WFDASH-WEBAPP/scripts/migrations/add-quote-items-user-edited.ts

/**
 * One-shot migration: add `user_edited` to the live `quote_items`
 * collection (idempotent) and backfill existing rows.
 *
 * Run with:
 *
 *     npx tsx scripts/migrations/add-quote-items-user-edited.ts
 *
 * Backfill rule:
 *   - Items with a non-empty description that were UPDATED after
 *     creation (i.e. a human saved them via the process page) are
 *     set to `user_edited = true`.
 *   - Items that were created and never updated (auto-extracted from
 *     the email pipeline) are left as `user_edited = false`. The
 *     rewriter gate will refuse them until the user edits + saves
 *     them in the process page — which is the intended behaviour.
 *
 * The script is safe to run multiple times.
 */
import PocketBase from "pocketbase";

const PB_URL = process.env.PB_URL || "http://127.0.0.1:8090";
const ADMIN_EMAIL = process.env.PB_ADMIN_EMAIL;
const ADMIN_PASSWORD = process.env.PB_ADMIN_PASSWORD;

async function main() {
  if (!ADMIN_EMAIL || !ADMIN_PASSWORD) {
    throw new Error("Set PB_ADMIN_EMAIL and PB_ADMIN_PASSWORD env vars.");
  }

  const pb = new PocketBase(PB_URL);
  await pb.collection("_superusers").authWithPassword(ADMIN_EMAIL, ADMIN_PASSWORD);
  console.log(`[user_edited] Authenticated against ${PB_URL}`);

  // 1. Ensure the field exists on the collection.
  const collection = await pb.collections.getOne("quote_items");
  const hasField = collection.fields.some(
    (f: { name: string }) => f.name === "user_edited"
  );

  if (!hasField) {
    console.log("[user_edited] Field missing — adding it to quote_items.");
    await pb.collections.update(collection.id, {
      fields: [
        ...collection.fields,
        {
          name: "user_edited",
          type: "bool",
          required: false,
          presentable: false,
        },
      ],
    });
    console.log("[user_edited] Field added.");
  } else {
    console.log("[user_edited] Field already exists — skipping schema update.");
  }

  // 2. Backfill. Pull in pages of 200 so we don't OOM on large tables.
  let page = 1;
  let scanned = 0;
  let updated = 0;
  let skippedAlreadyTrue = 0;
  let skippedMachine = 0;

  while (true) {
    const batch = await pb.collection("quote_items").getList(page, 200, {
      fields: "id,description,created,updated,user_edited",
    });
    if (batch.items.length === 0) break;

    for (const item of batch.items) {
      scanned++;
      if (item.user_edited === true) {
        skippedAlreadyTrue++;
        continue;
      }
      // Heuristic: a non-empty description AND a meaningful gap
      // between created and updated timestamps (PB stores ISO; Date
      // diff > 1000ms is enough) means a human saved this row at
      // least once. Empty description or created==updated means it's
      // still in its auto-extracted state.
      const hasDesc =
        typeof item.description === "string" && item.description.trim().length > 0;
      const humanSaved =
        hasDesc &&
        new Date(item.updated).getTime() - new Date(item.created).getTime() >
          1000;

      if (!humanSaved) {
        skippedMachine++;
        continue;
      }

      await pb.collection("quote_items").update(item.id, { user_edited: true });
      updated++;
    }

    if (batch.page >= batch.totalPages) break;
    page++;
  }

  console.log(`[user_edited] Scanned ${scanned} items.`);
  console.log(`[user_edited]   Flipped to true: ${updated}`);
  console.log(`[user_edited]   Already true:      ${skippedAlreadyTrue}`);
  console.log(`[user_edited]   Left as false (machine / no description): ${skippedMachine}`);
  console.log("[user_edited] Done.");
}

main().catch((err) => {
  console.error("[user_edited] Migration failed:", err);
  process.exit(1);
});