From ff40d09116eb3f10925318bb51ff444201ff2bc6 Mon Sep 17 00:00:00 2001
From: zkldi <ktchidev@gmail.com>
Date: Fri, 1 Oct 2021 18:22:43 +0100
Subject: [PATCH] Update anonymise DB scripts to be way more stable.

---
 ...nymise-db.ts => clone-and-anonymise-db.ts} | 45 ++++++++++++++-----
 1 file changed, 34 insertions(+), 11 deletions(-)
 rename server/scripts/{anonymise-db.ts => clone-and-anonymise-db.ts} (57%)

diff --git a/server/scripts/anonymise-db.ts b/server/scripts/clone-and-anonymise-db.ts
similarity index 57%
rename from server/scripts/anonymise-db.ts
rename to server/scripts/clone-and-anonymise-db.ts
index f7250001a..f4aa265b6 100644
--- a/server/scripts/anonymise-db.ts
+++ b/server/scripts/clone-and-anonymise-db.ts
@@ -1,8 +1,20 @@
-//
-//
-
 import CreateLogCtx from "lib/logger/logger";
 import monk from "monk";
+import { execSync } from "child_process";
+
+import { Command } from "commander";
+
+const program = new Command();
+program
+	.option("--nsFrom <Database to clone from.>")
+	.option("--nsTo <Database to clone and anonymise to.>");
+
+program.parse(process.argv);
+const options = program.opts();
+
+if (!options.nsTo || !options.nsFrom) {
+	throw new Error(`Expected --nsFrom and --nsTo options.`);
+}
 
 const logger = CreateLogCtx(__filename);
 
@@ -10,15 +22,14 @@ const logger = CreateLogCtx(__filename);
  * Strips all name and private information from the database.
  * For use for things like future tachi-db-dumps?
  *
- * @param connectionString The database to anonymise. DO NOT PASS THE PRODUCTION DATABASE TO THIS!
- * Clone the database with mongodump->mongorestore (This is the documented way to clone a db.)
+ * @param nsTo The database to anonymise. DO NOT PASS THE PRODUCTION DATABASE TO THIS!
  */
-async function CloneAndAnonymiseDB(connectionString: string) {
-	logger.info(`Connecting to ${connectionString}.`);
+async function AnonymiseDB(nsTo: string) {
+	logger.info(`Connecting to ${nsTo}.`);
 
-	const clonedDB = monk(connectionString);
+	const clonedDB = monk(`127.0.0.1/${nsTo}`);
 
-	logger.info(`Connected to ${connectionString}.`);
+	logger.info(`Connected to ${nsTo}.`);
 
 	const r1 = await clonedDB.get("user-private-information").update(
 		{},
@@ -75,9 +86,21 @@ async function CloneAndAnonymiseDB(connectionString: string) {
 
 		logger.info(`Removed all ${collection} documents.`, { r3 });
 	}
+
+	logger.info(`Done! Closing.`);
+
+	process.exit(0);
+}
+
+function CloneDB(nsFrom: string, nsTo: string) {
+	logger.info(`Cloning Database. This will take a while.`);
+	execSync(`mongodump --gzip --archive=temp-dump --db=${nsFrom}`);
+	execSync(`mongorestore --archive=temp-dump --gzip --nsFrom='${nsFrom}.*' --nsTo='${nsTo}.*'`);
+	execSync(`rm temp-dump`);
+	logger.info(`Cloned!`);
 }
 
 if (require.main === module) {
-	// Maybe turn this into a commander program, it doesn't really matter.
-	CloneAndAnonymiseDB(process.argv[2]);
+	CloneDB(options.nsFrom, options.nsTo);
+	AnonymiseDB(options.nsTo);
 }