From 9a69ed9d902b4ec9904d02048aee1a3b4e87714b Mon Sep 17 00:00:00 2001
From: Brian Gough <briangough@users.noreply.github.com>
Date: Fri, 31 Jan 2025 15:55:53 +0000
Subject: [PATCH] Merge pull request #23169 from
 overleaf/bg-export-global-blobs

Add script to export global blobs to CSV

GitOrigin-RevId: 86c885981b150aae4541770a094c822e1fb364b9
---
 .../storage/scripts/export_global_blobs.mjs   | 69 +++++++++++++++++++
 1 file changed, 69 insertions(+)
 create mode 100644 services/history-v1/storage/scripts/export_global_blobs.mjs

diff --git a/services/history-v1/storage/scripts/export_global_blobs.mjs b/services/history-v1/storage/scripts/export_global_blobs.mjs
new file mode 100644
index 0000000000..ccbb1237dd
--- /dev/null
+++ b/services/history-v1/storage/scripts/export_global_blobs.mjs
@@ -0,0 +1,69 @@
+/**
+ * A script to export the global blobs from mongo to a CSV file.
+ *
+ * node storage/scripts/export_global_blobs.mjs --output global_blobs.csv
+ *
+ * The output CSV has the following format:
+ *
+ * hash,path,byteLength,stringLength,demoted
+ *
+ * hash: the hash of the blob
+ * path: the path of the blob in the blob store
+ * byteLength: the byte length of the blob, or empty if unknown
+ * stringLength: the string length of the blob, or empty if unknown
+ * demoted: true if the blob has been demoted to a reference, false otherwise
+ */
+
+// @ts-check
+import { ObjectId } from 'mongodb'
+import { GLOBAL_BLOBS, loadGlobalBlobs } from '../lib/blob_store/index.js'
+import { client } from '../lib/mongodb.js'
+import commandLineArgs from 'command-line-args'
+import fs from 'node:fs'
+
+// Enable caching for ObjectId.toString()
+ObjectId.cacheHexString = true
+
+function parseArgs() {
+  const args = commandLineArgs([
+    {
+      name: 'output',
+      type: String,
+      alias: 'o',
+    },
+  ])
+  const OUTPUT_STREAM = fs.createWriteStream(args['output'], { flags: 'wx' })
+
+  return {
+    OUTPUT_STREAM,
+  }
+}
+
+const { OUTPUT_STREAM } = parseArgs()
+
+async function main() {
+  await loadGlobalBlobs()
+  OUTPUT_STREAM.write('hash,path,byteLength,stringLength,demoted\n')
+  for (const [hash, { blob, demoted }] of GLOBAL_BLOBS) {
+    const { hash: blobHash, byteLength, stringLength } = blob
+    if (blobHash !== hash) {
+      throw new Error(`hash mismatch: ${hash} !== ${blobHash}`)
+    }
+    const path = blobHash.slice(0, 2) + '/' + blobHash.slice(2)
+    const byteLengthStr = byteLength === null ? '' : byteLength
+    const stringLengthStr = stringLength === null ? '' : stringLength
+    OUTPUT_STREAM.write(
+      `${hash},${path},${byteLengthStr},${stringLengthStr},${demoted}\n`
+    )
+  }
+}
+
+main()
+  .then(() => console.log('Done.'))
+  .catch(err => {
+    console.error('Error:', err)
+    process.exitCode = 1
+  })
+  .finally(() => {
+    client.close().catch(err => console.error('Error closing MongoDB:', err))
+  })