d-k-e / c-atf-feature-extractor Goto Github PK

Some determinatives have quirks like {d}en, this should have been conceived as {d}-en, currently i have no reason to why this occurs, because bracket separator methods should have been handling it.

Words and determinatives are handled at a line level. Whereas some words are most probably dispersed to multiple lines. Use the algorithm that i've described for oracc.

Before closing the 0.1 beta milestone include an outlier text checker, to make sure that the text that would be treated falls within the perimeters of a reasonable text.

• Nested object part structures like:
  @surface a
  @column 1

@column 2
1.
2.

Are not handled well. They are going to be probably flattened to something like:
@surface a - @column 1
1.
2.
@surface a - @column 2
etc.
Flattening was a bad idea, because it risks of loosing potentially useful information for later on. I'll add "isPartOf", and "relationType" keys to dictionaries and use a controlled vocabulary for values.
Some of the values are:
text_part
part_part
part_AL_Oc

Signs and words that belong to Another Language Occurrence or determinative is not marked in the sign and word dictionaries.

The Extractor is in its alpha stage DO NOT attempt to use it for scholarly work. The closing of this issue would mark the beta stage. It is by then that you can attempt to use it for scholarly work.

• Very rarely in achaemenid royal inscriptions some complex signs, like ANSZE are dispersed into multiple lines. If they are presented as a specified compound sign, something like ANSZE(|AN.\newline SZE|). This is most likely an outlier for other languages. Thus doesn't fall into the scope of 0.1b

Tests will cover the following classes:
cAtfLineGetter
cAtfLineDictBuilder
cAtfALHandler
cAtfWordDictBuilder
cAtfSignDictBuilder
cAtfTextBuilder

I probably won't write tests for tester classes, like cAtfLineTester, etc.

Currently, at the cAtfTextBuilder.

If someone sees any documentation on how lexical texts are implemented, feel free to modify the code base as you see fit. If you don't want to deal with the implementation, at least post it here so that i can take a look at it myself.