docs

Test text discovery

For Linguana contributors and beta evaluators.

Extraction does not require a token or provider.

Configure a free build

Omit token and translation, and keep upload: false. Then run source guards around the build:

bun run source:before
bun run build
bun run source:after

Inspect the artifact

Open .linguana/manifest.json and verify schema version, project, environment, source locale, source revision, messages, diagnostics, and locations. Run the build twice; the hash and stable IDs should not change without semantic copy changes.

Test a policy corpus

Add fixtures for literal text, allowed attributes, data-no-translate, data-translate, hidden values, URL-like strings, ICU placeholders, rich content, and conflicting explicit IDs. Assert eligible messages appear and excluded content does not.

CI strictness

Start with an object strict policy so duplicate IDs fail while expected conservative skips stay visible. Promote additional categories only after the current diagnostic set is clean.

Failure and next

If a build attempts upload, check both upload and translation; either remote mode requires API URL and token. See Build diagnostics.

Did this page get you to a working result?

On this page

No Headings