Closed
Description
We should filter out candidates with short names because otherwise they can match when we'll use LIKE
and leads to a wrong category:
r.m.w.s.SeriesInfoExtractorServiceImpl : Determining category from a fragment: 'ДАГОМЕЯ ? М? ДОИСТОРИЧЕСКАЯ ФАУНА ДИНОЗАВРЫ АВИАПОЧТА 3 БЕЗЗУБЦОВЫЕ МАРКИ С ПОЛЯМИ КУПИТЬ! (39)'
r.m.w.s.SeriesInfoExtractorServiceImpl : Possible candidates: [С, М?, ФАУНА, (39), 3, БЕЗЗУБЦОВЫЕ, ПОЛЯМИ, ДОИСТОРИЧЕСКАЯ, ДИНОЗАВРЫ, МАРКИ, АВИАПОЧТА, ДАГОМЕЯ, ?, КУПИТЬ!]
r.m.w.s.SeriesInfoExtractorServiceImpl : Found categories: []
r.m.w.s.SeriesInfoExtractorServiceImpl : Possible candidate: 'С%'
r.m.w.s.SeriesInfoExtractorServiceImpl : Found categories: [2]
The result is wrong because 2
is id of the Sport
category, while I expect that it should found "Prehistoric animals" category.
Also let's filter out candidates with long names.
Perhaps, after fixing #819 it will find "Prehistoric animals" category from the first attempt (via aliases) and won't produce wrong result (because it won't do lookup with LIKE
).