A large number of "-Kaisho" files had the horizontal stroke going left
to right but were labelled as ㇒ in kvg:type. This changes the easy
cases to ㇐. There are some remaining cases of different types which
will have to be checked individually.
Before this commit, the radical element contained only one
stroke. I've altered it to add stroke 5 which seems to fit better the
supposed 儿 radical, with two strokes. I removed the
kvg:partial="true" from that as well, and I also changed the type of
stroke 5 from CJK stroke S to CJK stroke SP since it bends leftward.
This adds to previous edits where a vertical was replaced with a dash
for "Kaisho" files. In all cases the line didn't appear to be a
vertical from its geometry, and formed part of a "hat" structure. The
judgement on what was vertical was made using an average over all
verticals in the KanjiVG data set, and a cut off of being over 3.3
standard deviations away from the mean of the vertical lines was used
as the criterion. Spot checks were performed on about five cases of
the total of about 500 changes. The previous similar edits were done
not with the statistics, but with numbers I had typed in.
Various characters had a stroke type of ㇔ for the 上 (ue) elemement,
whereas most of them had a stroke type of ㇐/㇔, where the actual
strokes were virtually identical. This changes the ㇔ ones into the ㇐
/㇔ format for consistency.
This edit is an automated repair of a large number of -Kaisho variants
with a stroke type within 曲 which is horizontal where it should be
vertical, or vice-versa. Some of the cases seemed to be swapped, and
some of them were just with a single line in the wrong direction.
These mostly occur in -Kaisho files where the sixth stroke of a 足
element has the type ㇒, presumably because there was a plan to draw
it that way, but the element doesn't actually have that type since it
wasn't drawn in the sloping form. This automatically changes all
instances to ㇑. A few spot checks were performed.
There are a lot of files with Hz/VtFstRiLe names where the third and
fourth strokes were swapped. Many of these also have Nelson radicals
based on the incorrect stroke. This swaps them around.
This is an automated edit and not every change was checked by human. I
checked about five characters in total afterwards.
This applies the same bulk changes as the previous edit to files where
the kaisho version has a dash rather than a vertical line, for the
cases of 广 and 疒.
A difference between the -Kaisho files and the standard ones is that
the top stroke of 宀 and 亠 is not a vertical line but a
down-and-left-going dash. This alters the type of these from ㇑ to ㇔.
Most of the change was done automatically, with about five spot checks
of the outputs.
This edit was mostly done automatically, with some spot checks and one
hand edit. It consists of looking for pairs of horizontal and vertical
strokes with the types the wrong way around in the "*HzFst*" and
similarly named files. There are about eighty examples. I have
spot-checked the changes on about ten cases.
This character was essentially a duplicate of 08956.svg, with little
graphical difference and the identical stroke order, but with
incorrect stroke numbering. There is also a "-Hyougai" version of this
character which differs in the lack of stroke 9.
The stroke type was missing for two 07725 variants, and the stroke
order in the HzFst variant was muddled. This keeps the stroke ordering
but changes the grouping and line numbering, since HzFst seems to be
"horizontal first", meaning the cross bar on 止 is the first stroke.
