fix: skip downloading CIDs for empty files (#617)

Some CIDs represent an empty file and Storacha returns HTTP 410 Gone
error response in such case.

In this patch, I am fixing the function `fetchMeasurements` to
skip the retrieval and return an empty list for such CIDs.

Signed-off-by: Miroslav Bajtoš <oss@bajtos.net>

Author: bajtos

lib/preprocess.js

@@ -156,6 +156,11 @@ export const preprocess = async ({
  * @returns {Promise<import('./typings.js').RawMeasurement[]>}
  */
 export const fetchMeasurements = async (cid, { signal, noCache = false } = {}) => {
+  if (cid === 'bafkqaaa' || cid === 'bafkreihdwdcefgh4dqkjv67uzcmw7ojee6xedzdetojuzjevtenxquvyku') {
+    // These CIDs represent an empty file
+    return []
+  }
+
   const res = await fetch(
     `https://${encodeURIComponent(cid)}.ipfs.w3s.link?format=car`,
     {

test/preprocess.js

@@ -4,7 +4,8 @@ import {
   preprocess,
   Measurement,
   parseMeasurements,
-  assertValidMeasurement
+  assertValidMeasurement,
+  fetchMeasurements
 } from '../lib/preprocess.js'
 import { Point } from '../lib/telemetry.js'
 import assert from 'node:assert'
@@ -86,6 +87,20 @@ describe('preprocess', () => {
   })
 })
 
+describe('fetchMeasurements', () => {
+  it('handles empty batch with CID bafkqaaa', async () => {
+    const cid = 'bafkqaaa'
+    const measurements = await fetchMeasurements(cid)
+    assert.deepStrictEqual(measurements, [])
+  })
+
+  it('handles empty batch with CID bafkreihdwdcefgh4dqkjv67uzcmw7ojee6xedzdetojuzjevtenxquvyku', async () => {
+    const cid = 'bafkreihdwdcefgh4dqkjv67uzcmw7ojee6xedzdetojuzjevtenxquvyku'
+    const measurements = await fetchMeasurements(cid)
+    assert.deepStrictEqual(measurements, [])
+  })
+})
+
 describe('parseMeasurements', () => {
   const measurements = [{ foo: 'bar' }, { beep: 'boop' }]
   it('parses a JSON array', () => {