Merge pull request #3306 from overleaf/jpa-script-validate-data-of-model

[scripts] validate-data-of-model: add the new script

GitOrigin-RevId: c432598a1efee58e95f3751ac0f63113ec837344

services/web/scripts/helpers/batchedUpdate.js

@@ -8,6 +8,7 @@ if (process.env.BATCH_LAST_ID) {
 }
 
 async function getNextBatch(collection, query, maxId, projection) {
+  maxId = maxId || BATCH_LAST_ID
   if (maxId) {
     query['_id'] = { $gt: maxId }
   }
@@ -35,7 +36,7 @@ async function batchedUpdate(collectionName, query, update, projection) {
   projection = projection || { _id: 1 }
   let nextBatch
   let updated = 0
-  let maxId = BATCH_LAST_ID
+  let maxId
   while (
     (nextBatch = await getNextBatch(collection, query, maxId, projection))
       .length
@@ -66,6 +67,7 @@ function batchedUpdateWithResultHandling(collection, query, update) {
 }
 
 module.exports = {
+  getNextBatch,
   batchedUpdate,
   batchedUpdateWithResultHandling
 }

services/web/scripts/validate-data-of-model.js

@@ -0,0 +1,47 @@
+const { getNextBatch } = require('./helpers/batchedUpdate')
+const { db, waitForDb } = require('../app/src/infrastructure/mongodb')
+
+const MODEL_NAME = process.argv.pop()
+const Model = require(`../app/src/models/${MODEL_NAME}`)[MODEL_NAME]
+
+function processBatch(batch) {
+  for (const doc of batch) {
+    const error = new Model(doc).validateSync()
+    if (error) {
+      const { errors } = error
+      console.log(JSON.stringify({ _id: doc._id, errors }))
+    }
+  }
+}
+
+async function main() {
+  await waitForDb()
+  const collection = db[Model.collection.name]
+
+  const query = {}
+  const projection = {}
+
+  let nextBatch
+  let processed = 0
+  let maxId
+  while (
+    (nextBatch = await getNextBatch(collection, query, maxId, projection))
+      .length
+  ) {
+    maxId = nextBatch[nextBatch.length - 1]._id
+    processed += nextBatch.length
+    console.error(maxId, processed)
+
+    processBatch(nextBatch)
+  }
+  console.error('done')
+}
+
+main()
+  .then(() => {
+    process.exit(0)
+  })
+  .catch(error => {
+    console.error({ error })
+    process.exit(1)
+  })