Dictionary Packager

How it works

1. Drop a StarDict dictionary — the three files name.ifo, name.idx, and name.dict (or all of them inside a .zip)
2. We parse the .ifo metadata, validate the word index, strip any macOS resource-fork junk (.DS_Store, ._name), and pre-compute the sparse offset .cache file so your device doesn't have to scan the whole index on first lookup
3. Download the cleaned .zip and extract it to /dictionary/ on your SUMI SD card
4. In SUMI, open any book → hold Center to open Reader Settings → pick Look up Word. Or use the Dictionary plugin from Home for keyboard search.

SUMI expects the layout /dictionary/<name>/<name>.idx etc. The <name> is what shows up in Settings — you can rename it below before downloading.

Why package dictionaries through the site

Raw StarDict files usually work, but there are three common problems we fix here:

• macOS resource forks. If you unzip a dictionary archive on macOS and copy it to SD card, finder creates ._name.idx shadow files. SUMI's multi-dictionary scanner treats every folder as a candidate dictionary — these shadows show up as ghost entries and break the folder list. We strip them.
• First-lookup latency. On a fresh dictionary, the first word lookup has to scan the whole .idx file to build a sparse offset table (one offset per 512 words). For a 200k-word dictionary that's a 2-3 second delay. We pre-compute the .cache file client-side so the very first lookup is instant.
• Bad .idx sort order. Some third-party dictionaries ship with non-sorted indexes, which breaks the binary search. We scan the whole index and flag this before you copy the files to your device.