How it works


A simple dictionary app built using Flutter. Like knadh/dictmaker but for mobile apps.

Powers Olam and upcoming Alar apps.

Olam, Datuk datasets are downloaded, processed and binary packed. This is added to the app assets.

To download data sets run:

$ make download
# output:
#   scripts/olam.csv
#   scripts/datuk.yaml

To process these and generate assets:

$ make db
# output:
#   assets/datuk_defs.bin
#   assets/datuk_words.bin
#   assets/olam_defs.bin
#   assets/olam_words.bin

Make targets olam, alar and ditto changes the app name, bundle ID and icon. This is to make use of the same codebase to build multiple apps.

To build Olam:

$ make olam
$ flutter build apk

To generate models ditto uses dartgen, a WIP inline code generator collection for dart.

Install dartgen:

pub global activate --source git

After changing models run:

$ dartgen

How it works

The generator script parses and combines datasets and exports to app's asset directory. The main Olam dataset is in the format:

# id	english_word	part_of_speech	malayalam_definition
174569	.net	n	പുത്തന്‍ കമ്പ്യൂട്ടര്‍ സാങ്കേതികത ഭാഷ

Datuk and Alar datasets are in a YAML format described here.

Since the format and size of the content is known before hand ditto binary packs these datasets into a word index (olam_words.bin) and a definition file(olam_defs.bin). Word index contains offsets to the definitions in the definition file. Words are sorted alphabetically in the index file. This is shown as it is in the app's index page.

For searching app first performs a binary search on the sorted word index and then ranks results based on the phonetic hash of corresponding langauge - MLPhone for Malayalam, KNPhone for Kannada and DoubleMetaphone for English.


ditto is licensed under the GPL 3.0 license