This commit does the following things:
* Remove python2 support
* Using a bytearray instead of chr list to build up the binary data.
this also removes the need of using char encoding.
LGTM identified an instance where a function was defined twice: in this
case, one version was intended to take a list of items and the second
version just a single item. Because they share the same name and number
of arguments, the second definition overrode the first. This causes no
problems in the current code because the version that takes a list is
never used. However, for consistency with the analogous
"globalize_vectors" and "globalize_vector" functions, the "localize*"
versions are changed to match that pattern. All calls in are
converted to the singular use.
When creating arrays of colors, the color information is often created
to be the same length as some array of objects. In three instances this
was achieved with a loop over that list of objects, even though the
objects themselves are never used. This commit eliminates those loops
and creates the required number of color instances directly.
Identified by LGTM.
LGTM is concerned about the unused loop variable in this algorithm.
While there was nothing wrong here, to eliminate the alarm the loop was
removed entirely, since Python allows the use of the multiplication
operator here to achieve the same effect.
LGTM objected to the re-use of the loop variable a in the algorithm.
Upon closer inspection, the algorithm in this function could be modified
to match the algorithm in sync_snap_statusbar_button (which does the
same thing to the statusbar that this does to the toolbar). This
eliminates the double-use. This change does not affect the functionality
of the routine.
An AttributeError is raised when `direction=Vector(0,0,0)` and obj is
an Arch::Space on line: bead9bb938/src/Mod/Draft/draftfunctions/svg.py (L799)
This patch checks if early on if the direction vector and raises a
ValueError with a description of what has gone wrong.
A caveat with this solution is that this new behaviour might break old
code which depends on that invalid directions can be used.
As discussed in https://forum.freecadweb.org/viewtopic.php?f=3&t=54842, if OpenSCAD creates a DXF with no layers in it, the code that is supposed to handle that in FreeCAD has a minor type error in it that prevents the import from working.
The code would only find a center snap on the face with index=0 of solids.
In V0.19 Part.getShape was introduced (line 399). But not all consequences were not fully implemented.
In the '# we are snapping to an edge' section (line 411) the code could be cleaned up. There is no need to check if the index of the edge is correct for the parent object since we are no longer dealing with a parent object. That portion was effectively dead code.
The '# we are snapping to a face' section (line 429 in the revised code) has been modified accordingly, which fixes the bug.
Forum discussion:
https://forum.freecadweb.org/viewtopic.php?f=23&t=54747
