Skip to main content

​
Usage

These commands are flag-only. Hyperlocalise does not read a Lokalise YAML config in this version. You can pass --config to reuse projectID and token settings from an i18n.yml storage block.

​
What this command family does

These commands talk to the Lokalise API v2 file and glossary endpoints. They are separate from Hyperlocalise native i18n.yml and sync push / sync pull workflows. Use native sync when you want Hyperlocalise-managed entry sync through the Lokalise adapter. Use this surface when you want flag-driven file upload and download against a Lokalise project. Use them for:
  • source-file upload (upload sources)
  • import of a pre-translated locale file (upload translations)
  • source download (download sources)
  • translation download (download translations)
  • glossary download (Hyperlocalise-shaped CSV)
Official lokalise2 uses file upload / file download with --lang-iso. Hyperlocalise keeps the same command family shape as Crowdin, Phrase, and Smartling (upload sources, download translations).

​
Credentials

Hyperlocalise reads --token-env (default LOKALISE_API_TOKEN). Do not put the API token in i18n.yml. Point apiTokenEnv at an environment variable instead.

​
Command notes

upload sources requires --project-id (or projectID in --config), --source-locale (or sourceLanguage in --config), and at least one --file. --dry-run checks files and flags and does not need a token. upload translations requires --target-locale on every run. It does not take the locale from sourceLanguage or targetLanguages in --config. Config may still supply the project id and token. The locale comes from the flag, not the filename: uploading en.json as French uses --target-locale fr. --replace-modified maps to Lokalise replace_modified. Off (the default) preserves translations Lokalise treats as modified. On replaces those modified translations from the uploaded file. It is not a guarantee that every existing key stays unchanged. Dry-run and the success line print modified_translations=preserved or modified_translations=replaced. Success means the import is queued, not that strings are already visible in Lokalise. The command prints a process id and stops. It does not wait for the background import to finish. Lokalise matches files by name. Use the same filename as the source file when you can. Language codes in the name (en.json / fr.json) are usually treated as one file. A totally different name can create a second file. --file can repeat for several files in that one locale. Use a separate invocation for each target locale. download translations can take more than one --target-locale. Use %locale% in --output when you download more than one locale to files. glossary download writes CSV to stdout unless you pass --output.

​
Crowdin and lokalise2 gap inventory

​
Examples

Upload a source file, then a French translation that reuses the source filename:
Download French translations: