Hello Emmanuel,
Looks to be a really nice app. I'm not sure I
fully understood all the
features, do you confirm it's not a tool to provide word translations
(except to English)?
it's not so much a translation software (like Google Translate), there
is no way to translate
sentences or parts of sentences, but you can look up single word
translations (or whatever the
stored unit on Wiktionary is). Its main purpose is to make a subset of
Wiktionary available offline
(and adapt it for mobile consumption).
Other question: is the software opensource? We, at
Kiwix, are in
particular interested in the scrapper.
The iOS app itself is not open source, but big parts of the extraction
software are. the main work is done by
JWKTL (Java-based Wiktionary Library,
https://github.com/dkpro/dkpro-jwktl), where I'm also a contributor.
JWKTL parses the data into a BerkeleyDB file; I wrote a small adapter to
use SQLite instead.
It should be easy to do something similar for the ZIM format.
However, more recently I've started to move away from JWKTL since a some
data cannot be extracted from
the raw mediawiki markup alone. A good example are pronunciations, which
are often entirely generated in Lua.
For this I use a combination of microformats2 and a simple Python
parser. See this thread on phabricator
for more details:
https://phabricator.wikimedia.org/T138709#3048681
This often requires some minor changes to templates or modules. My plan
is to add special
markup for the most important features, which should also help with the
transition to Wikidata.
Jan