Gå til hovedinnhold

Translation Table format (.tab, .tsv)

TransTable v1 is the tab-separated bilingual exchange table that Okapi's generic writer produces for its "Translation Table" output. The first line is a header declaring the signature (TransTableV1) and the source and target locales as three tab-separated cells. Each following line is a context crumb (okpCtx:tu=<id>) followed by a source cell and an optional target cell.

The reader extracts one translatable block per text unit, pairing the source and target cells. Cells may be wrapped in double quotes, which are stripped on read; the literal sequences \t and \n inside a cell are unescaped to tab and newline. Whitespace-only lines are skipped. A crumb may carry a segment suffix (:s=<seg-id>); rows that share the same tu=<id> and carry segment suffixes are merged into a single segmented text unit. A missing or malformed header aborts the read.

Configuration is supplied on the collection's format: block in a .kapi project. The one key is allowSegments (default true): when enabled, :s=<seg-id> suffixes are honoured and rows sharing a tu=<id> merge into one segmented unit; when disabled, every row is its own text unit regardless of any segment suffix. It mirrors Okapi's allowSegments parameter.

IDtranstable
SourceBuilt-in
Extensions.tab, .tsv
MIME Typestext/tab-separated-values
CapabilitiesRead + Write

This format has no configurable parameters.

Examples

Read a bilingual table without segment grouping

Disable segment grouping so each row becomes its own text unit, regardless of any :s=<seg-id> suffix on the crumb.

format:
  name: transtable
  config:
    allowSegments: false

Processing notes

  • One block per text unit; source and target cells pair into source and target segments.

  • Surrounding double quotes are stripped and \t / \n escapes are decoded on read.

  • With allowSegments enabled, rows sharing a tu=<id> and carrying :s=<seg-id> suffixes merge into one segmented block.

Limitations

  • The header must be a valid TransTableV1 signature line; otherwise the read fails.

  • Only the v1 table shape is supported.

← Back to the Format Reference