Improve text alignment

[egrim]

I found some unexpected text alignment results when cutting text into the faces of a cube and took a swing at improving the results. Here's some code the shows the issue by creating a box and cutting a hole in the center of the >Z face for a reference point. It then cuts the letter "I" into that same face with various alignment options:

import cadquery as cq
from cadquery import exporters

face = cq.Workplane().box(20, 20, 20).faces(">Z").hole(.5).faces(">Z").workplane()

bottom_left = face.text("I", 20, -2, halign="left", valign="bottom")
centered = face.text("I", 20, -2, halign="center", valign="center")
top_right = face.text("I", 20, -2, halign="right", valign="top")

exporters.export(bottom_left, "bottom_left.svg", opt=dict(projectionDir=(0, 0, 1)))
exporters.export(centered, "centered.svg", opt=dict(projectionDir=(0, 0, 1)))
exporters.export(top_right, "top_right.svg", opt=dict(projectionDir=(0, 0, 1)))

Here is the output before and after the proposed change in this PR.

bottom_left before:

bottom_left after:

centered before:

centered after:

top_right before:

top_right after:

[egrim]

Oh man, those SVGs really don't work with dark mode. Please let me know if I need to convert them to JPGs.

[lorenzncode]

The spacing is defined by the font. Shouldn't the font spacing be respected?

import cadquery as cq

font1 = "/usr/share/fonts/liberation-mono/LiberationMono-Regular.ttf"
font2 = "/usr/share/fonts/google-droid-sans-fonts/DroidSans.ttf"

a = cq.Workplane().text("A", 20, 2, halign="left", valign="bottom", fontPath=font1)
i = cq.Workplane().text("I", 20, 2, halign="left", valign="bottom", fontPath=font2)

This results in:

FontForge:

Oh man, those SVGs really don't work with dark mode. Please let me know if I need to convert them to JPGs.

I think it's OK. SVG export provides strokeColor. Perhaps missing is a background color option.

[egrim]

The spacing is defined by the font. Shouldn't the font spacing be respected?

Perhaps! I think there's definitely a case to be made that left aligning should respect that spacing, but it was unexpected when I wanted to center a letter on a face and it wasn't actually centered.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (6869dc6) 94.35% compared to head (1dc1555) 94.37%.
Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1455      +/-   ##
==========================================
+ Coverage   94.35%   94.37%   +0.01%     
==========================================
  Files          28       28              
  Lines        5809     5809              
  Branches      993      993              
==========================================
+ Hits         5481     5482       +1     
  Misses        197      197              
+ Partials      131      130       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[lorenzncode]

Changes for halign="center" look good to me and agree with a couple of other tools that I tested.

The PR halign left, right behavior does not match with other tools. I would suggest to revert the changes for left, right.

Regarding valign, I would suggest not remove the current bottom (it might be renamed say "bottom-base" then add the new bottom change). The valign="bottom" CQ current behavior matches for example Blender "Bottom Baseline".

valign="top" - Perhaps add as a new valign option? I don't know that the existing top behavior should be completely replaced.

[egrim]

Good news! We can rely on OCCP to do the alignment for us instead of using a bounding box based method. I've just pushed a couple of commits that should do what's needed.

[egrim]

Marking as draft while I work on the test failure that popped up.

[egrim]

Good to go now.

[lorenzncode]

LGTM. I confirm alignment is improved compared to what I might expect with other tools.

Example with halign="right", valign="center":

CQ 2.3.1:

PR, blender:

[jmwright]

@adam-urbanczyk @lorenzncode It sounds like this is ready for formal review, correct?

[adam-urbanczyk]

LGTM, nice that the solution is based on what OCCT provides!

[jmwright]

Since @lorenzncode has previously stated approval of this PR, I'll go ahead and improve it and get it merged.

[jmwright]

Thanks @egrim