The vertical line was set to be a subgroup of 三 in two of these, and
in the other one the 三 group was split into two parts but the parts
were adjacent.
There is a very complex and not very useful-looking nest of subgroups
which I think could just be completely removed from all of these
cases, but I'm not tackling that in this change, I've just put the
nest of subgroups under a new group with a single child.
There were a very large number of cases where visually the element was
攵, but it was marked as kvg:original="攵" and kvg:element="攴". I can
only guess that this was an error in the script which was used to
create the files originally.
Six svgs used 巣 (05de3) where they should have been 巢 (05de2).
There was also an extraneous group around the top three strokes of
this containing kvg:element="⺍" and kvg:original of a hiragana つ
which is probably from つかんむり. However these characters do not
contain either of these elements, so I've removed that group.
The last six strokes were made the first six, a number was added for
stroke 21, and strokes 7 and 8 were swapped. I checked the stroke
order here:
https://kakijun.jp/page/E39A200.html
For some reason the left vertical of 非 had stroke type ㇒ in
KanjiVG. Given the error I made with 心 and 灬 I did check against the
CJK stroke type documentation but I cannot see any evidence why ㇒ is
the correct stroke path, so I'm changing all of these to ㇓. 非 is
also used as a "visual" subelement of 韭 (nira), but almost all of
these nira elements, except two which seem to contain errors, see
https://github.com/KanjiVG/kanjivg/issues/296
and
https://github.com/KanjiVG/kanjivg/issues/297,
have a vertical stroke, so those are all unchanged.
This should fix
https://github.com/KanjiVG/kanjivg/issues/286
Some attributes had got into the wrong order, and some of the IDs on
the outermost groups (StrokePaths_ and StrokeNumbers_) had diverged
from the file name.
This runs a script which moves files around and deletes old files, as
well as editing files where there is no HzFst version to move the
strokes around.
This change consists of searching for 𠕁 elements with IDS, then
searching the KanjiVG file for a group with element 冊, then looking
at this group's paths to see if they correspond to the type, and
swapping them around if so. The single-element groups on these paths
were in a mess already, so I have just removed them completely.
From the groupings, it seems as if this was done to add another stroke
order variant for this 冊 piece, but in fact all of the found elements
in KanjiVG had the same stroke order.
