From 20bd57e9bd005fb7ada0eb54f1239a485044db00 Mon Sep 17 00:00:00 2001
From: Eric Mc Sween <5454374+emcsween@users.noreply.github.com>
Date: Mon, 8 Jan 2024 13:34:54 -0500
Subject: [PATCH] Merge pull request #16314 from
 overleaf/em-migrate-docops-versions

Migrate doc versions from docOps to docs

GitOrigin-RevId: 34b0130e9f0849b8e16f1febc38c78b55d51a4c1
---
 ...0_move_doc_versions_from_docops_to_docs.js | 45 +++++++++++++++++++
 1 file changed, 45 insertions(+)
 create mode 100644 services/web/migrations/20231219081700_move_doc_versions_from_docops_to_docs.js

diff --git a/services/web/migrations/20231219081700_move_doc_versions_from_docops_to_docs.js b/services/web/migrations/20231219081700_move_doc_versions_from_docops_to_docs.js
new file mode 100644
index 0000000000..e4a37093db
--- /dev/null
+++ b/services/web/migrations/20231219081700_move_doc_versions_from_docops_to_docs.js
@@ -0,0 +1,45 @@
+const { ReadPreference } = require('mongodb')
+
+const BATCH_SIZE = parseInt(process.env.BATCH_SIZE || '1000', 10)
+
+exports.tags = ['server-ce', 'server-pro', 'saas']
+
+exports.migrate = async ({ db }) => {
+  const records = db.docOps.find(
+    {},
+    { readPreference: ReadPreference.secondaryPreferred }
+  )
+
+  let docsProcessed = 0
+  let batch = []
+  for await (const record of records) {
+    const docId = record.doc_id
+    const version = record.version
+    batch.push({
+      updateOne: {
+        filter: {
+          _id: docId,
+          version: { $exists: false },
+        },
+        update: { $set: { version } },
+      },
+    })
+    if (batch.length >= BATCH_SIZE) {
+      await db.docs.bulkWrite(batch, { ordered: false })
+      batch = []
+    }
+    docsProcessed += 1
+    if (docsProcessed % 100000 === 0) {
+      console.log(`${docsProcessed} docs processed`)
+    }
+  }
+  if (batch.length > 0) {
+    await db.docs.bulkWrite(batch, { ordered: false })
+  }
+}
+
+exports.rollback = async ({ db }) => {
+  // Nothing to do on rollback. We don't want to remove versions from the docs
+  // collection because they might be more current than the ones in the docOps
+  // collection.
+}