gen_fb_glyph_data.py

import csv
import json
from fontTools.ttLib import TTFont


class SetEncoder(json.JSONEncoder):
    def default(self, obj):
       if isinstance(obj, set):
          return list(obj)
       return json.JSONEncoder.default(self, obj)


def get_glyph_contours(font, glyph):
    """
    Flatten a glyph's components and count each contour
    """
    contour_count = 0
    components = [glyph]
    while components:
        g = components.pop(0)
        if g.isComposite():
            for comp in g.components:
                components.append(font['glyf'][comp.glyphName])
        if g.numberOfContours != -1:
            contour_count += g.numberOfContours
    return contour_count


def get_font_data(font):
    contour_map = []
    cmap = font['cmap'].getcmap(3,1).cmap
    cmap_reversed = dict(zip(cmap.values(), cmap.keys()))

    for glyph_name in font.getGlyphSet().keys():
        if glyph_name in cmap_reversed:
            uni_glyph = cmap_reversed[glyph_name]
            contours = get_glyph_contours(font, font['glyf'][glyph_name])
            contour_map.append({
                'unicode': uni_glyph,
                'name': glyph_name,
                'contours': set([contours])
            })

    return contour_map


def combine_fonts_data(fonts_data):
    unnested_data = []
    glyph_pos = {}
    g_count = 0

    for family in fonts_data:
        for glyph in family:
            if glyph['unicode'] not in glyph_pos:
                glyph_pos[glyph['unicode']] = g_count
                g_count += 1
                unnested_data.append(glyph)
            else:
                unnested_contours = unnested_data[glyph_pos[glyph['unicode']]]['contours']
                unnested_data[glyph_pos[glyph['unicode']]]['contours'] = unnested_contours | glyph['contours']
    return unnested_data

def main():
    fonts_path = [
        '/Library/Fonts/Arial.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/prozalibre/ProzaLibre-SemiBoldItalic.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/sourceserifpro/SourceSerifPro-Bold.ttf',
        '/Users/marc/Documents/googlefonts/hotfix/out/repo_cp/ofl/rosarivo/Rosarivo-Regular.ttf',
        '/Users/marc/Documents/googlefonts/hotfix/out/repo_cp/ofl/rosarivo/Rosarivo-Italic.ttf',
        '/Users/marc/Documents/googlefonts/hotfix/out/repo_cp/ofl/raleway/Raleway-BlackItalic.ttf',
        '/Users/marc/Documents/googlefonts/hotfix/out/repo_cp/ofl/librebaskerville/LibreBaskerville-Bold.ttf',
        '/Users/marc/Documents/googlefonts/hotfix/out/repo_cp/ofl/worksans/WorkSans-Regular.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ufl/ubuntu/Ubuntu-BoldItalic.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/vollkorn/Vollkorn-BlackItalic.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/breeserif/BreeSerif-Regular.ttf',
        '/Users/marc/Documents/googlefonts/old_repo/old-googlefontdirectory/ofl/cantarell/Cantarell-Oblique.ttf',
        '/Users/marc/Documents/googlefonts/hotfix/out/repo_cp/ofl/cantoraone/CantoraOne-Regular.ttf',
        '/Users/marc/Documents/googlefonts/hotfix/out/repo_cp/ofl/carme/Carme-Regular.ttf',
        '/Users/marc/Documents/googlefonts/hotfix/out/repo_cp/ofl/creteround/CreteRound-Regular.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/cuprum/Cuprum-Regular.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/eczar/Eczar-Bold.ttf',
        '/Users/marc/Documents/googlefonts/hotfix/out/repo_cp/ofl/enriqueta/Enriqueta-Regular.ttf',
        '/Users/marc/Documents/googlefonts/hotfix/out/repo_cp/ofl/faunaone/FaunaOne-Regular.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/hind/Hind-Light.ttf',
        '/Users/marc/Documents/googlefonts/hotfix/out/repo_cp/ufl/ubuntumono/UbuntuMono-Bold.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/belgrano/Belgrano-Regular.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/benchnine/BenchNine-Regular.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/trirong/Trirong-Light.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/archivonarrow/ArchivoNarrow-Bold.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/mitr/Mitr-Regular.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/overpass/Overpass-Regular.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/jura/Jura-Regular.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/overpass/Overpass-Black.ttf',
        '/Users/marc/Documents/googlefonts/fonts/apache/roboto/Roboto-Regular.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/montserrat/Montserrat-Regular.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/montserrat/Montserrat-Black.ttf',
        '/Users/marc/Documents/googlefonts/fonts/ofl/montserrat/Montserrat-Thin.ttf',
        '/Users/marc/Documents/googlefonts/manual_font_cleaning/20160628-gfnt/99 prepublic/fonts/desktop/Spectral-Regular.ttf',
    ]
    fonts_data = []
    for font_path in fonts_path:
        font = TTFont(font_path)
        fonts_data.append(get_font_data(font))

    m_data = combine_fonts_data(fonts_data)

    with open('glyph_data.json', 'w') as glyph_file:
        json.dump(m_data, glyph_file, indent=4, cls=SetEncoder)
    print 'done'


if __name__ == '__main__':
    main()