WFDASH-WEBAPP/scripts/migrations/2026-06-30-backfill-price-list-search-text.ts

/**
 * One-time backfill — rebuild `price_list.search_text` for every
 * existing row using the comprehensive builder in
 * `src/lib/price-list-search.ts`, AND fix the `supplier` column on
 * legacy rows where it stores a PB id instead of the resolved name.
 *
 * Why this exists alongside the search_text migration:
 *   The search_text migration only ADDS the column + index. It
 *   does not rewrite rows — adding the column left the existing
 *   46 rows with `search_text = ""` and `supplier = "q6hpwbxsicyoiuu"`.
 *
 *   This script:
 *     1. Resolves the supplier name from raw_price_list →
 *        suppliers expansion (so `supplier` becomes the human
 *        name, not the PB id).
 *     2. Rebuilds `search_text` from EVERY text column on the
 *        row PLUS the full `specifications` JSON PLUS the
 *        tokenised number columns PLUS the resolved supplier name.
 *
 *   Together with `getPriceLists` (which now ORs against every
 *   text column AND the search_text blob), typing anything —
 *   "jeisa", "12mm", "118", "slugger", "turbo steel", "each" —
 *   matches.
 *
 * Idempotent: re-runs only PATCH a row if the rebuilt value differs
 * from what's currently stored. Safe to run on every deploy.
 *
 * Run with:
 *   npm run pb:backfill:price-list-search-text
 */
import PocketBase from "pocketbase";
import {
  buildPriceListSearchText,
  type SearchTextSource,
} from "../../src/lib/price-list-search";

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

if (!ADMIN_EMAIL || !ADMIN_PASSWORD) {
  throw new Error("PB_ADMIN_EMAIL and PB_ADMIN_PASSWORD are required.");
}

type PriceListRow = SearchTextSource & {
  id: string;
  supplier?: string;
  source_file?: string;
};

type RawPriceListRow = {
  id: string;
  supplier_name?: string | null;
  expand?: {
    supplier_name?: {
      id: string;
      suppliername?: string;
    };
  };
};

/**
 * Look up the supplier name for a price_list row by following
 * `source_file` → `raw_price_list` → `expand.supplier_name`.
 * Returns the resolved name (e.g. "JEISA - WESTSCOTT GROUP") or
 * null if the chain is broken (manually-created row with no
 * source_file, or the raw_price_list was deleted).
 */
async function resolveSupplierName(
  pb: PocketBase,
  sourceFile: string | undefined,
  fallback: string | undefined,
): Promise<string | null> {
  if (!sourceFile) return fallback ?? null;
  try {
    const raw = await pb
      .collection("raw_price_list")
      .getOne<RawPriceListRow>(sourceFile, { expand: "supplier_name" });
    const name = raw.expand?.supplier_name?.suppliername;
    if (name) return name;
  } catch {
    // raw_price_list row is gone — fall back to whatever the
    // price_list row stored (could be an old id or a real name).
  }
  return fallback ?? null;
}

/**
 * PB ids are 15 lowercase alphanumeric characters (a-z, 0-9).
 * Used to decide whether a legacy `supplier` value is actually
 * an id that the row expects the API to expand.
 */
function looksLikePbId(value: string | undefined): boolean {
  return !!value && /^[a-z0-9]{15}$/.test(value);
}

async function main() {
  const pb = new PocketBase(PB_URL);
  await pb.collection("_superusers").authWithPassword(ADMIN_EMAIL!, ADMIN_PASSWORD!);

  let page = 1;
  const perPage = 200;
  let scanned = 0;
  let patchedSearch = 0;
  let patchedSupplier = 0;
  let unchanged = 0;
  const sampleRebuilds: Array<{ id: string; before: string; after: string }> = [];

  // One pagination loop over every price_list row.
  while (true) {
    const result = await pb.collection("price_list").getList<PriceListRow>(page, perPage, {
      sort: "-updated",
    });
    if (result.items.length === 0) break;

    for (const row of result.items) {
      scanned++;

      // 1. Resolve supplier name from raw_price_list.
      const resolvedSupplier = await resolveSupplierName(
        pb,
        row.source_file,
        looksLikePbId(row.supplier) ? "" : row.supplier,
      );

      // 2. Build the new search_text.
      const newSearchText = buildPriceListSearchText(row, resolvedSupplier);

      // 3. Decide what to PATCH.
      const patch: Record<string, string> = {};
      if ((row as unknown as Record<string, unknown>).search_text !== newSearchText) {
        patch.search_text = newSearchText;
      }
      if (
        resolvedSupplier &&
        resolvedSupplier !== row.supplier
      ) {
        patch.supplier = resolvedSupplier;
      }

      if (Object.keys(patch).length === 0) {
        unchanged++;
      } else {
        if (patch.search_text !== undefined) patchedSearch++;
        if (patch.supplier !== undefined) patchedSupplier++;
        if (sampleRebuilds.length < 3 && patch.search_text !== undefined) {
          sampleRebuilds.push({
            id: row.id,
            before: String((row as unknown as Record<string, unknown>).search_text ?? ""),
            after: newSearchText,
          });
        }
        await pb.collection("price_list").update(row.id, patch);
      }
    }

    if (result.items.length < perPage) break;
    page++;
  }

  console.log(`[backfill] scanned ${scanned} rows across ${page} page(s).`);
  console.log(`[backfill] patched search_text: ${patchedSearch}, supplier: ${patchedSupplier}, unchanged: ${unchanged}`);
  if (sampleRebuilds.length > 0) {
    console.log("[backfill] sample rebuilds:");
    for (const s of sampleRebuilds) {
      console.log(`  ${s.id}:`);
      console.log(`    before: ${s.before.slice(0, 120)}${s.before.length > 120 ? "…" : ""}`);
      console.log(`    after:  ${s.after.slice(0, 240)}${s.after.length > 240 ? "…" : ""}`);
    }
  }
  console.log("[backfill] done.");
}

main().catch((err) => {
  console.error(err);
  process.exit(1);
});