fix(file-search): store workspace file revisions at millisecond precision (#7941)

* fix(file-search): store workspace file revisions at millisecond precision

The content revision round-trips through JavaScript Date and JSON, both of
which truncate to milliseconds, while now() stores microseconds. A
sub-millisecond revision stops comparing equal to its own round-trip, so
every SQL equality keyed on it matches zero rows and the file is never
claimed, indexed, or cleaned up.

Truncate at the column default and enforce it with a BEFORE trigger for the
writers a default cannot reach, then retire the rows minted before the
invariant existed in a batched script migration.

* fix(file-search): normalize revisions promoted by metadata-only writes

A chat upload is materialized into a workspace file by setting context
alone, so the revision is never written and the search-index trigger would
key the promoted file to a value nothing can claim. Fire the normalizer on
every update; `UPDATE OF` is evaluated against the statement's target
columns, so this does not demote provenance on a metadata write.

Accept the shared integration database in the repair suite the way the
sibling suites do, and record the new script migration in the registry
order assertions.

* chore(db): drop the 0357 migration ahead of renumbering

* chore(db): renumber the content revision migration to 0358

Staging took 0357 for the outbox autovacuum migration. The SQL is byte for
byte the previously reviewed file; only the number and the references to it
changed.

Author: waleedlatif1

.github/workflows/test-build.yml

@@ -222,7 +222,10 @@ jobs:
         working-directory: packages/db
         env:
           KNOWLEDGE_ACL_TEST_DATABASE_URL: postgresql://postgres:postgres@127.0.0.1:5432/sim_auth_scim
-        run: bunx vitest run script-migrations/0016_backfill_search_vectors.postgres.test.ts
+        run: >-
+          bunx vitest run
+          script-migrations/0016_backfill_search_vectors.postgres.test.ts
+          script-migrations/0018_repair_workspace_file_content_revision.postgres.test.ts
 
       - name: Verify Search progress, pagination, and outbox scheduling in PostgreSQL
         working-directory: apps/sim

packages/db/migrations/0358_workspace_file_content_version_precision.sql

@@ -0,0 +1,29 @@
+ALTER TABLE "workspace_files" ALTER COLUMN "content_updated_at" SET DEFAULT date_trunc('milliseconds', now());--> statement-breakpoint
+-- The trigger is the invariant; the truncating default above only keeps the common insert path from
+-- paying for it. A default cannot cover the writers that matter here: explicit `CURRENT_TIMESTAMP`
+-- expressions and raw SQL inserts.
+--
+-- It fires on every UPDATE rather than `UPDATE OF content_updated_at`, because a row can enter the
+-- search index without its revision being written: materializing a chat upload sets `context` alone, and
+-- `workspace_files_search_index_pending` then indexes whatever revision the row already carried. `UPDATE
+-- OF` is evaluated against the statement's target columns, so widening it does not make
+-- `workspace_files_secret_provenance_demote` fire on those metadata writes — normalizing a legacy row
+-- keeps its tracked provenance. The WHEN guard keeps the plpgsql call off the hot path for the compliant
+-- writes that are the norm, and the name sorts before both other triggers so they see the normalized
+-- value. Rows minted before this runs are retired by script migration
+-- `0018_repair_workspace_file_content_revision`, which the runner applies after every SQL migration.
+CREATE OR REPLACE FUNCTION workspace_file_content_version_millisecond()
+RETURNS trigger
+LANGUAGE plpgsql
+AS $$
+BEGIN
+	NEW.content_updated_at := date_trunc('milliseconds', NEW.content_updated_at);
+	RETURN NEW;
+END;
+$$;--> statement-breakpoint
+DROP TRIGGER IF EXISTS workspace_files_content_version_millisecond ON workspace_files;--> statement-breakpoint
+CREATE TRIGGER workspace_files_content_version_millisecond
+BEFORE INSERT OR UPDATE ON workspace_files
+FOR EACH ROW
+WHEN (NEW.content_updated_at <> date_trunc('milliseconds', NEW.content_updated_at))
+EXECUTE FUNCTION workspace_file_content_version_millisecond();