' + inspect(oldValue));
+    }
+  }
+
+  async anonymizeStringValue(oldValue: string): Promise<string> {
+    const mappedNewValue = this._valueMap.get(oldValue);
+    if (mappedNewValue !== undefined) {
+      return mappedNewValue;
+    }
+
+    const encoder = new TextEncoder();
+    const hash = await this._hashFunction(
+      encoder.encode(this._hashSalt + oldValue).buffer,
+    );
+    const newValue = generateNewValue(hash, oldValue);
+    this._valueMap.set(oldValue, newValue);
+    return newValue;
+  }
+}
+
+function isSafeListedName(
+  safeListedSchema: GraphQLSchema,
+  typeInfo: TypeInfo,
+  nameNode: NameNode,
+  key: string | number | undefined,
+  parent: ASTNode | ReadonlyArray<ASTNode> | undefined,
+): boolean {
+  invariant(isNode(parent) && typeof key === 'string');
+
+  switch (parent.kind) {
+    case Kind.FIELD:
+      if (key === 'name') {
+        return typeInfo.getFieldDef() != null;
+      }
+      return false;
+    case Kind.ARGUMENT:
+      return typeInfo.getArgument() != null;
+    case Kind.OBJECT_FIELD:
+      return typeInfo.getInputType() != null;
+    case Kind.DIRECTIVE:
+      return typeInfo.getDirective() != null;
+    case Kind.NAMED_TYPE:
+      return safeListedSchema.getType(nameNode.value) !== undefined;
+
+    case Kind.DIRECTIVE_DEFINITION:
+      return safeListedSchema.getDirective(nameNode.value) !== undefined;
+    case Kind.SCALAR_TYPE_DEFINITION:
+    case Kind.OBJECT_TYPE_DEFINITION:
+    case Kind.INTERFACE_TYPE_DEFINITION:
+    case Kind.UNION_TYPE_DEFINITION:
+    case Kind.ENUM_TYPE_DEFINITION:
+    case Kind.INPUT_OBJECT_TYPE_DEFINITION:
+    case Kind.SCALAR_TYPE_EXTENSION:
+    case Kind.OBJECT_TYPE_EXTENSION:
+    case Kind.INTERFACE_TYPE_EXTENSION:
+    case Kind.UNION_TYPE_EXTENSION:
+    case Kind.ENUM_TYPE_EXTENSION:
+    case Kind.INPUT_OBJECT_TYPE_EXTENSION:
+      return safeListedSchema.getType(nameNode.value) !== undefined;
+
+    case Kind.FIELD_DEFINITION:
+    case Kind.INPUT_VALUE_DEFINITION:
+    case Kind.ENUM_VALUE_DEFINITION:
+      return false;
+
+    default:
+      return false;
+  }
+}
+
+function isPlainObject(object: Object) {
+  const prototype = Object.getPrototypeOf(object);
+  return prototype === Object.prototype || prototype === null;
+}
+
+function generateNewValue(hash: ArrayBuffer, oldValue: string): string {
+  const hashNumber = typedArrayToBigInt(hash);
+  try {
+    const parsedValue = parseValue(oldValue);
+    const first32Bit = hashNumber % BigInt(2 ** 32);
+    switch (parsedValue.kind) {
+      case Kind.INT:
+        return first32Bit.toString();
+      case Kind.FLOAT:
+        return '0.' + first32Bit.toString();
+      default:
+    }
+  } catch (_e) {
+    // ignore errors
+  }
+  return 'h_' + encodeBase62(hashNumber);
+}
+
+function typedArrayToBigInt(array: ArrayBuffer): bigint {
+  let result = 0n;
+  const bytes = new Uint8Array(array);
+  for (const [index, byte] of bytes.entries()) {
+    result += BigInt(byte) << BigInt((bytes.length - 1 - index) * 8);
+  }
+  return result;
+}
+
+const b62CharacterSet =
+  '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz';
+function encodeBase62(number: bigint): string {
+  let result = '';
+
+  let leftOver = number;
+  do {
+    const reminder = leftOver % 62n;
+    result = b62CharacterSet.charAt(Number(reminder)) + result;
+    leftOver = (leftOver - reminder) / 62n;
+  } while (leftOver > 0n);
+
+  return result;
+}

src/utilities/__tests__/Anonymizer-test.ts

@@ -0,0 +1,64 @@
+import { webcrypto } from 'node:crypto';
+
+import { expect } from 'chai';
+import { describe, it } from 'mocha';
+
+import { dedent } from '../../__testUtils__/dedent';
+
+import { parse } from '../../language/parser';
+import { print } from '../../language/printer';
+
+import { Anonymizer } from '../Anonymizer';
+
+async function expectAnonymized(document: string) {
+  const anonymizer = new Anonymizer({
+    hashSalt: 'graphql-js/',
+    hashFunction: (data) => webcrypto.subtle.digest('SHA-256', data),
+  });
+  return expect(print(await anonymizer.anonymizeDocumentNode(parse(document))));
+}
+
+// test with schema, query => snapshot + test the same result
+// test with invalid query due to arg mismatch (if argument replaced to the same became valid)
+// test with coercion from string to int/float
+// test with introspection query with type
+describe('Anonymizer', () => {
+  it('can be Object.toStringified', () => {
+    const anonymizer = new Anonymizer({});
+
+    expect(Object.prototype.toString.call(anonymizer)).to.equal(
+      '[object Anonymizer]',
+    );
+  });
+
+  it('work', async () => {
+    const anonymizer = new Anonymizer({
+      hashSalt: 'graphql-js/',
+      hashFunction: (data) => webcrypto.subtle.digest('SHA-256', data),
+    });
+    const hashed = await anonymizer.anonymizeStringValue('test');
+    expect(hashed).to.equal('h_dBtROL5GGqP7VAoLl1CvQzrdgLUtOFRuqWCAhvWK8H0');
+  });
+
+  it('work', async () => {
+    (
+      await expectAnonymized(`
+      query TestQuery($arg: String) {
+        foo(arg: $arg)
+        bar {
+          baz @skip(if: false)
+        }
+      }
+    `)
+    ).to.equal(dedent`
+      query h_SJZgrQ0qER6XA2In0BvjgikiGyzS947FiPj0KVuWuqo($h_nVaZrh9Oups9oZLouxgQFpHNTo1kxlaa3dI8D5PBIdc: String) {
+        h_qbuzhEKLs429KQhe60wLZOP746k8mU69s6K3YN8sORO(
+          h_nVaZrh9Oups9oZLouxgQFpHNTo1kxlaa3dI8D5PBIdc: $h_nVaZrh9Oups9oZLouxgQFpHNTo1kxlaa3dI8D5PBIdc
+        )
+        h_I7FTvKWpQa6jnQb8LeHDG2AlHuFUkDwt2n2c4WA7efV {
+          h_uua9CNLJ9uByOi4HQXucOoTNwop41vn8bLShit1u9in
+        }
+      }
+    `);
+  });
+});

src/utilities/index.ts

@@ -1,3 +1,5 @@
+export { Anonymizer } from './Anonymizer';
+
 // Produce the GraphQL query recommended for a full schema introspection.
 export { getIntrospectionQuery } from './getIntrospectionQuery';
 

tsconfig.json

@@ -6,7 +6,7 @@
     "benchmark/benchmark.ts"
   ],
   "compilerOptions": {
-    "lib": ["es2020"],
+    "lib": ["es2020", "dom"],
     "target": "es2020",
     "module": "commonjs",
     "moduleResolution": "node",