MattiLehtinen / postgrator-cli / 29697552225

93%

push

github

feat(sqlite3)!: use built-in node:sqlite instead of the sqlite3 package (#370)

* feat(sqlite3)!: use built-in node:sqlite instead of the sqlite3 package

The `sqlite3` npm package pulls in native prebuilt binaries and hasn't
been needed since Node gained its own built-in SQLite bindings. Swap
lib/clients/sqlite3.js to node:sqlite's DatabaseSync, which also
handles multi-statement scripts (`db.exec`) without the workaround
added in #313.

BREAKING CHANGE: node:sqlite requires Node.js >= 22.13.0 (or >= 23.4.0
on odd-numbered releases), so this drops support for Node 20 and raises
the floor from Node 22.0.0. `sqlite3` is no longer an optional peer
dependency; users no longer need to `npm install sqlite3` for sqlite3
support.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

* fix(lint): raise node floor to 22.16 to drop import.meta.dirname suppressions

import.meta.dirname/filename only became non-experimental at Node
22.16.0/24.0.0, so bumping the engines floor from 22.13.0 to 22.16.0
(and 23.4.0 to 24.0.0, since 23.x is EOL anyway) lets eslint-plugin-n
confirm it's safe across the whole supported range, removing all 15
now-unused eslint-disable directives in the test file.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

* refactor(sqlite3): drop unnecessary async from the node:sqlite client

DatabaseSync's methods are synchronous (node:sqlite has no async
database API at all, only a plain function), and every real call site
(Postgrator's Client.js runQuery/runSqlScript, and this repo's own
test helpers) already invokes query/execSqlScript/end from within
another async function, so a synchronous throw here still surfaces
as a normal rejection to callers. No need to wrap these in async.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>

134 of 153 branches covered (87.58%)

Branch coverage included in aggregate %.

6 of 6 new or added lines in 1 file covered. (100.0%)

582 of 619 relevant lines covered (94.02%)

538.06 hits per line