From 2a61f3d4424b85ce2752455f0dafac8a1e1dbfa7 Mon Sep 17 00:00:00 2001
From: TomasC <49659410+tx0c@users.noreply.github.com>
Date: Thu, 29 Feb 2024 23:45:01 +0000
Subject: [PATCH] for back fill short-hash for past existing articles

for thematters/matters-server#3808

eval run:

```console
fill 1000 rows in 3854ms;
```
---
 bin/fill-short-hash.ts | 71 ++++++++++++++++++++++++++++++++++++++++++
 lib/db.ts              |  2 ++
 2 files changed, 73 insertions(+)
 create mode 100644 bin/fill-short-hash.ts

diff --git a/bin/fill-short-hash.ts b/bin/fill-short-hash.ts
new file mode 100644
index 0000000..f9c7867
--- /dev/null
+++ b/bin/fill-short-hash.ts
@@ -0,0 +1,71 @@
+#!/usr/bin/env -S node --trace-warnings --loader ts-node/esm
+
+import { sql, ARRAY_TYPE } from "../lib/db.js";
+
+// as same as https://github.com/thematters/matters-server/blob/develop/src/common/utils/nanoid.ts
+import { customAlphabet } from "nanoid";
+
+const ALPHABET = "0123456789abcdefghijklmnopqrstuvwxyz";
+
+const nanoid = customAlphabet(ALPHABET, 12); // ~35 years or 308M IDs needed, in order to have a 1% probability of one collision // https://zelark.github.io/nano-id-cc/
+
+const BATCH_SIZE = Number.parseInt(process.env.BATCH_SIZE || "", 10) || 100;
+
+async function main() {
+  const args = process.argv.slice(2);
+
+  // const [row] = await sql`SELECT VERSION(), CURRENT_TIMESTAMP`;
+  // console.log(new Date(), `get version:`, row);
+
+  const updated = await batchUpdateShortHashes(BATCH_SIZE);
+  // if (updated === 0) // set non null at the end;
+}
+
+async function batchUpdateShortHashes(BATCH_SIZE = 100) {
+  const started = Date.now();
+
+  const rows = await sql`-- batch fill short-hash for article table
+WITH last_batch AS (
+  SELECT id, title, short_hash,
+    ROW_NUMBER() OVER(ORDER BY id DESC) ::int seq
+  FROM article
+  WHERE short_hash IS NULL
+  ORDER BY id DESC LIMIT ${BATCH_SIZE}
+), last_batch_with_hash AS (
+  SELECT id, title, cte.*
+  FROM last_batch JOIN (
+    SELECT *
+    FROM UNNEST(
+      ${sql.array(
+        Array.from({ length: BATCH_SIZE }).map((_, seq) => seq + 1),
+        ARRAY_TYPE
+      )} ::int[],
+      ${sql.array(
+        Array.from({ length: BATCH_SIZE }).map(() => nanoid()),
+        ARRAY_TYPE
+      )} ::text[]
+    ) AS t(seq, short_hash)
+  ) cte USING(seq)
+)
+
+UPDATE article
+SET short_hash=last_batch_with_hash.short_hash
+FROM last_batch_with_hash
+WHERE article.id=last_batch_with_hash.id
+RETURNING article.* ; `;
+
+  const ended = new Date();
+  console.log(
+    ended,
+    `filled ${rows.length} rows in ${+ended - started}ms:`
+    // rows.slice(0, 3)
+  );
+
+  return rows.length;
+}
+
+async function alterSchemaNonNullable() {
+  await sql`ALTER TABLE article ALTER COLUMN short_hash SET NOT NULL ; `;
+}
+
+main().catch((err) => console.error(new Date(), "ERROR:", err));
diff --git a/lib/db.ts b/lib/db.ts
index 05ff21b..f6d1cae 100644
--- a/lib/db.ts
+++ b/lib/db.ts
@@ -33,6 +33,8 @@ export const sqlSIW = getPostgresJsClient(pgSearchDatabaseURL);
 
 // import { Article } from "../lib/pg-zhparser-articles-indexer.js";
 
+export const ARRAY_TYPE = 1009; // Postgres internal value shouldn't be here;
+
 export interface Article {
   id: string;
   articleId: string;