gf draft rearchitecture

fontbakery_sample.py

@condition
...
def gf_family(ttFont):
  from gfcore.font import family_name
  from gfcore.site import download_family, family_exists
  name = family_name(ttFont)
  if family_exists(name):
    return download_family(name)
  return None


@condition
...
def expected_name_table(ttFont):
  from gfcore.font import build_name_table
  return build_name_table(ttFont)


@condition
...
def expected_metadata(family_dir):
  return gfcore.addfont.MakeMetadata(family_dir)


@check
...
def check_nametable(ttFont, expected_name_table):
  current = ttFont['name']
  expected = expected_name_table
  # Simply compare them. No need for 400 checks

@check
...
def check_metadata(metadata, expected_metadata):
  ...

gftools_sample.py

## FIXING
# For fixing we use the constants and helper funcs from the gfcore.font
from gfcore.font import *

def fix_fstype(ttFont):
  ttFont['OS/2'].fsType = GF_FS_TYPE


def fix_nametable(ttFont):
  from gftools.font import family_name, style_name
  ttFont['name'] = build_name_table(family_name(ttFont), style_name(ttFont)


# QA
# for gftools QA we can use gfcore.site to fetch fonts from fonts.google.com
from gfcore.site import download_family, family_exists

if family_exists("Comfortaa"):
  existing_family = download_family("Comfortaa")

It would be interesting to see how you'd tackle tables which do not have constants such as the name table.

Sure. I don't imagine you would be able to do this for everything. But what you could do it for, you should. I think this is what Dave is getting at here:

...in a way the googlefonts profile is a coded version of the GF Spec...
Given topically-unrelated efforts... - to make better structured docs for the font format spec... - perhaps there's something we can do to deduplicate and pattern things so there's a common approach that makes contribution skills more transferrable.