fix: ensure purls are deduped and encoded

src/api/nes/nes.client.ts

@@ -1,4 +1,5 @@
 import type * as apollo from '@apollo/client/core/index.js';
+import { PackageURL } from 'packageurl-js';
 
 import { ApolloClient } from '../../api/client.ts';
 import type {
@@ -59,7 +60,8 @@ export const batchSubmitPurls = async (
   batchSize = DEFAULT_SCAN_BATCH_SIZE,
 ): Promise<ScanResult> => {
   try {
-    const batches = createBatches(purls, batchSize);
+    const dedupedAndEncodedPurls = dedupeAndEncodePurls(purls);
+    const batches = createBatches(dedupedAndEncodedPurls, batchSize);
     debugLogger('Processing %d batches', batches.length);
 
     if (batches.length === 0) {
@@ -80,6 +82,24 @@ export const batchSubmitPurls = async (
   }
 };
 
+export const dedupeAndEncodePurls = (purls: string[]): string[] => {
+  const dedupedAndEncodedPurls = new Set<string>();
+
+  for (const purl of purls) {
+    try {
+      // The PackageURL.fromString method encodes each part of the purl
+      const encodedPurl = PackageURL.fromString(purl).toString();
+      if (!dedupedAndEncodedPurls.has(encodedPurl)) {
+        dedupedAndEncodedPurls.add(encodedPurl);
+      }
+    } catch (error) {
+      debugLogger('Error encoding purl: %s', error);
+    }
+  }
+
+  return Array.from(dedupedAndEncodedPurls);
+};
+
 export const createBatches = (items: string[], batchSize: number): string[][] => {
   const numberOfBatches = Math.ceil(items.length / batchSize);
 

test/api/nes.client.test.ts

@@ -1,35 +1,67 @@
 import assert from 'node:assert';
 import { describe, it } from 'node:test';
-import { createBatches } from '../../src/api/nes/nes.client.ts';
+import { createBatches, dedupeAndEncodePurls } from '../../src/api/nes/nes.client.ts';
 import { DEFAULT_SCAN_BATCH_SIZE } from '../../src/api/types/hd-cli.types.ts';
 
-describe('createBatches', () => {
-  it('should handle empty array', () => {
-    const result = createBatches([], DEFAULT_SCAN_BATCH_SIZE);
-    assert.deepStrictEqual(result, []);
-  });
+describe('nes.client', () => {
+  describe('createBatches', () => {
+    it('should handle empty array', () => {
+      const result = createBatches([], DEFAULT_SCAN_BATCH_SIZE);
+      assert.deepStrictEqual(result, []);
+    });
 
-  it('should create single batch when items length is less than batch size', () => {
-    const items = ['a', 'b', 'c'];
-    const result = createBatches(items, 5);
-    assert.deepStrictEqual(result, [['a', 'b', 'c']]);
-  });
+    it('should create single batch when items length is less than batch size', () => {
+      const items = ['a', 'b', 'c'];
+      const result = createBatches(items, 5);
+      assert.deepStrictEqual(result, [['a', 'b', 'c']]);
+    });
 
-  it('should create single batch when items length equals batch size', () => {
-    const items = ['a', 'b', 'c', 'd', 'e'];
-    const result = createBatches(items, 5);
-    assert.deepStrictEqual(result, [['a', 'b', 'c', 'd', 'e']]);
-  });
+    it('should create single batch when items length equals batch size', () => {
+      const items = ['a', 'b', 'c', 'd', 'e'];
+      const result = createBatches(items, 5);
+      assert.deepStrictEqual(result, [['a', 'b', 'c', 'd', 'e']]);
+    });
+
+    it('should create multiple batches when items length exceeds batch size', () => {
+      const items = ['a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j'];
+      const result = createBatches(items, 3);
+      assert.deepStrictEqual(result, [['a', 'b', 'c'], ['d', 'e', 'f'], ['g', 'h', 'i'], ['j']]);
+    });
 
-  it('should create multiple batches when items length exceeds batch size', () => {
-    const items = ['a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j'];
-    const result = createBatches(items, 3);
-    assert.deepStrictEqual(result, [['a', 'b', 'c'], ['d', 'e', 'f'], ['g', 'h', 'i'], ['j']]);
+    it('should handle batch size of 1', () => {
+      const items = ['a', 'b', 'c'];
+      const result = createBatches(items, 1);
+      assert.deepStrictEqual(result, [['a'], ['b'], ['c']]);
+    });
   });
 
-  it('should handle batch size of 1', () => {
-    const items = ['a', 'b', 'c'];
-    const result = createBatches(items, 1);
-    assert.deepStrictEqual(result, [['a'], ['b'], ['c']]);
+  describe('dedupeAndEncodePurls', () => {
+    const inputs = [
+      {
+        purls: ['pkg:npm/@angular/core@14.3.0', 'pkg:npm/npm-bundled@2.0.1', 'pkg:npm/%40angular/core@14.3.0'],
+        expected: ['pkg:npm/%40angular/core@14.3.0', 'pkg:npm/npm-bundled@2.0.1'],
+        description: 'should dedupe angular core purls',
+      },
+      {
+        purls: ['pkg:npm/@angular/core@14.3.0', 'pkg:npm/npm-bundled@2.0.1', 'pkg:npm/rxjs@6.6.7'],
+        expected: ['pkg:npm/%40angular/core@14.3.0', 'pkg:npm/npm-bundled@2.0.1', 'pkg:npm/rxjs@6.6.7'],
+        description: 'should not dedupe unique purls',
+      },
+      {
+        purls: [
+          'pkg:maven/org.apache.commons/commons-lang3@3.12.0',
+          'pkg:maven/org.apache.commons/commons-lang3@3.12.0',
+          'pkg:maven/org.apache.commons/commons-lang3@3.12.0',
+        ],
+        expected: ['pkg:maven/org.apache.commons/commons-lang3@3.12.0'],
+        description: 'should dedupe maven purls',
+      },
+    ];
+    for (const input of inputs) {
+      it(`should dedupe and encode purls: ${input.description}`, () => {
+        const result = dedupeAndEncodePurls(input.purls);
+        assert.deepStrictEqual(result, input.expected);
+      });
+    }
   });
 });