two tables, dict and mean, joined by dict.id = mean.did

• dict contains the thai word
• mean contains one or more english translations for each thai word

dict.g is type

• o - one syllable word or syllable
• m - multi-syllable word
• s - symbol, ie, ๆ

pos Part of Speech, included in mean table

• n noun
• v verb
• c conjunction
• p preposition
• j adjective
• e adverb
• r pronoun
• a particle
• i interjection
• s syllable - note: this is not a part of speech

One meaning can have multiple pos.

• cp, ru, tl have been manually edited by the user, because the parser failed
• click the pm checkbox in the editor to open these fields to editing
• replaces and combines tlm and cpm
• Note: pm = true indicates the parser failed. this field must be updated when the parser succeeds

• the record cannot be re-parsed
• cp, ru, tl cannot be changed by a user
• the testsuite is built dynamically selecting these records from the dict
• when we train an AI to do parsing, these records become the training set
• Note: when pf = true, pm may be true or false

in addition to search exact and search partial:

• find words which have some known glyphs (replacing Noam::collect())
• find words which have only known glyphs
• pf = true/false
• pm = true/false
• lc = ก
• lc.length > 1
• lc includes ก
• ditto fc, tm, t
• includes rule
• type = m/o/s

options

• include only known words
• include only new words

implement via sql or dictionary iterate

• rewrite translate()
• rewrite translit()
• rewrite iterate()
• repurpose tlm/cpm to pm/pf (see above)
• implement search special (see above)